Verifiable reconciliation feedback-based behavior event excitation calculation and closed-loop regulation and control method and system, and storage medium

Abstract

The invention discloses a behavior event excitation calculation and closed-loop regulation and control method and system based on verifiable reconciliation feedback and a storage medium. Supply side behavior event data and use side (consumption / deduction / use / cancel-after-verification / consumption) event data are collected and desensitized and anonymized, and a trusted computing node under an uploading chain completes compliance and anti-cheating verification and calculates incentive output. And generating excitation by a global control function under the dynamic total amount upper limit constraint, and outputting a bounded smooth adjustment item by a controller based on the participation scale, the accumulated new increment and the use side deduction settlement feedback / risk state. Maintaining a Mint Legger and a Settlement Legger under the chain, generating an abstract commitment value in a window, and writing the abstract commitment value into a tamper-resistant auditing medium; and performing consistency reconciliation verification on the new increment, the deducted settlement amount and the accumulated balance state based on the commitment, and feeding back a result for online parameter adjustment / gating updating to realize closed-loop adaptive regulation and control.

Description

Technical Field

[0001] This invention relates to the technical fields of data security and trusted computing, distributed ledger, audit evidence storage, consistency verification and closed-loop control, and in particular to a method, system and computer-readable storage medium for generating incentive accounting units, deduction settlement, verifiable audit evidence storage and closed-loop adaptive parameter tuning for multi-source behavioral event data. Background Technology

[0002] In multi-source data environments such as IoT terminals, mobile applications, in-vehicle systems, and payment and settlement systems, various quantifiable events (e.g., mileage, energy consumption, learning behavior, device operation, merchant verification / consumption, etc.) are constantly generated. To encourage compliant behavior, suppress abnormal behavior, and support subsequent auditing and governance, it is usually necessary to reliably collect, verify, record, and settle these events, forming a closed loop among different entities. Existing solutions generally suffer from the following technical problems: Insufficient credibility of multi-source events: The sources of events are complex, and there is a lack of unified entry point verification, anti-replay, and traceability mechanisms, making it difficult to prove that the event actually occurred and has not been tampered with; Inconsistent on-chain / off-chain data: If new minting and deduction settlement are executed in different systems, inconsistencies in balance, cumulative minting amount, and settlement deduction amount are likely to occur, resulting in high auditing costs; Difficulty in achieving both auditing and privacy: Uploading all data in plaintext to the chain will cause privacy leaks and high storage / bandwidth costs; relying solely on off-chain records lacks third-party verifiability; Difficulty in system stability and anomaly suppression: Fluctuations in participation scale, abnormal injections, and attack replays can lead to output fluctuations or risk spread, and there is a lack of closed-loop parameter tuning mechanisms based on system consistency and secure operating status; Therefore, there is an urgent need for a technical solution that is feasible in engineering, verifiable in auditing, protects privacy, allows for closed-loop governance, and has a more comprehensive architecture. Summary of the Invention

[0003] Technical problems to be solved This invention aims to solve at least the following technical problems: providing a trusted collection, verification, and computation mechanism for multi-source behavioral events, reducing the risks of tampering, forgery, and replay; in scenarios where new casting and deduction settlements coexist, providing a mechanism based on dual ledgers + verifiable digest commitments + consistent reconciliation to achieve auditable, traceable, and low-cost third-party verification; in on-chain / off-chain collaborative or fully on-chain scenarios, providing a closed-loop adaptive parameter tuning control mechanism to ensure boundedness and smoothness of outputs while suppressing abnormal fluctuations; and providing an equivalent implementation of an immutable auditable ledger not limited to blockchain implementations to cover different engineering architectures. Technical solution To achieve the above objectives, the present invention provides the following technical solutions (corresponding to claims 1-21):

[0004] (1) General architecture: Immutable audit ledger + dual ledger consistency + verifiable reconciliation feedback closed-loop control (corresponding to claim 1) Behavioral event data records are collected and generated at the behavior collection terminal. These behavioral events are all quantifiable physical or virtual behavioral events, including supply-side and usage-side (consumption / deduction / use / redeem / consumption) behavioral events. Each behavioral event data record includes at least: a basic behavioral event value B, a credibility parameter V, a time period weight T, a persistence parameter S, a category label W, and an anonymous identifier ID or its hash of the behavioral subject. After anonymization / desensitization, the behavioral event data records are uploaded to an off-chain trusted computing node. The off-chain trusted computing node performs compliance and anti-fraud verification on the behavioral events and calculates the incentive output value (e.g., GCU(t)) upon successful verification. The incentive calculation is generated by the global control function F under the upper limit constraint of the dynamic total control function A(t). The controller / regulator Φ outputs an adjustment term U(t) based on the state quantities related to the participation scale, the state quantities related to cumulative casting, and / or the settlement feedback / risk state quantities on the user side (consumption / deduction / use / reversal / consumption), and the state quantities related to system consistency and safe operation. This ensures that the output is bounded and smooth, suppressing the abnormal amplification of small-scale participation and maintaining stability when the scale changes. Two types of append-only ledgers are maintained off-chain: the Mint Ledger for new incentive casting and the Settlement Ledger for deduction settlement. Verifiable digest commitment values ​​(such as Merkle roots, hash chain digests, or equivalent commitment structures) of the two ledgers are generated within a preset time window and written into the immutable audit ledger. Based on the digest commitment values, the periodic new casting amount, periodic deduction settlement amount, issuance speed, and cumulative balance status are reconciled and verified for consistency. The parameters of Φ and F are adjusted online accordingly to achieve closed-loop adaptive control. The immutable audit ledger is an append-only recording medium that supports third-party verification. Its implementation forms include, but are not limited to, consortium blockchains, centralized chains, WORM audit logs, trusted timestamps / notarized evidence storage, or combinations thereof.

