Real-time electric charge settlement and intelligent charge deduction method for intelligent electric meter

Abstract

The invention discloses an electric charge real-time settlement and intelligent deduction method for an intelligent electric meter, and belongs to the technical field of intelligent power grids, and the method comprises the steps: obtaining current waveform data through high-frequency sampling, generating a carrying feature vector through differential exponential smoothing, and achieving the dynamic classification of equipment load types based on the spectral characteristics and a lightweight convolutional neural network. Generating a dynamic rate parameter in combination with the power grid peak regulation demand index and the historical credit score of the user, and constructing a mixed charging bill; a cross-chain transaction message is formed by analyzing the hash value and embedding the hash value into a block chain timestamp, and a three-section settlement smart contract is activated; and monitoring load fluctuation data after fee deduction, generating an abnormal fluctuation mark, and performing optimization through federated learning. According to the method, comprehensive optimization of high-precision real-time charging, tamper-proof trusted transaction and adaptive anomaly monitoring is realized, and the peak regulation efficiency and the power utilization fairness of the power grid are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of smart grid technology, and in particular to a method for real-time electricity fee settlement and intelligent fee deduction of a smart meter. Background Art

[0002] In recent years, with the development of smart grid technology, smart meters, as the core equipment for collecting electricity consumption information, have been widely used in power systems. The traditional electricity bill settlement model is based on scheduled meter reading and fixed rate mechanism, which is difficult to adapt to the needs of real-time electricity price response, user credit assessment and dynamic peak load regulation of the power grid.

[0003] At present, the sampling frequency of traditional electric meters cannot capture the transient characteristics of current waveforms, and mean filtering or simple frequency domain analysis is often used in feature extraction, resulting in insufficient recognition rate of new electrical appliances by load classification algorithms, and frequent billing errors. The existing dynamic electricity price mechanism is mostly based on the preset peak and valley period division, and fails to integrate the real-time peak-shaving demand of the power grid, user credit scores and transient overload risks of equipment, resulting in low efficiency of power grid load balancing and lack of fairness in user electricity costs, especially when the contradiction between supply and demand is prominent, which can easily lead to failure of power regulation. Electricity transaction data relies on centralized server storage and processing, and there are risks of bill tampering, delayed settlement and malicious arrears; although some systems have introduced blockchain technology, cross-chain transactions lack efficient verification mechanisms, and the execution logic of smart contracts is single (such as relying only on balance threshold judgment), which is difficult to adapt to multi-level fund management needs. The detection of abnormal electricity consumption behaviors such as electricity theft and equipment failure usually relies on manual verification or offline data analysis, resulting in long response cycles and high audit costs. In addition, the existing algorithms are not sensitive enough to load fluctuation patterns, making it difficult to detect hidden anomalies in a timely manner. In response to the above problems, the industry urgently needs an intelligent electricity fee management method that can realize high-frequency real-time data collection, dynamic multi-dimensional rate fusion, tamper-resistant trusted settlement and self-closed loop of abnormal fluctuations. Summary of the invention

[0004] To solve the above problems, the present invention provides a real-time settlement and intelligent deduction method for electricity charges of smart meters, which adopts high-frequency sampling data coupled with waveform fingerprint compression, multi-dimensional dynamic rate generation and blockchain cross-chain verification technology, and can realize millisecond-level analysis of load characteristics, tamper-proof real-time settlement and adaptive closed-loop monitoring of abnormal electricity consumption behavior, taking into account both the peak-shaving efficiency of the power grid and the fairness of electricity consumption of users.

[0005] The above objectives can be achieved through the following solutions: A method for real-time electricity bill settlement and intelligent deduction of an intelligent electricity meter, including: obtaining continuous sampled current waveform data on a power line, and generating an original electrical parameter sequence at a frequency of no less than 100 sampling points per second; receiving the original electrical parameter sequence, and generating a feature vector carrying waveform fingerprints by using differential exponential smoothing processing on a TSA chip based on the sampling timestamps of the original electrical parameter sequence; according to the spectral distribution characteristics of the feature vector, outputting a classification result of the device load type through a lightweight convolutional neural network; receiving the load type classification result, and generating dynamic rate parameters in combination with the grid peak shaving demand index; receiving the dynamic rate parameters, and generating a mixed billing bill containing a transient overload coefficient by weighting the user's historical credit score; parsing the hash value of the mixed billing bill, and embedding a blockchain timestamp to form a cross-chain transaction message; after monitoring the cross-chain transaction message, parsing the status of the account pre-deposited amount to activate a three-stage settlement intelligent contract; capturing the load fluctuation data after deduction according to the three-stage settlement intelligent contract, and generating an abnormal fluctuation mark by comparing with the device shutdown characteristic curve; receiving the abnormal fluctuation mark, updating the parameters of the federated learning model and writing them into an anti-tampering ledger to complete the closed-loop processing.

[0006] Optionally, the generating a feature vector carrying waveform fingerprints by using differential exponential smoothing processing on a TSA chip based on the sampling timestamps of the original electrical parameter sequence includes: receiving the original electrical parameter sequence, extracting the differential value of adjacent point power and compressing it into a Huffman dictionary; parsing the Huffman dictionary, and superimposing the voltage phase mutation amount to generate a feature summary with a check bit; performing non-linear filtering based on the feature summary, and outputting a denoised feature vector and irreversibly restoring the original waveform.

[0007] Optionally, the outputting a classification result of the device load type through a lightweight convolutional neural network according to the spectral distribution characteristics of the feature vector includes: preprocessing the Fourier transform result of the feature vector to generate a device fingerprint mapping dictionary; receiving real-time current harmonic distribution data, and traversing the mapping dictionary to output a device identification result; parsing the power curve pattern that has not been recognized and writing it into a knowledge base to be optimized.

[0008] Optionally, the generating dynamic rate parameters in combination with the grid peak shaving demand index includes: parsing the JSON data of the grid peak shaving demand index to generate a basic rate matrix; correcting the weight distribution of the basic rate matrix according to the credit adjustment coefficient generated by the user's historical electricity consumption behavior; correcting the weight distribution based on the loss compensation factor of the real-time temperature of the distribution transformer, and outputting the final dynamic rate parameters.

[0009] Optionally, parsing the hash value of the hybrid billing statement and embedding the blockchain timestamp to form a cross-chain transaction message includes: parsing the JSON structure of the hybrid billing statement, separating the business logic data stream and the fund flow; unidirectionally pushing the separated business logic data stream to the consortium chain node for pre-confirmation; after receiving the pre-confirmation result, permanently writing the fund flow hash value into the genesis block of the transaction chain.

[0010] Optionally, parsing the account pre-deposited amount status to activate the three-stage settlement smart contract includes: when the account pre-deposited amount is greater than or equal to a preset first threshold, activating the full deduction module and generating a unique deduction voucher; when the account pre-deposited amount is less than the first threshold and greater than or equal to a preset second threshold, generating a phased pre-deduction protocol and locking the fund pool; when the account pre-deposited amount is less than the second threshold, triggering the relay control sequence and freezing the electricity usage permission.

[0011] Optionally, triggering the relay control sequence and freezing the electricity usage permission includes: receiving the grid load capacity data within the geofence, calculating the preset allowable power gradient; generating a device load reduction curve based on the preset allowable power gradient, and outputting a relay control instruction queue; burning the relay control instruction queue into the meter firmware and erasing the original control code.

