A dynamic pricing response and settlement processing method in a power retail scenario

Abstract

This invention relates to the field of electricity retail and settlement processing technology, and more specifically, to a dynamic pricing response and settlement processing method in electricity retail scenarios. This method acquires multi-source business data and generates a unified record sequence; identifies missing anomalies, misaligned anomalies, conflicting anomalies, and time-sensitive anomalies, forming an anomaly type combination code; identifies states based on dual time windows and generates response energy consumption segments, calculates segment reliability scores and segment reliability evolution sequences; generates segment-level dynamic pricing response budget vectors through anomaly action constraint matrices and scenario-object-time period settlement impact tensors, and performs segment-level settlement processing; and performs dominant component inheritance recalculation when a rollback segment occurs. This invention improves the linkage, reliability, and continuous output capability under abnormal scenarios in dynamic pricing response and settlement processing.

Description

Technical Field

[0001] This invention relates to the field of electricity retail and settlement processing technology, and more specifically, to a dynamic pricing response and settlement processing method in the context of electricity retail. Background Technology

[0002] In the electricity retail market, electricity retailers typically need to implement dynamic pricing responses for target retailers based on metered load, contractual constraints, market clearing results, historical response execution, and historical settlement records, and then process settlements based on the actual response results.

[0003] Existing technologies can typically perform load analysis, price adjustment, and settlement calculations, but they still have the following shortcomings in the electricity retail business scenario: Existing technologies typically treat missing, misaligned, conflicting, and timeout issues in multi-source business data only as preliminary cleaning problems. This results in the differential impact of different anomaly types on status identification, budget constraints, and settlement processing failing to propagate to subsequent processing links, thus affecting the reliability of dynamic pricing response strategies and settlement results. Existing technologies often use decentralized rules to handle anomalies, lacking a differentiated constraint structure that unifies anomaly identification, status correction, budget contraction, and settlement constraints. This leads to a loose processing chain in complex business scenarios, and a lack of stable linkage between dynamic pricing response and settlement processing.

[0004] Therefore, we propose a dynamic pricing response and settlement processing method for electricity retail scenarios to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a dynamic pricing response and settlement processing method in an electricity retail scenario to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a dynamic pricing response and settlement processing method in an electricity retail scenario, comprising the following steps: Obtain multi-source business data corresponding to the target retail object. The multi-source business data includes at least metering load data, contract parameter data, electricity price rule data, historical response execution logs, historical settlement details, market clearing results, and interface status data. Multi-source business data is mapped to a unified settlement benchmark timeline to generate a unified record sequence. Each record in the unified record sequence includes at least a timestamp, object identifier, scenario identifier, time period type identifier, load value, price benchmark value, contract constraint value, settlement rule index, source identifier, and validity identifier. Data quality processing is performed on the unified record sequence to identify key field missing, timestamp misalignment, cross-source key field conflict and continuous interface timeout, and the identification results are marked as missing anomalies, misalignment anomalies, conflict anomalies and time-sensitive anomalies, respectively, and then data credibility scores, source consistency flag sets and anomaly type combination codes are generated; Based on the unified record sequence, short and long analysis time windows are constructed respectively. Load fluctuation characteristics, price response characteristics, execution stability characteristics and settlement disturbance characteristics are extracted. The state deviation index, response sensitivity coefficient and settlement disturbance coefficient are calculated, and the state label, data credibility level and current effective parameter spectrum are determined. Using the state deviation index crossing threshold change point, time period type switching point, and execution stability coefficient change point as the dividing boundary, the continuous energy consumption process of the target retail object is dynamically divided to generate multiple response energy consumption segments, and the transferable load capacity range, trigger window, response time delay range, and execution stability coefficient are determined for each response energy consumption segment. Based on the distribution of anomaly types, anomaly duration, anomaly intensity, source consistency flag set status, and data credibility score within each response energy segment, the segment credibility score of each response energy segment is calculated, and a segment credibility evolution sequence is generated in chronological order. An anomaly action constraint matrix is ​​constructed to establish the mapping relationship between missing anomalies, misaligned anomalies, conflicting anomalies, and time-sensitive anomalies and state feature correction channels, fragment boundary correction channels, budget component contraction channels, and settlement item constraint channels. Differentiated constraints are applied to each response energy fragment based on the anomaly type combination encoding and fragment credibility evolution sequence. Based on response energy consumption segments, contract parameter data, electricity price rule data, historical settlement details and market clearing results, a scenario-object-time period settlement impact tensor is constructed. Each tensor unit stores at least the basic electricity charge impact component, deviation assessment impact component, incentive rebate impact component and default correction impact component. The abnormal type combination encoding and segment category of each response energy consumption segment are mapped to the corresponding tensor unit. Based on the current effective parameter spectrum, the segment credibility score of each response energy segment, the segment credibility evolution sequence, the constraint results of the anomaly effect constraint matrix, and the impact components of the corresponding tensor units, a dynamic pricing response budget vector is generated for each response energy segment. The dynamic pricing response budget vector includes at least the upper limit of price adjustment range, execution duration, notification advance, response limit, and settlement risk limit. Based on the dynamic pricing response budget vector, each response energy segment is divided into executable segments, restricted segments, and rollback segments. Based on the actual execution results and the dynamic pricing response strategy corresponding to each response energy segment, segment-level settlement processing is performed. The effective response electricity and settlement correction coefficient of each response energy segment are calculated to obtain the deviation responsibility allocation result and the incentive rebate or default processing result. Based on the intensity ranking result of the impact component corresponding to each response energy segment and the intensity of abnormal propagation impact, the dominant impact component identifier, dominant impact transfer sequence identifier, key propagation path identifier, propagation blocking point identifier, and dominant distortion source identifier are extracted. When the response energy segment is divided into a backoff segment, based on the anomaly type combination encoding, dominant distortion source identifier, and impact component of the corresponding tensor unit, a representative impact component template is selected from the scene-object-time period settlement impact tensor. Combined with the dominant impact transfer sequence of the previous valid period and the status label of the previous valid period, the dominant component inheritance recalculation is performed to generate a response settlement result with a low confidence mark.

[0007] In a preferred embodiment, the anomaly type combination encoding is used at least to characterize the existence status, dominant anomaly type, anomaly intensity level, and anomaly duration level of missing anomalies, misaligned anomalies, conflicting anomalies, and time-sensitive anomalies within the corresponding response energy segment, and serves as a common input for generating segment confidence scores, calling the anomaly effect constraint matrix, and filtering representative impact component templates.

[0008] In a preferred embodiment, the segment confidence evolution sequence is obtained by performing continuity constraint calculation on the segment confidence scores of multiple response energy segments arranged in chronological order. This is used to limit the change magnitude of budget components, the frequency of segment category switching, and the backoff switching threshold between adjacent response energy segments, so as to avoid oscillations in the dynamic pricing response strategy between adjacent segments.

[0009] In a preferred embodiment, in the anomaly constraint matrix, missing anomalies correspond to state feature correction channels, misaligned anomalies correspond to segment boundary correction channels, conflicting anomalies correspond to settlement item constraint channels, and time-sensitive anomalies correspond to budget component contraction channels. Furthermore, when two or more anomaly types exist simultaneously within the same response energy segment, the corresponding composite constraint path is invoked based on the anomaly type combination encoding to perform joint correction on the state deviation index, trigger window, settlement item confirmation status, and dynamic pricing response budget vector.

[0010] In a preferred embodiment, the scenario-object-time period settlement impact tensor is used not only to characterize the impact components of each settlement item under different scenarios, different objects and different time periods, but also to record the segment category, anomaly type combination code and dominant distortion source identifier corresponding to each response energy segment, so that the same tensor structure can simultaneously undertake the functions of segment mapping, settlement impact assessment, distortion source location and rollback template screening.

[0011] In a preferred embodiment, the dominant distortion source identifier is obtained by jointly sorting the anomaly type combination encoding, the fragment credibility evolution sequence, the intensity of the impact component in the corresponding tensor unit, and the propagation influence intensity of the key propagation path, and is used to determine the priority of budget component contraction, the priority of settlement item freezing, and the priority of recalculating back-off fragments.

[0012] In a preferred embodiment, the dominant component inheritance recalculation of the rollback segment does not directly use the historical settlement results. Instead, based on the anomaly type combination code and dominant distortion source identifier of the rollback segment, the target template with the highest matching degree with the current segment is selected from multiple representative impact component templates. Combined with the dominant impact transfer sequence of the previous effective period and the status label of the previous effective period, the basic electricity charge impact component, deviation assessment impact component, incentive rebate impact component and default correction impact component are recalculated item by item.

[0013] A dynamic pricing response and settlement processing system for electricity retail scenarios, characterized in that it includes: The data access module is used to acquire multi-source business data corresponding to the target retail object; The timeline mapping module is used to map the multi-source business data to a unified settlement benchmark timeline and generate a unified record sequence. The data quality processing module is used to identify missing key fields, misaligned timestamps, cross-source key field conflicts, and continuous interface timeouts, and to generate data credibility scores, source consistency flag sets, and anomaly type combination codes. The dual-time-window state recognition module is used to calculate the state deviation index, response sensitivity coefficient, and settlement disturbance coefficient, and to determine the state label, data credibility level, and current valid parameter spectrum. The state-driven fragment generation module is used to generate multiple response energy fragments based on the state deviation exponent cross-threshold change points, time period type switching points, and execution stability coefficient change points. The fragment credibility assessment module is used to calculate the fragment credibility score of each response energy fragment and generate a fragment credibility evolution sequence; The anomaly action constraint module is used to construct the anomaly action constraint matrix and apply differentiated constraints to each response energy segment based on the anomaly type combination encoding and the segment credibility evolution sequence. The settlement shock tensor construction module is used to construct the scenario-object-time period settlement shock tensor and complete the tensor mapping of anomaly type combination encoding, segment category and dominant distortion source identification; The dynamic pricing response budget generation module is used to generate dynamic pricing response budget vectors for each response energy segment and divide them into executable segments, restricted segments, and rollback segments based on the current effective parameter spectrum, segment confidence score, segment confidence evolution sequence, abnormal action constraint results, and the impact components of the corresponding tensor units. The segment-level settlement processing module is used to calculate the effective response power and settlement correction coefficient of each response energy segment, and extract the dominant impact component identifier, dominant impact transfer sequence identifier, critical propagation path identifier, propagation blocking point identifier, and dominant distortion source identifier. The rollback settlement module is used to, when a rollback segment occurs, select representative impact component templates from the scene-object-time period settlement impact tensor based on the anomaly type combination encoding and the dominant distortion source identifier, and perform dominant component inheritance recalculation to output a response settlement result with a low confidence marker.

