A method and system for generating aggregated evidence of physical constraints, ciphertext binding aggregation verification, and anomaly attribution for novel energy storage dispatch fulfillment.

By establishing a verifiable technical closed loop of energy storage scheduling task context, single-site performance evidence generation, and model ciphertext binding, the problem of model and performance binding in new energy storage scheduling is solved. It realizes the homology verification and anomaly attribution of model ciphertext and performance evidence after aggregation, thereby enhancing the credibility and traceability of scheduling performance.

CN123117701APending Publication Date: 2026-11-17CHINA ELECTRIC POWER RESEARCH INSTITUTE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611438661.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-16
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In new energy storage scheduling, there is a lack of verifiable binding between model updates and physical performance. After aggregation, it is difficult to confirm the authenticity of the performance evidence of a single node. Privacy protection and performance verification are difficult to balance. Evidence objects are easy to replace or impersonate across tasks. The participating sets and weights lack synchronous solidification during the aggregation process. It is difficult to locate the cause of the anomaly after verification failure.

Method used

Establish a verifiable technical closed loop for generating and publishing scheduling task context, generating single-site performance evidence, binding model ciphertext, synchronous aggregation verification, and anomaly attribution. Through residual bucketing, ciphertext header binding, and aggregable verification tags, ensure the inseparability of model ciphertext and performance evidence, and selectively open and audit trace when verification fails.

Benefits of technology

It ensures the homology and consistency between the model ciphertext and the performance evidence without disclosing the original data of the energy storage nodes, and enables consistency verification and anomaly attribution after aggregation, thereby enhancing the credibility and traceability of scheduling performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117701A_ABST
    Figure CN123117701A_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for generating aggregated evidence of physical constraints, encrypted binding and aggregation verification, and anomaly attribution for novel energy storage scheduling compliance, belonging to the field of data security and trustworthy verification technology for novel energy storage scheduling. The method includes: system initialization, energy storage asset registration to bind power station identity, metering anchor point, and version vector; generating and publishing scheduling task context; generating single-station compliance evidence, binding encrypted model data, performing edge pre-verification, synchronously aggregating encrypted model data and verification tags, and performing dual-link verification on the scheduling side; if the dual-link verification on the scheduling side fails, an anomaly attribution result is generated, a failure vector is output, and selective opening and audit tracing are performed to verify disputed residual components. This invention establishes a verifiable technical closed loop from scheduling task context generation and publishing, single-station compliance evidence generation, encrypted model binding, synchronous aggregation verification to anomaly attribution.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of novel energy storage scheduling data security and trusted verification technology, and more specifically, to a method and system for generating aggregated evidence of physical constraints, ciphertext binding aggregation verification, and anomaly attribution for novel energy storage scheduling performance. Background Technology

[0002] As the proportion of new energy sources integrated into the new power system continues to increase, electrochemical energy storage, shared energy storage, virtual power plant aggregated energy storage, user-side energy storage, source-grid-load-storage coordinated resources, and vehicle-grid interaction resources are gradually participating in dispatching operations such as frequency regulation, reserve, peak shaving and valley filling, demand response, spot market ancillary services, and capacity compensation. These operations not only require energy storage resources to meet performance requirements such as power, quantity, response time, state boundaries, and market rules within a designated dispatch window, but also require that the performance results be reliably verified by the dispatching side, the market side, or the auditing side.

[0003] In actual business systems, the performance data of energy storage power stations participating in scheduling can come from BMS, PCS, EMS, or energy storage monitoring systems, gateway metering devices, and related operation and business systems. The energy storage node referred to in this invention is a logical entity participating in scheduling tasks and undertaking local evidence generation. It can correspond to an energy storage power station or an authorized data processing entity representing that energy storage power station in the process of this invention, and is not limited to a physical device integrating BMS, PCS, EMS, and metering functions. The operating data, control logs, metering details, and revenue declaration information of energy storage nodes typically involve trade secrets, operational security, and data ownership boundaries. Scheduling platforms or market entities should not directly obtain the original cell data, PCS response curves, EMS control logs, gateway metering details, and revenue declaration details of each energy storage node. The statement in this invention that "not directly obtaining or uploading the original operating data as a whole" means that it is not necessary to additionally and centrally upload all fine-grained original data for the purpose of aggregated performance verification of this invention, and it does not affect the necessary data collection, transmission, and use of existing monitoring, scheduling, protection, metering, and other businesses in accordance with existing regulations. However, during the processes of scheduling settlement, performance evaluation and penalties, compensation disbursement, and dispute resolution, the dispatching side needs to determine whether each energy storage node has truly completed the designated scheduling task. Especially when multiple electrochemical energy storage power stations participate in demand response aggregation scheduling through aggregators, virtual power plant platforms, or third-party operating platforms, the dispatching platform typically first publishes the scheduling window, target power curve, target power volume, participation constraints, and settlement rules, and then the aggregation platform organizes multiple energy storage power stations to execute the response separately. After scheduling is completed, the platform may only submit the aggregated overall response curve, aggregation model results, metering summary, or settlement declaration results to the dispatching side. If there is a lack of binding verification between individual station performance evidence, model encrypted data, metering anchors, task context, and rule versions, the dispatching side cannot confirm whether the aggregation results truly originate from the same batch of participating nodes, the same scheduling task, the same metering window, and the same rule version.

[0004] In existing technologies, one type of approach focuses on federated learning or distributed learning to train energy storage state assessment models, SOC estimation models (SOC estimation models estimate the remaining battery capacity based on data such as battery voltage, current, temperature, and charge / discharge history), SOH estimation models (SOH estimation models estimate battery aging and health status based on data such as battery capacity decay, internal resistance changes, cycle count, temperature history, and charge / discharge behavior), or risk prediction models without centralized raw data. This type of approach can reduce the privacy risks associated with centralized transmission of raw data, but its verification focuses primarily on model updates, model parameters, or training results themselves, often making it difficult to prove a verifiable binding relationship between model updates and the physical performance of energy storage within a specified scheduling window. Another type of approach focuses on verifiable federated learning, gradient signatures, model aggregation verification, or contribution auditing. This type of solution can verify the completeness of the participant set, model update, or aggregation process to a certain extent. However, its verification object usually remains at the model calculation level, without fully considering the physical or business constraints of energy storage devices such as SOC, SOH, temperature, PCS response, metered electricity at the gateway, revenue declaration, and data quality. Therefore, it is difficult to determine whether the energy storage resources behind the model update truly meet the scheduling performance requirements. Another type of solution focuses on distributed energy response commitments, performance verification, metering verification, or on-chain notarization. This type of solution can record or verify some response data, metering summaries, or business credentials, but often relies on plaintext metering curves, plaintext device response data, or data directly accessible to the platform. Even with blockchain notarization, it can usually only prove that a certain data summary was saved at a certain point in time, and it is difficult to further prove that there is an inseparable technical binding relationship between model ciphertext, performance residuals, metering anchors, scheduling task context, rule versions, and aggregation weights.

[0005] Therefore, the existing technology has at least the following shortcomings: There is a lack of verifiable binding between model updates and energy storage scheduling fulfillment. Energy storage nodes may use historical windows, data generated by other tasks or other nodes to form model updates that are complete in form but inconsistent in origin.

[0006] After aggregation, it is difficult to verify the authenticity of the physical performance of a single node. After obtaining the aggregation results, the scheduling side finds it difficult to further confirm whether the SOC, temperature, response curve, metered electricity, revenue declaration and data quality of each participating node meet the scheduling rules of this round.

[0007] Privacy protection and performance verification are difficult to balance. Publicly disclosing original BMS (Battery Management System), PCS (Power Conversion System), EMS (Energy Management System), metering and revenue data would expose operational information, while simply uploading encrypted model updates or metering summaries is insufficient to support the judgment of the authenticity of performance.

[0008] Evidence objects are prone to replacement, reuse, or cross-task impersonation. If the model ciphertext, task context, econometric anchor, version vector, aggregate weights, and performance residuals are not jointly bound, objects from different sources may be spliced ​​together into a formally complete evidence package.

[0009] The aggregation process lacks a synchronous solidification mechanism for the participating set, aggregation weight, and single-site performance evidence. During model aggregation, weight calculation, evidence forwarding, or settlement declaration, the aggregation platform may experience replacement of the participating set, inconsistency of weight, or exchange of evidence objects.

[0010] After verification failure, there is a lack of fine-grained anomaly attribution. Existing solutions can usually only output whether the verification passed or failed, and it is difficult to distinguish between task context anomalies, encrypted binding anomalies, aggregation anomalies, measurement anomalies, version anomalies, physical constraint anomalies, revenue declaration anomalies, and data quality anomalies. Summary of the Invention

[0011] The purpose of this invention is to address the technical problems in novel energy storage scheduling and fulfillment scenarios, such as the ability to aggregate encrypted model data but difficulty in verifying physical fulfillment, the ability to upload evidence objects but difficulty in confirming source consistency, the need for confidentiality of original operational data but the need for verifiability in fulfillment audits, and the difficulty in locating the cause of verification failure after aggregation. This invention provides a method and system for generating aggregateable evidence of physical constraints, performing encrypted binding aggregation verification, and attributing anomalies in novel energy storage scheduling and fulfillment scenarios. This method and system aim to establish a verifiable technical closed loop from scheduling task context generation and publication, single-site fulfillment evidence generation, encrypted model binding, synchronous aggregation verification to anomaly attribution, without directly disclosing the original BMS, PCS, EMS, gateway metering, and revenue reporting data of energy storage nodes. Specifically, the technical problems to be solved by this invention include: To address the lack of verifiable binding between model updates and energy storage scheduling performance, a task binding mechanism is established, which is jointly constrained by scheduling task context, metering window, rule version, node pseudo-identity, metering anchor, weight descriptor, and version vector. This mechanism ensures that the model ciphertext component, verifiable ciphertext header, and performance evidence object are all bound to the same energy storage scheduling task.

[0012] To address the issue that raw operational data of energy storage nodes should not be made public while the dispatching side still needs to verify the authenticity of physical performance, physical or business constraints such as SOC, SOH, temperature, PCS response, gate metering, revenue reporting, and data quality are transformed into multi-dimensional performance residuals. Through residual bucketing, residual component commitment, performance evidence commitment root, and proof object, the dispatching side can verify the performance risk level or residual range without directly obtaining the raw operational details.

[0013] To address the issue of difficulty in confirming whether the performance evidence of a single node is included in the aggregation process along with the model ciphertext after the aggregation of multiple energy storage nodes, a synchronous generation and aggregation mechanism is constructed for homomorphic model ciphertext components, ciphertext header binding tags, and aggregable verification tags, so that post-aggregation verification can simultaneously cover the model aggregation link and the physical performance link.

[0014] To address issues such as evidence replacement, reuse of historical evidence, and cross-task impersonation, a hierarchical binding relationship is established between task challenge values, verifiable ciphertext headers, performance evidence commitment roots, ciphertext header binding tags, and aggregate binding summaries. This reduces the risk of objects formed from different tasks, windows, or nodes being pieced together into a formally complete evidence package.

[0015] To address the potential disconnect between the participating set, aggregation weight, and single-site performance evidence in the new energy storage aggregation scheduling process, the participating set, aggregation weight, model ciphertext package, verifiable ciphertext header, performance evidence commitment root, metering anchor, weight descriptor, and task challenge value are synchronously written into the aggregation binding summary. This ensures that the aggregation results submitted by the aggregation platform are in the same batch, the same task, and the same rule version as the single-site evidence objects.

[0016] To address the difficulty in locating the cause of anomalies after verification failures, an anomaly attribution mechanism is established based on the verification failure item, residual component category, rule version, measurement anchor point, and selective opening results of evidence objects. This mechanism is used to distinguish between task context anomalies, encrypted binding anomalies, aggregation anomalies, measurement anomalies, version anomalies, physical constraint anomalies, revenue declaration anomalies, and data quality anomalies.

[0017] Based on this, the present invention proposes a method for generating aggregated evidence of physical constraints, verifying encrypted bindings, and attributing anomalies for novel energy storage scheduling and fulfillment, including: The system is initialized to generate public parameters, key parameters, and residual bucketing rules, and energy storage assets are registered to bind power station identity, metering anchor point, and version vector. Scheduling task context is also generated to generate task context and task challenge value. Single-site performance evidence generation is performed to form residual buckets, measurement summaries and proofs. The model ciphertext is bound to generate a ciphertext header and an aggregable verification label. Edge pre-verification is performed. If it passes, synchronous aggregation is performed to aggregate the model ciphertext and verification label for dual-link verification on the scheduling side. If the dual-link verification on the scheduling side fails, an anomaly attribution result is generated, a failure vector is output, and selective opening and audit tracing are performed to verify the disputed residual components.

[0018] Optionally, if edge pre-validation fails, a single-site anomaly is logged to prevent entry into the aggregation.