[0005] (2) On-chain mint-only state change path (corresponding to claim 2) In a preferred embodiment, the immutable audit ledger comprises a consortium blockchain and its smart contracts: on-chain events only support state change events (mint events) that add new incentives and evidence events that do not change the balance state; transfer, destruction, and update-type balance state changes are not supported. After verification by the off-chain trusted computing node, a minting request is generated and submitted to the smart contract. Upon contract verification, the mint accounting is executed. The minting request includes at least event_id, event_digest, anonymous identifier hash, trustworthiness parameter, incentive output value (e.g., GCU(t)), timestamp, authorization proof (gateway_id and gateway_sig or equivalent proof), and policy_hash (a rule version commitment). (3) On-chain full-process closed-loop route (corresponding to claim 3) In another embodiment, the immutable audit ledger is implemented by the blockchain and its smart contracts. The smart contracts on the blockchain simultaneously support changes in the state of new incentives and changes in the state of deduction settlement. The changes in the state of deduction settlement include at least one of burning, transferring, or deduction based on the state machine. Specifically: new incentives are generated by off-chain trusted computing nodes to generate a minting request and trigger on-chain mint event accounting; the deduction settlement corresponding to the user side (consumption / deduction / use / redeem / consumption) can be directly executed by the on-chain smart contract to deduct, burn, and / or transfer based on the state machine, or the settlement module can complete the actual deduction settlement and generate a settlement result, and form a settlement summary or proof to be submitted to the chain. After verification by the smart contract, it is recorded as an on-chain deduction settlement event or an on-chain evidence storage event. New incentive events and deduction settlement events recorded on the chain are all subject to periodic reconciliation verification and are compared with the summary commitment values ​​of the Mint Ledger and Settlement Ledger maintained off-chain. The reconciliation results are fed back to the controller Φ to update the adjustment parameters, thereby achieving closed-loop stable control based on on-chain / off-chain collaboration.

[0006] (4) System and storage medium (corresponding to claims 4 and 5) The present invention also provides a system for implementing the above method, comprising: a behavior acquisition module, a data gateway, an off-chain trusted computing node, a dual ledger module, an immutable audit ledger interface module, and a closed-loop control module; and provides a computer-readable storage medium storing a computer program, which implements the above method when the program is executed.

[0007] (5) Expansion of accounting unit format (corresponding to claim 6) The incentive accounting unit can be either an account balance accounting unit or a voucher accounting unit; among them, the voucher accounting unit is a verifiable signature voucher or a verifiable commitment voucher, which is verified and voided or marked as used to achieve deduction settlement when used (consumption / deduction / use / redeem / consumption).

[0008] (6) In one implementation of the controller family and computing structure (corresponding to claims 7-12, 19), the excitation output value satisfies: GCU(t) = Gbehavior × C × U(t) U(t) = μ × R(t) × Φ(Ω(t)) Where: Gbehavior = B×V×T×S×W; ; Ω(t) includes at least the following state variables: those related to participation scale, those related to cumulative casting, those related to user-side (consumption / deduction / use / reimbursement / consumption) settlement, and / or those related to system consistency and secure operation. User-side settlement-related state variables include, but are not limited to, consumption amount, consumption rate, number of deduction settlement events, deduction settlement success rate, inventory level, budget execution progress, and remaining budget. System consistency and secure operation-related state variables include, but are not limited to, reconciliation inconsistency rate, verification failure rate, replay detection count, proportion of abnormal events, evidence storage delay, and node load. Φ The output satisfies boundedness and smoothness and suppresses amplification of anomalies under dynamic total limit constraints. It can be viewed as a weight vector / factor vector The components of each dimension, The aggregation is a scalar gating result obtained by aggregating the vectors; the aggregation is not limited to the product form, but can also be a weighted sum, norm, normalization, function composition or model scoring, etc. To adapt to different implementation environments and control requirements, this invention does not limit the combination form of F, Φ, and U(t). It can be multiplicative coupling, additive coupling, normalized amortization, piecewise combination, function composition, or equivalent transformations thereof. As long as the combination achieves bounded and smooth output under dynamic total quantity control constraints and implements feedback adjustment based on the consistency reconciliation verification results, thereby obtaining the same technical effect as the above implementation methods, it is considered an alternative implementation method of this invention. Meanwhile, Φ can have multiple implementation families: g(Nt, Ns(t)) family of gate functions (fractional saturation, power saturation, exponential saturation, sigmoid / hyperbolic or piecewise squeeze); dynamic difficulty D(t) (PoW isomorphism); budget allocation BUD(t); In this invention, the so-called "damping state quantity" is used to characterize the system's suppression strength or bearing pressure on excitation release. It can be composed of one or more indicators such as queue length, settlement delay, node load, failure rate, risk score, and reconciliation anomaly degree, and can be used as input to congestion gating, supply-side gating and / or use-side gating to reduce excitation strength when pressure increases or risk rises; In at least one implementation, a normalized allocation mechanism can be used within the budget window to normalize the weights of events, ensuring that the total amount of new castings within the window does not exceed the budget; and / or a squeeze / saturation shaping mapping can be used to limit Φ or U(t) within a preset range, thereby further ensuring that the output is bounded, smooth, and suppressing abnormal amplification; and / or the controller Φ can use feedback control structures such as PI / PID to adjust the stable target error, consume feedback quantity, and / or consistent reconciliation results; and / or a segmented threshold, melting factor, and / or release pool smoothing mechanism can be used to implement segmented adjustment, speed limiting, delayed release, batch release, or sliding window smoothing of the new casting intensity, in order to suppress short-term fluctuations and improve the stability of closed-loop control.