[0014] The technical effects and advantages of this invention are as follows: This invention classifies and identifies missing, misaligned, conflicting, and time-sensitive anomalies in multi-source business data and forms anomaly type combination codes. This allows different anomaly types to continue to play a role in subsequent status identification, budget constraints, and settlement processing, rather than just remaining at the pre-cleaning stage. Therefore, it improves the reliability of dynamic pricing response and settlement processing.

[0015] This invention dynamically segments the continuous energy consumption process to generate multiple response energy consumption segments, and further calculates the segment reliability score and segment reliability evolution sequence of each response energy consumption segment. This enables differentiated budget control, differentiated settlement processing, and differentiated rollback switching to be performed separately in different local time periods, thereby improving the response control accuracy and overall settlement robustness of local time periods.

[0016] This invention constructs an anomaly action constraint matrix and establishes a unified mapping relationship between the anomaly type combination encoding and the state feature correction channel, fragment boundary correction channel, budget component contraction channel, and settlement item constraint channel. This enables anomaly identification, state correction, strategy constraint, and settlement processing to form a unified processing chain, thereby improving the consistency and engineering level of the system's processing logic.

[0017] This invention constructs a scenario-object-time period settlement impact tensor and maps the segment category, anomaly type combination encoding, and dominant distortion source identifier to the corresponding tensor unit, thereby achieving unified structured support for dynamic pricing response budget generation, segment-level settlement evaluation, and rollback template screening. This improves the linkage between dynamic pricing response and settlement processing and reduces the probability of settlement risk getting out of control.

[0018] This invention performs a dominant component inheritance recalculation by combining anomaly type combination encoding, dominant distortion source identifier, representative impact component template, dominant impact transfer sequence of the previous effective period, and status label when a rollback segment occurs. This avoids the shortcomings of the prior art of stopping calculation or simply using historical results, thus improving the continuous output capability, result interpretability, and business processing efficiency under abnormal conditions. Attached Figure Description

[0019] Figure 1 This is a system framework module diagram of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Reference Figure 1 A dynamic pricing response and settlement processing method for electricity retail scenarios includes the following steps: Obtaining multi-source business data corresponding to the target retail object refers to extracting data records corresponding to the target retail object from multiple business systems, data collection systems, transaction systems, and interface operation monitoring systems, based on the target retail object's dynamic pricing response and settlement processing needs in the current business scenario. The extracted data records are then used as the basic input for subsequent unified record sequence generation, anomaly identification, state identification, fragment credibility assessment, anomaly effect constraint, settlement impact tensor construction, dynamic pricing response budget vector generation, and fragment-level settlement processing.

[0022] The target retail object can be a single electricity retailer, a user group under the same contract entity, an aggregated load unit managed uniformly by the electricity sales entity, or a park load object participating in demand response business. To ensure consistency in subsequent processing, when acquiring multi-source business data, it is preferable to first determine the association between the target retail object and its corresponding metering point, contract number, settlement unit identifier, market transaction unit identifier, and response execution unit identifier. For cases where one target retail object corresponds to multiple metering points or multiple business identifiers, they can be merged according to a pre-established object mapping table, ensuring that all data entering the unified record sequence are associated with the same object identifier. This avoids subsequent status identification deviations, settlement item mismatches, or misjudgments of abnormal types due to inconsistent object identifiers.

[0023] Multi-source business data includes at least metering load data, contract parameter data, electricity pricing rule data, historical response execution logs, historical settlement details, market clearing results, and interface status data. These data types are not isolated but collectively support the entire processing chain in this invention, from data quality identification to pricing response, settlement processing, and anomaly rollback recalculation. Therefore, when acquiring various types of data, it is preferable to acquire not only the business value itself but also its corresponding time information, source information, and validity information simultaneously to facilitate subsequent identification of missing anomalies, misaligned anomalies, conflicting anomalies, and time-sensitive anomalies.

[0024] Metered load data is used to characterize the actual electricity consumption behavior of target retail entities at different time granularities. It serves as the foundation for subsequent load fluctuation feature extraction, state deviation index calculation, response energy segment generation, and effective response electricity calculation. Metered load data can originate from smart meters, electricity information collection systems, load monitoring terminals, park energy management platforms, or distribution-side data acquisition devices. The acquired metered load data preferably includes at least a timestamp, object identifier, active power value, electricity value, demand value, sampling interval, data quality marker, and source identifier. In practical applications, if metered load data from different sources differ in sampling period, time precision, or unit format, the original records can be retained, along with their source identifier and original time information. This allows for aggregation, splitting, interpolation, inheritance, or unit conversion processing during the subsequent unified settlement benchmark time axis mapping stage. This approach ensures that the metered load data meets both the continuity requirements for state identification and provides original evidence for anomaly identification.

[0025] Contract parameter data is used to characterize the agreed-upon and settlement boundaries of the target retailer in the electricity retail business. It is a crucial basis for generating the current effective parameter spectrum, determining the transferable load capacity range, forming budget constraints, and executing segment-level settlement processing. Contract parameter data can be sourced from the electricity sales contract management system, customer file system, business rule configuration system, or settlement rule maintenance system. The acquired contract parameter data preferably includes at least the contract number, contract start and end times, agreed-upon capacity, response participation type, applicable price adjustment conditions, allowed response period, advance notice constraint, deviation tolerance rules, incentive rebate rules, default handling rules, and settlement cycle information. When different contract terms change at different effective periods, it is preferable to retain their effective period information so that the corresponding contract parameters can be accurately associated with the corresponding time period type and response energy segment after time axis mapping, thereby preventing subsequent budget vector generation and settlement correction coefficient calculation from deviating from the actual contract boundaries.

[0026] Electricity price rule data is used to characterize the applicable price formation rules and price adjustment boundaries in the current retail business scenario, and is the foundation for generating dynamic pricing response budget vectors and outputting dynamic pricing response strategies. Electricity price rule data can originate from electricity sales-side quotation systems, price strategy engines, time-of-use pricing configuration systems, market rule management systems, or temporary price adjustment strategy configuration systems. The acquired electricity price rule data preferably includes at least the base electricity price, time-of-use segmentation rules, peak / off-peak or high-peak time period identifiers, price fluctuation upper limit, price fluctuation lower limit, event-triggered price adjustment conditions, temporary price adjustment rules, price effective time, and applicable scope. For situations where multiple sets of electricity price rules apply to the same target retail object, it is preferable to retain the priority, mutual exclusion, or superposition relationships of each electricity price rule to accurately determine the upper limit of price adjustment, execution duration, and notification lead time when generating the dynamic pricing response budget vector.

[0027] Historical response execution logs are used to record the actual execution status of target retail objects in past dynamic pricing or demand response processes. They are a crucial foundation for calculating response sensitivity coefficients, execution stability characteristics, response time lag intervals, and subsequent dominant impact transfer sequences. Historical response execution logs can originate from demand response management platforms, load control systems, message distribution systems, terminal execution feedback systems, or electricity sales operation platforms. The acquired historical response execution logs preferably include at least the response event identifier, object identifier, response instruction issuance time, target time period, corresponding price instruction, execution start time, execution end time, actual response electricity volume, response time lag, execution success flag, execution failure reason, and source identifier. Ideally, the historical response execution logs should cover at least the current long analysis time window and the previous effective period to ensure sufficient continuity of historical data required for recalculating response sensitivity coefficients, execution stability coefficients, and rollback segments.

[0028] Historical settlement details are used to characterize the settlement results of target retail objects within a historical period. They are crucial for constructing the scenario-object-time period settlement impact tensor, extracting the dominant impact component, calculating settlement correction coefficients, and determining deviation responsibility allocation results. Historical settlement details can originate from electricity billing systems, deviation assessment systems, rebate settlement systems, or financial clearing systems. The acquired historical settlement details should ideally include at least the settlement period, object identifier, basic electricity bill amount, deviation assessment amount, incentive rebate amount, default correction amount, total settlement amount, item confirmation status, settlement rule index, and settlement timestamp. Ideally, the original compositional relationships of each settlement item should be preserved, enabling subsequent acquisition not only of the total settlement result but also identification of the impact strength of different settlement items under different response energy consumption segments. This supports the extraction of dominant impact component identifiers, key propagation path identifiers, and dominant distortion source identifiers.

[0029] Market clearing results characterize market prices, cleared electricity volumes, or transaction boundaries relevant to the business scenario of the target retailer. They serve as a crucial data source connecting external market changes with internal retail pricing response strategies. Market clearing results can originate from medium- and long-term trading platforms, spot market clearing platforms, ancillary service trading platforms, or internal market mapping services. Preferably, the acquired market clearing results include at least the clearing time, market period identifier, clearing price, cleared electricity volume, transaction direction, market scenario identifier, and applicable interval information. By combining market clearing results with metering load data, electricity price rule data, and historical settlement details, the dynamic pricing response budget vector and settlement impact tensor in this invention can consider not only the historical behavior of the target retailer but also the impact of external market scenario changes on price responses and settlement results.

