US 11002640

I'll search for authoritative information on this patent number.

Let me verify litigation/docket activity specifically.

Summary — US Patent 11,002,640 B2

Searches performed: Google Patents full-text record (authoritative source supplied), Justia Patents, patentleaderboard.com, plus targeted queries combining "11002640" with litigation/CAFC terms. Findings below are grounded in the full-text record, which I treat as authoritative.

Bibliographic data (as stated in the record)

Abstract

A device abnormality diagnosis method for a device to be diagnosed constituting a plant includes: obtaining time-series data of a plurality of state amounts of the plant which are correlated to an abnormality of the device to be diagnosed; obtaining abnormality diagnosis data on the plurality of state amounts by performing pre-processing on at least one state amount of the plurality of state amounts to exclude, from the time-series data on the state amounts, data of the at least one state amount obtained in an exclusion period which is at least a part of a transient state period during which the device to be diagnosed is affected by a state change of another constituent device of the plant; and performing abnormality diagnosis on the device to be diagnosed on the basis of the abnormality diagnosis data of the plurality of state amounts.

Plain-language overview of the independent claims

Claim 1 (method). A three-step method for diagnosing a device in a plant:

1. Collect time-series data for several state amounts (e.g., temperatures, pressures) that correlate with abnormality of the device being diagnosed.
2. Drop data only for the affected state amount(s), only during an "exclusion period." The exclusion period is at least part of the "transient state period" during which the device under diagnosis is disturbed because another plant device changed state (e.g., a soot blower upstream firing, or a neighboring unit operating). Pre-processing removes those particular data points from the time series.
3. Diagnose the device's abnormality based on the resulting abnormality-diagnosis data across all the state amounts.
   Key point: unlike the cited prior art (which discarded an entire state amount wholesale), this keeps every state amount in the analysis and excises only the time window contaminated by the other device's transient.

Claim 8 (apparatus). The system counterpart of claim 1, comprising:

• an input/output part that pulls time-series state-amount data from plant sensors and outputs the diagnosis result;
• a diagnosis data acquisition part that performs the same exclusion pre-processing for the exclusion period within the transient state period caused by another constituent device operating in the plant; and
• an abnormality diagnosis part that diagnoses the device from the resulting multi-state-amount data.

Dependent claims worth noting

The dependent claims flesh out how the exclusion period is fixed. Claim 2 specifies that the "other constituent device" is an abnormality-prevention device on the device itself or on a preceding-stage device, and that the transient period is the period affected by that device's operation. Claim 3 sets the exclusion period length from the response characteristic of the state amount after the state change begins. Claim 4 determines it via a pre-set correlation between a measured time constant and exclusion length. Claim 5 sets the exclusion length so as to reduce the difference between the frequency distribution and a normal distribution. Claim 6 uses an index of consistency between those distributions. Claim 7 covers the Mahalanobis-distance diagnosis: compute the distance against a unit space of normal-state data and flag abnormality when it exceeds a threshold.

Working example in the specification

The device to be diagnosed is a carbon-containing-fuel heat exchanger 1 (heat exchanger 2, fuel flow passage 3, heat-transfer tube 4, heat transfer surface 6) inside the gas cooler (SGC) 32b of an IGCC plant 30 (coal gasification furnace 32 with gasification part 32a, gas turbine facility 34, steam turbine facility 36, HRSG 38). The "abnormality prevention device" is a soot removing device 5 (oscillation, hard-ball drop, or injection/soot-blower type) sitting immediately above each exchanger. Because soot blows intermittently create sharp, non-stationary temperature excursions on the primary and secondary sides, the patent excludes those data windows before computing Mahalanobis distance, using response characteristics (e.g., a 63.2% ‑of‑stabilized‑value criterion), time constants, or histogram-vs-normal consistency to size the window. FIGS. 8A–8E extend this to a chain of soot blowers (5C, 5D, 5E) whose disturbances propagate downstream to exchanger 1E.

CAFC / litigation check — explicit uncertainty

I found no Court of Appeals for the Federal Circuit docket, appeal, IPR, or district-court complaint involving US 11,002,640 in any of my searches. I cannot treat that as a definitive negative — it likely reflects limits of the search tooling for docket databases rather than proof of zero activity. I recommend verifying directly against USPTO PatentCenter / USPTO Assignment, PACER/Docket Alarm for the Federal Circuit and district courts, and the PTAB (for any IPR/PGR) before relying on the absence of litigation.

One data-quality caution

A search result at business.patenthub.cn displayed "US11002640" alongside a Ford DISI-engine patent (US 7,165,529 B2, application filed 2004-12-02). I am interpreting this literally and not auto-correcting it: it appears to be a third-party database field mismatch (patent number vs. application/serial number collision), not the same document. The Justia, patentleaderboard, and Google Patents records all consistently tie 11,002,640 to the Mitsubishi Kubo et al. document summarized above, and it is also cited as prior art in US 11,977,373 — which corroborates the identity.

Confidence: high on bibliographic data, claims, and abstract (taken from the authoritative full text). Low-to-uncertain on the absence of any 2026 CAFC docket activity.

I'll search litigation databases for US patent 11002640.

Initial searches were unproductive. Let me query litigation-specific databases directly.

Let me try more targeted litigation databases.

Findings: Litigation Involving U.S. Patent No. 11,002,640

No litigation involving U.S. Patent No. 11,002,640 was found.

I was unable to identify any district court case, ITC Section 337 investigation, PTAB proceeding, or Federal Circuit appeal in which U.S. Patent No. 11,002,640 was asserted, challenged, or otherwise at issue.

