Blockchain-based green power consumption and storage data processing system

Abstract

The application discloses a green power consumption and storage data processing system based on a blockchain, relates to the technical field of data processing systems, and comprises an edge preprocessing module, a blockchain storage verification module, an energy storage health management module, a multi-scenario prediction module, a cross-chain permission management module and a federal collaborative processing module, wherein each module realizes data interaction through a distributed node network; the blockchain storage verification module comprises a lightweight consensus unit and a data chaining unit. The lightweight consensus unit of the blockchain storage verification module screens edge nodes as verification nodes, limits the verification of only adjacent edge node data, and combines standardized packaging and distributed storage of the data chaining unit, so that efficient verification and reliable storage of power data are realized. Compared with the prior art, the power data chaining efficiency and storage security can be improved, and thus the problems of heavy node verification burden and non-uniform data storage format in traditional blockchain storage can be solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing systems, in particular to a green power consumption and storage data processing system based on a blockchain. BACKGROUND

[0002] The power consumption and storage data processing system is a core tool of the power system, which is used for full life cycle management of power consumption and energy storage operation data, has data collection, cleaning and integration, analysis and mining and visualization presentation functions, can provide support for power dispatch, user power optimization and energy storage maintenance, and ultimately improve the stability of the power grid and reduce energy waste, and help new energy consumption.

[0003] At present, for the power consumption and storage data processing system, there are problems of low data storage verification efficiency, heavy node verification burden and non-uniform data storage format, which leads to slow power data chaining, poor security, affects the accuracy of power grid dispatch, the effectiveness of user power optimization and the scientificity of energy storage maintenance, and restricts the improvement of new energy consumption efficiency and power grid stability.

[0004] Therefore, the present application proposes a green power consumption and storage data processing system based on a blockchain to solve the above problems. SUMMARY

[0005] The main purpose of the present application is to provide a green power consumption and storage data processing system based on a blockchain to solve the problems raised in the background.

[0006] To achieve the above purpose, the technical scheme adopted by the present application is as follows: a green power consumption and storage data processing system based on a blockchain, comprising an edge preprocessing module, a blockchain storage verification module, an energy storage health management module, a multi-scenario prediction module, a cross-chain permission management module and a federal collaborative processing module, each module realizes data interaction through a distributed node network;

[0007] The edge preprocessing module is used for preprocessing the original power data generated by the green power terminal, including a data cleaning unit and a data compression unit;

[0008] The blockchain storage verification module is used for verifying and storing the preprocessed power data in a distributed manner, generating a green power traceability certificate, including a lightweight consensus unit and a data chaining unit;

[0009] The energy storage health management module is used for monitoring and residual value assessment of the health status of green power energy storage equipment, providing data support for energy storage asset accounting, including a health data acquisition unit and a residual value calculation unit;

[0010] The multi-scenario prediction module is used for dynamic prediction of green power output and power load, and includes a scenario identification unit and a prediction model switching unit.

[0011] The cross-chain permission management module is used for realizing permission control and traceability of green power cross-chain data, and includes a dynamic permission control unit and a cross-chain traceability unit.

[0012] The federal collaborative processing module is used for realizing collaborative analysis of multi-agent green power encrypted data, and includes a local model training unit and a global model aggregation unit.

[0013] Preferably, the data cleaning unit of the edge preprocessing module includes a power data receiving subunit and an outlier elimination subunit.

[0014] The power data receiving subunit is used for receiving voltage and current raw data output by a green power terminal, and the outlier elimination subunit uses a sliding window algorithm to identify and eliminate transient fault data that exceeds the normal range.

[0015] The data compression unit uses a difference encoding technology to compress the volume of the cleaned power data, and obtains preprocessed data.

[0016] Preferably, the lightweight consensus unit of the blockchain storage verification module selects edge nodes from a blockchain network as verification nodes, and limits each verification node to verify only the power data output by adjacent edge nodes.

[0017] The data on-chain unit includes a data packaging subunit and an on-chain storage subunit.

[0018] The data packaging subunit standardizes and packages the preprocessed power data, and the on-chain storage subunit uploads the packaged data to a blockchain network for distributed storage.

[0019] Preferably, the health data acquisition unit of the energy storage health management module includes a sensor group and a data encryption subunit.

[0020] The sensor group is used for acquiring index data of energy storage equipment, and the data encryption subunit uploads the encrypted index data to a blockchain after encryption processing by using an encryption algorithm.

[0021] The residual value calculation unit includes a coefficient determination subunit and a calculation subunit.