[0030] Interface status data characterizes the data transmission and interface operation status of each data source in the current and historical periods. It is an important basis for identifying time-sensitive anomalies, assisting in the identification of missing anomalies, and forming anomaly type combination codes. Interface status data can originate from data access middleware, interface monitoring systems, task scheduling systems, message buses, or log monitoring systems. The acquired interface status data preferably includes at least the data source identifier, request time, response time, number of consecutive timeouts, timeout duration, number of retries, most recent successful update time, data age, disconnection marker, and anomaly log summary. Since this invention does not simply treat interface anomalies as a pre-interception condition but needs to further incorporate them into the generation of dynamic pricing response budget vectors and fragment-level fallback switching logic, it is preferable to retain complete time and continuity information in the interface status data to facilitate subsequent identification of time-sensitive anomalies and determine their specific constraints on notification lead time, price adjustment range upper limit, execution duration, and response upper limit.

[0031] To ensure the consistent use of these data types in subsequent steps, it is preferable to simultaneously acquire or generate timestamps, source identifiers, object identifiers, scenario identifiers, and validity identifiers corresponding to each data record when acquiring multi-source business data. The timestamp is used for subsequent mapping to a unified settlement benchmark timeline; the source identifier is used for subsequent cross-source conflict identification and source consistency flag set generation; the object identifier is used to unify data from different systems to the same target retail object; the scenario identifier is used to distinguish between daily retail scenarios, temporary price intervention scenarios, demand response scenarios, or settlement verification scenarios; and the validity identifier is used to record whether the record is complete, has been corrected, or has an pending confirmation status. By simultaneously retaining the above auxiliary information during the acquisition phase, problems such as unclear data sources, unclear object boundaries, or unclear time correspondences can be avoided during subsequent timeline mapping and anomaly identification.

[0032] After acquiring multi-source business data, it is preferable not to immediately delete data with anomalies, but to retain its original business values, anomaly-related metadata, and source information. This allows subsequent steps to continue identifying missing, misaligned, conflicting, and time-sensitive anomalies after a unified record sequence is generated, and to further form anomaly type combination codes. In other words, the acquisition of multi-source business data in this embodiment not only serves traditional data preparation but also supports subsequent anomaly impact constraint matrix construction, fragment credibility score calculation, fragment credibility evolution sequence generation, scenario-object-time period settlement impact tensor mapping, and dominant component inheritance recalculation of rollback fragments. In this way, anomaly information in multi-source business data is no longer limited to the pre-cleaning level but continues to participate in the differentiated control of subsequent dynamic pricing response and settlement processing, thus maintaining consistency with the overall technical solution of this invention.

[0033] Furthermore, to ensure a continuous and traceable data foundation for subsequent short analysis windows, long analysis windows, the previous effective period, and rollback segment recalculation, it is preferable that the acquired historical response execution logs, historical settlement details, market clearing results, and interface status data at least cover the current analysis period, the long analysis window preceding the current analysis period, and the time range corresponding to the previous effective period. For cases where historical records are missing or only partial source data exists for certain time periods, existing records and their missing markers can be retained and included in the criteria for judging missing or time-sensitive anomalies during the subsequent anomaly identification stage, rather than being directly removed during the acquisition stage. This approach ensures the integrity of the data chain and also helps improve the feasibility of subsequent anomaly judgment, segment-level control, and rollback calculation.

[0034] In summary, "obtaining multi-source business data corresponding to the target retail object" is not simply reading raw values ​​from multiple systems. Rather, it involves acquiring a complete dataset that simultaneously supports state identification, fragment credibility assessment, anomaly differentiation constraints, settlement impact assessment, and anomaly rollback recalculation, under the premise of unified object mapping, preservation of source information, preservation of time information, preservation of validity identifiers, and guarantee of historical continuity. Therefore, subsequent steps can be processed in a coordinated manner based on the same data foundation, ensuring the consistency, reliability, and engineering applicability of the dynamic pricing response and settlement processing of this invention in the electricity retail scenario. Mapping multi-source business data to a unified settlement benchmark time axis means using the time granularity, settlement cycle boundaries, and time period division rules adopted in settlement processing as a unified reference to align data records from different data sources, different sampling cycles, different update times, and different business semantics in terms of time, object, and rules. This allows the originally scattered data to form a unified record sequence under the same time coordinate, which can be directly used for subsequent anomaly identification, status identification, fragment credibility assessment, anomaly constraint, construction of scenario-object-time period settlement impact tensor, and generation of dynamic pricing response budget vector.

[0035] The unified settlement benchmark time axis is not simply a time series from a single original data source, but rather a unified time reference constructed according to the settlement processing requirements of current electricity retail business. Preferably, the unified settlement benchmark time axis is determined by combining at least the settlement cycle, minimum settlement granularity, time-of-use division rules, market clearing period boundaries, and the control granularity required for dynamic pricing response. In other words, the unified settlement benchmark time axis must reflect both the settlement side's requirements for time boundaries and the dynamic pricing response side's requirements for identifying trigger windows, execution duration, and response lag intervals. In practical applications, if the settlement system performs settlement at a 15-minute granularity, market clearing results are released at a 30-minute granularity, and metering load data is collected at a 5-minute granularity, then the settlement granularity or its compatible minimum common time granularity is preferably used as the basic granularity of the unified settlement benchmark time axis. In subsequent mapping processes, high-frequency data is aggregated, and low-frequency data is extended or inherited to ensure that all types of data can form comparable and calculable unified records on the same time axis.

[0036] When constructing a unified settlement benchmark timeline, it is preferable to first determine the start and end times of the settlement cycle to which the current processing object belongs, time zone information, date boundary rules, holiday rules, and time period division rules such as peak, flat, valley, and peak periods. For data spanning days, months, or settlement cycles, it is preferable to segment according to the settlement cycle boundaries, so that the same original data can be decomposed into multiple mapped segments aligned with the unified settlement benchmark timeline when it crosses multiple settlement intervals, thereby avoiding the distortion of item attribution caused by cross-cycle data directly participating in subsequent settlement impact assessments. For daylight saving time adjustments, special holiday periods, or temporary policy price adjustment periods, it is preferable to pre-write the corresponding scenario boundaries and time period boundaries in the unified settlement benchmark timeline, so that the scenario identifiers and time period type identifiers in the subsequent unified record sequence can accurately reflect the external business environment.

[0037] When mapping multi-source business data to a unified settlement benchmark time axis, it is preferable to execute the corresponding time alignment strategy according to the data source type. For metering load data with a sampling frequency higher than the granularity of the unified settlement benchmark time axis, it is preferable to perform summation, averaging, peak extraction, last value inheritance, or weighted aggregation processing according to the target time slot to obtain the load value corresponding to the target time slot one-to-one. Among them, when it is necessary to calculate the actual response electricity in the future, the electricity accumulation method or power integration method can be used for conversion. When more attention is paid to the deviation of the real-time status, the average power value or representative power value can be used as the load value. For electricity price rule data, contract parameter data, and market clearing results with a sampling frequency lower than the granularity of the unified settlement benchmark time axis, it is preferable to use interval expansion, time period inheritance, or effective interval mapping to expand a low-frequency record into a mapping result covering multiple unified time slots within its effective interval. For data with event start and end times in the historical response execution log, it is preferable to map it to multiple corresponding unified time slots according to the effective time of the response instruction, the actual execution start time, and the actual execution end time, and record its execution status, time delay status, or execution intensity in each time slot. For data such as historical settlement details that are summarized periodically, it is preferable to first break it down into the corresponding settlement period according to the settlement rule index and settlement items, and then map it to the corresponding interval of the unified settlement benchmark time axis through the period attribution relationship, which is used for subsequent settlement impact tensor construction and dominant impact component identification.

[0038] In addition to time alignment, it is preferable to perform object alignment and rule alignment simultaneously. Object alignment refers to unifying the association of records from different data sources to the same target retail object. Since the object numbers used in the metering system, contract system, market transaction system, and settlement system may differ, it is preferable to pre-establish an object mapping table, which should at least include the correspondence between metering point identifiers, contract subject identifiers, settlement unit identifiers, market transaction unit identifiers, response execution unit identifiers, and the unified identifier of the target retail object. During the mapping process, when an original record cannot be directly matched to the target retail object, object attribution can be determined based on the mapping table, historical associations, or primary key combinations of the source system, and the attribution result can be written into the object identifier field. Rule alignment refers to unifying business rule information from different sources into a field expression method that can be directly used in subsequent steps. For example, capacity constraints, advance notice constraints, deviation tolerance rules, incentive rebate rules, and default handling rules in contract terms are uniformly expressed as contract constraint values ​​and settlement rule indexes; the base electricity price, floating upper limit, floating lower limit, and effective conditions in electricity price rules are uniformly mapped as price benchmark values ​​and their applicable ranges; and the market price, clearing segment, and scenario conditions in market clearing results are uniformly mapped as scenario identifiers and corresponding price reference conditions.

[0039] Based on the aforementioned time alignment, object alignment, and rule alignment processes, a unified record sequence is generated. The unified record sequence preferably consists of multiple unified records arranged in order along a unified settlement benchmark timeline. Each unified record corresponds to the comprehensive business status of a specific target retail object within a unified time slot. To ensure that subsequent anomaly type identification, fragment credibility assessment, anomaly effect constraint matrix invocation, settlement impact tensor mapping, and fragment-level settlement processing can be performed based on the same data foundation, each record in the unified record sequence includes at least a timestamp, object identifier, scene identifier, time period type identifier, load value, price benchmark value, contract constraint value, settlement rule index, source identifier, and validity identifier.