[0019] Optionally, if the dual-link verification on the scheduling side passes, the reliable performance result is output, and the process proceeds to scheduling acceptance or settlement.

[0020] Optionally, it also includes: consistency binding of scheduling tasks, metering windows, and rule versions, including: At the start of each round of new energy storage scheduling tasks, the scheduling task identifier, scheduling execution window, gate metering window, scheduling rule version, residual threshold version, and anti-replay information are incorporated into the same task binding relationship.

[0021] Optional, single-site performance evidence generation, including: physical constraint performance evidence generation; The generation of the physical constraint performance evidence includes: Based on the equipment status, response behavior, gateway metering results, and business application requirements of energy storage nodes, multi-dimensional performance residuals are constructed, and the residuals are converted into residual buckets with finite levels. Energy storage nodes generate component commitments for different residual categories, and the component commitments, metering anchors, weighting basis, version information and task binding information together form the performance evidence commitment root.

[0022] Optionally, it also includes: an inseparable binding between the model encrypted package and the performance evidence, including: When generating the model update ciphertext, a verifiable ciphertext header is generated, and the performance evidence commitment root, task binding information, measurement anchor, version information, residual commitment root, and weight basis are written into the ciphertext header. The energy storage node regenerates a ciphertext header and binds a tag, making the model ciphertext component and the aforementioned performance evidence object form an inseparable model ciphertext package.

[0023] Optionally, it also includes: aggregateable verification tags, aggregated binding digests, and two-link verification, including: The platform aggregates the model ciphertext of multiple energy storage nodes, and aggregates the model ciphertext with the ciphertext of a single node under the same participant set and the same weight rule. The participant set, aggregation weight, model ciphertext package, ciphertext header, performance evidence commitment, metering anchor, weight basis, version information and task binding information are synchronously solidified through aggregation binding digest. After receiving the aggregation results, the scheduling platform performs model aggregation link verification and physical fulfillment link verification respectively to confirm the consistency of the aggregation model results, participating sets, weights and aggregable verification labels, as well as the consistency of the encrypted header, fulfillment evidence, metering anchor, version information and residual proof.

[0024] Optional features include anomaly attribution, selective enabling, and system-wide collaborative deployment, including: The anomaly is attributed to the following reasons based on the location of the failure: task binding anomaly, encrypted header binding anomaly, tag anomaly, metering anomaly, version anomaly, residual range anomaly, aggregation anomaly, data quality anomaly, or revenue declaration anomaly. In disputed scenarios, selective opening is performed on specified residual categories, and the consistency of the opening results with the original component commitment and the root of the performance evidence commitment is verified.

[0025] Furthermore, this invention also proposes a system for generating aggregated evidence of physical constraints, performing encrypted binding aggregation verification, and attributing anomalies for novel energy storage scheduling compliance, comprising: The initialization unit is used for system initialization to generate public parameters, key parameters and residual bucketing rules, and to register energy storage assets to bind power station identity, metering anchor point and version vector, as well as to generate scheduling task context to generate task context and task challenge value. The verification unit is used to generate single-station performance evidence to form residual buckets, measurement summaries and proofs. It binds the model ciphertext to generate a ciphertext header and an aggregable verification tag for edge pre-verification. If it passes, it performs synchronous aggregation to aggregate the model ciphertext and verification tag for dual-link verification on the scheduling side. The attribution unit is used to generate anomaly attribution results, output failure vectors, and selectively enable and audit trace the dual-link verification on the scheduling side when it fails, in order to verify disputed residual components.

[0026] Optionally, if edge pre-validation fails, a single-site anomaly is logged to prevent entry into the aggregation.

[0027] Optionally, if the dual-link verification on the scheduling side passes, the reliable performance result is output, and the process proceeds to scheduling acceptance or settlement.

[0028] Optionally, the initialization unit is also used for: binding the consistency of scheduling tasks, metering windows, and rule versions, including: At the start of each round of new energy storage scheduling tasks, the scheduling task identifier, scheduling execution window, gate metering window, scheduling rule version, residual threshold version, and anti-replay information are incorporated into the same task binding relationship.

[0029] Optional, single-site performance evidence generation, including: physical constraint performance evidence generation; The generation of the physical constraint performance evidence includes: Based on the equipment status, response behavior, gateway metering results, and business application requirements of energy storage nodes, multi-dimensional performance residuals are constructed, and the residuals are converted into residual buckets with finite levels. Energy storage nodes generate component commitments for different residual categories, and the component commitments, metering anchors, weighting basis, version information and task binding information together form the performance evidence commitment root.

[0030] Optionally, the verification unit is also used for: the inseparable binding of the model ciphertext package and the performance evidence, including: When generating the model update ciphertext, a verifiable ciphertext header is generated, and the performance evidence commitment root, task binding information, measurement anchor, version information, residual commitment root, and weight basis are written into the ciphertext header. The energy storage node regenerates a ciphertext header and binds a tag, making the model ciphertext component and the aforementioned performance evidence object form an inseparable model ciphertext package.

[0031] Optionally, the verification unit is also used for: aggregating verification tags, aggregating binding digests, and two-link verification, including: The platform aggregates the model ciphertext of multiple energy storage nodes, and aggregates the model ciphertext with the ciphertext of a single node under the same participant set and the same weight rule. The participant set, aggregation weight, model ciphertext package, ciphertext header, performance evidence commitment, metering anchor, weight basis, version information and task binding information are synchronously solidified through aggregation binding digest. After receiving the aggregation results, the scheduling platform performs model aggregation link verification and physical fulfillment link verification respectively to confirm the consistency of the aggregation model results, participating sets, weights and aggregable verification labels, as well as the consistency of the encrypted header, fulfillment evidence, metering anchor, version information and residual proof.

[0032] Optional features include anomaly attribution, selective enabling, and system-wide collaborative deployment, including: The anomaly is attributed to the following reasons based on the location of the failure: task binding anomaly, encrypted header binding anomaly, tag anomaly, metering anomaly, version anomaly, residual range anomaly, aggregation anomaly, data quality anomaly, or revenue declaration anomaly. In disputed scenarios, selective opening is performed on specified residual categories, and the consistency of the opening results with the original component commitment and the root of the performance evidence commitment is verified.

[0033] In another aspect, the present invention also provides a computing device, comprising: one or more processors; A processor is used to execute one or more programs; When the one or more programs are executed by the one or more processors, the method described above is implemented.

[0034] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the method described above.

[0035] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a method for generating aggregateable evidence of physical constraints, ciphertext binding and aggregation verification, and anomaly attribution for novel energy storage scheduling performance. The method includes: system initialization to generate public parameters, key parameters, and residual bucketing rules; energy storage asset registration to bind power station identity, metering anchor points, and version vectors; and scheduling task context generation to generate task context and task challenge values. Single-station performance evidence generation is performed to form residual buckets, metering summaries, and proofs. The model ciphertext is bound to generate a ciphertext header and aggregateable verification tags. Edge pre-verification is performed; if successful, synchronous aggregation is performed to aggregate the model ciphertext and verification tags for scheduling-side dual-link verification. If the scheduling-side dual-link verification fails, anomaly attribution results are generated, a failure vector is output, and selective opening and audit tracing are performed to verify disputed residual components. This invention enables the inclusion of scheduling task context, metering anchor point, rule version, aggregation weight, model ciphertext, and physical performance residuals into a single verifiable evidence chain without directly disclosing the original BMS, PCS, EMS, metering, and revenue data of energy storage nodes. Furthermore, this evidence chain can enter the aggregation process along with the model ciphertext, and after aggregation, consistency verification can still be performed on the participating sets, weight rules, model aggregation results, and the physical performance status of individual stations. Simultaneously, when verification fails, it supports anomaly attribution based on categories such as task, ciphertext, aggregation, metering, version, physical constraints, revenue declaration, and data quality. Attached Figure Description

[0036] Figure 1 is a schematic diagram of the system structure and data interaction according to an embodiment of the present invention; Figure 2 is a flowchart of the overall method according to an embodiment of the present invention; Figure 3 is a flowchart of single-station evidence generation and ciphertext binding according to an embodiment of the present invention; Figure 4 is a schematic diagram of the relationship between model ciphertext embedded authentication and evidence binding in an embodiment of the present invention; Figure 5 is a flowchart of the aggregation-based dual-link verification and anomaly attribution in an embodiment of the present invention; Figure 6 is a flowchart of selective opening and audit traceability according to an embodiment of the present invention. Detailed Implementation

[0037] Exemplary embodiments of the invention will now be described with reference to the accompanying drawings. However, the invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to fully and completely disclose the invention and to fully convey its scope to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the drawings is not intended to limit the invention. In the drawings, the same units / elements are referred to by the same reference numerals.

[0038] Unless otherwise stated, the terms used herein (including technical terms) have their common meaning as understood by one of ordinary skill in the art. Furthermore, it is understood that terms defined in commonly used dictionaries should be understood to have a meaning consistent with the context of their relevant field, and not to be interpreted as having an idealized or overly formal meaning.

[0039] Example 1: This invention proposes a method S100 for generating aggregated evidence of physical constraints, verifying encrypted bindings, and attributing anomalies for novel energy storage scheduling compliance, comprising: S101. System initialization to generate public parameters, key parameters and residual bucketing rules, and to register energy storage assets to bind power station identity, metering anchor point and version vector, and to generate scheduling task context to generate task context and task challenge value. S102. Generate single-station performance evidence to form residual buckets, measurement summaries and proofs. Bind the model ciphertext to generate ciphertext headers and aggregable verification tags. Perform edge pre-verification. If it passes, perform synchronous aggregation to aggregate the model ciphertext and verification tags and perform dual-link verification on the scheduling side. S103. If the dual-link verification on the scheduling side fails, an anomaly attribution result is generated, a failure vector is output, and selective opening and audit tracing are performed to verify the disputed residual components.

[0040] If edge pre-verification fails, a single-site anomaly is recorded to prevent it from entering the aggregation.

[0041] If the dual-link verification on the scheduling side passes, a reliable performance result is output, and the process proceeds to scheduling acceptance or settlement.

[0042] This also includes: the consistency binding of scheduling tasks, metering windows, and rule versions, including: At the start of each round of new energy storage scheduling tasks, the scheduling task identifier, scheduling execution window, gate metering window, scheduling rule version, residual threshold version, and anti-replay information are incorporated into the same task binding relationship.

[0043] Among them, the generation of single-site performance evidence includes: generation of physical constraint performance evidence; The generation of the physical constraint performance evidence includes: Based on the equipment status, response behavior, gateway metering results, and business application requirements of energy storage nodes, multi-dimensional performance residuals are constructed, and the residuals are converted into residual buckets with finite levels. Energy storage nodes generate component commitments for different residual categories, and the component commitments, metering anchors, weighting basis, version information and task binding information together form the performance evidence commitment root.

[0044] This also includes: the inseparable binding of the model ciphertext package and the performance evidence, including: When generating the model update ciphertext, a verifiable ciphertext header is generated, and the performance evidence commitment root, task binding information, measurement anchor, version information, residual commitment root, and weight basis are written into the ciphertext header. The energy storage node regenerates a ciphertext header and binds a tag, making the model ciphertext component and the aforementioned performance evidence object form an inseparable model ciphertext package.

[0045] This also includes: aggregateable verification tags, aggregated binding digests, and dual-link verification, including: The platform aggregates the model ciphertext of multiple energy storage nodes, and aggregates the model ciphertext with the ciphertext of a single node under the same participant set and the same weight rule. The participant set, aggregation weight, model ciphertext package, ciphertext header, performance evidence commitment, metering anchor, weight basis, version information and task binding information are synchronously solidified through aggregation binding digest. After receiving the aggregation results, the scheduling platform performs model aggregation link verification and physical fulfillment link verification respectively to confirm the consistency of the aggregation model results, participating sets, weights and aggregable verification labels, as well as the consistency of the encrypted header, fulfillment evidence, metering anchor, version information and residual proof.

[0046] Among these, anomaly attribution, selective opening, and system collaborative deployment include: The anomaly is attributed to the following reasons based on the location of the failure: task binding anomaly, encrypted header binding anomaly, tag anomaly, metering anomaly, version anomaly, residual range anomaly, aggregation anomaly, data quality anomaly, or revenue declaration anomaly. In disputed scenarios, selective opening is performed on specified residual categories, and the consistency of the opening results with the original component commitment and the root of the performance evidence commitment is verified.