[0009] (7) Equivalent implementation of anti-replay, version tracing and authorization proof (corresponding to claims 13 and 14) The newly added casting request / event shall include at least: event_id, event_digest, anonymous identifier hash, timestamp, authorization proof, and policy_hash. event_id is used for anti-replay; policy_hash is used for rule change tracing and audit recalculation.

[0010] Authorization proofs are not limited to gateway signatures, but also include threshold signatures, multisignatures, remote proofs, zero-knowledge proofs, or combinations thereof, as long as the verifiable request is generated by the authorized entity and is bound to the rule version commitment.

[0011] (8) Batch root commitment and Merkle proof (corresponding to claims 15 and 16) New casting and / or deduction settlement events can be batch-processed off-chain and generated into a periodic root commitment written to an immutable audit ledger; each event is verified through proof-based verification. A Merkle proof is generated for any casting or settlement record, including a leaf digest, a proof path hash sequence, and a root hash identifier. The leaf digest consists of at least the hash combination of event_id, anonymous identifier hash, digest, value (incentive / deduction), timestamp, and policy_hash.

[0012] (9) Centralized chain governance (corresponding to claim 17) A central chain can be set up to anchor the periodic root commitments of consortium chains / business chains, and to manage the registration and revocation of policy_hash and authorized entity identities.

[0013] (10) Dual ledger consistency reconciliation (corresponding to claim 18) Based on the summary commitment values ​​of Mint Ledger and Settlement Ledger, the system performs reconciliation and verification of the periodic increase in casting volume, periodic deduction in settlement volume, and cumulative balance status, and feeds back the reconciliation results to the controller Φ for online parameter adjustment.

[0014] (11) Triggering of governance action (corresponding to claim 20) Consistent reconciliation includes verifying the consistency of the periodic new casting volume, the periodic deduction settlement volume, and the cumulative balance status. When the reconciliation inconsistency rate, verification failure rate, or abnormal event ratio exceeds the threshold, governance actions are triggered, including reducing the new casting rate, adjusting the gain / threshold of Φ, enabling segmented circuit breaking or releasing pool smoothing, restricting gateway requests from abnormal sources, suspending new casting, and generating audit alarms.

[0015] (12) Equivalent audit medium as a fallback (corresponding to claim 21) Immutable audit ledgers are not limited to blockchain ledgers, but also include equivalent audit media that can provide append-only writing, immutability or detectable tampering, and third-party verifiability. These include WORM audit log systems, trusted timestamps / notarized evidence storage, TEE / HSM signature logs, threshold signatures / multi-party consensus joint audit ledgers or combinations thereof, used to write periodic summary commitment values ​​and version commitment information and support integrity verification and audit traceability.