[0040] The timestamp uniquely identifies the unified record's corresponding unified time slot position, preferably represented by the slot start and end times on the settlement baseline time axis, or by adding granularity length to the slot start time. The object identifier characterizes the target retail object to which the unified record belongs, ensuring that subsequent state identification, fragment generation, and settlement processing all revolve around the same object. The scenario identifier characterizes the business or market environment to which the current unified record belongs, such as daily retail scenarios, temporary price intervention scenarios, demand response scenarios, settlement verification scenarios, or high-volatility market scenarios, enabling accurate positioning of the scenario dimension when constructing the scenario-object-time period settlement impact tensor. The time period type identifier characterizes the peak period, normal period, valley period, peak period, holiday period, or other preset business time period types corresponding to the current unified record, allowing pricing budget generation and settlement impact assessment to distinguish processing differences under different time attributes.

[0041] The load value characterizes the energy consumption status of the target retailer within the current unified time slot. It can be an energy consumption value, average power value, representative power value, demand value, or an equivalent load value calculated from sampled data. The appropriate load representation method should be selected based on the subsequent processing objectives: when more attention is paid to load changes before and after pricing response, the average power value can be used; when more attention is paid to effective response energy consumption or settlement correction coefficients, the energy consumption value or the load value calculated based on integration can be used. The price benchmark value characterizes the applicable base price level or base pricing reference within the current unified time slot. It can be determined comprehensively from the base electricity price, time-of-use electricity price, temporary price adjustment rules, or market clearing price. The contractual constraint value characterizes the contractual boundary conditions that the target retailer must follow within the current unified time slot. This can include agreed capacity, allowable response boundaries, advance notice requirements, response limits, deviation tolerance thresholds, etc. Multiple contractual constraint parameters can also be compressed or indexed for direct use when generating the dynamic pricing response budget vector. The settlement rule index is used to characterize the set of settlement rules applicable within the current unified time slot. Through this index, it can be linked to detailed configurations such as basic electricity fee rules, deviation assessment rules, incentive rebate rules, and default correction rules, so that subsequent segment-level settlement processing and settlement impact tensor construction can call on consistent rule bases.

[0042] The source identifier records the composition of the original data source corresponding to the current unified record. Since a unified record is usually mapped from data from multiple sources, the source identifier can be a single source identifier, a source combination identifier, a source priority identifier, or a source tracking index. Preferably, the source information of each key field in the unified record is retained through the source identifier to facilitate subsequent cross-source key field conflict identification, source consistency flag set generation, and dominant distortion source identifier extraction. The validity identifier is used to characterize whether the current unified record is complete, whether there is a pending confirmation status, whether it has undergone interpolation, whether it has undergone inheritance extension, whether there are missing fields, whether there is a time misalignment, or whether it needs to enter the subsequent anomaly identification process. The validity identifier does not simply indicate whether the record is usable or unusable, but preferably can at least distinguish the states of original valid records, corrected valid records, partially missing records, pending confirmation records, and low-confidence warning records for subsequent differentiated processing.

[0043] When generating a unified record sequence, it is preferable to preserve the correspondence before and after mapping. That is, for each unified record in the unified record sequence, it is preferable to retain its corresponding original data record set, mapping rule information, and correction information. This allows for tracing back to the source field, source system, and mapping path if missing, misaligned, conflicting, or time-sensitive anomalies are subsequently discovered. In this way, anomaly identification no longer stops at the result level but can trace back to the original data level, providing support for subsequent anomaly type combination encoding generation, anomaly effect constraint matrix invocation, and extraction of dominant distortion source identifiers.

[0044] To avoid insufficient disclosure, this embodiment also preferably provides clear processing principles for different mapping scenarios. When multiple candidate original records exist within the same unified time slot, the target value can be determined based on source priority, time proximity, data quality markers, or business confirmation status. Information on unselected candidate records is retained using source identifiers or validity identifiers for subsequent conflict identification. When a certain type of key field is completely missing within a unified time slot, the unified record is not directly deleted. Instead, the current time slot and object identifier are retained, and the corresponding field is set to null or marked as missing. Simultaneously, a pending identification status is written to the validity identifier, enabling it to enter the missing anomaly judgment stage in subsequent anomaly identification. When an original record spans multiple unified time slots, it is preferable to split the mapping by time slot according to its effective interval, ensuring that each split unified record inherits the same object identifier, rule index, or source identifier. When the timestamp of an original record falls near the boundary of two adjacent unified time slots, it is preferable to determine its assigned time slot based on a preset tolerance range, and retain the offset between the original timestamp and the mapped time slot for subsequent identification of misalignment anomalies and correction of trigger windows and response delay intervals.

[0045] After the unified record sequence is generated, it is preferably arranged in chronological order and grouped according to object identifier, scene identifier, and time period type identifier, thus providing direct input for the subsequent construction of short and long analysis time windows. Furthermore, the load value, price benchmark value, contract constraint value, and settlement rule index in the unified record sequence are not merely used for static recording, but collectively form the basis for calculating the subsequent state deviation index, response sensitivity coefficient, settlement disturbance coefficient, and current effective parameter spectrum; while the source identifier and validity identifier continue to serve as important inputs for identifying anomaly types, generating anomaly type combination codes, calculating segment confidence scores, forming segment confidence evolution sequences, and invoking the anomaly effect constraint matrix. Therefore, the unified record sequence is not merely a simple data summary result, but a core intermediate data structure that runs through all subsequent processing steps of this invention.

[0046] By mapping multi-source business data to a unified settlement benchmark timeline and generating a unified record sequence, the previously scattered, inconsistent time-granularity, non-uniform object identification, and inconsistent rule expressions of data can be transformed into a unified data foundation that can be processed within the same time frame, object frame, and rule frame. On the one hand, this method provides complete and traceable input conditions for subsequent identification of missing, misaligned, conflicting, and time-sensitive anomalies. On the other hand, it also enables subsequent dynamic pricing response budget vector generation, segment-level settlement processing, and dominant component inheritance-based recalculation of rollback segments to be built on a unified data semantics, thereby improving the consistency, reliability, and engineering applicability of this invention in the electricity retail scenario.

[0047] Performing data quality processing on the unified record sequence goes beyond simply filtering out anomalous data. It involves a comprehensive process: subsequent state identification, segment credibility assessment, invocation of the anomaly constraint matrix, scenario-object-time period settlement impact tensor mapping, and inheritance-based recalculation of the dominant components of rollback segments. This process jointly identifies, classifies, and structures the integrity, temporal consistency, cross-source consistency, and interface timeliness of key fields within the unified record sequence. Through this processing, anomalous information from multi-source business data no longer remains at the initial cleaning layer but continues to participate in the differentiated constraint processes of subsequent dynamic pricing responses and settlement handling.

[0048] Specifically, key fields include at least the load value, price benchmark value, contract constraint value, settlement rule index, timestamp, object identifier, and source-related fields. When performing data quality processing on a unified record sequence, key field missingness is first identified based on preset field integrity rules. If a unified record contains a key field that is empty, has an unresolvable field value, has a field value exceeding the business allowable range, or has a field that exists but cannot be effectively associated with the current object identifier or timestamp, then the corresponding issue for that unified record is identified as a key field missing and marked as a missing anomaly. Missing anomalies primarily reflect the degree of incompleteness of information in the current record during status feature extraction, budget component generation, or settlement item confirmation; their subsequent application prioritizes status feature correction and parameter conservatism processing.

[0049] Timestamp misalignment refers to an offset exceeding a preset tolerance range between the original time information of data from various sources in the unified record and the mapped unified settlement benchmark timeline, or inconsistencies in the time correspondence between different sources for the same object or the same business event. During identification, the differences between the original sampling time, rule effective time, response execution start and end time, interface reception time, and the unified time slot boundary are preferentially compared. When the difference exceeds the preset tolerance range, or when the deviation affects trigger window determination, response delay determination, or segment boundary attribution, it is identified as timestamp misalignment and marked as a misalignment anomaly. Misalignment anomalies primarily reflect uncertainties in time alignment, and their subsequent effects primarily affect trigger window contraction, response delay interval correction, and segment boundary correction.

[0050] Cross-source critical field conflicts refer to inconsistencies between field values ​​from different data sources that should correspond to the same business status in terms of object identifier and timestamp, and where such inconsistency exceeds a preset conflict threshold. Conflict identification can be performed separately for load values, price benchmark values, contract constraint values, settlement rule indexes, or execution status fields. When field values ​​from two or more sources within the same time slot show numerical differences, status differences, or rule pointing differences, and these differences cannot be directly resolved through source priority or confirmation status, they are identified as cross-source critical field conflicts and marked as conflict-type anomalies. Conflict-type anomalies primarily reflect uncertainty in settlement attribution or business fact judgment, and their subsequent effects primarily include de-weighting, freezing, delayed confirmation, and locating the main source of distortion in settlement items.

[0051] Interface timeout refers to a situation where an interface corresponding to a data source experiences consecutive response timeouts, persistent disconnections, retry failures, or data ages exceeding the maximum valid duration within a preset observation window. Identification can be based on request time, response time, number of consecutive timeouts, most recent successful update time, and data age. When the consecutive timeout length or data age reaches a preset threshold, it is identified as an interface timeout and marked as a time-sensitive anomaly. Time-sensitive anomalies primarily reflect a decline in data timeliness, and their subsequent impact is primarily on adjusting notification lead times, reducing price adjustment magnitudes, shortening execution duration, and handling rollback / switchover decisions.