What I searched

• General web/litigation queries for patent number "11002640" and "US 11,002,640"
• Queries pairing the patent/assignee (Mitsubishi Power / Mitsubishi Hitachi Power Systems) with litigation terms
• Queries aimed at litigation aggregators (Unified Patents litigation portal, CourtListener/PACER-style sources, Docket Navigator)

All results returned were either unrelated subject matter that happened to contain the number string (e.g., an EPO application EP 11002640, a registration number list, a Bull hardware manual) or unrelated Mitsubishi litigation concerning wind turbine patents (e.g., General Electric Co. v. Mitsubishi Heavy Industries, Ltd., N.D. Tex. 3:10-cv-00276-F; ITC Inv. No. 337-TA-641) — none of which involved U.S. 11,002,640. I did not return results for similar-but-different patent numbers, per your instruction.

Caveats on confidence

• I was not able to complete direct lookups against the specific litigation databases (the Unified Patents caselist page, PACER, and Docket Navigator) within the available search budget — those attempts returned no substantive results rather than a confirmed "no cases" signal from the database itself.
• Accordingly, my conclusion is based on the absence of any indexed evidence of litigation, not on a confirmed empty docket retrieved directly from PACER or a litigation analytics platform. A negative finding of this kind is always weaker than a positive one.

Context about the patent (from the record)

• Patent: U.S. 11,002,640 B2 — "Device abnormality diagnosis method and device abnormality diagnosis device"
• Application: US 16/080,811 (national stage); priority date 2016-03-31; filed 2016-09-13; granted 2021-05-11
• Original assignee: Mitsubishi Hitachi Power Systems, Ltd.; current assignee: Mitsubishi Power, Ltd.
• Inventors: Hiroyoshi Kubo, Hiromi Aota, Yasunari Shibata, Katsuhiko Yokohama, Yasunori Ishizu, Yoshinori Koyama
• Legal status: Expired – Fee Related (adjusted expiration 2037-06-23)
• Tech area: Plant/equipment abnormality diagnosis using Mahalanobis distance, with exclusion of transient-state data (e.g., IGCC gas cooler / soot-blower disturbances)

The "Expired – Fee Related" status is consistent with a patent that was never commercially enforced or litigated.

Recommended next step for confirmation

To convert this into a definitive negative, a direct query should be run in a litigation analytics tool (Docket Navigator, Lex Machina, or RPX Insight) and/or PACER for the patent number "11002640," plus the Unified Patents litigation caselist. If you can provide access or a capture from one of those sources, I can reconcile it against these findings.

Bottom line: Based on the searches performed, there is no known litigation involving U.S. Patent No. 11,002,640. I cannot state with high confidence that litigation is definitively absent, because I was unable to complete direct database confirmation.

PTAB Proceedings Report — U.S. Patent No. 11,002,640 B2

Patent: Device abnormality diagnosis method and device abnormality diagnosis device
Patent Owner (current): Mitsubishi Power, Ltd. (recorded 2020-12-31, change of name from Mitsubishi Hitachi Power Systems, Ltd.)
Inventors: Hiroyoshi Kubo; Hiromi Aota; Yasunari Shibata; Katsuhiko Yokohama; Yasunori Ishizu; Yoshinori Koyama
Priority: 2016-03-31 (JP) · Filing: 2016-09-13 (PCT) · § 371(c): 2018-08-29 · Granted: 2021-05-11
Adjusted expiration (Google Patents): 2037-06-23

Verification caveat: The USPTO Open Data Portal returns zero AIA trial proceedings for this patent. I ran targeted web searches against PTAB dockets, Docket Alarm PTAB records, CourtListener, and Federal Circuit materials and found no IPR, PGR, CBM, or Director Review proceeding naming US 11,002,640. No proceeding number appears below because none exists to report — I have not invented any. If a recently-filed petition has not yet propagated to public indexes, it would not appear here; confirming via USPTO PTAB E2E (Patent Trial and Appeal Board End-to-End) and the Patent Center "Proceedings" tab is the only way to close that residual gap with certainty.

---

Proceedings overview

Total AIA trial proceedings on file: 0 (0 active; 0 with claims invalidated; 0 with claims sustained; 0 settled; 0 with institution denied).

There is no PTAB record to summarize, and therefore no claim of US 11,002,640 has ever been canceled, narrowed, or even adjudicated at the Board. The patent is completely untested — no AIA-trial estoppel attaches to anyone, no petitioner has consumed the § 315(b) window on the merits, and every prior-art ground a defendant could raise today was available the day the patent issued.

The defensive posture is the opposite of "hardened by litigation survival": it is unlitigated and unadjudicated. That cuts both ways for a defendant. On the plus side, there is no estoppel, no adverse IPR record, and no Federal Circuit precedent construing the claims. On the minus side, there is no claim-level kill shot to hand a court, no prosecution-disclaimer-free claim construction from a PTAB panel, and no invalidity record to leverage in licensing discussions. The patent's value as an assertion vehicle is also undercut by a significant ownership/lifecycle signal flagged below.

---

Per-proceeding detail

None. No AIA trial proceeding has been filed on this patent.

---

Strategic summary

(1) Claim status: entirely UNTESTED. No claim of US 11,002,640 has been addressed by the PTAB, an examiner on reexamination, or a district court. Nothing is CANCELED, nothing is SUSTAINED by an adjudicative body, and nothing has been narrowed through a certificate of cancellation or disclaimer that this record shows. The eight aspects recited in the "Solution to the Problems" section of the specification — (1) the core method claim covering exclusion-period pre-processing; (2) the "other constituent device is an abnormality prevention device" limitation; (3) exclusion period set by response characteristic; (4) exclusion period set from a time constant; (5) exclusion period set to reduce deviation from a normal distribution; (6) exclusion period set from an index of distributional consistency; (7) the Mahalanobis-distance/unit-space comparison; (8) the device claim — appear to map to the granted claim set, but the mapping of those numbered paragraphs to issued claim numbers should be verified against the printed claims in Patent Center before relying on it in any validity chart. Do not assume the numbering.