[0016] Beneficial effects Compared with the prior art, the present invention has at least the following technical effects: 1. Verifiable auditing and non-repudiation: By generating periodic summary commitment values ​​for Mint Ledger and Settlement Ledger within a preset time window, and combining this with a proof-based verification mechanism, authorized auditors / third parties can verify the inclusion of events and consistency of results without obtaining plaintext data, thereby achieving verifiable auditing, evidence retention, and non-repudiation. 2. Consistency assurance and state risk reduction: By maintaining the new minting ledger and the settlement ledger off-chain, and performing consistency reconciliation verification based on the periodic summary commitment value, the consistency verification of the periodic new minting amount, periodic settlement amount and cumulative balance status is carried out, thereby reducing the risk of state inconsistency caused by on-chain / off-chain parallelism, multi-system collaboration or cross-domain settlement; 3. Anti-replay, traceable and recalculated: By setting event_id for events to support uniqueness verification / anti-replay, and by using policy_hash to commit to rule versions and combining authorization proof (such as gateway signature or equivalent verifiable authorization proof) to bind the request source to the rule version, the system security is improved, and audit traceability and recalculation verification by version rule are supported; 4. Balancing privacy and performance: By writing digest commitment structures (such as Merkle roots, hash chain digests, or equivalent commitment structures) instead of writing all plaintext on the chain / ledger, storage and bandwidth costs are reduced and system load is lessened, while the plaintext exposure surface is reduced and privacy protection capabilities are improved. 5. Economic closed loop and sustainable total control: By incorporating feedback status quantities such as deduction settlement amount, consumption rate, budget execution progress, and inventory level formed on the user side (consumption / deduction / use / reverse / consumption) into the gating and parameter adjustment of the controller / regulator Φ, and combining the constraints of dynamic total upper limit A(t), remaining total R(t) or budget window budget BUD(t), the newly added casting intensity is adaptively adjusted according to the use and system status, thereby realizing the economic closed loop of "supply-use-feedback", avoiding unrestrained expansion and improving long-term sustainability; 6. Stability Improvement and Anomaly Suppression: The controller / regulator Φ performs closed-loop parameter tuning based on the participation scale state quantity, cumulative casting state quantity, user side feedback / risk state quantity, and system consistency and safe operation state quantity (such as reconciliation inconsistency rate, verification failure rate, abnormal event ratio, etc.), and applies boundedness and smoothness constraints to the output, thereby suppressing the amplification of anomalies under small-scale participation, reducing short-term fluctuations, and maintaining stable system operation. 7. Scalable Architecture: This invention does not limit the unique chain structure and unique computational expression form. It covers both the limited state change route of mint-only and the full process route that supports deduction and settlement on the chain. At the same time, it provides equivalent implementation backstop for tamper-proof audit media (including consortium blockchain, central chain anchoring, WORM audit logs, trusted timestamps / notarized evidence storage, TEE / hardware signature logs, etc.), and allows control functions and gating functions to adopt a variety of equivalent combinations and shaping methods (such as normalized amortization, squeeze / saturation shaping, segmented circuit breaking and release pool smoothing, etc.). Attached Figure Description

[0017] This instruction manual does not include any drawings. Detailed Implementation

[0018] The specific embodiments are described below in conjunction with the basic principles of the present invention. It should be understood that the scope of protection of the present invention is not limited to the following embodiments; any equivalent substitutions made to the structure, step sequence or parameter form without departing from the spirit of the present invention shall fall within the scope of protection of the present invention.

[0019] Behavioral event data recording: The behavior collection terminal generates event records for supply-side behavioral events and usage-side (consumption / deduction / use / redeem / consumption) behavioral events. The event records include at least: the basic value of the behavioral event B, the credibility parameter V, the time period weight T, the persistence parameter S, the category label W, and the anonymous identifier ID of the behavioral subject or its hash identifier; in at least one implementation, the event records also include fields such as device fingerprint digest, location / environment digest, business context label, and collection terminal signature digest for anti-fraud and audit recalculation.

[0020] Anonymization / Desensitization and Trusted Entry Point: Before event records enter the behavior control platform, the data gateway performs desensitization / anonymization processing and generates a unique event identifier (event_id) and an event digest (event_digest). The data gateway generates a verifiable authorization certificate (such as gateway_id and gateway_sig, or an equivalent verifiable authorization certificate) and binds the authorization certificate to the rule version commitment (policy_hash) to support rule change tracing and audit recalculation.

[0021] Off-chain trusted computing and verification: Off-chain trusted computing nodes perform a verification process on events that includes at least compliance verification, anti-fraud verification, and anti-replay verification. Among them, anti-replay verification can be implemented based on the uniqueness check of event_id, timestamp window constraints, challenge-response, or equivalence mechanisms. After the verification is passed, the off-chain trusted computing node performs incentive calculation to obtain the incentive output value (e.g., GCU(t)).

[0022] Excitation Calculation and Dynamic Total Quantity Control: In at least one implementation, the global control function F generates the basic issuance intensity under the upper limit constraint of the dynamic total quantity control function A(t), and the controller / regulator Φ outputs the adjustment term U(t); wherein, A(t) can be a fixed upper limit, a phased release upper limit, or an upper limit function that updates with the system state. To ensure that the total quantity is controlled, the cumulative issuance amount S(t) can be counted within a preset time window, and the remaining total quantity R(t) = A(t) - S(t) can be calculated, and R(t) ≥ 0; R(t) can be used as a gating input to participate in the calculation of U(t) or to impose restrictions on the output.

[0023] The set of state variables Ω(t) and the symmetric closed loop: Ω(t) includes at least: state variables related to participation scale (such as supply-side participation scale, user-side participation scale), state variables related to cumulative casting (such as cumulative casting quantity, inventory level), state variables related to user-side settlement (such as consumption quantity, consumption rate, number of deduction settlement events, success rate, budget execution progress, remaining budget), and state variables related to system consistency and safe operation (such as reconciliation inconsistency rate, verification failure rate, abnormal event ratio, evidence storage delay, node load). The controller Φ can implement symmetric closed-loop adjustment of the supply side and user side based on Ω(t): when user-side consumption is insufficient or risk increases, the intensity of new casting is reduced; when user-side consumption is sufficient and reconciliation consistency is good, the casting intensity can be smoothly increased or restored.