[0012] Optionally, capturing the post-deduction load fluctuation data according to the three-stage settlement smart contract and comparing it with the device shutdown characteristic curve to generate an abnormal fluctuation mark includes: capturing the post-deduction load fluctuation data and extracting the power decay slope; comparing the power decay slope with the reference value in the device characteristic knowledge base to generate a deviation coefficient vector; parsing the line impedance mutation parameter in the deviation coefficient vector to construct an abnormal score matrix associated with multi-source data; when the abnormal score matrix exceeds the preset safety threshold, locking the fund transfer of the associated transaction and initiating a manual review, and generating an inspection report including the abnormal fluctuation mark.

[0013] Optionally, updating the federated learning model parameters and writing them into the tamper-proof ledger to complete the closed-loop process includes: receiving the abnormal fluctuation mark and the corresponding inspection report, using the time period when the abnormal fluctuation mark occurs as an index, extracting the feature vector from the edge node to generate a data fingerprint, and outputting the global optimization requirement; according to the global optimization requirement, reconstructing the convolutional kernel attention weight matrix, and at the same time writing the old parameter hash value into the tamper-proof ledger; compiling the updated weight matrix into a binary instruction set carrying the version number of the abnormal fluctuation mark, and encrypting and transmitting it to the specified meter cluster.

[0014] Based on the same inventive concept, the present invention also provides a real-time electricity bill settlement and intelligent deduction system for an intelligent electricity meter. The system includes: a data acquisition module for acquiring continuous sampled current waveform data on a power line and generating an original electrical parameter sequence at a frequency of not less than 100 sampling points per second; a feature vector generation module for receiving the original electrical parameter sequence and generating a feature vector carrying waveform fingerprints by using differential exponential smoothing processing on a TSA chip based on the sampling timestamps of the original electrical parameter sequence; a device load classification module for outputting a device load type classification result through a lightweight convolutional neural network according to the spectral distribution characteristics of the feature vector; a rate parameter generation module for receiving the load type classification result and generating dynamic rate parameters in combination with the power grid peak shaving demand index; a billing statement generation module for receiving the dynamic rate parameters and generating a mixed billing statement containing a transient overload coefficient by weighting the user's historical credit score; a transaction message generation module for parsing the hash value of the mixed billing statement and forming a cross-chain transaction message by embedding a blockchain timestamp; an intelligent contract generation module for monitoring the cross-chain transaction message and parsing the account pre-deposited amount status to activate a three-stage settlement intelligent contract; an abnormal fluctuation marking generation module for capturing post-deduction load fluctuation data according to the three-stage settlement intelligent contract and generating an abnormal fluctuation marking by comparing the device shutdown characteristic curve; a feedback record module for receiving the abnormal fluctuation marking, updating the parameters of the federated learning model and writing them into an anti-tampering ledger to complete the closed-loop process.

[0015] Compared with the prior art, the present invention has the following advantages: 1. By combining high-frequency sampling technology with intelligent algorithms, refined analysis of current waveforms and real-time dynamic classification of load types are achieved; differential exponential smoothing processing effectively extracts waveform features, and in combination with a lightweight convolutional neural network, significantly improves the recognition ability of nonlinear devices and transient loads, solving the problem that traditional methods are difficult to capture complex electricity consumption characteristics and providing a reliable data basis for accurate billing. 2. A dynamic rate model is constructed based on real-time peak shaving requirements, user credit assessment, and device operating status, generating a mixed billing strategy that integrates multiple regulatory factors; by flexibly responding to power grid load fluctuations and personalized electricity consumption behaviors, it not only improves the power grid peak shaving efficiency but also ensures the fairness of user rights and interests, effectively balancing the supply-demand contradiction and optimizing the energy distribution efficiency. 3. The blockchain cross-chain architecture is adopted to decouple and verify the billing process, and the business logic and capital flow are separated to ensure the traceability of the entire transaction link; the three-stage intelligent contract combined with the hierarchical account management mechanism realizes hierarchical prevention and control of arrears risks and automatic execution, enhances the anti-repudiation of transactions through the federated chain pre-confirmation and genesis block anchoring technology, and constructs a transparent and reliable electricity bill settlement ecosystem. The real-time capture device shuts down features and changes in line status, establishing an abnormal fluctuation marking mechanism to quickly identify concealed electricity consumption problems; the federated learning model continuously optimizes the load classification algorithm through distributed collaborative training, combines the blockchain ledger to solidify the iterative process, forms an autonomous evolution system for abnormal warning, parameter update, and closed-loop disposal, and comprehensively improves the system's robustness and long-term adaptability.

[0016] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be realized and obtained by the structures pointed out in the specification, the claims, and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is a schematic flowchart of a method for real-time electricity bill settlement and intelligent deduction of an intelligent electricity meter according to an embodiment of the present invention.

[0019] Figure 2 It is a schematic structural diagram of a system for real-time electricity bill settlement and intelligent deduction of an intelligent electricity meter according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0021] Refer to Figure 1 , an embodiment of the present invention proposes a method for real-time electricity bill settlement and intelligent deduction of an intelligent electricity meter. Through high-frequency sampling data coupling waveform fingerprint compression, multi-dimensional dynamic rate generation, and blockchain cross-chain verification technologies, it can achieve millisecond-level analysis of load characteristics, tamper-proof real-time settlement, and adaptive closed-loop monitoring of abnormal electricity consumption behavior, taking into account both the power grid peak shaving efficiency and the fairness of user electricity consumption.

[0022] The specific steps of the method in this embodiment include: Obtain the current waveform data sampled continuously on the power line, and generate an original electrical parameter sequence at a frequency of no less than 100 sampling points per second; Receive the original electrical parameter sequence, and generate a feature vector carrying waveform fingerprints by using differential exponential smoothing processing on the TSA chip based on the sampling timestamps of the original electrical parameter sequence; Specifically, obtain the original voltage and current waveform data through high-speed sampling no less than 100 times per second. While using the differential exponential smoothing algorithm to remove high-frequency noise, retain the local mutation characteristics of the waveform, such as harmonic phase jumps or steep waveforms at the moment of equipment startup and shutdown. The differential calculation obtains the gradient value by weighted averaging of the front and back data windows, and generates a fingerprint feature vector containing waveform trends and details. This method is more suitable for the accurate expression of short-term impact loads compared with the traditional mean filtering. Differential exponential smoothing is a dynamic filtering method that combines a sliding window and gradient calculation, and adjusts the smoothing intensity according to the time weight of the data points, highlighting the waveform change trend during noise reduction. Waveform fingerprints are a set of characteristic parameters extracted from waveform segments formed by multiple samplings, which can uniquely characterize specific types of electrical appliances or abnormal states.

[0023] Among them, the sliding window has a value range of 50 - 200 sampling points, and the dynamic adjustment formula of the smoothing coefficient is , wherein is the standard deviation of the power differential within the current window, is the mean value of the historical window standard deviation, is an adjustment parameter, which can be preset ; The calculation formula of the backward difference gradient value is , wherein, is the current value at time, is the sampling interval, which can be set to 0.01 seconds; The calculation formula of the weighted smoothing output is .