(2) Estoppel landscape: wide open. Because no IPR has been instituted and no final written decision has issued, no § 315(e)(2) estoppel applies to any party on any ground. Every defendant retains the full universe of §§ 102/103 grounds — including art that was before the examiner and art that was not. The only estoppel constraints in the system are ordinary: IPR-estoppel would attach prospectively only to a petitioner who takes a case to FWD, and prosecution-history estoppel/§ 325(d) discretion is a set of discretionary denial levers, not estoppel.

(3) Vehicle availability and the one-year clock. Available AIA vehicles are now narrow:

• IPR — available, subject to the § 315(b) one-year bar running from service of a complaint alleging infringement on the defendant or a real party in interest/privy. If your client was served more than one year ago, IPR is time-barred absent joinder.
• PGR — closed. The nine-month post-grant window ran out on approximately 2022-02-11. No petitioner can file one now.
• CBM — unavailable. The transitional CBM program sunset on 2020-09-16, and the subject matter here (industrial plant abnormality diagnosis) is a technical invention that would not have qualified as a "covered business method" in any event.

So for a currently-asserted defendant, IPR is the only AIA trial lever, and the § 315(b) date governs everything.

(4) Pattern signals: none to report. No petitioner has filed multiple IPRs against this patent — no one has filed once. The patent owner has not pursued any PTAB appeal (there is no FWD to appeal) and has no Federal Circuit docket on this patent that my searches surfaced. There is no defensive aggregator (Unified Patents, RPX, AST, Open Invention Network) in the chain — that follows a fortiori from there being no petition at all. Note that the patent owner is Mitsubishi Power, Ltd., a large operating company, not a non-practicing entity; the classic "well-asserted patent eventually attracts an IPR (and often a Unified Patents petition)" dynamic has not been triggered because there is no visible assertion campaign driving it.

(5) Non-PTAB signal worth verifying — potential fee lapse. Google Patents carries this patent's legal status as "Expired – Fee Related" with an adjusted expiration of 2037-06-23. Taken literally, that label suggests a maintenance-fee lapse rather than a full-term expiry. The first maintenance fee (due at 3.5 years from grant, i.e. ~2024-11-11, with a six-month surcharge grace period to ~2025-05-11) would have come due within the window preceding this record. If the 3.5-year maintenance fee was in fact not paid and the surcharge period lapsed, the patent is in the public domain and there is no infringement case to defend against at all — that is the single highest-value item to confirm. Google's legal-status labels are algorithmically generated and are not authoritative; I am flagging this as a lead to verify, not as an established fact. Check the Patent Center maintenance-fee tab and the USPTO fee record directly, and if a lapse is confirmed, check whether a petition to accept an unintentionally delayed payment under 37 C.F.R. § 1.378 has been filed (which would revive it).

---

Recommended next steps

1. Confirm the absence of PTAB activity directly. Run the patent number through USPTO PTAB E2E and the Patent Center "Proceedings" tab. The ODP feed is the canonical structured source and it is empty; E2E is the belt-and-suspenders check for very recent filings.

2. Confirm the maintenance-fee status first — it dominates everything else. Pull the fee history for US 11,002,640 in Patent Center. If the 3.5-year fee lapsed and the grace period closed without a § 1.378 revival, the patent is unenforceable/expired and no validity or non-infringement work is necessary. If it was paid, or if a revival petition was granted, proceed to step 3. Cite the fee-record date in any demand-letter response.

3. If the patent is live and your client has been served within the last year, treat IPR as an open but time-boxed option. The statutory clock is unforgiving: a petition must be filed within § 315(b)'s one year of service, the Board's institution decision is due within 6 months of the preliminary response/§ 313 due date, and any FWD must issue within 1 year of institution (35 U.S.C. § 316(a)(11)), extendable only for good cause. Diarize the § 315(b) date immediately — it is the single most common way defendants lose the IPR option.

4. Because the claims are untested, do not assume an IPR is a slam dunk. There is no PTAB construction to borrow; you will need to build the claim-construction and § 112 record from scratch, and the specification is heavy on functional/pre-processing language ("exclusion period which is at least a part of a transient state period," "response characteristic," "index indicating consistency") that is fertile ground for both indefiniteness and, more usefully, narrow constructions that knock out an infringement theory.

5. Coordinate with district court. With no IPR filed, there is no § 315(e)(2) estoppel risk yet, but filing an IPR creates it for grounds raised or reasonably raisable. If parallel litigation is pending, decide up front whether to stipulate to forego duplicative invalidity grounds (Sand Revolution style) to blunt a § 314(a) Fintiv-successor discretionary-denial argument — noting that the Board's current practice under the Fintiv memo framework weighs the parallel-trial posture heavily.

6. Ground every statement above against a primary source before filing. The statements about zero proceedings rest on the ODP structured data plus a negative web-search result. Cite the ODP record (or PTAB E2E screenshot) and the Patent Center fee record in your work product, and re-verify immediately before any deadline — a petition filed after this report was generated would change the analysis materially.

I'll reconstruct the assignment chain, starting with targeted searches for the recorded assignment records.

Ownership & Assignment Reconstruction — US 11,002,640 B2

Source discipline note: the authoritative full text supplied for this patent contains a Google Patents legal events block but no reel/frame numbers. My searches for the recorded assignment cover sheets for application 16/080,811 returned nothing retrievable (the queries hit name-collision noise, unrelated PDFs, and a step limit). Below I separate what the record states from what I could not retrieve, and I do not invent reel/frame pairs.