[0047] The following is in conjunction with the appendix Figure 1-6 The working principle and specific implementation methods of the present invention will be further explained below: The basic working principle of this invention is as follows: In each round of new energy storage scheduling tasks, the scheduling platform generates task context and task challenge value. The energy storage node generates local model updates and physical constraint performance evidence based on the BMS, PCS, EMS, gateway metering anchor points and business declaration data within the scheduling window. The performance evidence does not directly disclose the original operating curves, control logs, or revenue declaration details. Instead, it transforms multi-dimensional physical or business residuals, such as model residuals, response residuals, SOC residuals, SOH residuals, temperature residuals, measurement residuals, revenue declaration residuals, and data quality residuals, into bucketed results (separating continuous residual values ​​into threshold ranges, each range being called a "bucket"; for example, the response residual bucket indicates which level the scheduling power response deviation belongs to; the SOC residual bucket indicates which level the SOC constraint deviation belongs to. Advantages: privacy protection: original curves and precise values ​​are not disclosed; verifiability: the scheduling side can know whether it is within the allowable range; ease of commitment proof: commitments and range proofs can be made for the "bucket level"; ease of anomaly attribution: which category falls into the anomaly bucket is attributed to that problem), and further generates residual component commitments, performance evidence commitment roots, and residual range proofs or consistency proofs. During the single-site evidence generation phase, the energy storage node generates a verifiable ciphertext header and writes the performance evidence commitment root, task challenge value, task context summary, rule or model version vector summary, metering anchor summary, and aggregation weight descriptor into the ciphertext header. At the same time, the verifiable ciphertext header is bound to the homomorphic model ciphertext component through the ciphertext header binding tag, so that the model ciphertext package cannot be replaced, reused, or impersonated across tasks without being separated from the performance evidence commitment root, metering anchor, and task context of this round.

[0048] Before aggregation, edge verification nodes perform pre-verification on single-node evidence objects. Verification includes at least the node's identity or certificate, task challenge value, verifiable ciphertext header, econometric anchor proof, residual proof, ciphertext header binding label, and single-node aggregable verifiable label. Only evidence objects that pass pre-verification are allowed to enter the aggregation stage. During aggregation, the aggregation platform synchronously aggregates the pre-verified homomorphic model ciphertext components and aggregable verifiable labels, generating an aggregated binding summary containing the binding participant set, aggregation weights, ciphertext header, performance proof commitment root, econometric anchor, and task challenge value. Upon receiving the aggregation results, the scheduling platform performs dual-link verification: the first is the model aggregation link, used to verify the consistency between the aggregated model result, the model ciphertext set, the aggregable verifiable labels, the aggregation weights, and the participating node set; the second is the physical performance link, used to verify the consistency between the verifiable ciphertext header, performance proof commitment root, econometric anchor, weight descriptor, version vector, and residual proof. When both links pass verification, the reliable result of the new energy storage scheduling is output; when either link fails verification, the scheduling platform generates an abnormal failure vector based on the failure location, and selectively opens the specified residual component in the case of disputed verification, without disclosing the complete original operation data.

[0049] Furthermore, this invention does not merely apply encryption, signature, commitment proof, or evidence storage to energy storage data separately, but rather organizes the tasks, evidence, ciphertext, aggregation, and traceability in a novel energy storage scheduling and fulfillment task into the following identity constraint chain: Among them, the task challenge value Constraints on this round of energy storage dispatch tasks, metering windows, and rule versions; residual component commitments. Constrain various performance residual buckets and support selective opening by category; proof object Summary of Measurement Anchor Points Aggregate weight descriptor and version vector The physical constraints, measurement sources, aggregation weighting basis, and rule versions of evidence for a single site are jointly defined; the commitment root of performance evidence. The above objects are aggregated into a single-site performance evidence root; the encrypted header can be verified. Tags are bound to the ciphertext header. Bind the performance evidence root to the model encrypted package Single-node aggregation verification of tags Aggregate verification tags and aggregation binding summary Constraints include the set of participants after aggregation, aggregation weights, and the set of model ciphertext; and the abnormal failure vector. The location of the aforementioned constraint chain break is mapped to a traceable anomaly category. Therefore, this invention addresses the complex technical problem of simultaneously verifying model credibility, physical performance credibility, metering credibility, and settlement audit credibility in aggregated scheduling scenarios.

[0050] To avoid this invention being misunderstood as a conventional parallel use of cryptographic algorithms, federated learning aggregation, or metrological evidence storage, this invention further defines four layers of identity constraints. The first layer is task identity, requiring that the task context, metrological window, rule version, and task challenge value remain consistent during evidence generation, ciphertext binding, and scheduling-side verification. The second layer is metrological identity, requiring that residual component commitments, metrological anchor digests, and metrological signatures correspond to the same gateway metrological window. The third layer is ciphertext identity, requiring that homomorphic model ciphertext components, ciphertext header binding labels, and performance evidence commitment roots be jointly bound, ensuring that model updates cannot be separated from the performance evidence commitment root for individual replacement. The fourth layer is aggregation identity, requiring that the participant set, aggregation weights, weight descriptors, model ciphertext set, single-node label set, and aggregation labels be constrained by the same aggregation binding digest. Only when all four layers of identity are simultaneously met will the reliable performance result of this round of novel energy storage scheduling be output; if any layer is interrupted, the abnormal failure vector records the interruption location and triggers selective opening or audit tracing. Compared to schemes that only verify the correctness of model aggregation, only verify the authenticity of measurement data, or only provide ex-post evidence of performance results, this invention organizes the physical constraint residual component commitment, performance evidence commitment root, verifiable ciphertext header, ciphertext header binding label, weight-aware aggregateable verification label, aggregation binding summary, and anomaly failure vector into a continuous technical chain within the same task context. This continuous technical chain enables single-site physical performance evidence to enter the aggregation process along with the model ciphertext, and remains verifiable, attributable, and selectively accessible after aggregation, thereby enhancing the technical feasibility and authorization differentiation of aggregation scheduling performance verification.

[0051] The entities involved in this invention include scheduling platforms. , No. Energy storage nodes Edge verification node Aggregation Platform metering anchor points and audit nodes .in, Used to generate and publish scheduling task contexts, generate challenge values, verify aggregation results, and output anomaly attributions; Used to generate local model updates and performance evidence objects; Used for pre-verifying single-node evidence objects; Used to synchronize aggregated model ciphertext and aggregable tags; Used to provide a meter summary or meter signature; Used to save verification summaries, anomaly attribution summaries, and selectively open records.

[0052] For ease of description, the main symbols are explained below. Throughout the formulas, residual categories, data source categories, and anomaly categories are all represented as upright labels in the subscripts, for example... , , , , To avoid misinterpreting category labels as superscripts as exponentiation, use upright font for entity abbreviations, e.g. ... , , and Variables such as round number, node index, category index, weight, and residual are represented in mathematical italics. To avoid confusion, the effective sampling rate is denoted as... The residual component commitment is recorded as The details are shown in Table 1 below: Table 1

[0053] The method provided by this invention is based on the principle that "novel energy storage dispatch performance evidence enters the aggregation process along with the model's encrypted text, and remains verifiable and attributable after aggregation." It forms a technical closed loop consisting of task binding, single-site physical constraint evidence generation, model encrypted text binding, edge pre-verification, synchronous aggregation, dual-link verification on the dispatch side, anomaly attribution, and selective opening. To enable those skilled in the art to accurately understand the system structure, data flow, and verification logic of this invention, the following description, in conjunction with accompanying drawings, illustrates the system structure, overall process, single-site evidence generation, encrypted evidence binding, post-aggregation verification, and anomaly tracing. Multiple electrochemical energy storage power stations are then used as examples. The preferred embodiment describes the process of accepting scheduling, generating local performance evidence, submitting model ciphertext, completing synchronous aggregation, accepting dual-link verification, and performing anomaly tracing in the same demand response task. The processing objects, verification conditions, output objects, and anomaly branches of each stage are explained step by step.

[0054] The demand response task described herein is only used to explain how the present invention is implemented in a typical novel energy storage dispatching service and does not constitute a limitation on the scope of protection of the present invention. For frequency regulation, reserve, peak shaving and valley filling, capacity compensation, spot market ancillary services or other market-based dispatching fulfillment scenarios, as long as there are requirements for dispatching task context, local fulfillment data of energy storage nodes, metering anchor points, model encrypted aggregation, post-aggregation verification and anomaly tracing, the task binding, physical constraint evidence commitment, encrypted binding, synchronous aggregation verification and anomaly attribution mechanism described in this invention can be adopted.

[0055] Figures 1 to 6 illustrate the system entity interaction, overall method flow, single-station evidence generation and ciphertext binding, the binding relationship between model ciphertext and performance evidence, post-aggregation dual-link verification and anomaly attribution, and selective opening and audit tracing in dispute scenarios, respectively, in embodiments of the present invention. The figures show the flow relationships between scheduling tasks, metering anchors, single-station performance evidence, model ciphertext packages, aggregated messages, dual-link verification results, and anomaly tracing records across different entities and processing stages, as well as the sequential dependencies between task binding, evidence commitment, ciphertext binding, synchronous aggregation, post-aggregation verification, and anomaly attribution.

[0056] The system shown in Figure 1 includes at least a scheduling platform. Multiple energy storage nodes metering anchor points Edge verification node Aggregation Platform and audit nodes The scheduling platform is used to generate and publish scheduling task context, generate task challenge values, verify aggregation results, and output anomaly attributions. The energy storage node is used to generate local model updates, physical constraint fulfillment evidence, and model encrypted packages. The metering anchor point is used to provide metering summaries or meter signatures corresponding to the scheduling window. The edge verification node is used to pre-verify single-station evidence objects before aggregation. The aggregation platform is used to synchronize and aggregate comorphic model encrypted components and aggregable verification tags. The audit node is used to save verification summaries, anomaly attribution summaries, and selectively open records.

[0057] It should be noted that, Figure 1 The connections between the various entities involved are used to represent the logical data interaction, authentication, or verification dependencies involved in this invention, and do not indicate that there is a direct physical connection between different security zones or different management domains; actual data interaction can be achieved through existing data collection and forwarding, boundary protection, or authorized data exchange mechanisms.

[0058] Figure 2 summarizes the overall execution sequence of the method of this invention. The method preferably includes stages such as system initialization, energy storage asset registration, scheduling task context generation and release, single-site performance evidence generation, model encrypted binding, edge pre-verification, synchronous aggregation, scheduling-side dual-link verification, anomaly attribution, and selective opening. These stages are not simply independent sequential processes; rather, they form a continuous constraint relationship within the same task context through task challenge values, metering anchor summaries, version vectors, performance evidence commitment roots, verifiable encrypted headers, encrypted header binding tags, aggregable verification tags, and aggregated binding summaries.

[0059] In the overall process, energy storage nodes only generate a submitable single-site evidence object after completing local window data alignment, residual bucketing, component commitment, metering anchor binding, and encrypted header binding. Edge verification nodes only allow the evidence object to enter the aggregation stage after confirming that the single-site evidence object meets the requirements of task binding, metering anchor, residual proof, and tag verification. After aggregation, the scheduling platform simultaneously executes model aggregation link and physical fulfillment link verification, and generates anomaly failure vectors based on the verification failure locations. Thus, this invention enables "whether the model is correctly aggregated" and "whether each energy storage node is truly fulfilling its obligations" to still be verified by the same technical framework after aggregation.

[0060] Figure 3 illustrates the single-site evidence generation and encrypted binding process at the energy storage node. (Energy storage node) Within this round of scheduling window, data from BMS, PCS, EMS, border metering, and business declaration are collected or read. This data is then time-aligned, anomaly markers are added, and missing value markers are applied according to a unified sampling window to form aligned local window data. Based on this, the energy storage nodes calculate the multidimensional performance residual vector. And based on the bucketing threshold sequence corresponding to each residual category Generate residual bucket number .

[0061] To support performance verification without disclosing raw operating data and complete residual values, energy storage nodes generate residual component commitments for each residual category. The residual bucket commitment root is formed by the commitments of each component. Subsequently, the energy storage node will commit the residual bucket root and metering anchor summary. Aggregate weight descriptor Version vector and task challenge value Jointly bound as the root of performance evidence commitment And generate residual range proof or consistency proof. During the model encryption binding phase, the energy storage node will... , , , and Write a verifiable ciphertext header And bind tags through the ciphertext header. Will Binding to homomorphic model ciphertext components This generates the model ciphertext package. Single-node aggregateable verification tags .

[0062] Through the above processing, the single-site evidence object is not merely a metrological summary or model update file, but is composed of a temporary pseudo-identity, metrological anchor point, residual component commitment, performance evidence commitment root, proof object, verifiable ciphertext header, model ciphertext package, and aggregateable verification tag. This structure prevents the separation of model ciphertext from physical performance evidence and also prevents evidence objects from historical windows, other tasks, or other energy storage nodes from being pieced together into a formally complete current task evidence.