[0024] According to the spectral distribution characteristics of the feature vector, output the classification result of the equipment load type through a lightweight convolutional neural network; Specifically, perform a fast Fourier transform on the feature vector, extract the power spectral density distribution in the range of 0.1 - 2 kHz, and identify the spectral differences between constant impedance type and electronic switch type loads; at the same time, input the standardized feature vector into a pre-trained lightweight convolutional neural network model. This network uses depthwise separable convolutions to reduce the computational amount and achieve efficient classification of nonlinear devices such as variable frequency air conditioners and charging piles. The classification results are updated every 50 milliseconds. The lightweight convolutional neural network is a deep learning model optimized by reducing the number of network layer parameters and using low-rank decomposition technology, suitable for real-time inference of embedded devices with limited computing power.

[0025] Among them, the lightweight convolutional neural network architecture includes: the input layer receives a spectral vector with a dimension of 128×1; the first convolutional layer has 4 groups of depthwise separable convolutions, with a kernel size of 3×1, a stride of 2, and an output channel number of 32; the attention module calculates the channel attention weights using a squeeze-and-excitation network (SENet) with a compression ratio r = 16; the second convolutional layer has 2 groups of low-rank decomposition convolutions, which decompose the original 5×1 kernel into two layers of 5×3 and 3×1, reducing the number of parameters by 40%; the fully connected layer has the number of output nodes corresponding to N known device types and is activated by Softmax; the training data set contains the harmonic feature data of typical industrial devices in the GB / T 17215 standard, such as electric welders and frequency converters.

[0026] Receive the classification result of the load type, and generate dynamic rate parameters in combination with the grid peak shaving demand index; Receive the dynamic rate parameters, and generate a mixed billing bill containing a transient overload coefficient by weighting the user's historical credit score; Specifically, according to the real-time peak shaving demand index released by the power grid dispatching center, such as the regional line load rate and the transformer temperature over-limit warning, calculate the power supply cost adjustment factor, and generate time-of-use differentiated rates in combination with the user credit score database. The user credit score database can include historical payment records and the number of defaults; for the transient overload of the equipment detected in the load classification result, such as when the instantaneous current exceeds 80% of the nominal value when the air conditioner compressor starts, introduce an overload coefficient to weight the over-limit electricity consumption for billing, and finally generate a multi-layer billing structure including basic fees, peak shaving surcharges, and credit discounts. The peak shaving demand index is a quantitative indicator reflecting the current supply-demand balance state of the power grid, which is dynamically updated according to real-time load fluctuations and is used to adjust the time-of-use electricity price ratio. The transient overload coefficient is a punitive billing weight for the short-term abnormal energy-consuming behavior of the equipment, and its value is determined by the product of the over-limit degree and the duration.

[0027] Analyze the hash value of the mixed billing bill, and form a cross-chain transaction message by embedding the blockchain timestamp; After monitoring the cross-chain transaction message, analyze the status of the pre-deposited amount in the account to activate a three-stage settlement smart contract; Specifically, the hash value of the bill core fields and the latest timestamp of the main chain of the power grid blockchain are embedded in the transaction message, and cross-chain verification is carried out with the user-side microgrid alliance chain to ensure the immutability and chronological consistency of the transaction data; when triggering the three-stage settlement smart contract, the current period's fees are preferentially deducted from the user's pre-deposited account. When the balance is lower than the threshold, part of the prepayment is frozen and a warning is issued. If the critical value is reached, a remote power-off instruction is automatically sent to the meter circuit breaker module. Cross-chain verification is to confirm the legality of the transaction between different blockchain networks through relay nodes to prevent double-spending attacks or forged bills. The three-stage settlement is a multi-level fund control mechanism for pre-deduction of fees, freezing buffer, and emergency power-off, and the response strategy is triggered hierarchically according to the balance status.

[0028] Capture the post-deduction load fluctuation data according to the three-stage settlement smart contract, and generate an abnormal fluctuation mark by comparing with the device shutdown characteristic curve; Receive the abnormal fluctuation mark, update the parameters of the federated learning model, and write them into the tamper-proof ledger to complete the closed-loop processing.

[0029] Specifically, after the deduction operation is executed, continuously monitor the current fluctuation curve, perform template matching on the actual waveform with the decay slope and oscillation frequency parameters in the standard feature library when the device is normally shut down. If the deviation exceeds 15%, it is marked as an abnormal event and an on-site inspection signal is triggered; at the same time, the anonymized abnormal data is uploaded to the federated learning central node to jointly update the load classification model of multiple meter terminals and optimize the sensitivity of the next detection. Federated learning update is to improve the abnormal recognition ability of each terminal model through distributed collaborative training on the premise of protecting user privacy and avoid the security risks of centralized data collection.

[0030] Exemplarily, this method forms a refined description of the device's electricity consumption behavior through high-precision waveform sampling and feature extraction, implements dynamic rate decision-making by combining the real-time peak shaving requirements of the power grid and the user's credit status, uses blockchain cross-chain verification to ensure the non-repudiation of billing data in the whole process, and realizes system self-optimization through anomaly detection and model iteration.

[0031] Optionally, the sampling timestamp based on the original electrical parameter sequence generates a feature vector carrying waveform fingerprints by using differential exponential smoothing processing in the TSA chip, including: Receive the original electrical parameter sequence, extract the power differential value of adjacent points, and compress it into a Huffman dictionary; Specifically, after receiving the original electrical parameter sequence, the system first calculates the power differential values of adjacent sampling points, and these differential values reflect the instantaneous rate of current change. Subsequently, Huffman coding is used to compress the differential values, and an optimal prefix code table is constructed based on the occurrence frequencies of different values to form a compact Huffman dictionary. This process effectively reduces the consumption of transmission bandwidth while retaining the timing characteristics of power mutations, such as the current step data at the moment of motor startup or switch. The Huffman dictionary is a variable-length coding structure based on the statistical characteristics of power differentials. Small-amplitude fluctuations with high frequency are represented by short codes, and large-amplitude mutations with low frequency retain long code configurations, achieving a balance between data volume compression and feature integrity.

[0032] Parse the Huffman dictionary and superimpose the voltage phase mutation amount to generate a feature summary with a parity bit. Specifically, after parsing the compressed data of the Huffman dictionary, the voltage phase mutation amount is superimposed, such as the zero-crossing offset and harmonic phase difference, to enhance the characterization ability of waveform features. To ensure data credibility, a cyclic redundancy check code is added as an additional bit to generate a feature summary with self-verification ability. This verification mechanism can detect data tampering or loss during transmission and prevent feature distortion caused by signal interference. The voltage phase mutation amount is the phase abnormal offset of the voltage waveform within a specific period and is used to capture the phase distortion caused by capacitor switching or the switching of power electronic devices.

[0033] Perform non-linear filtering based on the feature summary, output the denoised feature vector and irreversibly restore the original waveform.

[0034] Specifically, perform non-linear filtering processing based on the feature summary, and use a piecewise function to differentially suppress noises of different amplitudes. Small-amplitude background noises are significantly attenuated, while large-amplitude mutation signals such as equipment startup and shutdown are enhanced. The feature vector output after filtering is encrypted and transformed through a one-way hash function to ensure that even if the feature vector is obtained, the original current waveform cannot be deduced, achieving double protection of user privacy and data security. Non-linear filtering is a data processing method that dynamically adjusts the filtering intensity according to the signal amplitude, adopts different suppression strategies for tiny noises and significant mutation signals, and avoids the problem of signal edge blurring caused by traditional linear filtering.