---

Inventors

Support: patentleaderboard.com lists Aota (3 patents, MHPS) and Koyama (41 patents, MHPS) as MHPS inventors, with 11002640 in both portfolios; a patent-review profile for Hiroyoshi Kubo lists his top assignees as Mitsubishi Heavy Industries 13, Mitsubishi Hitachi Power Systems 4, Mitsubishi Power 1, Canon 1.

Unusual-pattern check — no fire-sale precursor found:

• All six inventors assigned to the same employer group, and Kubo is still filing in the Mitsubishi orbit as recently as a 2026-07-09 publication (US 2026/0194886 A1, "Facility abnormality determination device…"). Continuous employment signals do not support the "all inventors departed within 12 months of filing" pattern that often precedes a portfolio sale.
• Data-quality caution (consistent with the prior section's caution): patentleaderboard.com also shows "Hiroyoshi Kubo — 25 Patents at Canon," and lists US 11002640 under Canon. That is a name collision (two different individuals), not evidence of an inventor move. The patents-review profile resolves Kubo's true weight to Mitsubishi. I interpret these records literally and do not auto-correct them, but I treat the Canon attribution as a third-party database artifact.

---

Original assignee

Named on the record as the assignee of the inventors' rights (recording of 2018-08-29): MITSUBISHI HITACHI POWER SYSTEMS, LTD. (MHPS)

• Legal name/address in the record: MITSUBISHI HITACHI POWER SYSTEMS, LTD., 3-1, Minatomirai 3-Chome, Nishi-ku, Yokohama-shi, Kanagawa 220-8401, Japan.
• Name-normalization flag: the Google Patents header displays Original Assignee: Mitsubishi Power Ltd and Current Assignee: Mitsubishi Power Ltd. That is Google back-filling today's corporate name onto the original assignee field — the actual entity that received the 2018 assignment was MHPS. This is exactly the kind of artifact the user should not read as a 2016 "Mitsubishi Power" filing.
• Primary line of business: thermal power generation equipment — gas turbines, steam turbines, boilers, HRSGs, and integrated coal gasification combined cycle (IGCC) plants. MHPS was formed by combining the thermal-power businesses of Mitsubishi Heavy Industries (MHI) and Hitachi; MHI went to 100% ownership when Hitachi's stake transferred, and the company renamed itself Mitsubishi Power, Ltd.
• Did they ship a product embodying the claims? Yes — the specification itself is written around MHPS's own equipment: a carbon-containing-fuel heat exchanger in the gas cooler (SGC) 32b of the coal gasification furnace 32 of an IGCC plant 30, with soot removing devices (soot blowers) as the "abnormality prevention device." This is a diagnostic method for the assignee's own IGCC heat exchangers, i.e., internal fleet-monitoring technology rather than a licensing asset.
• Current status: operating. A core MHI group company (MHI is listed on the Tokyo Stock Exchange). The 2020-12-31 "Change of Name" event corresponds to the Hitachi-stake transfer / MHPS → Mitsubishi Power transition announced by MHI (April 2020).

---

Assignment timeline

Recorded events as displayed in the Google Patents legal-events block for US 11,002,640 / application 16/080,811. Execution dates are not exposed by this source; the dates below are the recordation/event dates as displayed. Reel/frame numbers were not retrievable from the sources available to me — each entry carries an explicit Reel/frame: not retrieved marker rather than a fabricated number.

• 2016-09-13 (event) / recorded 2016-09-13 — Reel not retrieved

  • Conveyance: Application filed (Google legal-event label "filed"; no separate assignment document surfaced for this date)
  • Assignor: n/a (filing event)
  • Assignee: displayed as Mitsubishi Power Ltd — name-normalized artifact; the filing applicant was Mitsubishi Hitachi Power Systems, Ltd.
  • Correspondent: not retrieved
  • Context: initial filing. The 16/080,8xx series and the 2016-09-13 filing date are consistent with entry of the US national stage of a PCT application (priority 2016-03-31, JP). The full text excerpt does not display the PCT number, so I do not state one.

• 2018-08-29 (event) / recorded 2018-08-29 — Reel not retrieved

  • Conveyance: Assignment (reassignment to employer) — Google label "reassignment"
  • Assignor: AOTA, Hiromi; ISHIZU, Yasunori; KOYAMA, Yoshinori; KUBO, Hiroyoshi; SHIBATA, Yasunari; YOKOHAMA, Katsuhiko (all six named inventors, individually listed in the record)
  • Assignee: MITSUBISHI HITACHI POWER SYSTEMS, LTD., 3-1 Minatomirai 3-Chome, Nishi-ku, Yokohama-shi, Kanagawa 220-8401, Japan
  • Correspondent: not retrieved. Inference flagged as unverified: sibling MHPS recordings of the same era (e.g., application 16/086,705, covered by reel 046925/0414, recorded 2018-09-20) name WENDEROTH, LIND & PONACK, L.L.P., Washington DC, submitter Janae Gallun. That firm's appearance on a different MHPS application is not a finding for this patent; a single appearance is not evidence of anything beyond routine outside counsel.
  • Context: standard employee/obligation assignment to the operating employer, executed around US national-stage entry (~3.3 years after the PCT filing date).

• 2020-12-31 (event) / recorded 2020-12-31 — Reel not retrieved

  • Conveyance: Change of Name (record literally: "CHANGE OF NAME (SEE DOCUMENT FOR DETAILS)")
  • Assignor: MITSUBISHI HITACHI POWER SYSTEMS, LTD.
  • Assignee: MITSUBISHI POWER, LTD.
  • Correspondent: not retrieved
  • Context: internal corporate renaming only — no change in beneficial ownership. MHPS → Mitsubishi Power, Ltd. following the Hitachi stake transfer to MHI. No consideration, no third party, no asset movement.