[0024] Gating Function Family and Congestion Gating: To suppress amplification of small-scale participation anomalies and maintain system stability, Φ(Ω(t)) may include a congestion gating term g(·). The congestion gating term is generated based at least on the user-side participation scale state quantity and / or user-side damping state quantity, and the supply-side participation scale state quantity and / or supply-side damping state quantity, to characterize the degree of system congestion; the damping state quantity may consist of queue length, settlement delay, node load, failure rate, risk score, reconciliation anomaly degree, etc. The congestion gating term satisfies at least two of the following: boundedness, smoothness, monotonicity, or piecewise monotonicity, and may take the form of saturation ratio, normalized ratio, power / exponential transformation, sigmoid / logistic, or piecewise continuous function.

[0025] Squeeze / Saturation Shaping: To further ensure bounded and smooth output, a squeeze / saturation shaping mapping can be applied to Φ or U(t) to limit it within a preset range; the squeeze can be in the form of hard truncation or smooth saturation. The upper and lower bounds of the preset range can be dynamically determined by R(t), budget progress, inventory limit, risk threshold, or reconciliation consistency threshold to push the output back to a stable region when an abnormal amplification trend occurs.

[0026] Normalized Allocation and Budget Window: To ensure that the total amount of new castings within the window is controlled, the budget BUD(t) or the castable amount ΔA(t) can be determined within a preset time window. The weighted score q_i(t) of each event within the window is calculated and then normalized to generate incentive values ​​for each event. The budget BUD(t) or ΔA(t) can be output by Φ and / or by a closed-loop control module equivalent to Φ, and is updated online based on user-side settlement feedback and reconciliation results. Normalized allocation ensures that the total casting amount within the window naturally does not exceed the budget and reduces the risk of abnormal amplification caused by excessive casting volume.

[0027] The dual-ledger module maintains two off-chain ledgers: a Mint Ledger for new incentive minting and a Settlement Ledger for deduction settlement. Both ledgers use append-only writing and do not modify existing records. The Mint Ledger records each new incentive minting result; the Settlement Ledger records the deduction settlement results corresponding to the usage side (consumption / deduction / use / redeem / consumption). In at least one implementation, the deduction settlement is completed by the off-chain settlement module, and a settlement summary or settlement proof is generated for on-chain verification or notarization.

[0028] Periodic Digest Commitment and Immutable Audit Ledger: Within a preset time window, verifiable digest commitment values ​​are generated for both the Mint Ledger and the Settlement Ledger. These commitment values ​​can be Merkle roots, hash chain digests, or equivalent commitment structures. The commitment values ​​are then written into an immutable audit ledger. This immutable audit ledger is an append-only recording medium that supports third-party verification. It can be implemented using a consortium blockchain, a central chain, WORM audit logs, trusted timestamps / notarized evidence storage, TEE / HSM signature logs, or a combination thereof.

[0029] Consistency reconciliation verification: The behavior control platform performs consistency reconciliation verification on the periodic new casting volume, periodic deduction settlement volume and cumulative balance status based on the above summary commitment value, and forms a reconciliation result; the reconciliation may include periodic summary consistency verification of Mint and Settlement, inclusion proof verification of single event, and constraint verification of balance / stock evolution relationship.

[0030] Verifiable proof unfolding: In at least one implementation, a Merkle proof can be generated for any casting or settlement record. The proof includes a leaf digest, a proof path hash sequence, and a root hash identifier. The leaf digest consists at least of a hash combination of event_id, anonymous identifier hash, event digest, numeric fields (incentive value and / or deduction value), timestamp, and policy_hash. Third parties can use this to verify event inclusion and commitment consistency without obtaining the plaintext.

[0031] On-chain mint-only path: In a preferred implementation, the immutable audit ledger includes the consortium blockchain and its smart contracts. On-chain, only new incentive state change events (mint events) and evidence events that do not change the balance state are supported. Transfer, destruction, and update-type balance state changes are not supported. Off-chain trusted computing nodes generate minting requests and submit them to the smart contract. The request includes at least event_id, event_digest, anonymous identifier hash, incentive output value, timestamp, authorization proof, and policy_hash. After the contract is verified, mint accounting is executed.

[0032] On-chain support for deduction settlement path: In another implementation, in addition to supporting mint events, on-chain smart contracts also support deduction settlement status change events corresponding to the user side (consumption / deduction / use / redeem / consumption). The events include at least one of burn, transfer, and / or quota deduction based on the state machine. Deduction settlement can be executed directly by the on-chain contract according to the preset state machine rules, or the off-chain settlement module can generate a settlement summary or proof and submit it on-chain. After contract verification, it is recorded as an on-chain deduction settlement event and / or an on-chain evidence storage event.

[0033] Restricted state change path: In another implementation, only new incentive events and evidence events that do not change the balance are recorded on the chain. The deduction settlement is only executed off-chain and a verified settlement summary or proof is stored on the chain to achieve tamper-proof auditing of the deduction settlement process without directly changing the balance on the chain.

[0034] Reconciliation Feedback and Online Parameter Adjustment: The reconciliation results are fed back to the controller Φ or the equivalent closed-loop control module for online updates of the global control function F, the gain parameters, threshold parameters, gating parameters and / or budget parameters of the controller Φ; thereby enabling closed-loop adjustment of new incentive intensity, deduction settlement rules or related gating parameters when the participation scale, consumption feedback, risk status or consistency status changes.