[0035] Exemplarily, the data storage efficiency is optimized through power differential compression, the reliability of feature representation is enhanced by combining phase mutation enhancement and verification mechanisms, and high-fidelity denoising is achieved through nonlinear filtering. The finally generated irreversible feature vector not only retains the key electricity consumption behavior fingerprints but also effectively prevents the risk of data leakage. This method significantly improves the real-time processing ability of the electric meter edge computing device, enhances the anti-interference ability of load feature analysis, and at the same time meets the privacy protection requirements, providing a high-precision and high-security data basis for subsequent dynamic rate calculation and anomaly detection. Its beneficial effects include improving the waveform feature generation efficiency, strengthening the sensitive data protection ability, and ensuring the robustness of feature extraction in complex electricity consumption scenarios.

[0036] Optionally, the output of the device load type classification result through the lightweight convolutional neural network according to the spectral distribution characteristics of the feature vector includes: Preprocess the Fourier transform result of the feature vector to generate a device fingerprint mapping dictionary; Specifically, preprocess the Fourier transform result of the feature vector, extract the key frequency band parameters through the spectral energy distribution and fundamental frequency harmonic components, and cluster and encode the typical spectral features of different device types into a hash index structure to form a device fingerprint mapping dictionary. This dictionary serves as a reference library for load recognition, supporting fast comparison and retrieval. The device fingerprint mapping dictionary is a device feature index table constructed through Fourier transform and data clustering, storing the power spectrum features of different devices in specific frequency bands in hash form.

[0037] Receive the real-time current harmonic distribution data and traverse the mapping dictionary to output the device recognition result; Specifically, receive the real-time current harmonic distribution data, calculate the amplitude and phase differences of its main harmonic components, and traverse the known device feature templates stored in the device fingerprint mapping dictionary. The cosine similarity algorithm is used to measure the matching degree between the real-time data and the template. When it exceeds the set threshold, the corresponding device type is output to achieve the rapid classification of electronic devices such as variable frequency air conditioners and LED drivers. The real-time current harmonic distribution data is a set of multiple harmonic components generated during the operation of the electrical equipment, and the spectral vector composed of the fundamental wave and high-frequency harmonic components such as the 3rd and 5th harmonics is separated through Fourier transform.

[0038] Analyze the unrecognized power curve pattern and write it into the knowledge base to be optimized.

[0039] Specifically, when analyzing the unrecognized power curve pattern, extract its characteristic parameters such as the transient start waveform and the steady-state operation harmonic distortion rate, mark the timestamp and environmental parameters, such as the grid voltage fluctuation value, and then write them into the knowledge base to be optimized. This knowledge base serves as the data source for offline training and supports the expansion of the device type recognition ability during subsequent model iteration. The knowledge base to be optimized is a structured database specifically storing the characteristics of unrecognized power consumption patterns, and generates new device labels through manual annotation or automated clustering for model retraining.

[0040] Exemplarily, an efficient matching of known loads is achieved by constructing a device fingerprint dictionary, combined with dynamic identification of harmonic distribution and autonomous accumulation of unknown patterns, to form an extensible load classification system. This method can improve the coverage of load classification, timely capture the characteristics of new electrical devices, and continuously iterate to optimize the model performance, enhancing the system's adaptability to new scenarios while reducing manual intervention. Its beneficial effects include improving the device classification accuracy, reducing the missed detection rate, and effectively supporting the refined management requirements of the power grid for diverse loads.

[0041] Optionally, the generation of dynamic rate parameters in combination with the grid peak shaving demand index includes: Parse the JSON data of the grid peak shaving demand index to generate a basic rate matrix; Specifically, parse the JSON format data of the grid peak shaving demand index, extract the key parameters including the regional load rate and the renewable energy consumption rate, and generate a two-dimensional matrix containing the basic electricity price and the peak shaving surcharge rate for different time periods according to the preset peak-valley period division rules. This matrix dynamically maps the current grid supply and demand status and provides an initial framework for subsequent rate adjustment. The basic rate matrix is an electricity price structure model constructed based on the two dimensions of time and load type. The rows represent time period intervals, and the columns correspond to the benchmark price units of different types of loads.

[0042] Modify the weight distribution of the basic rate matrix according to the credit adjustment coefficient generated based on the user's historical electricity consumption behavior; Specifically, read the user's historical electricity consumption behavior dataset, calculate the credit score through the number of late payment times and the record of over-capacity electricity consumption violations, and normalize it into a credit adjustment coefficient. Apply this coefficient as a weighting factor to the corresponding column weights of the basic rate matrix, reduce the peak shaving surcharge rate for users with excellent credit, and implement a premium strategy for high-risk users to form a preliminarily adjusted rate distribution. The credit adjustment coefficient is a dynamic weight parameter reflecting the user's credit level, and quantifies the user's historical behavior into a rate discount or penalty ratio through an algorithm.

[0043] Modify the weight distribution based on the loss compensation factor of the distribution transformer's real-time temperature, and output the final dynamic rate parameters.

[0044] Specifically, the temperature data of the distribution transformer winding is collected in real time, and the loss compensation factor is calculated by combining the loss model under the rated working condition. The compensation factor corresponding to the period with excessive temperature is superimposed on the row weights of the basic rate matrix to allocate the electricity bill for the additional operation and maintenance costs caused by equipment overload, and multi-dimensional dynamic rate parameters integrating grid peak shaving, user credit and equipment loss are output. The loss compensation factor is a rate correction coefficient for compensating the additional energy loss of distribution equipment due to overload or aging, and is dynamically calculated according to the deviation between the real-time temperature and the standard working condition.

[0045] Exemplarily, a multi-dimensional dynamic rate model is constructed by integrating data on grid peak shaving requirements, user credit assessment, and distribution equipment loss. This method realizes the refined control of rates, responds to the real-time load fluctuations of the grid, takes into account the differences in user behavior and the health status of equipment, and promotes the efficient allocation of electric power resources. Its beneficial effects include improving the fairness and transparency of rate setting, motivating users to optimize their electricity consumption habits, reducing the risk of grid equipment overload, and supporting the coordinated optimization of the economy and security of the power system.

[0046] Optionally, the parsing of the hash value of the hybrid billing bill and the embedding of the blockchain timestamp to form a cross-chain transaction message includes: Parse the JSON structure of the hybrid billing bill and separate the business logic data stream and the fund flow; Specifically, the JSON structure is a lightweight data exchange format, which stores information such as rate parameters, electricity consumption, and credit discounts of the billing bill in a hierarchical manner in the form of key-value pairs. The business logic data stream contains the billing rules associated with the electricity price policy and the load type, and the fund flow involves the actual settlement amount and the payment path information. This step ensures the independence and traceability of business configuration and fund transactions during subsequent blockchain processing through structured field separation. The hybrid billing bill is a multi-layer billing data set generated by integrating dynamic rates, credit scores, and transient overload coefficients. The JSON structure is a nestable tree-like data format that supports efficient parsing and cross-platform transmission. The business logic data stream is a set of non-financial information representing electricity price calculation rules and electricity consumption strategies. The fund flow is a financial operation record containing transaction amounts, account identifiers, and settlement timestamps.