[0063] Figure 4 illustrates the constraints between the task challenge value, residual component commitment, performance evidence commitment root, verifiable ciphertext header, model ciphertext package, single-node aggregatable verification label, aggregated verification label, and anomaly failure vector in this invention. The arrows between the objects in the figure represent hash bindings, ciphertext header bindings, aggregatable label bindings, or verification dependencies between the corresponding objects, rather than simply indicating the data transmission direction. Through these binding relationships, the model ciphertext, physical performance evidence, metering anchors, version vectors, and aggregated weight descriptors can be constrained within the same scheduling task context.

[0064] Figure 5 illustrates how the scheduling platform performs synchronous verification of the model aggregation link and the physical fulfillment link after receiving the aggregation model results, aggregation verification tags, participant sets, and aggregation binding summaries submitted by the aggregation platform. The model aggregation link verifies the consistency between the aggregation model results, participant sets, aggregation weights, and aggregable verification tags; the physical fulfillment link verifies the consistency between the encrypted header, fulfillment proof commitment root, metering anchor, version vector, residual proof, and weight descriptor. If any link verification fails, the scheduling platform generates an anomaly failure vector based on the failure location and attributes the anomaly to categories such as task context, encrypted header, tag, metering, version, physical constraints, aggregation, data quality, or revenue declaration.

[0065] Figure 6 illustrates how, when a scheduling platform or audit node needs to verify disputed residual components, it can only open the bucketed components corresponding to a specified residual category and necessary commitment auxiliary information without disclosing the complete BMS, PCS, EMS, metering curves, and revenue details, and verify the consistency between this opening result and the original performance evidence commitment. This process enables the auditor to conduct limited verification for anomaly categories while avoiding the overall disclosure of energy storage operation data, business strategies, or revenue details unrelated to the dispute.

[0066] To establish a correspondence between the accompanying drawings, processing steps, and protection points, the correspondence between Figures 1 to 6 and the following steps is shown in Table 2 below.

[0067] Table 2

[0068] The method steps of the present invention will be described below with reference to Figures 1 to 6.

[0069] Step S1: System Initialization The system initialization phase is handled by the scheduling platform. The execution phase is used to determine the public parameters, key parameters, residual bucketing parameters, and verification rule parameters that are commonly relied upon for subsequent energy storage node registration, scheduling task context generation and publication, single-site performance evidence generation, model encrypted binding, aggregation verification, and anomaly attribution. After initialization, the model encrypted, performance evidence, metering anchors, aggregation weights, and anomaly attribution in the same round of scheduling tasks are all verified based on unified system public parameters.

[0070] Step S1.1: Algorithm parameter setting. Scheduling platform Select security parameters and determine the hash function. Model encryption algorithms that support linear aggregation of ciphertext Signature Algorithm Commitment Algorithm Proof Algorithm And weight-aware aggregateable verification label algorithm .in, Used to protect and aggregate local model updates; ciphertext header binding tags are used to establish an authentication association between homomorphic model ciphertext components and verifiable ciphertext headers; commitment algorithms are used to hide the contents of residual buckets; proof algorithms are used to prove the consistency between residual buckets, econometric anchors, version vectors, weight descriptors, and task challenge values. This is used to support post-aggregation verification by multiple nodes under the same participant set and weight rule. Regular signatures are used for identity verification and message integrity verification, and do not replace aggregable verification tags.

[0071] Step S1.2: Key parameter generation. Scheduling platform Generate signature key pairs for the scheduling platform, aggregation platform, and edge verification nodes respectively. , and And generate an aggregation public key for model update ciphertext aggregation. .in, Used for encrypted local model updates of energy storage nodes, the aggregation platform is based on Aggregation is performed on the ciphertext components of the homomorphic model without decrypting the plaintext of any single-node model update. Each private key is securely stored by its respective entity and is not published as a public parameter.

[0072] Step S1.3: Setting residual categories and bucketing thresholds. Scheduling platform Define the set of residual categories in, Represents the model residuals. This represents the scheduling response residual. Represents the residual of the state of charge. Indicates residual health status. Indicates temperature residual. Indicates measurement residuals, Indicates the residual in income reporting. This represents the data quality residual. For each class... Dispatch platform Set bucketing threshold sequence The bucketing threshold is used to convert continuous performance residuals into finite levels, enabling the verifier to determine the performance risk level or residual range, while avoiding the public disclosure of the original operating curves, control logs, measurement details, and revenue reporting details.

[0073] Step S1.4: System public parameter release. Scheduling platform. Generate system public parameters in, The allowed timestamp error threshold is then set. Subsequently, the scheduling platform... release , , , and and store them securely respectively. , , And the private key or decryption control material corresponding to the aggregated encryption system. At this point, the system initialization phase is complete.

[0074] Step S2: Energy Storage Asset Registration The energy storage asset registration phase is used in the dispatch platform Establish energy storage nodes on the side The registration benchmark enables node identity keys and metering anchors. Rated capacity, rated power, equipment certificate, initial version vector, and pseudo-identity derived materials are incorporated into the same registration record. This registration record serves as the prerequisite for subsequent scheduling task context generation and release, task challenge value generation, single-site evidence generation, encrypted header binding, and anomaly tracing. Energy storage nodes that are not registered or whose registration record verification fails will not enter the subsequent performance evidence generation and aggregation verification process.

[0075] Step S2.1: Registration request generation. Energy storage node. Based on publicly available system parameters Generate or load node signature key pairs and collect or read node identifiers. Metering anchor point marking Rated capacity Rated power Equipment Certificate and the initial version vector .in, It is used at least to indicate the initial version status of BMS, PCS, EMS, metering interface, model rules, and residual rules related to the generation of performance evidence for this node.

[0076] Energy storage node construction registration payload And generate a registration signature Subsequently, the energy storage node submits a registration request to the scheduling platform. in A timestamp is generated for the registration request. Through the above signature, the node incorporates its identity, public key, metering anchor, rated operating parameters, and initial version status into the registration request, avoiding subsequent registration based solely on a single device identifier or a single metering identifier.

[0077] Step S2.2: Registration request verification and registration record generation. (Schedule platform) take over Next, first verify the timestamp condition. And verify the equipment certificate The validity, unrevoked status, and its relationship with the node identifier This involves the correspondence between device sequence information and other data. Subsequently, the scheduling platform verifies whether the registration signature meets the requirements. In a preferred embodiment, the scheduling platform also verifies... , and Check if it has been occupied or revoked, and verify. , Whether it falls within the allowable range consistent with the equipment certificate, grid connection data, or metering records.

[0078] If any verification fails, the scheduling platform terminates the node's registration and records the registration exception. If the verification passes, the scheduling platform generates a pseudo-identity derivation seed for the energy storage node. Alternatively, the seed can be generated through negotiation with the energy storage nodes, and a metering binding summary can be calculated. To facilitate subsequent verification that the materials derived from the false identity have not been replaced, the scheduling platform further generates seed commitments. Forming a registration record root in, Used to constrain the correspondence between energy storage nodes and metering anchors, rated capacity, rated power, and initial version status; Used to constrain the overall consistency of the node's registration records.

[0079] Step S2.3: Node Certificate Issuance and Local Confirmation. The scheduling platform issues node certificates for the energy storage nodes. in For the certificate validity period, This is the certificate serial number. The scheduling platform generates a registration response. in Sending In the implementation of the method, the Transmitted via a secure channel; in implementations where seeds are not directly transmitted, the scheduling platform and energy storage nodes can generate equivalent pseudo-identity derived materials through key negotiation, retaining only the derived materials in the registration response. .

[0080] Energy storage node received Then, verify the signature of the scheduling platform. and recalculate as well as when and At that time, the energy storage node confirms that its node identity, metering anchor point, rated operating parameters, initial version status, and pseudo-identity derived materials have been correctly bound and securely stored. If the above confirmation fails, the energy storage node rejects the registration response and returns a registration confirmation error to the scheduling platform.

[0081] After step S2 is completed, the scheduling platform saves the registration status. in This indicates whether the registration is valid, suspended, revoked, or pending review. (The following is a continuation of the previous sentence.) In round-robin scheduling tasks, energy storage nodes can, according to , Task Challenge Value and registration record root Derived temporary pseudo-identity This ensures that subsequent performance evidence can be verified as originating from registered energy storage assets, while avoiding the direct exposure of unnecessary long-term identity information during the aggregation and verification phase.

[0082] Step S3: Generating and Publishing the Scheduling Task Context The scheduling task context generation and release phase is used in the first stage. Before the commencement of a new round of energy storage scheduling and fulfillment tasks, the task context, scheduling window, metering window, scheduling rule version, settlement or assessment rule version, residual bucketing threshold version, and task challenge value are solidified. After this stage of processing, the energy storage nodes, edge verification nodes, aggregation platform, and scheduling platform have a consistent understanding of the time window, metering caliber, rule version, and verification parameters used in this round of tasks. Subsequently generated fulfillment evidence, model encrypted text, aggregable verification tags, and anomaly attribution results are all based on the task release message.

[0083] The scheduling task context generation and publishing described in this invention refers to generating and publishing task contexts and related verification parameters for the generation and verification of performance evidence for existing scheduling business tasks, without changing the existing scheduling task or scheduling instruction generation, issuance and execution process.

[0084] Step S3.1: Task context generation. When the scheduling platform... Initiating the first When scheduling energy storage tasks, generate a task context: in, As a task identifier; For scheduling service type; and These are the start and end times of the scheduling window, respectively. The execution window is composed of the sampling times within the scheduling window; This is the metering verification window corresponding to the scheduling window; This serves as the reference power curve or target response curve for this round. For target power or power constraint; To participate in constraints, the rated capacity, rated power, SOC range, device availability, communication status, registration status, and version compatibility conditions of energy storage nodes are defined. For settlement or assessment rules; This is a version vector for scheduling rules, metering rules, settlement rules, model rules, and residual threshold rules.

[0085] In a preferred embodiment, the scheduling platform further generates a set of candidate nodes. This set consists of energy storage nodes that are validly registered, have not been revoked, and meet the basic access conditions. The candidate node set only represents the range of nodes eligible to participate and is not the same as the final participating set that enters the aggregation verification.

[0086] Step S3.2: Solidify the metering window summary and threshold version. The scheduling platform obtains the relevant information from the metering system or metering anchor management system. Corresponding task measurement window summary The task metering window summary can be a metering sampling root, metering log root, metering device signature summary root, or an aggregated root composed of multiple metering anchor point summaries. To ensure that the residual bucketing rule cannot be unilaterally replaced within this round of tasks, the scheduling platform calculates a threshold version summary. in, The bucketing threshold for each residual category defined in step S1 is in the 1st step. Values ​​or version references in round tasks.

[0087] Subsequently, the scheduling platform randomly selects a random number for this round of tasks. And calculate the task challenge value. The task challenge value simultaneously constrains the task content, scheduling window, metering window, rule version, bucketing threshold, and scheduling platform random number, ensuring that historical window evidence, other task evidence, or other metering window evidence cannot be directly reused in this round of tasks.

[0088] Step S3.3: Task Release Message Generation. The scheduling platform generates a task release message. in, Generate a timestamp for the task publishing message. The scheduling platform will Send to candidate energy storage node set It can also be sent synchronously to the edge verification node. and aggregation platform This is to ensure that subsequent single-site evidence pre-verification and aggregate verification use the same task benchmark.

[0089] Step S3.4: Energy storage node task verification and temporary identity generation. Energy storage node receive Then, verify the timestamp condition. And verify the signature of the scheduling platform. Subsequently, the energy storage node recalculated. and The results are then verified to ensure they match the corresponding values ​​in the task release message. The energy storage nodes also consider their local registration status. Node Certificate Rated capacity Rated power Current SOC, device alarm status, communication status, metering anchor status, and local version vector. Determine whether it meets the participation constraints. .

[0090] If any of the above verifications fail, the energy storage node will not generate a performance evidence object for this round and will record the reason for task rejection. If the verification passes, the energy storage node generates a temporary pseudo-identity for this round based on the pseudo-identity derivation material saved in step S2. in, Seeds for pseudo-identity derivation of energy storage nodes. This serves as the root of the registration record. The temporary pseudo-identity is used for subsequent performance evidence commitments, encrypted header binding, and selective opening processes, enabling the evidence in this round to be traced back to registered assets, while avoiding the direct exposure of long-term node identifiers in unnecessary scenarios.

[0091] Step S3.5: Task Acceptance Confirmation and Participation Benchmark Recording. For energy storage nodes that pass the task verification, the energy storage nodes calculate the task benchmark summary. And generate a task and accept confirmation. in, Energy storage nodes will Submit to the scheduling platform or edge verification node. Verify via the scheduling platform or edge verification node. Then, a task acceptance set is formed. And generate a task participation benchmark summary in, The task acceptance set is locked for a certain period of time. The task acceptance set does not require all candidate nodes to eventually participate in the aggregation, but it provides a benchmark for participation in subsequent pre-validation, synchronous aggregation, and anomaly attribution in this round; the actual set of nodes entering the aggregation should be a subset of the task acceptance set that has passed single-site evidence pre-validation.