[0035] Governance Action Trigger: When the reconciliation inconsistency rate, verification failure rate, replay detection count, or abnormal event ratio exceeds a preset threshold, the behavior control platform triggers at least one governance action, including: reducing the new casting rate, adjusting the gain and / or threshold of Φ, enabling segmented circuit breaking or release pool smoothing mechanism, restricting requests from the gateway of the abnormal source, suspending new casting and generating audit alarms, in order to reduce system risks and restore consistency.

[0036] Equivalent Implementation: To adapt to different implementation environments and control requirements, this invention does not limit the combination form of F, Φ, and U(t), which can be multiplicative coupling, additive coupling, weighted combination, normalized amortization, piecewise combination, function composition, or equivalent transformations thereof; wherein, the equivalent transformation is to mathematically rewrite or remap the calculation expression without changing the set of input state variables, the reconciliation verification link, and the dynamic total constraint mechanism; as long as the constraints of dynamic total control, boundedness, and smoothness are met, and feedback parameter tuning based on consistent reconciliation verification results is achieved, all are alternative implementations of this invention.

[0037] The above description is only a preferred embodiment of the present invention. Those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and all such modifications and improvements should fall within the protection scope of the present invention.

Claims

1. A method, system, and storage medium for behavioral event incentive calculation and closed-loop control based on verifiable reconciliation feedback, characterized in that, Includes the following steps: (1) Collect and generate behavioral event data records at the behavior collection terminal. The behavioral events are all quantifiable physical or virtual behavioral events, including supply-side behavioral events and user-side consumption / deduction / use / refund / consumption behavioral events. The behavioral event data record includes at least: behavioral event base value B, credibility parameter V, time period weight T, persistence parameter S, category label W, and anonymous identifier ID of the behavioral subject or its hash identifier; (2) The behavioral event data records are uploaded to the off-chain trusted computing node of the behavior control platform after being desensitized / anonymized; the off-chain trusted computing node performs compliance and anti-fraud verification on the behavioral events, and performs incentive calculation to obtain incentive output value after the verification is passed; (3) The incentive calculation is generated by the global control function F under the upper limit constraint of the dynamic total control function A(t), and the controller / regulator Φ outputs the incentive intensity adjustment term U(t) based on the participation scale state quantity, the cumulative casting state quantity and / or the consumption / deduction / use / refund / consumption feedback risk state quantity on the use side, so that the output has boundedness and smoothness, so as to suppress the abnormal amplification of small-scale participation and maintain stability when the scale changes; (4) Maintain at least two types of append-only ledgers off-chain: Mint Ledger and Settlement Ledger, which record the new incentive results and the corresponding deduction settlement results for consumption / offsetting / use / reverse / consumption respectively in an append-only manner; the deduction settlement results can be generated by the off-chain settlement module and synchronously form a settlement summary or proof for on-chain verification or storage, without modifying or deleting existing records; (5) Generate verifiable digest commitment values ​​for the Mint Ledger and Settlement Ledger within a preset time window. The digest commitment values ​​include Merkle roots, hash chain digests, or equivalent commitment structures, and are written into an immutable audit ledger to achieve audit traceability. The immutable audit ledger is an append-only recording medium that supports third-party verification. Its implementation forms include, but are not limited to, consortium blockchains, central chains, WORM storage audit logs, trusted timestamps / notarized evidence storage, or combinations thereof. (6) The behavior control platform performs consistency reconciliation verification on the periodic new casting quantity, periodic deduction settlement feedback indicators and cumulative balance status based on the summary commitment value, and dynamically adjusts the parameters of controller Φ and function F online based on the reconciliation results, thereby realizing closed-loop adaptive control of the incentive accounting unit.

2. The method as described in claim 1, characterized in that: In at least one implementation, the immutable audit ledger includes at least one consortium blockchain and its smart contract; the consortium blockchain only supports new incentive state change events (mint events) and hashed evidence events that do not change the balance state, and does not support balance state change events such as transfers, destruction, and updates; after verification, the off-chain trusted computing node generates a minting request and submits it to the smart contract, the minting request including at least: event unique identifier event_id, behavior event digest event_digest, anonymous identifier hash, trust parameter, incentive output value (e.g., GCU(t)), timestamp, authorization proof (including gateway identifier gateway_id and its signature gateway_sig or equivalent verifiable authorization proof) and rule version commitment policy_hash; after the smart contract verifies, it executes the mint event accounting.