[0047] Unidirectionally push the separated business logic data stream to the consortium blockchain node for pre-confirmation; Specifically, the consortium blockchain nodes are the blockchain network nodes composed of authorized members such as power grid enterprises and regulatory agencies, which verify the legality of business logic through a consensus mechanism. The pre-confirmation process checks the compliance of data formats and the validity of permissions. For example, it verifies whether the user credit adjustment coefficient is within the range agreed upon in the contract. The one-way push uses asymmetric encryption to ensure the irreversible transmission of data from the electricity meter to the blockchain, preventing tampering. The consortium blockchain nodes are the authorized blockchain member server clusters participating in the verification of business logic. Pre-confirmation is a preliminary audit mechanism for data legality before formal accounting, avoiding invalid transactions being recorded on the chain.

[0048] After receiving the pre-confirmation result, the hash value of the fund flow is permanently written into the genesis block of the transaction chain.

[0049] Specifically, the transaction chain is a blockchain branch dedicated to fund settlement. The genesis block is the initial block of this chain, recording an immutable hash value fingerprint. The hash value of the fund flow is a unique digital digest generated through the SHA-3 algorithm. After being written, it forms a chain association with the timestamp and the hash of the adjacent block, ensuring the audibility of the entire life cycle of the transaction record. The transaction chain is a financial-specific blockchain network independent of the business chain, optimizing the processing efficiency of high-frequency transactions. The genesis block is the first block of the blockchain network, storing the initial state of the chain and the core verification rules. Permanently writing the hash value is an irreversible operation that solidifies the transaction summary to the blockchain in a cryptographically bound manner.

[0050] Among them, the cross-chain transaction verification mechanism is the business logic chain, that is, the consortium blockchain adopts the improved PBFT consensus, and the node response threshold is set to a 2 / 3 majority; the genesis block of the transaction chain defines the hash anchoring rule: the hash value of the newly generated block is synchronized to the Hyperledger Fabric main chain through the Merkle tree every 10 minutes; the trigger condition of the smart contract can be that when the user balance is less than the second threshold, the Chaincode function FreezeAccount(bytes32 userID) is called, and an SGX encrypted log is generated. Exemplarily, a separated and trusted processing of business logic and fund settlement is achieved through a cross-chain transaction architecture. When parsing the billing statement, the JSON structure separation mechanism ensures the independent transfer of business rules and financial data. The pre-confirmation of the consortium chain ensures the compliance of business strategies, and the writing of the genesis block of the transaction chain solidifies the fingerprint of fund operations. At the technical level, combining the distributed ledger feature of the blockchain with the one-way data pipeline design effectively isolates the impact of business changes and transaction history, forming an anti-tampering full-link evidence chain. Its beneficial effects include: enhancing the credibility and non-repudiation of billing transactions, and shortening the verification cycle of cross-institutional settlement; the decoupled processing of the business flow and the fund flow enhances the system scalability, supporting the dynamic adjustment of rate strategies without disturbing the existing transaction records; the genesis block anchoring mechanism strengthens the transaction traceability ability, providing an irrefutable audit basis for electricity bill disputes. This method ultimately realizes a highly secure and efficient real-time electricity bill settlement system, taking into account the compliance requirements of grid operation and the protection of user rights and interests.

[0051] Optionally, the parsing of the account pre-deposited amount status to activate the three-stage settlement smart contract includes: When the account pre-deposited amount is greater than or equal to a preset first threshold, activate the full deduction module and generate a unique deduction voucher; Specifically, the full deduction module is a program component that directly deducts all payable amounts from the user's account according to the electricity bill amount, applicable to the scenario where the user has sufficient funds. The unique deduction voucher is an electronic receipt containing a transaction serial number, a timestamp, and an encrypted signature, ensuring the immutability and uniqueness of the transaction record. This step prevents the accumulation of arrears through immediate deduction, and at the same time generates a unique voucher to provide a traceable basis for subsequent inquiries and disputes. The full deduction module is a software functional unit that executes real-time one-time deduction of the electricity bill amount. The unique deduction voucher is an encrypted data packet that identifies a single transaction and has uniqueness, used to verify the authenticity of the transaction.

[0052] When the account pre-deposited amount is less than the first threshold and greater than or equal to a preset second threshold, generate a phased pre-deduction agreement and lock the fund pool; Specifically, the phased pre-deduction agreement is a temporary contract that deducts the amount to be deducted in batches according to a preset cycle, such as deducting in phases by the hour or by electricity consumption. The locking of the fund pool is to freeze a part of the balance in the user's account as a performance bond to ensure the reliability of subsequent deduction execution. This step alleviates the electricity consumption pressure of users with short-term insufficient funds through flexible deduction rules, and at the same time avoids the cash flow risk of the grid enterprise. The phased pre-deduction agreement is a temporary fund management plan that allows installment deductions. The locking of the fund pool is an operation that restricts the liquidity of a specific amount in the account, used to ensure the feasibility of phased deductions.

[0053] When the account pre-deposited amount is less than the second threshold, trigger the relay control sequence and freeze the electricity usage permission.

[0054] Specifically, the relay control sequence is a set of hardware switch instructions arranged in order of priority, and the power usage permission is frozen by cutting off or restricting the power supply line. For example, the non-essential load circuit is preferentially disconnected or the low-power consumption mode is entered. This step directly blocks the delinquent user from continuing to use electricity through hardware-level control, and forcibly ensures the recovery of electricity charges and the safety of power grid operation. The relay control sequence is an ordered instruction set for controlling the switch state of the relay in the electricity meter. Freezing the power usage permission is an operation to terminate the user's continued electricity usage by physically disconnecting the energized circuit or restricting the power supply power.

[0055] Exemplarily, a hierarchical response mechanism is adopted to implement stepped fund management, and the settlement strategy is dynamically adjusted according to the account balance status. Its core principle lies in dividing the credit risk level through preset thresholds. When the balance is higher than the first threshold, the settlement is quickly completed to ensure the recovery of funds. When the balance is in the middle range, phased deductions are enabled to balance the user experience and the rights and interests of the power grid. When it is lower than the second threshold, the permission freezing is enforced to eliminate the delinquent risk. The technical effects are reflected in preventing the capital chain break caused by arrears at multiple levels and enhancing the active control ability of electricity charge recovery; the phased agreement reduces the short-term capital pressure of users and improves the continuity of electricity usage; the hardware-level permission freezing ensures the mandatory and timeliness of arrears management. The overall method combines flexible billing and rigid control, avoiding the drawbacks of frequent power outages in the traditional prepaid mode and strengthening the security of electricity charge transactions through hierarchical response, and finally achieving the dual optimization of user rights and interests protection and power grid enterprise risk control.