• 2021-05-11 (event) / recorded 2021-05-11 — Reel n/a

  • Conveyance: Application granted / publication of US11002640B2
  • Assignor/Assignee: same entity (Mitsubishi Power, Ltd.)
  • Correspondent: n/a
  • Context: issuance — not an assignment event; included for chain completeness.

Legal-status datum (affects the NPE screen): the record lists status "Expired – Fee Related," adjusted expiration 2037-06-23. Read literally, the patent lapsed for non-payment of maintenance fees (the first 3.5-year fee would have fallen due ~2024-11-11 for a 2021-05-11 grant). Verify in PatentCenter before relying on this — Google's status field is an inference, not a legal conclusion. If accurate, it means no one is currently monetizing or asserting this patent.

Total recorded assignments in the chain: two (2018-08-29 assignment to MHPS; 2020-12-31 change of name). There is no post-issuance transfer of any kind.

---

Timeline diagram

timeline
    title Ownership of US 11002640
    2016 : PCT filing by Mitsubishi Hitachi Power Systems
    2018 : Six inventors assign rights to MHPS
    2019 : US pre-grant publication 20190086294 A1
    2020 : Name change only to Mitsubishi Power Ltd
    2021 : Patent issued as US 11002640 B2
    2024 : First maintenance fee window opens

---

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT.
The only assignee-side transfer is the 2020-12-31 Change of Name from MITSUBISHI HITACHI POWER SYSTEMS, LTD. to MITSUBISHI POWER, LTD. The record itself labels it a name change "SEE DOCUMENT FOR DETAILS," the Yokohama address is unchanged, and the Japanese-language corporate announcement confirms an intra-MHI reorganization with no third-party purchaser. No "IP / Holdings / Licensing / Ventures" entity appears anywhere in the chain.

2. Known asserter in the chain — NOT PRESENT.
Neither assignee matches Acacia, Marathon, Intellectual Ventures, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, or any Spangenberg entity. Both assignees are Mitsubishi Heavy Industries group operating companies.

3. Repeat correspondent across the chain — UNCLEAR (data not retrievable).
I could not retrieve the correspondent of record for either of this patent's two recordings, so I cannot compare correspondents across the chain. What I can say with a citation is that a different MHPS application in the same window (16/086,705) was recorded via Wenderoth, Lind & Ponack, L.L.P., submitter Janae Gallun, at reel 046925/0414, recorded 2018-09-20 (source: https://legacy-assignments.uspto.gov/assignments/assignment-pat-046925-0414.pdf). Wenderoth is a general-practice IP firm serving operating companies across many technologies; recurrence on MHPS filings would indicate routine outside-counsel handling of a corporate portfolio, not an NPE recording mill. No conclusion drawn.

4. Cascading transfers — NOT PRESENT.
Two events across 4 years and 3 months (2018-08-29 → 2020-12-31), both within one corporate group. The "<24 months through chained LLCs" pattern is absent; there are no LLCs in the chain at all.

5. Pre-litigation transfer — NOT PRESENT.
No assignment within 6 months before any suit, because no infringement suit, IPR/PGR, or CAFC appeal naming US 11,002,640 was surfaced in any search (carried forward from the prior section, with the same explicit low-confidence caveat about docket-database coverage). The nearest transfer-in-time to issuance is the 2020-12-31 name change, ~4.5 months pre-issuance but with no litigation downstream of it.

6. Bankruptcy fire-sale — NOT PRESENT.
No Chapter 7/11 proceedings for MHPS or Mitsubishi Power. The Hitachi-stake transfer to MHI was a solvent parent-level reorganization, not an insolvency sale. Contrast with the Kodak/Nortel/Polaroid pattern, which requires a docketed bankruptcy estate selling patents to third parties — nothing analogous here.

7. Privateering — NOT PRESENT.
No transfer to any asserting entity, and no SEC/8-K disclosure (MHI is a Japanese TSE-listed issuer filing under a different regime) of a patent-transfer monetization program. The patent reads as in-house fleet diagnostics for the assignee's own IGCC equipment.

8. Defensive aggregator (anti-NPE) — NOT PRESENT.
The chain terminates at Mitsubishi Power, Ltd., not at RPX, AST, LOT Network, Unified Patents, or OIN. I found no evidence that this specific patent was contributed to a defensive aggregator — I am not asserting MHI's LOT membership one way or the other, because I could not verify it.

---

Verdict

Insufficient data — construed per the rubric as "no records, or only the original assignment," which is the accurate description of this chain, and as a negative finding: there is no NPE chain here.

The entire ownership chain consists of the inventors' 2018-08-29 assignment to MITSUBISHI HITACHI POWER SYSTEMS, LTD. and the 2020-12-31 Change of Name to MITSUBISHI POWER, LTD. — two intra-group events over four-plus years, terminating at a solvent operating company that manufactures the very IGCC heat exchangers the specification diagnoses. No NPE-listed assignee, no shell LLC, no third-party transfer, no cascading chain, and no litigation were identified; the record-level status "Expired – Fee Related" (if confirmed) further indicates the patent is not being monetized. The one genuinely unresolved item is the correspondent of record and reel/frame citations, which I could not retrieve and which must be read directly off the Assignment Center.

Verification links:

• Assignment Center search (search by patent number 11002640 or application 16/080,811): https://assignmentcenter.uspto.gov/
• Legacy assignment search: https://assignment.uspto.gov/patent/index.html
• Companion record confirming the MHPS recording pattern (reel 046925/0414): https://legacy-assignments.uspto.gov/assignments/assignment-pat-046925-0414.pdf
• Corporate rename confirmation (MHI, 2020-04-24): https://www.mhi.com/jp/news/grp200424.html
• Patent record: https://patents.google.com/patent/US11002640/en

What to pull manually to close the gap: run application 16/080,811 in Assignment Center, capture (a) executed date vs. recorded date for the 2018 assignment, (b) reel/frame for both events, and (c) the correspondent name and firm of record for each — then check whether that correspondent also appears on the MHPS/Mitsubishi Power sibling filings (e.g., 16/086,705) before drawing any recurrence conclusion.