3. The method as described in claim 1, characterized in that: In at least one implementation, the immutable audit ledger includes at least one blockchain and its smart contract, wherein the smart contract at least supports on-chain state change events (mint events) for newly added incentives; and, in at least one implementation, the smart contract further supports deduction settlement state change events corresponding to consumption / deduction / use / reimbursement / consumption, wherein the deduction settlement state change events include at least one of burn and / or transfer and / or state machine-based quota deduction; wherein: (1) After the off-chain trusted computing node completes compliance verification, anti-fraud verification and strategy calculation for the behavioral event data record, it generates a new incentive minting request and submits it to the chain. After verification by the smart contract, the new incentive event (mint event) is triggered on the chain. (2) The deduction settlement corresponding to consumption / credit / use / redeem / consumption shall satisfy at least one of the following implementation paths: (2.1) On-chain deduction settlement path: The on-chain smart contract performs at least one of the following on the target account balance based on the state machine: deduction, burning and / or transfer, according to the preset state machine rules, and generates an on-chain deduction settlement event; (2.2) Off-chain settlement - on-chain accounting path: The off-chain settlement module generates deduction settlement results based on consumption / deduction / use / redeem / consumption events, and forms a settlement summary or settlement certificate to be submitted to the chain; After the smart contract verifies the settlement summary or settlement certificate, it records it as an on-chain deduction settlement event and / or an on-chain evidence storage event; (2.3) Restricted state change path: Only new incentive events and evidence events that do not change the balance are recorded on the chain. Deduction settlement is only executed off-chain and a verified settlement summary or proof is recorded on the chain to achieve tamper-proof audit of the deduction settlement process without directly changing the balance on the chain. (3) The on-chain new incentive event, the on-chain deduction settlement event (if present) and / or the on-chain evidence storage event jointly participate in the periodic reconciliation verification, generate the reconciliation result and feed it back to the controller Φ or an equivalent closed-loop control module, which is used to make closed-loop adjustments to the new incentive intensity, deduction settlement rules or related gating parameters to achieve system stability.

4. A behavioral event-driven computation system for implementing the method of any one of claims 1 to 3, characterized in that, include: (1) Behavior acquisition module; (2) Anonymization / desensitization and trusted entry module (data gateway), used to generate unique event identifiers, event summaries and authorization certificates; (3) Off-chain trusted computing nodes are used to perform verification, incentive calculation and generate casting / deduction requests; (4) Dual ledger module, including Mint Ledger and Settlement Ledger; (5) An immutable audit ledger interface module, used to write summary commitment values ​​and support third-party verification; (6) Closed-loop control module, used to adjust the parameters of controller Φ and global control function F online based on the reconciliation results.

5. A computer-readable storage medium having a computer program stored thereon, said computer program, when executed by a processor, implementing the method of any one of claims 1 to 3.

6. The method according to any one of claims 1 to 3, characterized in that: The incentive accounting unit includes an account balance accounting unit and / or a voucher accounting unit; wherein, the voucher accounting unit is a verifiable signature voucher or a verifiable commitment voucher, which is verified and voided or marked as used when consumed / deducted / used / redeemed to achieve deduction.

7. The method according to any one of claims 1 to 3, characterized in that: GCU(t) = Gbehavior × C × U(t), and U(t) = μ × R(t) × Φ(Ω(t)); where Gbehavior = B × V × T × S × W, and C is the gating result obtained by aggregation operation of a set of weight factors or multidimensional weight vectors; in at least one embodiment, C = ∏(i=1..n) ω_i; The Ω(t) includes at least the state quantities related to participation scale, the state quantities related to cumulative casting, the state quantities related to consumption / deduction / use / reverse / consumption settlement, and / or the state quantities related to system consistency and safe operation; wherein the state quantities related to consumption / deduction / use / reverse / consumption settlement include, but are not limited to, at least one of consumption amount, consumption rate, number of deduction settlement events, deduction settlement success rate, stock level, budget execution progress, and remaining budget; The output Φ satisfies boundedness and smoothness and suppresses amplification of anomalies under the constraint of dynamic total upper limit; furthermore, Φ implements negative feedback adjustment on the gating factors involved by U(t) and / or R(t) based on the state variables related to system consistency and safe operation, so that the new casting intensity is reduced when the reconciliation inconsistency rate, verification failure rate or abnormal event ratio increases, and the new casting intensity is restored or increased when the system consistency and safe operation state is restored; wherein, the state variables related to system consistency and safe operation include, but are not limited to, reconciliation inconsistency rate, verification failure rate, replay detection times, abnormal event ratio, evidence storage delay or node load, etc.

8. The method as described in claim 7, characterized in that: The Φ is characterized by the dynamic difficulty D(t), and U(t) = μ × R(t) × ψ(D(t)), where D(t) is updated in a closed loop according to the participation scale-related state variables and / or system state variables, and the value of D(t) is within a preset interval; ψ(·) is a bounded gating function that is monotonically non-increasing or piecewise monotonically non-increasing as D(t) increases; in at least one embodiment, ψ(D(t)) is at least one of the following: reciprocal form, fractional form, exponential decay form, or sigmoid form.

9. The method as described in claim 7, characterized in that: Within a preset time window, the cumulative amount of newly added incentives is statistically analyzed to obtain the cumulative amount S(t). Based on the dynamic total limit A(t), the remaining total amount R(t) = A(t) - S(t) is calculated, and R(t) ≥ 0. The R(t) participates in the gating calculation of U(t) to achieve dynamic total control.

10. The method as described in claim 7, characterized in that: The Φ is based on the consumption feedback amount and / or consumption rate formed by consumption / deduction / use / reimbursement / consumption and / or deduction settlement events, or on the stable target error, and implements feedback control, satisfying bounded and smooth constraints.