[0052] After completing the above identification, a data credibility score is further generated. The data credibility score quantifies the usability of the data foundation corresponding to the current unified record or fragment, and is preferably calculated by weighting the degree of missing key fields, the magnitude of timestamp misalignment, the intensity of cross-source conflicts, and the degree of interface timeliness decay. The degree of missing fields can be characterized by the number of missing fields or the proportion of missing key fields; the magnitude of misalignment can be characterized by the time offset; the intensity of conflicts can be characterized by the number of conflicting fields, the difference in conflict values, or the duration of conflict; and the degree of timeliness decay can be characterized by the data age or the length of consecutive timeouts. The lower the data credibility score, the more conservative constraints need to be applied when the corresponding record subsequently enters status identification, budget generation, and settlement processing.

[0053] Simultaneously, a source consistency flag set is generated. This set is used to record, field by field, whether data from different sources is consistent, confirmed, has a pending verification status, and whether a priority source coverage relationship exists. Preferably, the source consistency flag set sets consistency flags for at least the load value, price benchmark value, contract constraint value, and settlement rule index, enabling subsequent anomaly constraint matrices to call different constraint channels according to field categories, rather than uniformly shrinking the entire record.

[0054] Building upon this foundation, anomaly type combination codes are further generated. These codes do not simply indicate the existence of a particular anomaly, but rather serve to structurally express the presence, dominant anomaly type, intensity level, and duration level of missing, misaligned, conflicting, and time-sensitive anomalies within a corresponding response energy segment. Specifically, the presence status of the four anomalies can be represented as whether they occur; the dominant anomaly type as the anomaly category with the greatest impact on the current segment; the anomaly intensity level as mild, moderate, or severe; and the anomaly duration level as single time slot, short duration, or long duration. For a response energy segment containing multiple unified records, the anomaly tags of each unified record within the segment can be aggregated first, and then an anomaly type combination code corresponding to that response energy segment can be formed.

[0055] Anomaly type combination encoding serves as a common input for multiple subsequent processing stages. Firstly, it generates segment credibility scores, combining the global data credibility score with the anomaly type composition, dominant anomaly type, and duration within the segment to create differentiated credibility evaluations for different response energy segments. Secondly, it invokes the anomaly effect constraint matrix, mapping missing, misaligned, conflicting, and time-sensitive anomalies to the state feature correction channel, segment boundary correction channel, settlement component constraint channel, and budget component contraction channel, respectively, triggering composite constraint paths when multiple anomalies occur concurrently. Thirdly, it filters representative impact component templates. When a rollback segment appears, instead of simply using historical results, it selects a target template with a higher matching degree from multiple representative impact component templates associated with the scenario-object-time period settlement impact tensor, based on the anomaly type combination characteristics of the current segment, to support the inherited recalculation of the dominant component.

[0056] Through the above processing method, abnormal information in the unified recording sequence can be clearly identified, classified and expressed, and continuously transmitted to the subsequent pricing response and settlement processing links. This solves the problem in the prior art that abnormalities are only treated as a pre-cleaning problem and cannot distinguish the differences in the impact of different abnormalities on subsequent business links. In this way, the reliability of dynamic pricing response, settlement processing and continuous output under abnormal conditions of the present invention in the power retail scenario are improved.

[0057] Based on a unified record sequence, short and long analysis time windows are constructed. The short analysis time window characterizes the immediate changes of the target retail object within the current adjacent time slot, while the long analysis time window characterizes its trend and stability changes over a longer continuous period. Based on these short and long analysis time windows, load fluctuation characteristics, price response characteristics, execution stability characteristics, and settlement disturbance characteristics are extracted. Load fluctuation characteristics reflect the amplitude, direction, and continuity of load value fluctuations within the window; price response characteristics reflect the correspondence between changes in the price benchmark value and load changes; execution stability characteristics reflect the response lag, execution persistence, and execution consistency in historical response execution logs; and settlement disturbance characteristics reflect the fluctuation intensity of sub-items such as basic electricity fees, deviation assessments, incentive rebates, and default corrections in historical settlement details. Furthermore, based on the above characteristics, a state deviation index, a response sensitivity coefficient, and a settlement disturbance coefficient are calculated. The state deviation index characterizes the degree of deviation of the current state from the historical stable state; the response sensitivity coefficient characterizes the target retail object's responsiveness to price changes; and the settlement disturbance coefficient characterizes the disturbance risk of the current state to subsequent settlement results. Then, by combining the data credibility score and the source consistency flag set, the state label, data credibility level and current effective parameter spectrum corresponding to the current unified record sequence are determined, so as to be called for subsequent response energy segment generation, abnormal action constraint and dynamic pricing response budget vector generation. Using the threshold change points of the state deviation index, the time period type switching points, and the execution stability coefficient change points as dividing boundaries, the continuous energy consumption process of the target retail object is dynamically segmented. Specifically, the threshold change points of the state deviation index characterize the moment when the target retail object transitions from a stable state to a deviation state or from a deviation state to a recovery state; the time period type switching points characterize the boundaries between peak, flat, valley, peak, or preset business periods; and the execution stability coefficient change points characterize the moments when the consistency, continuity, or time lag characteristics of historical response execution change significantly. Based on these boundaries, the continuous energy consumption process can be divided into multiple response energy consumption segments that are relatively consistent in terms of state characteristics, time period attributes, and execution capabilities. For each response energy consumption segment, the transferable load capacity range, trigger window, response time lag range, and execution stability coefficient are further determined by combining the load value, price benchmark value, contract constraint value, historical response execution log, and data reliability level within the corresponding time period. Among them, the transferable load capacity range is used to characterize the load range within the segment that can participate in price response without violating contractual constraints and settlement risk constraints; the trigger window is used to characterize the effective time interval for issuing dynamic pricing response strategies; the response time delay range is used to characterize the time delay range between the issuance of price signals and actual load changes; and the execution stability coefficient is used to characterize the consistency and repeatability of response execution within the segment, so as to provide subsequent segment credibility assessment, anomaly constraint, and dynamic pricing response budget vector generation. The segment credibility score characterizes the reliability of a single response energy segment for dynamic pricing response and settlement processing under current data conditions. The calculation uses the anomaly type distribution, anomaly duration, anomaly intensity, source consistency flag set status, and data credibility score corresponding to each unified record within the response energy segment as input. Specifically, the anomaly type distribution characterizes the composition of missing, misaligned, conflicting, and time-sensitive anomalies within the response energy segment; the anomaly duration characterizes the number of time slots or duration of a continuous existence of a particular anomaly within the response energy segment; the anomaly intensity characterizes the impact of anomalies on current segment status identification, segment boundary determination, budget contraction, and settlement confirmation; the source consistency flag set status characterizes the consistency of key fields such as load value, price benchmark value, contract constraint value, and settlement rule index across different data sources; and the data credibility score provides a reliable reference for the overall data foundation of the segment. Preferably, the above factors can be weighted separately and then summed to obtain the segment reliability score of the corresponding response energy segment; when the proportion of missing anomalies, conflicting anomalies or time-sensitive anomalies increases, or the duration of anomalies increases or the source consistency decreases, the segment reliability score will decrease accordingly.

[0058] After obtaining the segment reliability score for each response energy segment, the multiple response energy segments are arranged in chronological order, and a segment reliability evolution sequence is further generated. The segment reliability evolution sequence is not simply a record of the temporal arrangement of the segment reliability scores, but is obtained by calculating the continuity constraints on the segment reliability scores of adjacent response energy segments. The continuity constraint calculation considers at least the temporal adjacency relationship between adjacent segments, changes in state labels, changes in execution stability coefficients, and continuous changes in anomaly type combinations to avoid unrealistic large jumps in segment reliability scores between adjacent segments due to local anomaly fluctuations. Preferably, when the state characteristics and time-period attributes of two adjacent response energy segments are continuous, but the segment reliability score shows a sudden increase or decrease, it can be constrained and corrected according to a preset smoothing rule, thereby obtaining a segment reliability evolution sequence that better reflects actual business continuity.

[0059] The segment credibility evolution sequence is used to limit the magnitude of budget component changes, segment category switching frequency, and rollback switching threshold between adjacent response energy segments. Specifically, when the segment credibility evolution results of adjacent segments change little and remain in a high range, the corresponding budget component is allowed to continue to adjust continuously or slowly; when the segment credibility evolution results continue to decline, the upper limit of price adjustment magnitude, execution duration, notification lead time, or response upper limit is gradually reduced; when the segment credibility evolution results are below the preset threshold and remain below it for a certain period, the corresponding response energy segment is then switched to a rollback segment. In this way, not only can differentiated control be performed based on the differences in data quality of each segment, but also the dynamic pricing response strategy can avoid frequent switching between normal execution, restricted execution, and rollback execution between adjacent segments, reducing strategy oscillation and improving the stability and engineering applicability of segment-level budget control, settlement processing, and abnormal rollback.

[0060] Constructing an anomaly impact constraint matrix involves establishing a predefined mapping relationship between missing, misaligned, conflicting, and time-sensitive anomalies identified during the data quality processing phase and different constraint objects in subsequent dynamic pricing responses and settlement processes. This ensures that anomaly information no longer remains solely at the pre-processing cleaning layer but continues to participate in state identification and correction, fragment boundary correction, budget control, and settlement confirmation. The anomaly impact constraint matrix is ​​preferably structured with anomaly types as rows and constraint channels as columns. Each matrix cell characterizes the manner, intensity, triggering conditions, and duration of the corresponding anomaly type's effect on the corresponding constraint channel. The State Feature Correction Channel is used to perform conservative correction on the State Deviation Index, Response Sensitivity Coefficient, Settlement Disturbance Coefficient, and Current Effective Parameter Spectrum; the Segment Boundary Correction Channel is used to perform position correction or contraction correction on segment boundaries, trigger windows, and response time lag intervals near the State Deviation Index cross-threshold change point, time period type switching point; the Budget Component Contraction Channel is used to contract at least one of the following: upper limit of price adjustment range, execution duration, notification advance, response upper limit, and settlement risk upper limit; the Settlement Item Constraint Channel is used to perform weight reduction, freezing, delayed confirmation, or low-confidence marking on the settlement item confirmation status corresponding to basic electricity charges, deviation assessment, incentive rebates, and default corrections.