[0092] After step S3, the scheduling platform, edge verification nodes, aggregation platform, and verified energy storage nodes all possess consistent task release messages, task challenge values, task baseline digests, and task participation baseline digests. The local performance evidence generated in subsequent steps S4, the verifiable ciphertext header generated in step S5, the aggregate binding digest generated in step S7, and the abnormal failure vector generated in step S9 all use... , and As a basis for verifying the identity of tasks.

[0093] Step S4: Generation of local performance evidence The local performance evidence generation phase is performed by energy storage nodes that have passed the task verification in step S3. Execution. The purpose of this stage is to convert the actual performance status within the current scheduling window into the task challenge value without uploading the original BMS, PCS, EMS, metering curves, and revenue declaration details. Task baseline summary Metric anchors, version vectors, and temporary pseudo-identities The bound physical constraint performance evidence object. After this stage of processing, the energy storage node outputs not a single metering summary or a single model update, but a combination of evidence including model update, residual bucket, component commitment, metering anchor summary, weight descriptor, performance evidence commitment root, and consistency proof.

[0094] Step S4.1: Local window data acquisition. In the scheduling window... and corresponding metrological verification window Internal, energy storage nodes Collect or read local performance data in, It should include at least SOC, SOH, temperature, protection actions, and alarm status; It should include at least active power, reactive power, power change time, and response time; It should include at least control commands, execution status, and local policy version; It should include at least the metered electricity volume at the point of entry, a summary of the metering sample, and the meter signature; This includes at least revenue reporting, settlement reporting, or performance evaluation reporting data. This data is stored locally at the energy storage node or in a controlled audit environment and is not uploaded as plaintext.

[0095] Step S4.2: Time Alignment and Data Quality Assessment. Energy storage nodes execute according to the task execution window. Metrological verification window and allowable sampling deviation ,right The process involves time alignment, outlier marking, missing value marking, and window boundary clipping to obtain aligned window data. in, This represents the data alignment function specified by system-public parameters or rule versions. Let... To obtain the set of sampling times that, after alignment, satisfy sampling integrity, timestamp validity, and device status validity, the energy storage node calculates the effective sampling rate. when At the same time, the energy storage node can still generate evidence objects, but the data quality residuals are mapped to a higher risk level, so that subsequent verification can distinguish between insufficient data quality, metering window mismatch and performance anomaly, instead of simply treating missing data as a normal response.

[0096] Step S4.3: Local model update generation. Energy storage nodes are generated based on the aligned window data. Previous round global model parameters Task context Task baseline summary and local version vector Calculate local model update The local model update can be used for energy storage response capability assessment, available capacity correction, anomaly prediction, or scheduling strategy modification. To protect energy storage node operating data and operational strategies, Submit the model in encrypted form in subsequent steps, not in plaintext.

[0097] Step S4.4: Calculation of Multidimensional Performance Residuals. The energy storage node calculates the multidimensional performance residual vector. Each residual component is non-negative and is used to characterize deviations in model adaptation, power response, battery status, metering consistency, revenue reporting consistency, and data quality during this round of scheduling. The original curves or complete details of these residual components are not required to be publicly disclosed; their calculation basis is retained locally and demonstrated through subsequent commitments and proofs.

[0098] Model residuals are used to characterize the degree of fit between the local model update and the current window data. Let... For based on The local loss function, If the upper bound of the allowable loss for this round is given, then Dispatch response residuals are used to characterize the actual power curve. Decompose the target of the node The tracking error and response delay error. Preferably, the power tracking error is calculated first. and response delay deviation Recalculate in, This indicates the time from when the energy storage node receives the dispatch command to when the actual power reaches the response criterion; To allow for response latency. If the business rules do not consider response latency, then... The SOC residual reflects both the state of charge exceeding limits and the energy conservation deviation, and is calculated as follows: in, This represents the actual change in SOC within the window recorded by the BMS. This refers to the change in SOC estimated based on metered electricity consumption or power integration.

[0099] The SOH residual is used to identify whether the change in health status within the current scheduling window exceeds the upper limit of the allowable decay: Temperature residuals are used to characterize whether the temperature of an energy storage battery exceeds the safety limit. Metering residuals are used to measure the electrical quantity on the PCS side. Electricity at the metering anchor point Consistency between them: Revenue reporting residuals are used to measure the revenue reported by energy storage operators. Consistency with accrued revenue calculated based on scheduling rules, settlement rules, and measurement data: in, Indicates basis The resulting benefit or performance evaluation function. Data quality residuals are used to reflect sampling integrity, time alignment, outlier ratio, and missing value repair. The above To prevent positive numbers with a denominator of zero, their values ​​are provided by the system's publicly available parameters or rule versions. In engineering boundary handling, if a sampling point simultaneously has both PCS power and BMS status missing values, this sampling point is not used to reduce response residuals but is included in the data quality residuals; if the metering anchor signature is valid but the metering sampling root is not... If a mismatch occurs, a metering anomaly flag is prioritized and the metering residual risk level is increased. If an energy storage node triggers a temperature or SOC-related limit exceedance or alarm within the scheduling window, the corresponding temperature or SOC residual is mapped to a high-risk bucket. If other equipment protection actions or alarms are triggered, the corresponding anomaly flag is set according to preset rules. Through the above processing, missing data, metering inconsistencies, and safety violations are categorized into different residual categories, facilitating subsequent anomaly attribution.

[0100] Step S4.5: Residual Bucketing and Component Commitment. Energy storage nodes are determined based on the bucketing threshold. Each residual component Mapped to residual bucket number like If the value exceeds the highest threshold range, it will be mapped to the highest risk bucket. This yields the bucketed vector. To support selective activation by category in disputed scenarios, energy storage nodes are configured for each residual category. Randomly select components and commit to random numbers And generate residual component commitments Subsequently, the energy storage node computes the residual bucket commitment root. This process binds each residual bucket to the same task, the same temporary pseudo-identity, and the same task baseline summary, while allowing only specified categories to be opened during subsequent dispute verification. Instead of disclosing other residual components or raw operating data.

[0101] Step S4.6: Metering Anchor Point Verification and Summary Generation. Energy storage nodes are generated from metering anchor points. Obtaining metering anchor point proof in, It can be a measurement sampling root, a measurement log root, or a signed measurement summary root. This is the signature of the metering device or system on the metering certificate. The energy storage node performs normalized serialization on the metering anchor certificate and calculates... in, The field order, encoding method, time format, and missing value filling method are determined by... or The regulations stipulate that this summary allows subsequent verification parties to verify the metering window and metering anchor linked to local performance evidence without obtaining the full metering details.

[0102] Step S4.7: Generation of Weight Descriptors and Performance Evidence Commitment Roots. Energy storage nodes generate aggregated weight descriptors based on the rated capacity, rated power, current residual bucket, and data quality status determined during the registration phase. in, and The public coarse-grained level is derived from the response residual bucket, data quality residual bucket, or corresponding proof result; when the business rules require hiding the specific bucket number, only the coarse-grained level or the level confirmed by the edge verification node is disclosed, and the original bucket number is not disclosed.

[0103] Energy storage node computational performance evidence commitment root This commitment root binds temporary pseudo-identities, task identifiers, task benchmark summaries, task participation benchmark summaries, component commitment roots, measurement anchor summaries, weight descriptors, version vectors, and task challenge values ​​together, making it difficult to piece together evidence objects from different tasks, different measurement windows, different nodes, or different rule versions into valid evidence for this round.

[0104] Step S4.8: Range and Consistency Proof Generation. The energy storage node generates a residual range proof or a consistency proof. This proof enables the verifier to confirm that each residual component commitment falls within a predefined bucket range and that there is consistency between the residual component commitment root, metering anchor summary, weight descriptor, version vector, task challenge value, and performance evidence commitment root. Based on this, the verifier can verify the formal integrity and task binding relationship of the single-site performance evidence without obtaining the original BMS, PCS, EMS, metering curves, or revenue details of the energy storage node.

[0105] After step S4 is completed, the energy storage node forms a local performance evidence object. This object serves as input for subsequent steps S5 (ciphertext header binding) and S6 (edge ​​pre-verification). The energy storage node is not publicly disclosed except for specified components required for selective opening or dispute auditing. , Complete residual value Complete bucketed vector and commitment to random numbers .

[0106] Step S5: Model encrypted binding and aggregateable tag generation The model ciphertext binding and aggregateable tag generation stages are handled by energy storage nodes. Execution. This stage is used to update the local model obtained in step S4. Local performance evidence objects Summary of Measurement Anchor Points Weight descriptor , Task Challenge Value Task baseline summary Summary of task participation benchmarks The data is bound into a single-site evidence package that can be submitted. After this stage of processing, the model ciphertext, performance evidence, measurement anchors, version vectors, and aggregate weights are no longer independent objects, but form a verifiable binding relationship within the same task context.

[0107] Step S5.1: Verifiable ciphertext header generation. The energy storage node randomly selects an encryption random number. And calculate the cryptographic random number commitment in, The encryption process is used only for local model updates and is not submitted as a plaintext field. This is used to prove that the ciphertext generation process of this round of model is bound to this round of task, thus avoiding exposing the encrypted random number itself to the verifier.

[0108] Energy storage nodes generate verifiable ciphertext headers: in, This is the encrypted header format version; Used to bind the root of the performance evidence commitment; Used to bind task challenge values; , and Used to bind task context, task baseline, and participation baseline; and Used to bind rule versions and node-side versions; Used to bind metering anchor point summaries; Used to bind the commitment root of the residual component; Used to bind aggregate weight descriptors; Used to bind the encrypted aggregate public key of the model used in this round.

[0109] The energy storage node further calculates the ciphertext header digest. in, This indicates normalized serialization. If the residual bucket level can be made public, the verification party can base it on the publicly available information. calculate If the residual bucket grade is not publicly disclosed, then As public input, by the object of proof Prove that it is consistent with the commitments of each residual component.

[0110] Step S5.2: Local model update encryption and header binding. The energy storage node uses the aggregated public key issued in step S1. Homomorphic encryption is applied to local model updates to obtain ciphertext components of the model that can participate in aggregation. Subsequently, the energy storage node computes a ciphertext header and binds a tag. Encrypted package of energy storage node formation model in, Used for subsequent model ciphertext aggregation; This is used to bind the model ciphertext components, verifiable ciphertext header, performance evidence commitment root, econometric anchor summary, residual component commitment root, weight descriptor, task baseline, and task challenge value to the same evidence object. If subsequently... , , , , , , Or encrypted components If replaced, at least one of the following will fail verification: the ciphertext header binding tag, the proof object, or the single-node aggregateable verification tag.

[0111] Step S5.3: Single-node aggregateable verification tag generation. To ensure that single-node evidence can still be verified after aggregation, the energy storage node constructs a tag message digest. Then, a single-node aggregateable verification tag is generated. in, This is to support tagging algorithms that aggregate and verify publicly available or verified weight descriptors. Ordinary digital signatures are only used for message origination and integrity verification and do not replace... The aggregated verification function. By... , , and Included Single-node labels can constrain the version of the participating set, weight rules, and scheduling rules in this round, avoiding the need to change weight descriptors or reuse labels across tasks during the aggregation phase.

[0112] Step S5.4: Single-site evidence package generation and submission. The energy storage node generates evidence objects. in, Generate a timestamp for the evidence object. Energy storage nodes will Send to edge verification node In step S6, the edge verification node performs pre-verification on the evidence object based on the task baseline, registration records, and metering rules published by the scheduling platform.

[0113] To facilitate engineering implementation and claim segmentation, the object of evidence The functions of the main fields are shown in the table below.

[0114] After step S5 is completed, the single-site evidence package It simultaneously possesses three types of binding relationships: the first type is the binding between the model ciphertext and the ciphertext header, which is... The second type is the binding between the encrypted header and physical evidence of performance, which is manifested by... , , , and The third type is the binding between single-node evidence and subsequent aggregated verification, manifested by... This is reflected in the fact that only when all three types of bindings mentioned above are met can a single-site model update be considered a valid candidate for entering the edge pre-validation and synchronous aggregation stages in this round of tasks.

[0115] Step S6: Edge Pre-verification The edge pre-verification phase is performed by edge verification nodes. Execution is used to eliminate single-site evidence packages before aggregation due to signature errors, inconsistent task contexts, abnormal encrypted header bindings, abnormal metering anchors, abnormal residual proofs, or abnormal aggregatable labels. Only evidence packages that pass this stage are allowed to enter the synchronous aggregation stage. This stage does not decrypt local model updates, nor does it require energy storage nodes to disclose the original BMS, PCS, EMS, metering curves, or revenue declaration details.