11. The method as described in claim 7 or 10, characterized in that: The Φ(Ω(t)) includes a congestion gating term g(·), which is generated based at least on the user-side participation scale state quantity and / or user-side damping state quantity, and the supply-side participation scale state quantity and / or supply-side damping state quantity, and is used to characterize the congestion level of the system and suppress the aberration amplification of the issuance gating; wherein, the congestion gating term satisfies at least two of the following constraints: boundedness, smoothness, monotonicity, or piecewise monotonicity; and the congestion gating term takes at least one of the following forms: (1) Saturated ratio form or normalized ratio form; (2) The ratio is transformed by a power function and / or an exponential function; (3) Sigmoid / Logistic form or piecewise continuous function form; The congestion gating term, together with the supply-side gating term and / or the user-side gating term, constitutes a gating function for generating the controller / regulator Φ. Each gating term is combined in at least one of the following ways: multiplicative coupling, additive coupling, weighted combination, normalized amortization, piecewise combination, or function composition. Furthermore, the gating function can be further used to make Φ satisfy the boundedness and smoothness within a preset interval through squeeze / saturation shaping mapping, thereby achieving symmetrical closed-loop regulation of the supply side and the user side and suppressing abnormal amplification.

12. The method as described in any one of claims 7, 10, or 11, characterized in that: The Φ includes a segmented threshold mechanism, a circuit breaker factor, and / or a release pool smoothing mechanism, which are used to implement at least one of the following on the release intensity of newly added excitation: segmented adjustment, rate limiting, delayed release, batch release, or sliding window smoothing, so as to suppress short-term fluctuations and improve the stability of closed-loop control.

13. The method according to any one of claims 1 to 3, characterized in that: A new casting request and / or a new incentive event must include at least event_id, event_digest, anonymous identifier hash, timestamp, authorization proof, and policy_hash; where event_id is used for anti-replay or uniqueness verification, and policy_hash supports rule change tracing and audit recalculation.

14. The method as described in claim 13, characterized in that: The authorization proof includes, but is not limited to, gateway signature, threshold signature, multisignature, remote proof, zero-knowledge proof or a combination thereof. The authorization proof can be verified to prove that the new incentive casting request was generated by the authorized entity and is bound to the rule version commitment policy_hash or equivalent version commitment information.

15. The method according to any one of claims 1 to 3, characterized in that: New casting and / or deduction settlement events are batch-processed off-chain to generate periodic root commitments, which are Merkle root or equivalent commitment structures and written into an immutable audit ledger; each event is verified through proof expansion or equivalent verification mechanisms.

16. The method as described in claim 15, characterized in that: For any casting or settlement record, generate a Merkle proof, which includes a leaf digest, a proof path hash sequence, and a root hash identifier; the leaf digest consists of at least a hash combination of event_id, anonymous identifier hash, event digest, numeric fields (incentive value and / or deduction value), timestamp, and policy_hash.

17. The method as described in claim 2 or 3, characterized in that: The central chain is set up to anchor the periodic root commitments of consortium chains and / or business chains, and is used for the registration, renewal or revocation management of policy_hash and authorized entity identities.

18. The method according to any one of claims 1 to 3, characterized in that: Based on the summary commitment values ​​of Mint Ledger and Settlement Ledger, the consistency of the newly added casting quantity, the deducted settlement quantity, and the cumulative balance status in the cycle is verified, and the reconciliation results are fed back to the controller Φ for online parameter tuning and / or policy gating updates.

19. The method as described in any one of claims 7, 10, or 11, characterized in that: The combination of the global control function F, the controller / regulator Φ, and the regulation term U(t) is not limited to multiplicative coupling, but also includes additive coupling, weighted coupling, normalized amortization, piecewise combination, function composition, or equivalent transformation. The equivalent transformation involves mathematically rewriting or parameter remapping the computational expression without changing the set of input state variables and constraints. As long as the combination satisfies the constraints of dynamic total control, boundedness, and smoothness, and achieves feedback parameter tuning based on consistent reconciliation verification results, it can achieve the same technical effect of suppressing abnormal amplification and maintaining stable system operation.

20. The method as described in claim 1 or 18, characterized in that: The consistency reconciliation verification includes the consistency verification of the periodic increase in casting volume, the periodic deduction in settlement volume, and the cumulative balance status. When the reconciliation inconsistency rate, verification failure rate, or proportion of abnormal events exceeds a preset threshold, the behavior control platform triggers at least one governance action, including: reducing the new casting rate, adjusting the gain parameter and / or threshold parameter of controller Φ, enabling segmented circuit breaking or release pool smoothing mechanism, restricting requests from gateways of abnormal sources, and / or suspending new casting and generating audit alarms, in order to reduce system risks and restore consistency.

21. The method as described in claim 1, characterized in that: The implementation of the immutable audit ledger is not limited to blockchain ledgers, but also includes equivalent audit media that can provide append-only writing, immutability or detectable tampering, and third-party verifiability. The equivalent audit media includes, but is not limited to: audit log systems using WORM media, evidence chain systems based on trusted timestamps or notarized evidence, immutable log systems generated and signed by hardware security modules and / or trusted execution environments, joint audit ledger systems using threshold signatures and / or multi-party consensus, or combinations thereof. The equivalent audit media is used to write the periodic summary commitment values ​​and version commitment information of the Mint Ledger and Settlement Ledger, and supports integrity verification and audit traceability of the summary commitment values, thereby achieving the same or equivalent technical effects as blockchain ledgers.