[0022] The coefficient determination subunit determines a health degree coefficient, a use time length coefficient, and an environmental adaptation coefficient of the energy storage equipment, and the calculation subunit calculates the residual value of the energy storage equipment based on the determined coefficients.

[0023] Preferably, the sensor group comprises a voltage sensor, a temperature sensor, a current sensor and a capacity detection sensor;

[0024] The voltage sensor is used to collect output voltage data of the energy storage device, the temperature sensor is used to collect internal temperature data of the energy storage device, the current sensor is used to collect charge and discharge current data of the energy storage device, and the capacity detection sensor is used to collect residual capacity data of the energy storage device to calculate the capacity attenuation rate.

[0025] Preferably, the scene recognition unit of the multi-scene prediction module comprises a scene data receiving subunit and a scene judgment subunit.

[0026] The scene data receiving subunit receives green power historical load data, temperature and humidity data, weather warning data and regional power consumption event information, and the scene judgment subunit judges the current green power prediction scene based on the received data.

[0027] The prediction model switching unit comprises a model calling subunit and a model calibration subunit.

[0028] The model calling subunit calls the corresponding prediction model according to the judged scene, and the model calibration subunit compares the prediction data and the actual data periodically, and calibrates the parameters of the prediction model.

[0029] Preferably, the dynamic permission control unit of the cross-chain permission management module comprises a permission division subunit and a permission management subunit.

[0030] The permission division subunit divides the data access permission into three levels of viewing permission, using permission and modifying permission, and the permission management subunit assigns the corresponding permission based on the user type, and recycles the assigned permission when the preset condition is triggered.

[0031] The cross-chain traceability unit comprises an identification generation subunit and a traceability query subunit.

[0032] The identification generation subunit generates a unique identification for each piece of cross-chain data, which contains source chain ID, original data hash, cross-chain time, receiving chain ID and usage record, and the traceability query subunit provides identification query function to obtain data flow path.

[0033] Preferably, the local model training unit of the federal collaborative processing module comprises a local encrypted data acquisition subunit and a sub-model training subunit.

[0034] The local encrypted data acquisition subunit acquires locally stored green power encrypted data, and the sub-model training subunit trains a green power prediction sub-model locally using the acquired encrypted data.

[0035] Preferably, the global model aggregation unit of the federal collaborative processing module comprises a parameter receiving subunit, a model aggregation subunit and a contribution evaluation subunit;

[0036] The parameter receiving subunit receives the sub-model parameters uploaded by each subject;

[0037] The model aggregation subunit aggregates the received sub-model parameters by weighting according to the proportion of the data amount of each subject, to generate a global prediction model.

[0038] Preferably, the contribution evaluation subunit calculates the data contribution of each subject based on the influence of the uploaded sub-model parameters on the accuracy of the global prediction model, and uploads the contribution result to the blockchain.

[0039] The present application has the following beneficial effects:

[0040] 1. The lightweight consensus unit of the blockchain storage verification module of the present application screens edge nodes as verification nodes, limits the verification of only adjacent edge node data, and realizes efficient verification and credible storage of power data by combining the standardized packaging and distributed storage of the data chaining unit, which can improve the efficiency of power data chaining and the security of data storage compared with the prior art, thereby solving the problems of heavy node verification burden and non-uniform data storage format in traditional blockchain storage.

[0041] 2. The local model training unit of the federal collaborative processing module trains sub-models locally, and the global model aggregation unit aggregates sub-model parameters by weighting according to the proportion of data amount and evaluates the contribution of each subject, which realizes model collaborative optimization under multi-subject data privacy protection, and can improve the security of multi-subject collaborative analysis and the accuracy of the model compared with the prior art, thereby solving the problems of high privacy leakage risk and unbalanced collaborative model optimization when sharing data among multiple subjects.

[0042] 3. The present application sets up an edge preprocessing module, a blockchain storage verification module, an energy storage health management module, a multi-scenario prediction module, a cross-chain permission management module and a federal collaborative processing module, and each module realizes data interaction through a distributed node network, which realizes the whole-process closed-loop management of green power data from collection and preprocessing to storage, analysis, prediction and cross-subject collaboration, and can improve the coherence and integrity of green power data processing compared with the prior art, thereby solving the problems of fragmented green power data processing links and low data flow efficiency in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 The figure is a schematic diagram of the overall system architecture of the present application;

[0044] Figure 2 The figure is a schematic diagram of the edge preprocessing module architecture of the present application;

[0045] Figure 3 A block chain storage verification module architecture diagram of the application is shown in the figure.

[0046] Figure 4 A storage health management module architecture diagram of the application is shown in the figure.