[0116] Step S6.1: Evidence Object Integrity Verification. The edge verification node receives the evidence object submitted by the energy storage node. Edge verification nodes obtain information based on task acceptance confirmation records, registration records, or controlled traceability tables. and The corresponding node public key and node certificates And confirm that the certificate is valid within the current task window. Then verify the timestamp condition. and node signature If timestamp, certificate status, or signature verification fails, the edge verification node rejects the evidence object and records the time anomaly, certificate anomaly, signature anomaly, or evidence integrity anomaly.

[0117] Step S6.2: Task binding and ciphertext header binding verification. Edge verification node parsing can verify the ciphertext header. It also verifies whether the temporary pseudo-identity, task identifier, task baseline digest, task participation baseline digest, task challenge value, task context digest, rule version digest, node version digest, metering anchor digest, residual component commitment root, weight descriptor digest, and aggregate public key in the encrypted header are consistent with the fields in the current round's task release message, task acceptance record, and evidence object, respectively. Meanwhile, edge verification node computation and from The header digest was confirmed to be consistent. Subsequently, the ciphertext header binding tags were verified to ensure they met the requirements. If any of the above conditions are not met, it indicates that there is a binding anomaly between the model's ciphertext component and the ciphertext header, the commitment root of the performance evidence, the task context, the measurement anchor, the version vector, or the weight descriptor, and the edge verification node rejects the evidence object.

[0118] Step S6.3: Metering Anchor Point Verification. Edge verification nodes verify the metering anchor point proof. Check whether the metering signature, metering window, and metering anchor point identifier are consistent with the node registration record and the metering verification window for this round, and recalculate. when Furthermore, if the metering signature verification passes, it confirms that the metering anchor point certificate is consistent with the current scheduling task and metering window; otherwise, it records the metering anchor point as abnormal.

[0119] Step S6.4: Verification of Performance Proof. Edge verification nodes verify residual range proof or consistency proof. When this verification is successful, it indicates that the residual component commitments submitted by the energy storage node are within the preset bucketing range, and that the residual component commitment root, metering anchor summary, weight descriptor, version vector, task challenge value, task baseline summary, and task participation baseline summary are consistent. If the residual bucket level is public, edge verification nodes can also directly verify it. If the residual bucket level is hidden, then only through Verify the above relationship.

[0120] Step S6.5: Single-node aggregatable label verification. The edge verification node recalculates the label message digest. And verify When the verification is successful, it means that the model ciphertext package, verifiable ciphertext header, performance evidence commitment root, proof object, metering anchor point, and weight descriptor are bound to the same energy storage node and can participate in subsequent aggregated label verification.

[0121] Step S6.6: Pre-verification pass message generation. When all verification steps S6.1 through S6.5 have passed, the edge verification node generates a pre-verification pass message. in, Edge verification nodes will Send to aggregation platform The pre-validation results summary is saved. Evidence objects that fail pre-validation do not proceed to the aggregation phase.

[0122] Step S7: Synchronous Aggregation The synchronous aggregation phase is handled by the aggregation platform. The execution is used to synchronously aggregate the homomorphic model ciphertext components and single-node aggregatable verification labels that have passed edge pre-validation, so that the aggregated model results, participating sets, aggregation weights, single-station evidence objects and aggregation labels maintain the same task context and the same weight rules.

[0123] Step S7.1: Pre-verification message confirmation and participant set determination. The aggregation platform receives the pre-verification pass message set from the edge verification nodes and verifies each one. The validity of the verification. The verified energy storage nodes constitute the actual aggregate participant set. in," This indicates the message payload signed by the edge verification node in step S6.6. The aggregation platform also verifies... And confirm the corresponding node for each participating node. , and Consistent. If If the number of participants is less than the minimum required for the task, or if there are nodes that do not belong to the task acceptance set, the aggregation platform will terminate the current round of aggregation and return a participant set exception.

[0124] Step S7.2: Aggregate Weight Calculation. The aggregation platform calculates the aggregate weights based on the scheduling task constraints and the pre-validated weight descriptors. and rule version Calculate the non-negative aggregation weights. A preferred implementation is as follows: in, To predefine the non-negative weighting function, only publicly available nominal parameters, validated coarse-grained response levels, validated data quality levels, and rule versions are used; hidden fine-grained residual buckets are not directly read. The above formula satisfies... like , or any If pre-validation fails, the aggregation platform must not generate aggregation results; instead, it should return a weight descriptor exception or a participation set exception. If specific business operations use fixed weights, capacity weights, or market clearing weights, the corresponding weight rule version should be written into... Or, as described later, the aggregation binding summary.

[0125] Step S7.3: Synchronously aggregate the model ciphertext and aggregable tags. The aggregation platform extracts data from each... Extracting homomorphic model ciphertext components And perform model ciphertext aggregation. Simultaneously, the aggregation platform performs synchronous aggregation on single-node aggregable verification tags. Both encrypted aggregation and tag aggregation use the same set of participants. and the same weight vector This avoids situations where the model aggregation set and the evidence verification set are inconsistent.

[0126] Step S7.4: Aggregate Binding Summary and Aggregate Message Generation. The aggregation platform calculates the aggregate binding summary. An aggregate message is then generated. in, The aggregation platform will Send to the dispatch platform The aggregation platform does not decrypt any single-node model updates, nor does it receive original running curves or revenue details.

[0127] Step S8: Dual-link verification on the scheduling side The dual-link verification phase on the scheduling side is handled by the scheduling platform. Execution. This phase simultaneously verifies the model aggregation link and the physical fulfillment link: the model aggregation link confirms the consistency between the aggregated model results, participant set, aggregation weights, model ciphertext set, and aggregation labels; the physical fulfillment link confirms the consistency between the ciphertext header, fulfillment proof commitment root, metering anchor, weight descriptor, version vector, and residual proof for each participant node. If both links pass, the scheduling platform confirms the reliability of this round of aggregation scheduling fulfillment; if either link fails, the anomaly attribution phase begins.

[0128] Step S8.1: Aggregated Message Integrity Verification. The scheduling platform verifies the aggregation platform signature and timestamp. If If the timestamp exceeds the allowed range, the scheduling platform will terminate the current round of aggregation verification and record the aggregation platform signature error or message timeliness error.

[0129] Step S8.2: Model aggregation link verification. The scheduling platform recalculates the aggregation binding summary. Then, verify that it matches the digest in the aggregate message. Then perform aggregate tag verification. When this verification is successful, it indicates that the aggregation model results are correct. With Participation Set Aggregate weights Model ciphertext set, aggregated tags The task baseline summary, task participation baseline summary, and rule version summary are consistent.

[0130] Step S8.3: Physical fulfillment link header verification. For each Dispatch platform analysis And verify the consistency of its fields with the current task and evidence objects: The scheduling platform simultaneously verifies In Is it equal to And verify the ciphertext header binding tag. Does it satisfy the equation described in step S6.2? If this group of verifications passes, it indicates that the model ciphertext component of this node is consistent with the commitment root of the current task, the performance evidence, the version vector, the econometric anchor, the weight descriptor, and the commitment root of the residual component.

[0131] Step S8.4: Physical Performance Verification. The scheduling platform continues to verify the performance proof of each participating node: When this verification is successful, it means The bound residual component commitment roots, econometric anchor summaries, weight descriptors, version vectors, task challenge values, task benchmarks, and task participation benchmarks must be consistent. If the verification fails, there is an anomaly in the physical fulfillment chain of that node.

[0132] Step S8.5: Acceptance flag generation. The scheduling platform defines the model aggregation link verification result as follows: Define nodes The physical fulfillment link verification result is And calculate the acceptance mark for this round. when At that time, the scheduling platform confirms that the performance of this round of aggregated scheduling is reliable and can... , , , and As one of the bases for settlement, assessment or audit; when At this point, the scheduling platform enters the anomaly attribution phase.

[0133] Step S8.6: Solidify the dual-link verification results. The scheduling platform generates a verification result record. It can also generate a signature or audit digest for the record. The technical role of the aforementioned dual-link verification is: homomorphic model ciphertext component Responsible for model aggregation, binding tags to the ciphertext header. Responsible for and Binding; Will , , , , , and Constraints must be within the same task context; and This ensures that the aforementioned single-node binding relationships remain verifiable after aggregation. Therefore, if single-node label verification, aggregated label verification, and performance verification are all successful, it can be confirmed that the aggregated model results and the current round of energy storage physical performance evidence share the same participant set, the same task context, and the same weighting rules.

[0134] Step S9: Anomaly Attribution The anomaly attribution phase is handled by the scheduling platform. This stage is executed when aggregation verification fails or physical fulfillment link verification fails at some nodes. Instead of simply outputting "verification failed," this stage generates a traceable anomaly failure vector based on the failure location, failure object, and failure category from steps S6 to S8. This allows the scheduling platform, aggregation platform, energy storage operator, and audit nodes to distinguish between task binding anomalies, encrypted header binding anomalies, tag anomalies, metering anomalies, version anomalies, physical residual anomalies, aggregation anomalies, data quality anomalies, and revenue declaration anomalies.

[0135] Step S9.1: Initialize the exception failure vector. For each participating node... The scheduling platform initializes the exception failure vector; The values ​​of each component are as follows: or The value is... This indicates that there is an anomaly in the corresponding category, and the value is [value]. This indicates that no anomalies of this category were found in this round of verification. The triggering conditions and technical meanings of each component are shown in the table below.

[0136] The aforementioned abnormal failure vectors are automatically generated by mapping failure locations in dual-link verification, without relying on manual experience. If multiple failure locations exist simultaneously at the same node, the corresponding multiple components can be set to [value missing]. .

[0137] Step S9.2: Task binding exception attribution. When the scheduling platform detects any node's exception... , , , or When inconsistent with the current task release message, task acceptance set, or aggregate message, set These anomalies are attributed to cross-task evidence reuse, historical evidence reuse, scheduling window mismatch, inconsistent task participation benchmarks, or task context tampering.

[0138] Step S9.3: Attribution of Ciphertext Header Binding Anomalies and Version Anomalies. This occurs when any header field consistency verification fails in step S8.3, or... In and Inconsistency, or ciphertext header binding tags If the binding formula is not satisfied, set If the failure location corresponds to , If the bucket threshold version, model version, device version, or metering interface version are inconsistent, further settings are required. This type of anomaly indicates an abnormal binding relationship between the model ciphertext and the commitment root of performance evidence, the summary of the econometric anchor, the weight descriptor, the commitment root of the residual component, or the version vector.

[0139] Step S9.4: Tag anomaly and aggregation anomaly attribution. When a single node's aggregateable validator tag verification fails, set... When aggregated tag verification, aggregated binding summary Participating in the group Aggregate weights Or aggregated model results When the consistency verification fails, set the following for the affected nodes: If the anomaly can only be located at the aggregation platform level and cannot be clearly pointed to a single node, the scheduling platform can... Generate aggregation exception markers for all nodes and record the failure items in the exception description as participation set exception, weight exception, aggregation label exception, or aggregation binding summary exception.

[0140] Step S9.5: Attribution of Meter Anchor Point Anomalies. This applies when the meter anchor point signature is invalid, or when the metering window... Inconsistency, discrepancies between the metering anchor point identifier and the registration record, or recalculation. Not equal to the evidence object When, set These anomalies are attributed to abnormal measurement anchor points, replaced measurement data, mismatched measurement windows, or inconsistent measurement summaries.

[0141] Step S9.6: Prove the attribution of failures. When At that time, the scheduling platform sets up an abnormal failure vector based on the proof failure items, the proof public input, and the proof error code. If the residual bucket value corresponding to the failure item exceeds the allowable range, response residual, SOC residual, SOH residual, temperature residual, or measurement residual, then an exception is set. If the failure item corresponds to the effective sampling rate, missing rate, time drift, outlier ratio, or data alignment condition, then set... If the failure item corresponds to a profit reporting residual or settlement report consistency, then set... If a failed proof involves multiple categories, then all corresponding components are simultaneously set to [value missing]. .

[0142] Step S9.7: Anomaly Attribution Result Generation and Audit Saving. The scheduling platform generates anomaly attribution results. in, The scheduling platform will Send to audit node Save. If long-term identity needs to be traced, the audit node or scheduling platform will only do so when preset authorization conditions are met, based on registration records and... Perform controlled mapping.

[0143] Step S10: Selective opening in disputed scenarios The selective opening phase is used during dispute verification or audit review to handle abnormal failure vectors. Limited access is granted for the specified residual category. This phase does not require energy storage nodes to disclose complete BMS, PCS, EMS, metering curves, complete residual vectors, or revenue reporting details; it only opens access for disputed categories. The relevant bucketing components, committed random numbers, and necessary auxiliary information.