[0056] Optionally, triggering the relay control sequence and freezing the power usage permission includes: Receiving the power grid load capacity data within the geographical fence and calculating the preset allowable power gradient; Specifically, the geographical fence is the virtual power supply area boundary divided according to the distribution network topology. The power grid load capacity data includes parameters such as the remaining capacity of the transformer in the current area, the upper limit of the line current, and the load volatility. Based on the real-time data, the preset allowable power gradient is calculated to determine the maximum power supply power reduction threshold available to the delinquent user during a specific period. For example, when the regional load rate exceeds the limit, the gradient calculation is used to limit the decrease rate of the instantaneous power consumption of the user, avoiding equipment damage caused by sudden voltage drop. The geographical fence is the power supply management geographical boundary delimited by the power grid node coordinates and the load distribution data. The power grid load capacity data is a set of dynamic parameters representing the current maximum bearable power of the distribution equipment. The preset allowable power gradient is the power adjustment rate limit value set based on the power grid safety constraints.

[0057] Generating a device load reduction curve based on the preset allowable power gradient and outputting a relay control instruction queue; Specifically, the device load reduction curve is an implementation plan for reducing the power supply in stages according to the time series, and a stepped power decline trajectory is generated according to the gradient threshold. For example, when the gradient limit is 5% of the rated power per second, an instruction queue for gradually cutting off non-critical loads in 10 seconds is generated, and the basic lighting power consumption is preferentially retained. The relay control instruction queue includes the switch action timing sequence, the target circuit number, and the execution delay parameter to ensure that the meter relay module accurately executes the power adjustment action. The device load reduction curve is a time-sequence control scheme for reducing the power supply in stages. The relay control instruction queue is a set of hardware switch operation instructions arranged according to the priority and time sequence.

[0058] Burn the relay control instruction queue into the meter firmware and erase the original control code.

[0059] Specifically, the burning operation is to write the compiled machine code instructions into the read-only storage area of the meter control chip through a secure channel to replace the original control logic. Erasing the original control code means clearing the storage area of the original relay driver program of the meter to prevent instruction conflicts or malicious code injection. For example, the page erasing technology of the FLASH memory is used to ensure that the new instruction queue exclusively occupies the hardware execution permission. Burning is an irreversible operation to solidify the control instructions into the meter memory through the programming interface. Erasing the original control code is to completely delete the data in the storage area of the original relay driver program of the meter.

[0060] Exemplarily, for the delinquent account scenario, refined power supply control is realized. The core principle lies in dynamically binding the real-time capacity of the power grid and the management of user power consumption rights. The geographical fence is used to accurately locate the load bottleneck area of the power grid, and a progressive load reduction strategy is generated in combination with the preset power gradient, which not only avoids the negative impact of direct power-off but also protects the safety of the distribution equipment through smooth power adjustment. The technical effects are reflected in the integration of the regional power grid state perception and the adaptive control ability, giving priority to ensuring the power supply continuity of critical loads; the time-sequential execution mechanism of the relay instruction queue improves the accuracy and safety of power adjustment; the firmware burning and code erasing operations ensure the irreversibility and anti-interference ability of the control logic. Finally, a flexible power-off management system that takes into account the power grid stability, equipment protection, and user experience is formed, significantly improving the intelligent level and execution reliability of power regulation in the delinquent account scenario.

[0061] Optionally, the capturing of the load fluctuation data after deduction according to the three-stage settlement smart contract and the generation of the abnormal fluctuation mark by comparing with the device shutdown characteristic curve include: Capturing the load fluctuation data after deduction and extracting the power decay slope; Specifically, the power decay slope is the rate of change of the current or power after the device is turned off with respect to time, which is obtained by sampling the millisecond-level current waveform data after deduction of the fee and calculating the first derivative of its decay curve or the slope value of the fitted linear stage. This slope reflects the physical response characteristics when the device is powered off, such as the difference between the inertial coasting of a motor and the instantaneous power-off of an electronic device, providing dynamic feature inputs for anomaly detection. The power decay slope is a quantitative index characterizing the rate of energy consumption decline when the device is powered off. The load fluctuation data is the instantaneous change record of the current or power on the power grid circuit after the electricity bill settlement is completed.

[0062] Compare the power decay slope with the reference value in the device feature knowledge base to generate a deviation coefficient vector; Specifically, the device feature knowledge base is a feature database that pre-stores the standard power decay parameters when various electrical appliances are normally powered off, including typical slope ranges, timing vibration modes, etc. By calculating the absolute difference and fluctuation variance between the decay slope collected in real time and the reference value of the corresponding device in the knowledge base, a deviation coefficient vector reflecting the degree of deviation is generated. This vector contains multi-dimensional anomaly indicators such as slope deviation degree and phase offset, providing basic data for subsequent correlation analysis. The device feature knowledge base is a reference library of device normal power-off characteristics established based on historical data or experiments. The deviation coefficient vector is a set of numerical indicators describing the difference between real-time data and reference features.

[0063] Analyze the line impedance mutation parameters in the deviation coefficient vector to construct an anomaly score matrix associated with multi-source data; Specifically, the line impedance mutation parameter is the rapid change amount of the impedance in the power grid line caused by switch operation, poor contact, or illegal access, which is obtained by calculating parameters such as the phase difference between voltage and current and the harmonic distortion rate when the power is off. Perform spatio-temporal correlation analysis on the impedance mutation data and other indicators in the deviation coefficient vector to construct an anomaly score matrix. Each element of this matrix represents the comprehensive weight score of different anomaly types, which is used to quantify the possibility and severity level of anomaly events. The line impedance mutation parameter is a physical quantity reflecting the instantaneous abnormal change of the impedance in the power grid line. The anomaly score matrix is a risk assessment quantification model formed by weighted fusion of multi-dimensional anomaly indicators.

[0064] When the anomaly score matrix exceeds the preset safety threshold, lock the fund transfer of the associated transaction and initiate manual review, generating an inspection report containing anomaly fluctuation marks.

[0065] Specifically, the preset security threshold is a dynamic threshold parameter set according to historical abnormal case statistics and grid safety regulations. By real-time monitoring whether the weighted sum of each dimension in the abnormal score matrix exceeds the threshold, the fund transfer process of suspicious transactions is triggered automatically. Meanwhile, the abnormal time period, device type, and score details are written into the inspection report to form a traceable abnormal event file, providing a basis for subsequent verification and model optimization. The preset security threshold is a hierarchical warning value for triggering the risk control mechanism. The inspection report is an audit document recording the characteristics of abnormal events and handling measures.

[0066] Exemplarily, an abnormal detection model is established by capturing the dynamic power consumption characteristics when the device is turned off. Its core principle lies in the collaborative analysis mechanism that integrates the physical characteristics of the device and the grid line state. By extracting the slope of power decay and comparing it with the standardized knowledge base, the rationality and consistency of the device power-off behavior are quantified; a multi-source correlation scoring model is constructed in combination with the line impedance parameters to effectively distinguish normal fluctuations from abnormal scenarios such as electricity theft and device failures. The technical effects are reflected in improving the recognition sensitivity of concealed abnormal power consumption behaviors, avoiding false alarms and missed detections caused by single-feature detection; the dynamic threshold mechanism realizes precise risk classification and control, reducing the workload of manual verification; the inspection report generation function improves the abnormal event traceability chain, strengthening the anti-electricity-theft and device health management capabilities. Finally, an intelligent closed-loop abnormal monitoring system is formed, significantly improving the security and fairness of the electricity settlement environment.