[0047] Figure 5 A multi-scenario prediction module architecture diagram of the application is shown in the figure.

[0048] Figure 6 A cross-chain permission management module architecture diagram of the application is shown in the figure.

[0049] Figure 7 A federal collaborative processing module architecture diagram of the application is shown in the figure. DETAILED DESCRIPTION

[0050] In order to make the technical means, creative features, purposes and effects of the application easy to understand, the application will be further described below in conjunction with specific embodiments.

[0051] Embodiment one, please refer to Figure 1 and Figure 2 The green power consumption and storage data processing system based on block chain includes an edge preprocessing module, a block chain storage verification module, a storage health management module, a multi-scenario prediction module, a cross-chain permission management module and a federal collaborative processing module, and each module realizes data interaction through a distributed node network.

[0052] The edge preprocessing module is used for preprocessing the original power data generated by the green power terminal, including a data cleaning unit and a data compression unit.

[0053] The data cleaning unit of the edge preprocessing module includes a power data receiving subunit and an outlier elimination subunit.

[0054] The power data receiving subunit is used for receiving the voltage and current raw data output by the green power terminal, and the outlier elimination subunit uses a sliding window algorithm to identify and eliminate transient fault data that exceeds the normal range.

[0055] The data compression unit uses differential encoding technology to compress the volume of the cleaned power data and obtain preprocessed data.

[0056] Further, the power data receiving subunit can communicate with various types of green power terminals, including but not limited to photovoltaic power generation terminals, wind power generation terminals, and hydroelectric power generation terminals, etc. The power data receiving subunit receives voltage data and current data output by the green power terminal in real time through the communication protocol, wherein the collection range of the voltage data covers 0V to 1000V, and the collection range of the current data covers 0A to 500A, which can meet the data source requirements of green power terminals of different power levels.

[0057] During data receiving, the power data receiving subunit also performs preliminary format regularization on the received data, converts non-standardized data output by different terminals into binary data format uniformly recognized by the system, and provides a unified data basis for subsequent processing of the outlier elimination subunit.

[0058] The outlier elimination subunit uses a sliding window algorithm to identify and eliminate abnormal data. By dynamically constructing a data window, the power data in the window is analyzed in real time to accurately identify transient fault data that exceeds the normal range.

[0059] The outlier elimination subunit first sets the normal fluctuation range of voltage and current data according to the historical operation data of the green power terminal and industry standard parameters, wherein the normal voltage range is set to ±5% of the rated output voltage, and the normal current range is set to ±8% of the rated output current. The sliding window slides on the real-time received power data stream with a time interval of 1 second, and the length of each window is set to contain 5 consecutive sampling data points.

[0060] During the sliding of each window, the outlier elimination subunit calculates the average value and standard deviation of the data in the window, and marks the data deviating from the average value by more than 3 times the standard deviation as suspected abnormal data. At the same time, if a data point exceeds the pre-set normal fluctuation range of voltage or current, it will also be marked as suspected abnormal data.

[0061] For the suspected abnormal data marked, the outlier elimination subunit further verifies the data trend in the previous and subsequent adjacent windows, which do not overlap in time to ensure clear verification logic. If the data before and after the suspected abnormal data is within the normal range and the data change conforms to the normal operation rules of the green power terminal, the data is determined to be transient fault data and is eliminated. If the data before and after the suspected abnormal data also has abnormal fluctuation, it is determined that there may be equipment failure, and the data is retained and a fault warning signal is sent to the system operation terminal to remind the staff to check the equipment.

[0062] The double verification mechanism of the outlier elimination subunit can effectively eliminate instantaneous failure data caused by electromagnetic interference, sensor instantaneous error and other factors, and can also avoid mistakenly deleting abnormal data caused by equipment failure, thereby ensuring the accuracy and reliability of data cleaning.

[0063] The data compression unit uses differential encoding technology to achieve data compression. Compared with traditional compression technologies such as Huffman encoding and LZW encoding, differential encoding technology has higher compression efficiency when processing continuously changing power data, and the decompression process is simple and can quickly restore the original data.

[0064] First, the data compression unit obtains the voltage and current time series data after data cleaning. The data sequence is sampled at an interval of 1 second, and includes the voltage value and current value at each sampling time. During compression processing, the data compression unit selects the first data in the data sequence as a reference baseline value, and then calculates the difference between each subsequent data and the previous data, i.e., the difference value.