[0144] Step S10.1: Selective Open Request Generation. When the scheduling platform or audit node needs to verify a certain abnormal component, the scheduling platform generates a selective open request. in, Indicates the residual category that needs to be opened. Indicate the reason for opening. The It should at least correspond to a set component in the abnormal failure vector. For example, when At that time, the category to open can be ;when At that time, the category to open can be .

[0145] Step S10.2: Limited activation of energy storage nodes. Energy storage nodes receive Next, verify the validity of the scheduling platform signature, task identifier, task baseline summary, open category, and open reason. Once verification is successful, the energy storage node only opens the category. Corresponding residual bucket number Residual component commitment Promise random numbers and necessary supporting information and generate a selective open response. in, remove Apart from the corresponding components, the energy storage node does not open other residual buckets, does not disclose the complete residual vector, and does not disclose the original operating data.

[0146] Step S10.3: Enable result verification. The scheduling platform or audit node is based on... Verify the opened result. The verification process includes at least the following: verify Belongs to The set of residual components under constraints is committed to, and verification is performed. Bind this , , , , , and The above process can be represented as follows: If the verification is successful, it confirms that the abnormal component can be traced back to the original performance evidence commitment; if the verification is unsuccessful, it confirms that the node has an abnormal commitment opening or that the disputed component is inconsistent with the original evidence object.

[0147] Step S10.4: Selectively open record saving. The scheduling platform or audit node generates a selectively open record: in, This indicates whether the verification was successful or failed. This record is used for dispute resolution, settlement review, or subsequent auditing, but does not expand the scope of the information that has been opened.

[0148] To illustrate the application of steps S1 to S10 in specific business scenarios, the following example uses multiple electrochemical energy storage power stations participating in demand response aggregation scheduling. This embodiment is only used to illustrate the implementation of the present invention in a typical novel energy storage scheduling fulfillment scenario and does not constitute a limitation on the scope of application of the present invention. For frequency regulation, reserve, peak shaving and valley filling, capacity compensation, spot market ancillary services, or other market-based scheduling fulfillment scenarios, as long as there is a task context, metering anchor point, single-station fulfillment evidence, model encrypted aggregation, and post-aggregation verification requirements, the same binding, aggregation verification, and anomaly attribution mechanisms can be adopted.

[0149] In this embodiment, the scheduling platform In the The demand response mission requires multiple electrochemical energy storage power stations exist to The discharge response is completed within the scheduling window. The scheduling platform determines the overall target response power curve. Total target response power Participation constraints and settlement or assessment rules Assign reference power curves to each candidate energy storage power station. The aforementioned business inputs are written into the task context. And further generate task challenge values. Task baseline summary Summary of task participation benchmarks Therefore, the performance evidence, model encrypted text, and aggregation tags generated by each power station subsequently are all associated with the same demand response task, the same scheduling window, and the same metering caliber.

[0150] In terms of system deployment, the scheduling task context generation and publishing module is deployed on the scheduling platform. It is used to generate the demand response task context, task challenge value, and rule version; the energy storage node evidence generation module and the encrypted binding module are deployed in each energy storage power station. It is used to generate local model updates, performance residuals, component commitments, performance evidence commitment roots, proof objects, and model ciphertext packages; the edge pre-validation module is deployed on edge validation nodes. It is used to perform signature, task binding, ciphertext header, metering anchor, proof object, and aggregateable tag verification on single-site evidence objects; the synchronous aggregation module is deployed on the aggregation platform. It is used to synchronously aggregate the pre-validated model ciphertext components and aggregable validation labels; the scheduling-side dual-link validation module, anomaly attribution module, and audit tracing module are deployed on the scheduling platform or audit node. It is used to store verification summaries, anomaly attribution summaries, and selectively opened records.

[0151] During the task execution phase, each energy storage power station first determines whether it meets the participation constraints for this round based on its registration records, rated power, rated capacity, SOC safety boundary, equipment operating status, metering anchor status, and local version vector. Energy storage power stations that pass verification generate a temporary pseudo-identity for this round based on pseudo-identity derivation materials. .exist to Within the execution window, each power station collects BMS data, PCS data, EMS command execution logs, gateway metering data, and business application data. Specifically, BMS data includes at least SOC, SOH, temperature, and alarm status; PCS data includes at least active power, reactive power, and response time; EMS logs include at least control commands, execution status, and local policy version; gateway metering data includes at least metered electricity consumption, metering window, and meter signature; and business application data includes at least demand response revenue or settlement application value.

[0152] The energy storage power station performs time alignment, sampling integrity checks, and outlier marking on the above multi-source data to obtain aligned window data. And calculate the effective sampling rate. When the sampling rate, missing rate, or time drift does not meet the preset thresholds, the energy storage power station does not directly upload the original curve, but instead reflects this situation in the data quality residual. In the middle. Subsequently, the energy storage power station was based on Previous round global model parameters Task context and local version vector Get local model update The model residuals, response residuals, SOC residuals, SOH residuals, temperature residuals, measurement residuals, revenue reporting residuals, and data quality residuals are calculated. These residuals are then mapped to residual buckets. and through component commitment and residual component commitment root Curing.

[0153] In terms of metering and weighting, energy storage power stations start from the metering anchor point. Obtaining metering anchor point proof And calculate the summary of the quantitative anchor points. Simultaneously, the energy storage power station forms an aggregated weight descriptor based on rated capacity, rated power, response level, data quality level, and version vector. Subsequently, the energy storage power station generated performance evidence and commitment root. This binds temporary pseudo-identities, task identifiers, task baseline summaries, task participation baseline summaries, residual component commitment roots, metering anchor summary, weight descriptors, version vectors, and task challenge values ​​together. If an attacker replaces the metering window, reuses historical residual buckets, replaces weight bases, replaces rule versions, or splices other power plant evidence into the encrypted model of this round, subsequent encrypted header verification, proof verification, or label verification will fail.

[0154] During the encrypted binding phase, the energy storage power station will , , , , , , , and cryptographic random number commitment Write a verifiable ciphertext header and update the local model. Encryption into model ciphertext components Ciphertext header bound to tag Will , , , , , , , and They are bound to the same evidence object. The energy storage power station further generates single-node aggregateable verification tags. and the evidence object Send to the edge verification node.

[0155] Edge verification node for evidence object Perform pre-verification. Verification includes node signature, task baseline, encrypted header fields, encrypted header binding tags, metering anchor proof, residual range proof or consistency proof, and single-node aggregateable verification tags. Evidence objects that pass pre-verification are sent to the aggregation platform; evidence objects that fail pre-verification are rejected and recorded as signature anomaly, task binding anomaly, encrypted header anomaly, metering anchor anomaly, proof anomaly, or tag anomaly, depending on the failure.

[0156] The aggregation platform is based on a collection of pre-validated energy storage power stations. and verified weight descriptors Calculate aggregate weights The weighting function uses only publicly available parameters, validated coarse-grained response levels, validated data quality levels, and rule versions, without directly reading the hidden fine-grained residual buckets. Subsequently, the aggregation platform performs homomorphic aggregation on the model's encrypted components to obtain the aggregated model result. It also performs synchronous aggregation on single-node aggregable verification tags to obtain aggregated tags. To prevent the aggregation platform from replacing participating sets, modifying weights, or swapping evidence objects, the aggregation platform generates an aggregation binding summary. The participating sets, weights, aggregation model results, aggregation labels, and evidence summaries for each individual station will be submitted to the scheduling platform.

[0157] After receiving the aggregation message, the scheduling platform first verifies the aggregation platform signature, participant set, aggregation weight, and aggregation binding digest, and then performs model aggregation link verification and physical fulfillment link verification. The model aggregation link is used for confirmation. , , , Consistency between the model's encrypted set and the physical fulfillment link; used to verify the consistency of each energy storage power station. , , , , , and Consistency between them. When both links pass verification, the scheduling platform outputs... The verification results can be used as one of the bases for demand response settlement, assessment or audit.

[0158] If the model aggregation link or physical fulfillment link verification fails, the scheduling platform generates an abnormal failure vector based on the failure location. For example, if the task challenge value or task baseline summary is inconsistent, it is attributed to an abnormal task binding; if the metering anchor signature is valid but the metering window is inconsistent with the current round of tasks, it is attributed to an abnormal metering anchor; if the response residual bucket proof fails, it is attributed to an abnormal physical fulfillment residual; if the effective sampling rate or time alignment condition does not meet the rules, it is attributed to an abnormal data quality; if the aggregate weight is inconsistent with the rule version, it is attributed to an aggregation anomaly. For disputed abnormal components, the scheduling platform or audit node only requests to open the component commitment, bucket number, and necessary auxiliary information for the corresponding residual category, without requiring the disclosure of the complete PCS power curve, BMS status curve, EMS control log, or revenue details.

[0159] As can be seen from the above embodiments, this invention does not merely provide post-hoc evidence for demand response results, nor does it simply verify the correctness of model aggregation. Instead, it solidifies task, measurement, and rule benchmarks during the generation and release phases of the demand response task context, generates physical constraint performance evidence on the energy storage power station side, completes the binding of ciphertext and evidence roots on the model ciphertext side, maintains synchronization between the model ciphertext and aggregable tags on the aggregation side, performs dual-link verification on the scheduling side, and outputs interpretable anomaly attributions and controllable selective opening results when verification fails. Therefore, single-station physical performance evidence can enter the aggregation process along with the model ciphertext and remain verifiable, traceable, and selectively openable after aggregation.

[0160] In a preferred embodiment Pedersen commitments, hash commitments, or other commitment algorithms with hiding and binding properties can be used; Range proofs, set membership proofs, or alternative zero-knowledge proof mechanisms can be used. It can be implemented using homomorphic encryption, function encryption, or secure aggregation that supports weighted or linear ciphertext aggregation; Signatures, authentication codes, or homomorphic tagging mechanisms that support aggregation verification can be used. This invention does not limit the specific cryptographic algorithm implementation, as long as it can realize the binding relationship between the model ciphertext components, ciphertext header binding tags, performance evidence commitment roots, measurement anchors, weight descriptors, task challenge values, and the aggregated two-link verification.

[0161] Compared with the prior art, the present invention has at least the following beneficial effects.

[0162] (1) Improve the reliability of the binding between model updates and actual energy storage performance. This invention incorporates the model ciphertext package, performance evidence commitment, scheduling task, metering anchor point, weight basis and version information into the same evidence object, so that model updates cannot be submitted or replaced separately from the corresponding scheduling task, metering window and performance evidence, thereby reducing the risks of historical evidence reuse, cross-task evidence transfer, cross-node evidence splicing and model ciphertext and performance evidence splitting and replacement.

[0163] (2) Performance verification without disclosing the original operating data. This invention transforms performance factors of energy storage nodes, such as model adaptability, scheduling response, battery status, temperature status, metering consistency, revenue declaration consistency, and data quality, into multidimensional residuals, which are expressed through bucketing, component commitment, and proof mechanisms. The scheduling platform can verify whether the performance status of energy storage nodes meets the preset rules without directly obtaining the original BMS, PCS, EMS, metering curves, and revenue declaration details.

[0164] (3) Balancing audit verifiability with operational data privacy protection. This invention adopts a component commitment generation method based on residual categories, which enables the disclosure of only the bucketed components and necessary auxiliary information corresponding to the specified residual categories when disputes occur, without disclosing the complete power curve, battery status curve, control log, metering details or revenue declaration details, thereby achieving a balance between audit verification and protection of commercially sensitive information.

[0165] (4) Enhanced post-aggregation verification capability. This invention enables the homomorphic model ciphertext components and aggregable verification tags to be aggregated synchronously according to the same participant set and the same aggregation weight. The participant set, weight, ciphertext header, performance evidence commitment, measurement anchor point, and version information are solidified through aggregation binding digest. As a result, the scheduling platform can still verify the correspondence between the aggregation model results and the performance evidence of each individual station after aggregation, reducing the risk of participant set replacement, weight tampering, and evidence object swapping.

[0166] (5) A dual-link verification closed loop is formed, encompassing model credibility, physical performance credibility, and measurement credibility. This invention not only verifies the correctness of the model aggregation process but also verifies the consistency of the physical performance evidence, measurement anchors, version information, and residual proofs bound to the model ciphertext. Thus, the credibility of model updates, the credibility of energy storage physical performance, the credibility of measurement, and the credibility of settlement audits can be jointly verified within the same task context.

[0167] (6) Enhanced capabilities in anomaly handling, accountability, and multi-scenario applicability. This invention categorizes verification failure reasons into task binding, encrypted header binding, tags, metering, version, residual range, aggregation, data quality, and revenue reporting, enabling the scheduling platform to output an explainable source of the anomaly after verification failure. This facilitates operation and maintenance handling, accountability differentiation, settlement review, and audit tracing. This invention can be applied to scenarios involving multiple electrochemical energy storage power stations participating in demand response aggregation scheduling, virtual power plant aggregation response, shared energy storage scheduling, user-side energy storage response, vehicle-to-grid interactive resource scheduling, and market-based auxiliary service settlement. It can provide verifiable evidence for scheduling acceptance, market settlement, anomaly handling, and dispute auditing.