[0067] Optionally, the process of updating the federated learning model parameters and writing them into the tamper-proof ledger to complete the closed-loop processing includes: Receiving the abnormal fluctuation mark and the corresponding inspection report, using the time period when the abnormal fluctuation mark occurs as an index, extracting the feature vector from the edge node to generate a data fingerprint, and outputting the global optimization requirement; Specifically, the data fingerprint is a unique identifier generated by compressing the feature vector through a hash algorithm, used for quickly indexing and matching the electricity consumption feature data of a specific time period. The global optimization requirement is a model update instruction generated according to the abnormal analysis result, marking the model modules that need to be optimized and the abnormal detection dimensions to be enhanced. This step accurately locates and associates the data through the abnormal time period index, generating the target parameters for collaborative optimization of the federated learning system. The data fingerprint is a feature summary formed by encrypting the electricity consumption characteristics of a specific time period. The global optimization requirement is a set of target instructions for triggering the update of the federated learning model parameters. The edge node is a local computing device deployed at the electricity meter terminal or the regional gateway.

[0068] According to the global optimization requirement, reconstruct the convolutional kernel attention weight matrix, and at the same time write the old parameter hash value into the tamper-proof ledger; Specifically, the convolutional kernel attention weight matrix is a set of parameters for allocating the attention degree of the input features in the spatial dimension in the lightweight neural network. The reconstruction operation adjusts the weight proportion of different feature channels according to the abnormal fluctuation features, such as enhancing the detection weight of the harmonic distortion parameters. The old parameter hash value is the parameter fingerprint generated by the one-way hashing algorithm, which is written into the blockchain anti-tampering ledger to form a version traceability baseline, ensuring the integrity and auditability of the model iteration process. The convolutional kernel attention weight matrix is a network parameter structure for allocating the importance of the feature space. The old parameter hash value is the encrypted digest record of the parameter set before the model update. The anti-tampering ledger is an immutable data storage system implemented by blockchain technology.

[0069] Compile the updated weight matrix into a binary instruction set carrying the version number of the abnormal fluctuation mark, and encrypt and transmit it to the specified electric meter cluster.

[0070] Specifically, the binary instruction set is the low-level machine code after compilation processing, which is suitable for the instruction set architecture of the electric meter terminal processor. The version number of the abnormal fluctuation mark is the unique identifier embedded in the update file, which is used to trace the abnormal type and occurrence time corresponding to the model optimization. The encrypted transmission uses the asymmetric encryption algorithm to protect the communication pipeline, ensuring that the update instruction is not maliciously intercepted or tampered with during the transmission process. The binary instruction set is a set of machine language codes that can be directly embedded in the electric meter firmware for execution. The version number of the abnormal fluctuation mark is the model update identifier associated with a specific abnormal event. The specified electric meter cluster is a set of grouped terminal devices that need to synchronously update the model.

[0071] Exemplarily, by constructing a federated learning dynamic update mechanism based on abnormal events, the continuous optimization of the power consumption monitoring model is realized. Its core principle is to convert the abnormal fluctuation features into model optimization signals, and jointly improve the feature recognition ability of multiple terminals while protecting the local privacy data. The technical effects are reflected in strengthening the detection sensitivity of the model to new electricity theft patterns and equipment failures through elastic weight adjustment; the version traceability function of the anti-tampering ledger ensures the credibility of the model iteration and prevents parameter tampering attacks; the encrypted compilation and directional transmission mechanism improves the security and device compatibility of the model update. This method forms a closed-loop optimization system with synchronous enhancement of privacy security and technical defense and attack capabilities, providing the intelligent electric meter with long-term self-evolving abnormal monitoring capabilities, and significantly improving the intelligent level of power grid anti-fraud and equipment health management.

[0072] Among them, the privacy protection mechanism and parameter update rule are: each terminal node uses its local abnormal data set to calculate the loss function with respect to the model parameters gradient ; To ensure differential privacy, for each gradient Add Gaussian noise , where is the standard deviation of the noise, which can be set to 0.01, represents Gaussian noise with a mean of 0 and a variance of , is the identity matrix.

[0073] Privacy budget of differential privacy and The relationship of is usually estimated by the following formula: , where is the L2 sensitivity of the gradient, which can usually be set to 1, is the number of iterations, is a small probability, usually set to 0.00001 or less, and can be set to 0.01 to satisfy the privacy budget constraint of = 1.5.

[0074] The server collects the noisy gradients of all terminal nodes and updates the parameters. For the new parameter , there is , where is the learning rate, which controls the step size of parameter update, is the total number of terminal nodes, is the gradient of the L2 norm, which is used for normalization.

[0075] Based on the same inventive concept, as Figure 2 shown, the present invention also provides a real-time electricity charge settlement and intelligent deduction system for an intelligent electricity meter. The system includes: A data acquisition module, configured to obtain current waveform data continuously sampled on a power line and generate an original electrical parameter sequence at a frequency of not less than 100 sampling points per second; A feature vector generation module, configured to receive the original electrical parameter sequence and generate a feature vector carrying waveform fingerprints by using differential exponential smoothing processing on a TSA chip based on the sampling timestamps of the original electrical parameter sequence; A device load classification module, configured to output a device load type classification result through a lightweight convolutional neural network according to the spectral distribution characteristics of the feature vector; A rate parameter generation module, configured to receive the load type classification result and generate dynamic rate parameters in combination with the power grid peak shaving demand index; A billing statement generation module, configured to receive the dynamic rate parameters and generate a mixed billing statement including a transient overload coefficient by weighting the user's historical credit score; A transaction message generation module, configured to parse the hash value of the hybrid charging bill and embed the blockchain timestamp to form a cross-chain transaction message; A smart contract generation module, configured to, after detecting the cross-chain transaction message, parse the status of the pre-deposited amount in the account to activate a three-stage settlement smart contract; An abnormal fluctuation mark generation module, configured to capture the load fluctuation data after deduction according to the three-stage settlement smart contract and generate an abnormal fluctuation mark by comparing with the device shutdown characteristic curve; A feedback record module, configured to receive the abnormal fluctuation mark, update the parameters of the federated learning model and write them into the tamper-proof ledger to complete the closed-loop processing.

[0076] It should be noted that the electrical connections between the above-mentioned various units do not necessarily represent direct connections of the lines. Indirect connection methods, as long as they can achieve the purpose of the present invention, are applicable to the embodiments of the present invention. The above are only exemplary embodiments of the present invention, and the scope of the present invention cannot be limited thereby.

[0077] That is, any equivalent changes and modifications made in accordance with the teachings of the present invention still fall within the scope covered by the present invention. Those skilled in the art will easily think of other embodiments of the present invention after considering the specification and the disclosure of the practical truth. This application aims to cover any variations, uses, or adaptive changes of the present invention, and these variations, uses, or adaptive changes follow the general principles of the present invention and include the common general knowledge or conventional technical means in the technical field not recorded in the present invention.