I'll search for the USPTO record and prior-art citations for this specific patent number.

I found the Justia citation list. Let me pull the complete cited-references list from both Justia and Google Patents.

Let me pull the complete citation list from the Google Patents record and Justia in more detail.

Prior-Art Analysis — U.S. Patent No. 11,002,640 B2

(Coordinates with the earlier sections: this section treats the bibliographic data, claims, and working example already established as authoritative and does not repeat them.)

---

1. Identification check (literal interpretation of "11002640")

I searched for the exact string 11002640. Two distinct records carry that string:

• US 11,002,640 B2 — "Device abnormality diagnosis method and device abnormality diagnosis device," Mitsubishi Power, Ltd. — this is the patent under analysis.
• A third-party database page (business.patenthub.cn) that displays "US11002640" as the application/serial number of a Ford DISI-engine patent (US 7,165,529 B2). Per instruction, I am not auto-correcting or merging these; this is a field-collision artifact and is excluded from the prior-art set below.

Confidence: high that the citation list below attaches to US 11,002,640 B2.

---

2. Source of the citation list, and a completeness caveat

The face-of-patent references I could retrieve come from the Justia record for this patent (https://patents.justia.com/patent/11002640, "Referenced Cited" section), which mirrors the USPTO front-page "References Cited" block. The specification itself separately names one of these as "Patent Document 1."

Important limitation: my attempt to pull the complete Google Patents "Patent Citations" table for US11002640 was cut off by the search-tool step limit, and the Justia snippet was itself truncated. The list below is therefore complete as to what I retrieved but should be re-verified against the full front page / Google Patents citation table before being treated as exhaustive. I flag this explicitly rather than presenting it as a closed set.

---

3. References cited on / against US 11,002,640

3.1 — JP 2013-200245 A (cited on the face of the patent AND discussed in the specification as "Patent Document 1")

3.2 — US 2015/0293531 A1 (Mikami)

3.3 — JP 5260343 (listed as a JP foreign patent document)

3.4 — Non-patent literature: Lv et al. (2012)

3.5 — International Search Report (procedural document, not prior art)

---

4. Consolidated § 102 assessment

Because US 11,002,640 B2 issued over these references, none of them was found by the examiner to anticipate the claims — that procedural fact should be stated up front, since it makes a bare "anticipation" finding in this section unlikely by construction. On the merits of the material I actually have:

The single most legally significant observation is the asymmetry between the cited art and claim 1: JP 2013-200245 A performs variable-level exclusion (drop a state amount entirely, for all time), whereas claim 1 performs temporal exclusion (keep every state amount; drop a bounded time window). The specification relies on exactly this distinction — its stated objective is to detect abnormalities "while taking account of data collected from the device to be diagnosed for all of the state amounts." That distinction is the likely reason the claims are patentable over the cited art, and it means the cited art is better recruited for a § 103 obviousness argument than for § 102.

---

5. Recommended completion steps (to close the gaps flagged above)

1. Pull the ISR for PCT/JP2016/076906 (dated 2016-12-06) — it is the ISA's own relevance-rated prior-art table for these exact claims and will resolve whether the list in § 3 is complete.
2. Retrieve the full front page of US 11,002,640 B2 (USPTO PatentCenter or the Google Patents "Patent Citations" table) to confirm the complete "References Cited" block and the exact U.S. classification fields.
3. Retrieve the full text of US 2015/0293531 A1 (Mikami) and JP 5260343 so their § 102/§ 103 positions can be assessed rather than left as "not confirmed."
4. Distinguish forward citations from prior art: US 11,977,373 (Kubo et al., later Mitsubishi filing) cites US 11,002,640, but a later patent is not prior art to this one. Likewise, JP 2017-182693 A ("Apparatus abnormality diagnosis method and apparatus abnormality diagnosis device") appears to be a same-family/same-applicant document, not prior art — do not treat it as a § 102 reference without first checking its priority claim.

---

6. Confidence statement

• High: the identity of US 11,002,640 B2; that JP 2013-200245 A is both a face citation and the specification's own "Patent Document 1"; the citation entries for US 2015/0293531 A1, JP 5260343, the Lv et al. article, and the PCT/JP2016/076906 ISR as listed.
• Low / explicitly uncertain: the completeness of the citation list (tool step limit truncated the Google Patents pull), the title/assignee/disclosure of US 2015/0293531 A1 and JP 5260343, and the exact filing dates for JP 2013-200245 A and the Mikami publication. I have not fabricated descriptions for these; where I did not retrieve content, I have said so.
• I found no anticipation-grade reference among those retrieved, and I recommend the ISR cross-check in § 5 before any § 102 conclusion is finalized.

I'll research the prior art landscape, starting with the reference cited on the face of the patent.

Let me look for the patent's citation lists and family members to identify the actual prior art of record.

Obviousness Analysis — U.S. Patent No. 11,002,640 B2

35 U.S.C. § 103; effective filing date 2016-03-31 (priority) / 2016-09-13 (actual PCT filing)

---

⚠️ Threshold caveat on the prior-art record (read first)

Before the analysis, an honesty flag required by your operating rules:

The instruction says to "use the results from the Prior Art section of this page." The supplied Google Patents full text (https://patents.google.com/patent/[US11002640](/patent/US11002640)/en) does not actually contain a "Citations," "Cited By," "Similar Documents," or "Prior Art" listing — the fetched rendering jumped from the bibliographic/classification block straight to the Definitions, Description, Claims, and Abstract. The only prior-art reference of record that appears anywhere in the document I was given is Patent Document 1: JP 2013-200245 A, cited in the patent's own Citation List and discussed in the Background and Summary.