[0061] Specifically, the missing anomaly corresponds to the state feature correction channel, which prioritizes reducing the confidence level of the state deviation index, response sensitivity coefficient, or settlement disturbance coefficient when the missing key field affects the feature integrity of the current segment, and generates a conservative current effective parameter spectrum accordingly; the misalignment anomaly corresponds to the segment boundary correction channel, which performs offset correction or contraction correction on the corresponding boundary when the timestamp misalignment affects the segment boundary, trigger window, or response delay determination; the conflict anomaly corresponds to the settlement item constraint channel, which performs downweight confirmation, freeze confirmation, or postponement confirmation on the corresponding settlement item when cross-source key field conflicts affect the confirmation of business facts; and the timeliness anomaly corresponds to the budget component contraction channel, which performs conservative contraction on at least one of the following when the interface times out continuously or the data age exceeds the limit: the upper limit of price adjustment range, the execution duration, the notification advance amount, and the response upper limit.

[0062] When two or more anomaly types exist simultaneously within the same response energy segment, they are not handled in isolation. Instead, the composite constraint path in the anomaly action constraint matrix is ​​invoked based on the anomaly type combination encoding, and the joint correction strength is determined by combining the segment credibility evolution sequence. The joint correction applies at least to the state deviation index, trigger window, settlement item confirmation status, and dynamic pricing response budget vector, enabling the coordinated transmission of the impact of multiple anomalies on the same segment under a unified constraint structure. This approach avoids the loose processing chain problem caused by handling different anomalies separately according to decentralized rules in existing technologies, thereby improving the consistency and engineering applicability of the dynamic pricing response strategy, segment-level settlement processing, and rollback switching.

[0063] Constructing a scenario-object-time period settlement impact tensor refers to a structured expression of the potential settlement impact of dynamic pricing responses in different business scenarios, target retail objects, and time periods, based on response energy consumption segments, contract parameter data, electricity price rule data, historical settlement details, and market clearing results. This ensures that settlement risks are no longer represented by scattered rules or single-amount estimates, but are organized into a unified intermediate data structure that can be used for subsequent budget generation, segment-level settlement processing, locating dominant distortion sources, and selecting rollback templates. Ideally, the scenario-object-time period settlement impact tensor should have at least three dimensions: scenario, object, and time period. The scenario dimension distinguishes between daily retail scenarios, demand response scenarios, temporary price intervention scenarios, high market volatility scenarios, or settlement verification scenarios. The object dimension distinguishes between different target retail objects or object categories. The time period dimension distinguishes between peak, flat, valley, peak, holiday, or other preset time period types. Through this three-dimensional index, different response energy consumption segments and their corresponding settlement impacts can be uniformly categorized into corresponding tensor units.

[0064] Each tensor unit stores at least the following impact components: basic electricity charge impact component, deviation assessment impact component, incentive rebate impact component, and default correction impact component. The basic electricity charge impact component characterizes the strength of the impact of price adjustments and load changes on the basic electricity charge amount under the current segment conditions; the deviation assessment impact component characterizes the strength of the impact on the deviation assessment result after the actual executed load deviates from the declared or agreed boundary; the incentive rebate impact component characterizes the degree of change in incentive rebates obtainable after meeting response conditions; and the default correction impact component characterizes the degree of impact on the default correction result when contractual constraints, execution boundaries, or confirmation conditions are not met. Each of these impact components can be represented by a change in amount, a normalized intensity value, a relative change ratio, or a graded score, as long as it reflects the degree to which the corresponding settlement item is affected by the dynamic pricing response under the current scenario, object, and time period. Preferably, the basic electricity charge impact component, deviation assessment impact component, incentive rebate impact component, and default correction impact component are all calculated jointly based on the itemized results in the historical settlement details, the rule boundaries in the contract parameter data, the price constraints in the electricity price rule data, and the market conditions in the market clearing results, thereby ensuring that the impact components in the tensor unit are consistent with the actual electricity retail business rules.

[0065] When mapping response energy consumption segments to corresponding tensor units, it is preferable to first determine the tensor index position based on the scene identifier, object identifier, and time period type identifier to which the response energy consumption segment belongs, and then map and write the contract constraint value, price benchmark value, historical execution characteristics, and segment-level settlement result or estimation result corresponding to the response energy consumption segment into the corresponding tensor unit. Unlike the traditional method of only recording settlement amounts, the tensor unit in this embodiment, in addition to storing each settlement impact component, also records the segment category, anomaly type combination code, and dominant distortion source identifier corresponding to each response energy consumption segment. The segment category is used to characterize whether the response energy consumption segment is currently an executable segment, a restricted segment, or a rollback segment; the anomaly type combination code is used to characterize the existence status, dominant anomaly type, anomaly intensity level, and anomaly duration level of missing anomalies, misaligned anomalies, conflicting anomalies, and time-sensitive anomalies within the segment; the dominant distortion source identifier is used to characterize the most important distortion source or risk source in the current segment. By recording the above information simultaneously in the same tensor structure, the tensor can no longer just passively store settlement impact values, but simultaneously undertake the functions of segment mapping, settlement impact assessment, distortion source location, and rollback template filtering.

[0066] Specifically, in the segment mapping function, the scenario-object-time period settlement impact tensor is used to aggregate different response energy consumption segments into corresponding tensor units according to a unified indexing method, providing a unified basis for subsequent comparisons of settlement impacts on similar scenarios, objects, and time periods. In the settlement impact assessment function, each impact component in the tensor is used to characterize the direction and intensity of the impact of settlement items such as basic electricity fees, deviation assessments, incentive rebates, and default corrections under the current segment conditions, thus providing constraints for the upper limit of price adjustment range, execution duration, notification lead time, response limit, and settlement risk limit in the dynamic pricing response budget vector. In the distortion source localization function, the segment category, anomaly type combination encoding, impact component intensity, and their time-series changes in the tensor are used to identify the main sources causing instability or unreliability in the current settlement results, thereby supporting the extraction of the dominant distortion source identifier. Regarding the rollback template filtering function, when a rollback segment appears, it can filter representative impact component templates that are similar to the one recorded in the tensor unit of the segment based on the scene, object, time period characteristics and abnormal type combination characteristics recorded in the tensor unit to which the segment belongs, and use them for subsequent dominant component inheritance recalculation.

[0067] The dominant distortion source identifier is not subjectively designated, but rather obtained by jointly ranking the anomaly type combination encoding, the segment credibility evolution sequence, the intensity of the impact component in the corresponding tensor unit, and the propagation influence intensity of the key propagation path. Specifically, the anomaly type combination encoding reflects the composition of the anomaly within the current segment and the dominant anomaly category; the segment credibility evolution sequence reflects the continuous trend of credibility change in the current segment relative to adjacent segments; the impact component intensity reflects which of the sub-items (basic electricity fee, deviation assessment, incentive rebate, and default correction) has a greater impact on the current settlement result; and the propagation influence intensity of the key propagation path reflects how the anomaly or impact gradually amplifies along the "state characteristics—budget component—settlement sub-item" link. Preferably, ranking values ​​or weight values ​​can be calculated for the above factors separately, and then the dominant distortion source identifier can be determined through weighted ranking, priority comparison, or comprehensive scoring. If the conflict-type anomalies reflected by the anomaly type combination coding are dominant, and the deviation assessment impact component and default correction impact component in the corresponding tensor unit are of high strength, while the critical propagation path shows that the conflict is mainly amplified along the settlement confirmation link, then the dominant distortion source of this segment can be identified as the settlement confirmation conflict-dominated distortion source. If the time-sensitive anomalies are dominant, and the price adjustment magnitude-related impact component increases significantly, then the dominant distortion source can be identified as the time-sensitive decay-dominated distortion source. In this way, the dominant distortion source identification can more accurately reflect the risk sources that need to be prioritized for control in the current segment.

[0068] After obtaining the dominant distortion source identifier, it is further used to determine the priority of budget component contraction, settlement item freezing, and rollback segment recalculation. Budget component contraction priority refers to prioritizing the contraction of budget components with the highest correlation to the dominant distortion source within the dynamic pricing response budget vector. For example, if the dominant distortion source is mainly manifested as a decrease in trigger effectiveness due to timeliness anomalies, then the notification lead time and price adjustment limit are prioritized for contraction; if the dominant distortion source is mainly manifested as boundary uncertainty caused by misalignment anomalies, then the execution duration and trigger window are prioritized for contraction. Settlement item freezing priority refers to prioritizing the freezing confirmation, delayed confirmation, or low-confidence marking of settlement items with higher correlation to the dominant distortion source during segment-level settlement processing, to prevent unreliable items from directly entering the final settlement result. Rollback segment recalculation priority refers to prioritizing the segment with a clearer dominant distortion source, more concentrated impact components, or stronger propagation impact when multiple segments simultaneously meet the rollback conditions, performing representative impact component template screening and dominant component inheritance-style recalculation on the segment with a clearer dominant distortion source, more concentrated impact components, or stronger propagation impact, thereby improving processing efficiency and result interpretability in abnormal scenarios.

[0069] Through the above approach, the scenario-object-time period settlement impact tensor not only characterizes the static changes in settlement results, but also further incorporates response energy consumption segments, anomaly type combination codes, segment categories, dominant distortion source identifiers, and the impact intensity of each settlement item into a unified structure. This allows the same tensor structure to continuously support segment mapping, settlement impact assessment, distortion source localization, and rollback template selection. This not only solves the problem of the existing technology's separation between the pricing and settlement stages and the lack of a unified, structured expression of settlement risk, but also enables anomaly information to continuously participate in subsequent budget control, settlement confirmation, and rollback recalculation processes. This improves the consistency, reliability, and continuous output capability of the dynamic pricing response and settlement processing in the electricity retail scenario.