Claims

1. A method for real-time settlement of electricity charges and intelligent deduction of an intelligent electricity meter, characterized in that The method includes: Obtain the current waveform data sampled continuously on the power line, and generate an original electrical parameter sequence at a frequency of no less than 100 sampling points per second; Receive the original electrical parameter sequence, and generate a feature vector carrying waveform fingerprints by using differential exponential smoothing processing on the TSA chip based on the sampling timestamps of the original electrical parameter sequence; According to the spectral distribution characteristics of the feature vector, output the classification result of the device load type through a lightweight convolutional neural network; Receive the load type classification result, and generate dynamic rate parameters in combination with the grid peak shaving demand index; Receive the dynamic rate parameters, and generate a mixed billing bill containing a transient overload coefficient by weighting the user's historical credit score; Analyze the hash value of the mixed billing bill, and embed the blockchain timestamp to form a cross-chain transaction message; After detecting the cross-chain transaction message, analyze the status of the account pre-deposited amount to activate the three-stage settlement smart contract; According to the three-stage settlement smart contract, capture the load fluctuation data after deduction, and generate an abnormal fluctuation mark by comparing with the device shutdown characteristic curve; Receive the abnormal fluctuation mark, update the parameters of the federated learning model, and write them into the tamper-proof ledger to complete the closed-loop processing.

2. The real-time electricity bill settlement and intelligent deduction method for an intelligent electric meter according to claim 1, characterized in that, The generating a feature vector carrying waveform fingerprints by using differential exponential smoothing processing on the TSA chip based on the sampling timestamps of the original electrical parameter sequence includes: Receive the original electrical parameter sequence, extract the differential value of adjacent point power, and compress it into a Huffman dictionary; Analyze the Huffman dictionary, and superimpose the voltage phase mutation amount to generate a feature digest with a check bit; Perform non-linear filtering based on the feature digest, and output the denoised feature vector and irreversibly restore the original waveform.

3. A method for real-time electricity bill settlement and intelligent deduction of an intelligent electric meter according to claim 1, characterized in that, The outputting the classification result of the device load type through a lightweight convolutional neural network according to the spectral distribution characteristics of the feature vector includes: Preprocess the Fourier transform result of the feature vector to generate a device fingerprint mapping dictionary; Receive the real-time current harmonic distribution data, and traverse the mapping dictionary to output the device identification result; Analyze the unrecognized power curve pattern, and write it into the knowledge base to be optimized.

4. A method for real-time settlement of electricity charges and intelligent deduction of an intelligent electricity meter according to claim 1, characterized in that, The generating dynamic rate parameters in combination with the grid peak shaving demand index includes: Analyze the JSON data of the grid peak shaving demand index to generate a basic rate matrix; According to the credit adjustment coefficient generated based on the user's historical electricity consumption behavior, correct the weight distribution of the basic rate matrix; Based on the loss compensation factor of the real-time temperature of the distribution transformer, correct the weight distribution, and output the final dynamic rate parameters.

5. A real-time electricity bill settlement and intelligent deduction method for an intelligent electricity meter according to claim 1, characterized in that, The analyzing the hash value of the mixed billing bill and embedding the blockchain timestamp to form a cross-chain transaction message includes: Analyze the JSON structure of the mixed billing bill, and separate the business logic data flow and the fund flow; Unidirectionally push the separated business logic data flow to the consortium chain node for pre-confirmation; After receiving the pre-confirmation result, permanently write the fund flow hash value into the genesis block of the transaction chain.

6. The real-time electricity bill settlement and intelligent deduction method for an intelligent electric meter according to claim 1, characterized in that, The analyzing the status of the account pre-deposited amount to activate the three-stage settlement smart contract includes: When the account pre-deposited amount is greater than or equal to the preset first threshold, activate the full deduction module and generate a unique deduction voucher; When the pre - deposited amount in the account is less than the first threshold and greater than or equal to the preset second threshold, a phased pre - deduction agreement is generated and the fund pool is locked; When the pre - deposited amount in the account is less than the second threshold, a relay control sequence is triggered and the electricity - using permission is frozen.

7. A method for real-time electricity bill settlement and intelligent deduction of an intelligent electricity meter according to claim 6, characterized in that The triggering of the relay control sequence and the freezing of the electricity - using permission include: Receiving the grid load capacity data within the geographical fence and calculating the preset allowable power gradient; Generating a device load - reduction curve based on the preset allowable power gradient and outputting a relay control instruction queue; Burning the relay control instruction queue into the meter firmware and erasing the original control code.

8. A method for real-time electricity bill settlement and intelligent deduction of an intelligent electricity meter according to claim 1, characterized in that, Generating an abnormal fluctuation mark by comparing with the device shutdown characteristic curve includes: Capturing the post - deduction load fluctuation data and extracting the power decay slope; Comparing the power decay slope with the reference value in the device characteristic knowledge base to generate a deviation coefficient vector; Analyzing the line impedance mutation parameters in the deviation coefficient vector and constructing an abnormal score matrix associated with multi - source data; When the abnormal score matrix exceeds the preset safety threshold, the fund transfer of the associated transaction is locked and a manual review is initiated, and an inspection report containing the abnormal fluctuation mark is generated.

9. A method for real-time electricity bill settlement and intelligent deduction of an intelligent electricity meter according to claim 8, characterized in that, The updating of the federated learning model parameters and writing them into the tamper - proof ledger to complete the closed - loop process includes: Receiving the abnormal fluctuation mark and the corresponding inspection report, using the time period when the abnormal fluctuation mark occurs as an index, extracting the feature vector from the edge node to generate a data fingerprint, and outputting the global optimization requirement; According to the global optimization requirement, reconstructing the convolutional kernel attention weight matrix, and at the same time writing the old parameter hash value into the tamper - proof ledger; Compiling the updated weight matrix into a binary instruction set carrying the version number of the abnormal fluctuation mark and encrypting and transmitting it to the specified meter cluster.

10. A real-time electricity bill settlement and intelligent deduction system for an intelligent electricity meter, which applies the real-time electricity bill settlement and intelligent deduction method for an intelligent electricity meter described in any one of claims 1-9, and is characterized in that, The system includes: A data acquisition module, which is used to obtain the current waveform data continuously sampled on the power line and generate an original electrical parameter sequence at a frequency of no less than 100 sampling points per second; A feature vector generation module, which is used to receive the original electrical parameter sequence, and generate a feature vector carrying a waveform fingerprint by using differential exponential smoothing processing on the TSA chip based on the sampling timestamp of the original electrical parameter sequence; A device load classification module, which is used to output the device load type classification result through a lightweight convolutional neural network according to the spectral distribution characteristics of the feature vector; A rate parameter generation module, which is used to receive the load type classification result and generate dynamic rate parameters in combination with the grid peak - shaving demand index; A billing statement generation module, which is used to receive the dynamic rate parameters and generate a mixed billing statement with a transient overload coefficient by weighting the user's historical credit score; A transaction message generation module, which is used to parse the hash value of the mixed billing statement and form a cross - chain transaction message by embedding the blockchain timestamp; A smart contract generation module, which is used to activate a three - stage settlement smart contract by parsing the status of the pre - deposited amount in the account after monitoring the cross - chain transaction message; An abnormal fluctuation mark generation module, which is used to capture the post - deduction load fluctuation data according to the three - stage settlement smart contract and generate an abnormal fluctuation mark by comparing with the device shutdown characteristic curve; A feedback record module, which is used to receive the abnormal fluctuation mark, update the federated learning model parameters and write them into the tamper - proof ledger to complete the closed - loop process.

Images