I therefore do not have a verified list of the examiner-cited references, and I will not fabricate one. The secondary references below were located by search and their dates are flagged for verification. Treat the combination theories as candidate § 103 attacks, not as a reconstruction of the actual prosecution record.

Contradiction check vs. prior sections: the earlier "Patent summary" section states that US 11,002,640 is cited as prior art in US 11,977,373. That is consistent and is not prior art to '640 (it is later art); it is useful only as evidence of the field's recognition of '640.

---

1. What claim 1 actually requires (and what it does not)

Claim 1, parsed into elements:

Two observations that materially drive the obviousness exposure:

1. Claim 1 does not recite the Mahalanobis distance at all. The MT method appears only in claim 7 (and in the specification's working example). Claim 1 therefore covers any abnormality-diagnosis technique applied to time-window-conditioned multi-channel data.
2. The exclusion is sample-wise, not channel-wise. Claim 1 keeps every state amount in the analysis and discards only the samples inside a time window. This is the sole point of departure from the cited art — and it is a narrow departure.

The specification's own Background/Summary admits that measuring state amounts of plurality of monitoring items at a predetermined interval and computing a Mahalanobis distance on the correlation between them is conventional (i.e., admitted prior art). The admitted problem with Patent Document 1 is that fully excluding a state amount "makes it difficult to find a particular abnormality whose diagnosis requires the excluded state amount." So the patent's asserted advance reduces to: don't discard the channel — discard only the contaminated time samples in that channel. That framing is important for § 103.

---

2. Reference inventory

Plus the general body of textbook control/statistics knowledge (first-order lag time constants, settling time ≈ 4–5τ, 63.2% step-response criterion, χ²/Kolmogorov–Smirnov/skewness–kurtosis goodness-of-fit tests). These are properly cited as evidence of what a PHOSITA knew, per KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007) and In re Kahn.

---

3. Primary § 103 combination

Combination A — D1 + D2 (+ D5) ⇒ claims 1, 2, 7, 8

D1 teaches the entire framework of claims 1(a), 1(c), 7, and 8: multiple state amounts of a plant measured in time series, correlation-based Mahalanobis distance, and a diagnosis decision. Critically, D1 also teaches the exclusion concept in principle: D1 computes the frequency distribution of each state amount's collected data and excludes state amounts whose distribution does not follow the normal distribution. D1 thus establishes, as a matter of record, the proposition that non-normally-distributed data degrades MT diagnosis and should be removed.

The gap D1 leaves is that D1's exclusion act is channel-level and permanent, whereas claim 1 requires sample-level and time-bounded exclusion. The specification itself frames this gap as the invention.

D2 supplies the bridge. D2 (Ihi) computes MD for a plant by first pre-processing the detection values themselves — extracting maximum/minimum over a predetermined period and defining a median "fluctuation value," then taking the difference between the newly input value and that median before computing MD. D2's stated purpose is that detection values are subject to periodic fluctuation (e.g., seasonal) and that "in abnormality diagnosis it is preferable to remove periodic fluctuation from the detection values in advance." D2 therefore teaches the general technique of conditioning the input to the unit space so that the MD reflects abnormality rather than a benign, known, systematic deviation — and, notably, conditions per-variable rather than discarding the variable.

D5 supplies the conceptual recognition. D5 expressly distinguishes the "status before the facility is stably activated" (an "initial status") from steady-state operation, teaching that data taken during a not-yet-stable period is different in kind and is treated under a separate determination rule.

Motivation to combine (articulable, D1-based):

• D1 and D2 are in the same field (plant/facility abnormality diagnosis by Mahalanobis distance) and address the same problem (false positives when benign, non-abnormal variation inflates the MD).
• D1's own logic — data that does not follow the normal distribution must not be fed to the MD — is a direct teaching to remove non-normal data. A PHOSITA reading D1 and observing that in a given device only a short transient window is non-normal (while the channel as a whole is fine) would naturally apply D1's criterion to the window rather than the channel, because that is the more faithful implementation of D1's stated rationale.
• The patent's own admitted drawback of D1 ("it becomes difficult to find a particular abnormality whose diagnosis requires the excluded state amount") supplies the design incentive to make the exclusion narrower rather than broader. Under KSR, "if a technique has been used to improve one device, and a person of ordinary skill in the art would recognize that it would improve similar devices in the same way, using the technique is obvious." Here the refinement is not even cross-device — it is a narrowing of the exclusion granularity to preserve diagnostic information, which is precisely what the applicant admits was desirable.
• Predictable result: excluding fewer data points while still removing the contaminating samples preserves diagnostic capability and reduces false positives. No new physical principle is invoked; claim 1 recites no algorithm, no formula, and no numerical criterion.

Claim 2 (other constituent device = abnormality prevention device on the device or on a preceding-stage device; transient period = period affected by its operation) is, given Combination A, an almost inevitable specification. The patent's own working example is a soot blower/soot removing device (oscillation type, hard-ball-drop, or injection/soot-blower for a carbon-containing-fuel heat exchanger in an IGCC gas cooler 32b). Soot blowing in boilers/HRSGs is among the most notoriously scheduled, intermittent, short-duration disturbances in power-plant thermal monitoring; any PHOSITA implementing Combination A on a fouling-monitored heat exchanger would identify the soot-blow event as the triggering event, because that is the event that both (i) is commanded by the control system and (ii) produces a large, sharp, non-stationary temperature excursion. Downstream propagation from a preceding-stage exchanger (the 5C/5D/5E chain in FIGS. 8A–8E) is likewise a straightforward consequence of series-connected heat exchangers exchanging sensible heat with the same fuel-gas stream.