[0070] The dynamic pricing response budget vector is used to characterize the permissible price response boundary of each response energy segment under current data conditions, anomaly constraints, and settlement risks. Specifically, for each response energy segment, a segment-level dynamic pricing response budget vector is generated by integrating its corresponding current effective parameter spectrum, segment credibility score, segment credibility evolution sequence, constraint results of the anomaly effect constraint matrix, and the impact component of the corresponding tensor unit in the scenario-object-time period settlement impact tensor. The current effective parameter spectrum provides the basic capability boundary of the segment in terms of response sensitivity, execution stability, transferable load capacity, and settlement disturbance risk; the segment credibility score characterizes the credibility of the current segment's data basis; the segment credibility evolution sequence constrains the continuity of budget component changes between adjacent segments; the anomaly effect constraint matrix transmits missing anomalies, misaligned anomalies, conflicting anomalies, and time-sensitive anomalies to the state correction, boundary correction, budget contraction, and settlement item constraint processes, respectively; the impact component of the corresponding tensor unit characterizes the strength of risk of the current segment in settlement items such as basic electricity charges, deviation assessment, incentive rebates, and default correction.

[0071] The dynamic pricing response budget vector includes at least the upper limit of price adjustment range, execution duration, notification lead time, response limit, and settlement risk limit. Specifically, the upper limit of price adjustment range defines the maximum allowable price adjustment range for this segment; the execution duration defines the maximum duration for which the pricing strategy remains effective within this segment; the notification lead time defines the minimum lead time between the issuance and actual execution of the pricing strategy; the response limit defines the maximum load boundary for this segment to participate in price response; and the settlement risk limit defines the maximum acceptable disturbance range for this segment in terms of settlement components. Preferably, when the segment's credibility is high, the anomaly constraints are weak, and the impact components in the tensor unit are within a controllable range, a more lenient budget vector is generated, and the segment is classified as an executable segment; when the segment's credibility decreases, the anomaly effect intensifies, or some settlement impact components increase, at least one item in the budget vector is contracted, and the segment is classified as a restricted segment; when the segment's credibility is below a preset threshold, the anomaly type combination is complex, and the settlement impact risk exceeds the settlement risk limit, the segment is classified as a rollback segment.

[0072] For rollback segments, this implementation does not directly use historical settlement results, but instead performs a dominant component inheritance-based recalculation. Specifically, based on the anomaly type combination code and dominant distortion source identifier of the rollback segment, the target template with the highest matching degree with the current segment is selected from multiple representative impact component templates; the matching degree is preferably determined jointly based on scenario consistency, object consistency, time period consistency, anomaly type combination similarity, and dominant distortion source consistency. Subsequently, combining the dominant impact transfer sequence of the previous effective period and the status label of the previous effective period, inheritance-based recalculation is performed on the basic electricity fee impact component, deviation assessment impact component, incentive rebate impact component, and default correction impact component, respectively, to obtain the rollback segment settlement result with a low-confidence tag. In this way, the problem of insufficient continuous output capability caused by stop calculation, simple use of historical values, or purely manual processing in the prior art can be avoided, and the interpretability and engineering applicability of the results under abnormal scenarios can be improved.

[0073] Segment-level settlement processing based on actual execution results and the dynamic pricing response strategy corresponding to each response energy segment means that instead of settling accounts uniformly for the target retailer on a full cycle basis, the actual execution effect and settlement impact under the corresponding dynamic pricing response strategy are verified separately for each response energy segment. Specifically, the target strategy information corresponding to each response energy segment is aligned with the actual execution results, which include at least the actual execution start time, actual execution end time, actual load change, actual response duration, and actual response lag. Based on this alignment result, combined with the contractual constraint value, settlement rule index, price benchmark value, and dynamic pricing response budget vector of the corresponding segment, the effective response electricity and settlement correction coefficient of the segment are calculated. Among them, the effective response electricity is used to characterize the response electricity that can be confirmed as an effective execution result under the constraints of trigger window, response lag interval, response upper limit, and execution duration; the settlement correction coefficient is used to characterize the degree of correction of the actual execution result of the segment to settlement items such as basic electricity fee, deviation assessment, incentive rebate, and default correction, which can be determined by the actual response deviation, execution stability coefficient, segment credibility score, and settlement rules.

[0074] After obtaining the effective response electricity and settlement correction coefficient for each response energy segment, the deviation responsibility allocation result and incentive rebate or default handling result for that segment are calculated respectively. The deviation responsibility allocation result is used to characterize the settlement responsibility share that the target retailer, electricity sales entity, or other relevant participants should bear under that segment due to response deviation, insufficient execution, or time mismatch; the incentive rebate or default handling result is used to characterize the incentive rebate amount that can be obtained or the default correction amount that should be borne when the segment meets or fails to meet the preset response requirements, contractual constraints, and price response boundaries. Preferably, in the segment-level settlement processing, the corresponding sub-item results for basic electricity fees, deviation assessment, incentive rebates, and default corrections are calculated separately, rather than just outputting a single total amount, so as to further perform impact component sorting and distortion source location in the subsequent process.

[0075] Furthermore, based on the intensity ranking of the impact components corresponding to each response energy consumption segment and the intensity of anomaly propagation impact, the dominant impact component identifier, dominant impact transfer sequence identifier, critical propagation path identifier, propagation blocking point identifier, and dominant distortion source identifier are extracted. The intensity ranking of the impact components is used to compare the relative impact magnitudes of the basic electricity charge impact component, deviation assessment impact component, incentive rebate impact component, and default correction impact component in the current segment. The identifier corresponding to the settlement item with the greatest impact intensity can be used as the dominant impact component identifier for that segment. The dominant impact transfer sequence identifier is used to characterize the process of the dominant impact component shifting from one settlement item to another in multiple response energy consumption segments arranged chronologically, such as from basic electricity charge dominance to deviation assessment dominance, or from incentive rebate dominance to default correction dominance. The critical propagation path identifier is used to characterize the main impact links between anomalies, state deviations, budget contraction, and settlement items. The propagation blocking point identifier is used to characterize nodes on this impact link that have a suppressive effect on impact amplification or can serve as locations for subsequent manual intervention, strategy contraction, or rule freezing. The dominant distortion source identifier is determined by combining the anomaly type combination code, the fragment credibility evolution sequence, the intensity of the impact component and its propagation influence, and is used to characterize the most important data distortion source or settlement risk source in the current fragment.

[0076] When a certain energy consumption segment is classified as a rollback segment, this implementation method does not directly remove the segment, nor does it directly use the entire settlement result of the previous period. Instead, it selects a representative impact component template from the scenario-object-time period settlement impact tensor based on the anomaly type combination code, the dominant distortion source identifier, and the impact component of the corresponding tensor unit for the rollback segment. The representative impact component template is preferably a tensor unit template that is historically similar to the current rollback segment in terms of scenario type, object category, time period attribute, anomaly composition, and dominant impact structure. It should at least include the reference distribution relationship of the basic electricity charge impact component, deviation assessment impact component, incentive rebate impact component, and default correction impact component. After template selection, the dominant impact transfer sequence and the status label of the previous effective period are further combined to perform dominant component inheritance recalculation on each of the above impact components to obtain an alternative settlement result for the rollback segment. Inheritance-based recalculation is not simply copying historical values. Instead, it estimates and corrects the basic electricity cost, deviation assessment, incentive rebate, and default correction separately based on the time period position, anomaly composition, dominant distortion source, and proportional relationship of each impact component in the target template of the current rollback segment. Finally, it generates a response settlement result with low-confidence markers.

[0077] By employing the aforementioned segment-level settlement processing and rollback segment recalculation methods, on the one hand, a direct correspondence can be established between the dynamic pricing response strategy and the settlement results at the segment level, improving the consistency between dynamic pricing response and settlement processing; on the other hand, even in scenarios with significant data anomalies or insufficient execution information, it can still output low-reliability response settlement results with reference value, avoiding the problem of insufficient continuous output capability caused by stop calculation, simply using historical results, or relying entirely on manual review in existing technologies. This improves the engineering feasibility, anomaly handling capability, and result interpretability of this invention in the electricity retail scenario.

[0078] A dynamic pricing response and settlement processing system for electricity retail scenarios includes: The data access module is used to acquire multi-source business data corresponding to the target retail object; The timeline mapping module is used to map the multi-source business data to a unified settlement benchmark timeline and generate a unified record sequence. The data quality processing module is used to identify missing key fields, misaligned timestamps, cross-source key field conflicts, and continuous interface timeouts, and to generate data credibility scores, source consistency flag sets, and anomaly type combination codes. The dual-time-window state recognition module is used to calculate the state deviation index, response sensitivity coefficient, and settlement disturbance coefficient, and to determine the state label, data credibility level, and current valid parameter spectrum. The state-driven fragment generation module is used to generate multiple response energy fragments based on the state deviation exponent cross-threshold change points, time period type switching points, and execution stability coefficient change points. The fragment credibility assessment module is used to calculate the fragment credibility score of each response energy fragment and generate a fragment credibility evolution sequence; The anomaly action constraint module is used to construct the anomaly action constraint matrix and apply differentiated constraints to each response energy segment based on the anomaly type combination encoding and the segment credibility evolution sequence. The settlement shock tensor construction module is used to construct the scenario-object-time period settlement shock tensor and complete the tensor mapping of anomaly type combination encoding, segment category and dominant distortion source identification; The dynamic pricing response budget generation module is used to generate dynamic pricing response budget vectors for each response energy segment and divide them into executable segments, restricted segments, and rollback segments based on the current effective parameter spectrum, segment confidence score, segment confidence evolution sequence, abnormal action constraint results, and the impact components of the corresponding tensor units. The segment-level settlement processing module is used to calculate the effective response power and settlement correction coefficient of each response energy segment, and extract the dominant impact component identifier, dominant impact transfer sequence identifier, critical propagation path identifier, propagation blocking point identifier, and dominant distortion source identifier. The rollback settlement module is used to, when a rollback segment occurs, select representative impact component templates from the scene-object-time period settlement impact tensor based on the anomaly type combination encoding and the dominant distortion source identifier, and perform dominant component inheritance recalculation to output a response settlement result with a low confidence tag.