[0168] In summary, the beneficial effects of this invention do not lie in the individual use of encryption, signature, commitment, metering digest, or evidence storage, but rather in integrating task binding, physical constraint performance evidence generation, model ciphertext binding, synchronous aggregation of aggregable tags, dual-link verification on the scheduling side, anomaly attribution, and selective opening into the same technical closed loop. This enables the novel energy storage aggregation scheduling performance to still possess verifiable, attributable, and auditable technical effects under privacy protection conditions.

[0169] Example 2: Furthermore, this invention also proposes a system 200 for generating aggregated evidence of physical constraints, verifying encrypted bindings, and attributing anomalies for novel energy storage scheduling compliance, comprising: Initialization unit 201 is used for system initialization to generate public parameters, key parameters and residual bucketing rules, and to register energy storage assets to bind power station identity, metering anchor point and version vector, as well as to generate scheduling task context to generate task context and task challenge value. Verification unit 202 is used to generate single-station performance evidence to form residual buckets, measurement summaries and proofs, bind the model ciphertext to generate ciphertext headers and aggregable verification tags, perform edge pre-verification, and if it passes, perform synchronous aggregation to aggregate the model ciphertext and verification tags for scheduling-side dual-link verification. The attribution unit 203 is used to generate an abnormal attribution result, output a failure vector, and selectively open and audit trace the dual-link verification on the scheduling side when it fails, in order to verify the disputed residual components.

[0170] If edge pre-verification fails, a single-site anomaly is recorded to prevent it from entering the aggregation.

[0171] If the dual-link verification on the scheduling side passes, a reliable performance result is output, and the process proceeds to scheduling acceptance or settlement.

[0172] The initialization unit 201 is also used for: binding the consistency of scheduling tasks, metering windows, and rule versions, including: At the start of each round of new energy storage scheduling tasks, the scheduling task identifier, scheduling execution window, gate metering window, scheduling rule version, residual threshold version, and anti-replay information are incorporated into the same task binding relationship.

[0173] Among them, the generation of single-site performance evidence includes: generation of physical constraint performance evidence; The generation of the physical constraint performance evidence includes: Based on the equipment status, response behavior, gateway metering results, and business application requirements of energy storage nodes, multi-dimensional performance residuals are constructed, and the residuals are converted into residual buckets with finite levels. Energy storage nodes generate component commitments for different residual categories, and the component commitments, metering anchors, weighting basis, version information and task binding information together form the performance evidence commitment root.

[0174] The verification unit 202 is also used for: the inseparable binding of the model ciphertext package and the performance evidence, including: When generating the model update ciphertext, a verifiable ciphertext header is generated, and the performance evidence commitment root, task binding information, measurement anchor, version information, residual commitment root, and weight basis are written into the ciphertext header. The energy storage node regenerates a ciphertext header and binds a tag, making the model ciphertext component and the aforementioned performance evidence object form an inseparable model ciphertext package.

[0175] The verification unit 202 is also used for: aggregating verification tags, aggregating binding digests, and dual-link verification, including: The platform aggregates the model ciphertext of multiple energy storage nodes, and aggregates the model ciphertext with the ciphertext of a single node under the same participant set and the same weight rule. The participant set, aggregation weight, model ciphertext package, ciphertext header, performance evidence commitment, metering anchor, weight basis, version information and task binding information are synchronously solidified through aggregation binding digest. After receiving the aggregation results, the scheduling platform performs model aggregation link verification and physical fulfillment link verification respectively to confirm the consistency of the aggregation model results, participating sets, weights and aggregable verification labels, as well as the consistency of the encrypted header, fulfillment evidence, metering anchor, version information and residual proof.

[0176] Among these, anomaly attribution, selective opening, and system collaborative deployment include: The anomaly is attributed to the following reasons based on the location of the failure: task binding anomaly, encrypted header binding anomaly, tag anomaly, metering anomaly, version anomaly, residual range anomaly, aggregation anomaly, data quality anomaly, or revenue declaration anomaly. In disputed scenarios, selective opening is performed on specified residual categories, and the consistency of the opening results with the original component commitment and the root of the performance evidence commitment is verified.

[0177] This invention enables the inclusion of scheduling task context, metering anchor point, rule version, aggregation weight, model ciphertext, and physical performance residuals into a single verifiable evidence chain without directly disclosing the original BMS, PCS, EMS, metering, and revenue data of energy storage nodes. Furthermore, this evidence chain can enter the aggregation process along with the model ciphertext, and after aggregation, consistency verification can still be performed on the participating sets, weight rules, model aggregation results, and the physical performance status of individual stations. Simultaneously, when verification fails, it supports anomaly attribution based on categories such as task, ciphertext, aggregation, metering, version, physical constraints, revenue declaration, and data quality.

[0178] Example 3: Based on the same inventive concept, this invention also provides a computer device, which includes a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to implement corresponding method flows or corresponding functions, thereby implementing the steps of the methods in the above embodiments.

[0179] Example 4: Based on the same inventive concept, this invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the method in the above embodiments.

[0180] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0181] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0182] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0183] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0184] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0185] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for generating aggregated evidence of physical constraints, performing encrypted binding aggregation verification, and attributing anomalies for novel energy storage dispatch fulfillment, characterized in that: include: The system is initialized to generate public parameters, key parameters, and residual bucketing rules, and energy storage assets are registered to bind power station identity, metering anchor point, and version vector. Scheduling task context is also generated to generate task context and task challenge value. Single-site performance evidence generation is performed to form residual buckets, measurement summaries and proofs. The model ciphertext is bound to generate a ciphertext header and an aggregable verification label. Edge pre-verification is performed. If it passes, synchronous aggregation is performed to aggregate the model ciphertext and verification label for dual-link verification on the scheduling side. If the dual-link verification on the scheduling side fails, an anomaly attribution result is generated, a failure vector is output, and selective opening and audit tracing are performed to verify the disputed residual components.

2. The method according to claim 1, characterized in that, If edge pre-validation fails, a single-site anomaly is recorded to prevent entry into the aggregation.

3. The method according to claim 1, characterized in that, If the dual-link verification on the scheduling side passes, a reliable performance result will be output, and the process will proceed to scheduling acceptance or settlement.

4. The method according to claim 1, characterized in that, Also includes: The consistency binding of scheduling tasks, metering windows, and rule versions includes: At the start of each round of new energy storage scheduling tasks, the scheduling task identifier, scheduling execution window, gate metering window, scheduling rule version, residual threshold version, and anti-replay information are incorporated into the same task binding relationship.

5. The method according to claim 1, characterized in that, Single-site performance evidence generation, including: physical constraint performance evidence generation; The generation of the physical constraint performance evidence includes: Based on the equipment status, response behavior, gateway metering results, and business application requirements of energy storage nodes, multi-dimensional performance residuals are constructed, and the residuals are converted into residual buckets with finite levels. Energy storage nodes generate component commitments for different residual categories, and the component commitments, metering anchors, weighting basis, version information and task binding information together form the performance evidence commitment root.

6. The method according to claim 1, characterized in that, Also includes: The inseparable binding of the model's encrypted package with the evidence of performance includes: When generating the model update ciphertext, a verifiable ciphertext header is generated, and the performance evidence commitment root, task binding information, measurement anchor, version information, residual commitment root, and weight basis are written into the ciphertext header. The energy storage node regenerates a ciphertext header and binds a tag, making the model ciphertext component and the aforementioned performance evidence object form an inseparable model ciphertext package.

7. The method according to claim 1, characterized in that, Also includes: It can aggregate verification tags, aggregate binding digests, and two-link verification, including: The platform aggregates the model ciphertext of multiple energy storage nodes, and aggregates the model ciphertext with the ciphertext of a single node under the same participant set and the same weight rule. The participant set, aggregation weight, model ciphertext package, ciphertext header, performance evidence commitment, metering anchor, weight basis, version information and task binding information are synchronously solidified through aggregation binding digest. After receiving the aggregation results, the scheduling platform performs model aggregation link verification and physical fulfillment link verification respectively to confirm the consistency of the aggregation model results, participating sets, weights and aggregable verification labels, as well as the consistency of the encrypted header, fulfillment evidence, metering anchor, version information and residual proof.

8. The method according to claim 1, characterized in that, Anomaly attribution, selective enabling, and collaborative system deployment include: The anomaly is attributed to the following reasons based on the location of the failure: task binding anomaly, encrypted header binding anomaly, tag anomaly, metering anomaly, version anomaly, residual range anomaly, aggregation anomaly, data quality anomaly, or revenue declaration anomaly. In disputed scenarios, selective opening is performed on specified residual categories, and the consistency of the opening results with the original component commitment and the root of the performance evidence commitment is verified.

9. A system for generating aggregated evidence of physical constraints, performing encrypted binding aggregation verification, and attributing anomalies for novel energy storage dispatch fulfillment, characterized in that: include: The initialization unit is used for system initialization to generate public parameters, key parameters and residual bucketing rules, and to register energy storage assets to bind power station identity, metering anchor point and version vector, as well as to generate scheduling task context to generate task context and task challenge value. The verification unit is used to generate single-station performance evidence to form residual buckets, measurement summaries and proofs. It binds the model ciphertext to generate a ciphertext header and an aggregable verification tag for edge pre-verification. If it passes, it performs synchronous aggregation to aggregate the model ciphertext and verification tag for dual-link verification on the scheduling side. The attribution unit is used to generate anomaly attribution results, output failure vectors, and selectively enable and audit trace the dual-link verification on the scheduling side when it fails, in order to verify disputed residual components.

10. The system according to claim 9, characterized in that, If edge pre-validation fails, a single-site anomaly is recorded to prevent entry into the aggregation.

11. The system according to claim 9, characterized in that, If the dual-link verification on the scheduling side passes, a reliable performance result will be output, and the process will proceed to scheduling acceptance or settlement.

12. The system according to claim 9, characterized in that, The initialization unit is also used for: binding the consistency of scheduling tasks, metering windows, and rule versions, including: At the start of each round of new energy storage scheduling tasks, the scheduling task identifier, scheduling execution window, gate metering window, scheduling rule version, residual threshold version, and anti-replay information are incorporated into the same task binding relationship.

13. The system according to claim 9, characterized in that, Single-site performance evidence generation, including: physical constraint performance evidence generation; The generation of the physical constraint performance evidence includes: Based on the equipment status, response behavior, gateway metering results, and business application requirements of energy storage nodes, multi-dimensional performance residuals are constructed, and the residuals are converted into residual buckets with finite levels. Energy storage nodes generate component commitments for different residual categories, and the component commitments, metering anchors, weighting basis, version information and task binding information together form the performance evidence commitment root.

14. The system according to claim 9, characterized in that, The verification unit is also used for: the inseparable binding of the model ciphertext package and performance evidence, including: When generating the model update ciphertext, a verifiable ciphertext header is generated, and the performance evidence commitment root, task binding information, measurement anchor, version information, residual commitment root, and weight basis are written into the ciphertext header. The energy storage node regenerates a ciphertext header and binds a tag, making the model ciphertext component and the aforementioned performance evidence object form an inseparable model ciphertext package.

15. The system according to claim 9, characterized in that, The verification unit is also used for: aggregating verification tags, aggregating binding digests, and two-link verification, including: The platform aggregates the model ciphertext of multiple energy storage nodes, and aggregates the model ciphertext with the ciphertext of a single node under the same participant set and the same weight rule. The participant set, aggregation weight, model ciphertext package, ciphertext header, performance evidence commitment, metering anchor, weight basis, version information and task binding information are synchronously solidified through aggregation binding digest. After receiving the aggregation results, the scheduling platform performs model aggregation link verification and physical fulfillment link verification respectively to confirm the consistency of the aggregation model results, participating sets, weights and aggregable verification labels, as well as the consistency of the encrypted header, fulfillment evidence, metering anchor, version information and residual proof.

16. The system according to claim 9, characterized in that, Anomaly attribution, selective enabling, and collaborative system deployment include: The anomaly is attributed to the following reasons based on the location of the failure: task binding anomaly, encrypted header binding anomaly, tag anomaly, metering anomaly, version anomaly, residual range anomaly, aggregation anomaly, data quality anomaly, or revenue declaration anomaly. In disputed scenarios, selective opening is performed on specified residual categories, and the consistency of the opening results with the original component commitment and the root of the performance evidence commitment is verified.

17. A computer device, characterized in that, include: One or more processors; A processor is used to execute one or more programs; When the one or more programs are executed by the one or more processors, the method described in any one of claims 1-8 is implemented.

18. A computer-readable storage medium, characterized in that, It contains a computer program, which, when executed, implements the method as described in any one of claims 1-8.