Claim 7 (unit space of normal-state data + threshold on MD) is directly and completely anticipated in substance by D1, and is also taught by D2, D3, D4 and D6. Claim 7 adds nothing to Combination A beyond what D1 already discloses; it is at best a prima facie obviousness rejection and arguably a § 102 rejection as to that element.

Claim 8 (apparatus: input/output part + diagnosis data acquisition part + abnormality diagnosis part) is the standard computer-implemented counterpart. Where the method is obvious, the apparatus reciting generic computer components (an I/O, a processing part, a memory part — per the specification) is obvious under In re Alappat / DDR Holdings framing, and the claimed parts map one-to-one onto the steps of claim 1.

---

4. Claims 3–6 — the exclusion-period sizing limitations

Claims 3–6 are where the strongest § 103 arguments lie, because each claims a parameter-selection methodology, and each such methodology is either expressly suggested by D1's own test or is textbook statistics/control theory.

Claims 3 and 4 — response characteristic and time constant

Motivation to combine: KSR holds that "a court must ask whether the improvement is more than the predictable use of prior art elements according to their established functions," and that where "design incentives and other market forces can prompt variations," a predictable variation is unpatentable. Sizing a blanking window from an identified time constant is a predictable variation of a known parameter, not a new result. The patent recites no criticality: claim 4 says only that the correlation is "pre-set," which is a matter of ordinary calibration. No numerical range is claimed. Under In re Applied Materials / In re Boesch, recitation of a method of arriving at a known parameter, without a demonstrated criticality or unexpected result, does not confer patentability.

Claims 5 and 6 — normal-distribution conformity of the residual data

Motivation to combine: Both claims 5 and 6 are the express suggestion of D1 applied iteratively. D1 teaches: non-normal data must not reach the MD. A PHOSITA who must choose a window length so that the retained data is normal would, as a matter of ordinary engineering, iterate the known normality test against candidate window lengths until the test result is acceptable. Claim 5 states the objective function ("reduce the difference"); claim 6 states that the objective function is computed with a consistency index. Neither recites a numerical threshold, a specific statistic, or any result that is more than the expected consequence of applying D1's own test. Under KSR, "obvious to try" squarely applies to claims 5–6, particularly where the art identifies the problem and supplies a finite number of identified, predictable solutions.

---

5. Claim chart summary

---

6. Rebuttals the applicant would raise, and my assessment

A. "D1 teaches away — it excludes whole state amounts, and the '640 specification criticizes that approach."
Weak. D1's criticism (as characterized in '640's own Background) is of over-broad, channel-level exclusion. A reference teaches away only if it "criticizes, discredits, or otherwise discourages" the claimed approach. D1 does not criticize time-windowed exclusion — it does not address it at all. If anything, '640's Summary contains the applicant's admission that the D1 approach's weakness is precisely that the excluded channel may be needed, which is a motivation to narrow the exclusion, i.e., toward the claim. This is the KSR "'better to know than to guess'" trap for the patentee.

B. "The 63.2% criterion and the γ index are specific, non-obvious refinements."
Weak as a matter of claim construction: neither appears in any claim. Claim 4 recites a pre-set correlation between time constant and exclusion length; claim 6 recites an "index indicating consistency." The specification's 63.2% figure and its γ expression are disclosed embodiments, and under Abbott Labs v. Sandoz and Phillips v. AWH, limitations appearing only in the specification are not read into the claims. Unclaimed subject matter cannot supply non-obviousness for claimed subject matter.

C. "The combination is impermissibly hindsight-driven."
Available but beatable. The defense holds up best against claims 5–6 if the examiner's only theory is "iterate until the histogram looks normal" with no record support. It holds up least against claim 1, because the motivation is drawn from D1's own stated rationale and from the applicant's own admission about D1's drawback — not from the '640 disclosure. Practitioners should expect this to be the battleground.

D. Graham factors / secondary considerations.
On this record: the scope and content of the prior art is dense (multiple MT-based plant diagnosis patents from the 2005–2014 window: D2 2014, D5 2005, D6 2012, plus the admitted prior art in the Background). The differences between the prior art and claim 1 amount to exclusion granularity. The level of ordinary skill is high — the named inventors are industrial-power-plant control engineers, and the art is replete with MD-based plant diagnostics. I found no evidence of secondary considerations: no commercial success evidence, no licensing, no long-felt-need narrative, no nexus. The patent's recorded legal status is Expired – Fee Related (adjusted expiration 2037-06-23), and the earlier litigation section of this analysis found no known assertion or challenge of '640 — consistent with a patent that was never commercially enforced. That absence cuts against a secondary-considerations defense (though it is not evidence of obviousness either).

---

7. Confidence and required verification

Recommended verification steps: pull the USPTO PatentCenter file wrapper for application 16/080,811 for the examiner's actual § 103 rejections, the Form 892 (References Cited), the applicant's remarks, and any submitted IDS; verify the JP family dates in J-PlatPat; and confirm no IPR/PGR was filed at the PTAB. If the examiner already made and withdrew a D1-based rejection, the prosecution history estoppel and the applicant's rebuttal arguments become the controlling material, and this analysis should be re-run against those remarks.

Bottom line: On the record I can verify, U.S. 11,002,640 is exposed to a strong § 103 challenge on claims 1, 2, 7 and 8 (framework art D1 plus routine input-conditioning art D2/D5, motivated by D1's own stated rationale and the applicant's admitted drawback of D1), and to an even stronger challenge on claims 3–6, which claim ordinary parameter-sizing methodologies — including the defining 63.2%/one-time-constant criterion of a first-order lag, which the specification recites as though it were an inventive refinement and which no claim actually recites.