[0079] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic pricing response and settlement processing method in an electricity retail scenario, characterized in that, Includes the following steps: Obtain multi-source business data corresponding to the target retail object. The multi-source business data includes at least metering load data, contract parameter data, electricity price rule data, historical response execution logs, historical settlement details, market clearing results, and interface status data. Multi-source business data is mapped to a unified settlement benchmark timeline to generate a unified record sequence. Each record in the unified record sequence includes at least a timestamp, object identifier, scenario identifier, time period type identifier, load value, price benchmark value, contract constraint value, settlement rule index, source identifier, and validity identifier. Data quality processing is performed on the unified record sequence to identify key field missing, timestamp misalignment, cross-source key field conflict and continuous interface timeout, and the identification results are marked as missing anomalies, misalignment anomalies, conflict anomalies and time-sensitive anomalies, respectively, and then data credibility scores, source consistency flag sets and anomaly type combination codes are generated; Based on the unified record sequence, short and long analysis time windows are constructed respectively. Load fluctuation characteristics, price response characteristics, execution stability characteristics and settlement disturbance characteristics are extracted. The state deviation index, response sensitivity coefficient and settlement disturbance coefficient are calculated, and the state label, data credibility level and current effective parameter spectrum are determined. Using the state deviation index crossing threshold change point, time period type switching point, and execution stability coefficient change point as the dividing boundary, the continuous energy consumption process of the target retail object is dynamically divided to generate multiple response energy consumption segments, and the transferable load capacity range, trigger window, response time delay range, and execution stability coefficient are determined for each response energy consumption segment. Based on the distribution of anomaly types, anomaly duration, anomaly intensity, source consistency flag set status, and data credibility score within each response energy segment, the segment credibility score of each response energy segment is calculated, and a segment credibility evolution sequence is generated in chronological order. An anomaly action constraint matrix is ​​constructed to establish the mapping relationship between missing anomalies, misaligned anomalies, conflicting anomalies, and time-sensitive anomalies and state feature correction channels, fragment boundary correction channels, budget component contraction channels, and settlement item constraint channels. Differentiated constraints are applied to each response energy fragment based on the anomaly type combination encoding and fragment credibility evolution sequence. Based on response energy consumption segments, contract parameter data, electricity price rule data, historical settlement details and market clearing results, a scenario-object-time period settlement impact tensor is constructed. Each tensor unit stores at least the basic electricity charge impact component, deviation assessment impact component, incentive rebate impact component and default correction impact component. The abnormal type combination encoding and segment category of each response energy consumption segment are mapped to the corresponding tensor unit. Based on the current effective parameter spectrum, the segment credibility score of each response energy segment, the segment credibility evolution sequence, the constraint results of the anomaly effect constraint matrix, and the impact components of the corresponding tensor units, a dynamic pricing response budget vector is generated for each response energy segment. The dynamic pricing response budget vector includes at least the upper limit of price adjustment range, execution duration, notification advance, response limit, and settlement risk limit. Based on the dynamic pricing response budget vector, each response energy segment is divided into executable segments, restricted segments, and rollback segments. Based on the actual execution results and the dynamic pricing response strategy corresponding to each response energy segment, segment-level settlement processing is performed. The effective response electricity and settlement correction coefficient of each response energy segment are calculated to obtain the deviation responsibility allocation result and the incentive rebate or default processing result. Based on the intensity ranking result of the impact component corresponding to each response energy segment and the intensity of abnormal propagation impact, the dominant impact component identifier, dominant impact transfer sequence identifier, key propagation path identifier, propagation blocking point identifier, and dominant distortion source identifier are extracted. When the response energy segment is divided into a backoff segment, based on the anomaly type combination encoding, dominant distortion source identifier, and impact component of the corresponding tensor unit, a representative impact component template is selected from the scene-object-time period settlement impact tensor. Combined with the dominant impact transfer sequence of the previous valid period and the status label of the previous valid period, the dominant component inheritance recalculation is performed to generate a response settlement result with a low confidence mark.

2. The dynamic pricing response and settlement processing method in an electricity retail scenario according to claim 1, characterized in that: The anomaly type combination encoding is used to characterize at least the existence status, dominant anomaly type, anomaly intensity level, and anomaly duration level of missing anomalies, misaligned anomalies, conflicting anomalies, and time-sensitive anomalies within the corresponding response energy segment, and serves as a common input for generating segment confidence scores, calling the anomaly effect constraint matrix, and filtering representative impact component templates.

3. The dynamic pricing response and settlement processing method in an electricity retail scenario according to claim 1, characterized in that: The segment credibility evolution sequence is obtained by calculating the continuity constraint of the segment credibility scores of multiple response energy segments arranged in chronological order. It is used to limit the change magnitude of budget components, the frequency of segment category switching and the backoff switching threshold between adjacent response energy segments, so as to avoid oscillations between adjacent segments in the dynamic pricing response strategy.

4. The dynamic pricing response and settlement processing method in an electricity retail scenario according to claim 1, characterized in that: In the anomaly constraint matrix, missing anomalies correspond to the state feature correction channel, misaligned anomalies correspond to the segment boundary correction channel, conflicting anomalies correspond to the settlement item constraint channel, and time-sensitive anomalies correspond to the budget component contraction channel. Furthermore, when two or more anomaly types exist simultaneously within the same response energy segment, the corresponding composite constraint path is invoked based on the anomaly type combination code to perform joint correction on the state deviation index, trigger window, settlement item confirmation status, and dynamic pricing response budget vector.

5. The dynamic pricing response and settlement processing method in an electricity retail scenario according to claim 1, characterized in that: In addition to representing the impact components of each settlement item under different scenarios, objects and time periods, the scenario-object-time period settlement impact tensor is also used to record the segment category, anomaly type combination code and dominant distortion source identifier corresponding to each response energy segment, so that the same tensor structure can simultaneously undertake the functions of segment mapping, settlement impact assessment, distortion source location and rollback template screening.

6. The dynamic pricing response and settlement processing method in an electricity retail scenario according to claim 1, characterized in that: The dominant distortion source identifier is obtained by jointly sorting the combination encoding of anomaly types, the evolution sequence of fragment credibility, the intensity of the impact component in the corresponding tensor unit, and the intensity of the propagation influence of the key propagation path. It is used to determine the priority of budget component contraction, the priority of settlement item freezing, and the priority of recalculating back-off fragments.

7. The dynamic pricing response and settlement processing method in an electricity retail scenario according to claim 1, characterized in that: The dominant component inheritance recalculation of the rollback segment does not directly use the historical settlement results. Instead, based on the anomaly type combination code and dominant distortion source identifier of the rollback segment, it selects the target template with the highest matching degree with the current segment from multiple representative impact component templates. Then, it performs itemized inheritance recalculation of the basic electricity charge impact component, deviation assessment impact component, incentive rebate impact component, and default correction impact component by combining the dominant impact transfer sequence of the previous effective period and the status label of the previous effective period.

8. A dynamic pricing response and settlement processing system for electricity retail scenarios, characterized in that, include: The data access module is used to acquire multi-source business data corresponding to the target retail object; The timeline mapping module is used to map the multi-source business data to a unified settlement benchmark timeline and generate a unified record sequence. The data quality processing module is used to identify missing key fields, misaligned timestamps, cross-source key field conflicts, and continuous interface timeouts, and to generate data credibility scores, source consistency flag sets, and anomaly type combination codes. The dual-time-window state recognition module is used to calculate the state deviation index, response sensitivity coefficient, and settlement disturbance coefficient, and to determine the state label, data credibility level, and current valid parameter spectrum. The state-driven fragment generation module is used to generate multiple response energy fragments based on the state deviation exponent cross-threshold change points, time period type switching points, and execution stability coefficient change points. The fragment credibility assessment module is used to calculate the fragment credibility score of each response energy fragment and generate a fragment credibility evolution sequence; The anomaly action constraint module is used to construct the anomaly action constraint matrix and apply differentiated constraints to each response energy segment based on the anomaly type combination encoding and the segment credibility evolution sequence. The settlement shock tensor construction module is used to construct the scenario-object-time period settlement shock tensor and complete the tensor mapping of anomaly type combination encoding, segment category and dominant distortion source identification; The dynamic pricing response budget generation module is used to generate dynamic pricing response budget vectors for each response energy segment and divide them into executable segments, restricted segments, and rollback segments based on the current effective parameter spectrum, segment confidence score, segment confidence evolution sequence, abnormal action constraint results, and the impact components of the corresponding tensor units. The segment-level settlement processing module is used to calculate the effective response power and settlement correction coefficient of each response energy segment, and extract the dominant impact component identifier, dominant impact transfer sequence identifier, critical propagation path identifier, propagation blocking point identifier, and dominant distortion source identifier. The rollback settlement module is used to, when a rollback segment occurs, select representative impact component templates from the scene-object-time period settlement impact tensor based on the anomaly type combination encoding and the dominant distortion source identifier, and perform dominant component inheritance recalculation to output a response settlement result with a low confidence marker.

Images