[0065] Since the voltage and current data output by the green power terminal presents a continuously smooth change trend under normal operating conditions, the difference between adjacent data is small, so the values in the difference value sequence are mostly in a small range. Compared with the original data sequence, the data distribution of the difference value sequence is more concentrated, which can effectively reduce the storage bit number of the data.

[0066] After obtaining the difference value sequence, the data compression unit optimizes the encoding of the difference value. According to the size range of the difference value, a variable length encoding method is used to encode it. For difference values with small absolute values, a shorter encoding length is used, and for difference values with large absolute values, a longer encoding length is used, to further reduce the overall volume of the data.

[0067] Embodiment two, please refer to Figure 3 The blockchain-based green power consumption and storage data processing system is shown. The blockchain storage verification module is used to verify and distribute the pre-processed power data, and generate green power traceability credentials, including a lightweight consensus unit and a data chaining unit.

[0068] The lightweight consensus unit of the blockchain storage verification module selects edge nodes from the blockchain network as verification nodes, and limits each verification node to only verify the power data output by adjacent edge nodes.

[0069] The data chaining unit includes a data packaging subunit and a chain storage subunit.

[0070] The data packaging subunit standardizes and packages the pre-processed power data, and the chain storage subunit uploads the packaged data to the blockchain network for distributed storage.

[0071] Further, the lightweight consensus unit first selects edge nodes from the blockchain network as verification nodes, and the selection process follows specific evaluation indicators to ensure that the selected verification nodes have reliable verification capabilities. The specific selection indicators include the hardware performance of the node, the stability of the network connection, the historical verification accuracy rate, and the distance between the node's geographic location and the green power terminal. The hardware performance requires the node to have at least a 4-core CPU, 8GB of memory, and 100GB of available storage space to ensure smooth operation of the data verification process. The network connection stability requires the node's average network delay to be no more than 50 milliseconds and the packet loss rate to be less than 1%, avoiding the impact of network problems on verification efficiency. The historical verification accuracy rate needs to be higher than 98% to ensure that the node can accurately identify data anomalies. The distance between the node's geographic location and the green power terminal should not exceed 50 kilometers to reduce the delay and loss in the data transmission process.

[0072] Through multi-dimensional indicator evaluation, the required edge nodes are selected from the blockchain network to form a verification node pool. The size of the verification node pool is dynamically adjusted according to the number of green power terminals, with 1 verification node corresponding to every 10 green power terminals to ensure that verification resources are reasonably allocated.

[0073] After completing the verification node selection, the lightweight consensus unit will limit each verification node to only verify the power data output by adjacent edge nodes, building a regional verification network. The determination of adjacent edge nodes is based on the topology of the nodes in the blockchain network. The blockchain network is divided into multiple sub-networks according to geographic regions, and the edge nodes within each sub-network are adjacent to each other. The division of sub-networks is based on administrative divisions, and the coverage range of each sub-network is controlled within 100 square kilometers to ensure that the data transmission distance between adjacent nodes is short.

[0074] After receiving the preprocessed power data transmitted by adjacent edge nodes, each verification node will verify it from three dimensions: data integrity, data consistency, and data authenticity. Data integrity verification is done by comparing the byte length of the data with the preset standard length. If they are consistent, the data is determined to be complete. If they are not consistent, the data is required to be retransmitted. Data consistency verification is done by comparing the data with the historical transmission data of the adjacent node. If the data change amplitude is within the preset threshold, the voltage change amplitude threshold is ±3%, and the current change amplitude threshold is ±5%, the data is determined to be consistent. Data authenticity verification is done by checking the digital signature of the data. Each edge node will attach a unique digital signature when transmitting data. The verification node reads the digital signature and matches it with the node's public key. If the match is successful, the data is determined to be authentic.

[0075] When the verification node completes the verification, the verification result is generated and fed back to the adjacent edge node, and the verification result is synchronized to the shared ledger of the blockchain network. If a data is verified by 3 or more adjacent verification nodes, it is determined that the data passes the consensus verification and can enter the subsequent data chaining link.

[0076] The data encapsulation subunit in the data chaining unit is responsible for standardizing and encapsulating the preprocessed power data, solving the problem of inconsistent output data formats of different edge nodes, and ensuring that the data can be normally transmitted and stored in the blockchain network. The encapsulation process follows the preset standardized data format, which includes data header, data body and data tail.

[0077] The data header includes data identification information, including green power terminal number, data collection time, and data type. The green power terminal number is encoded in 16-bit binary, ensuring that each terminal has a unique identifier. The data collection time is accurate to the millisecond level and uses UTC time format to avoid time record confusion due to time zone differences. The data type is distinguished by 2-bit binary coding, with 01 representing voltage data and 10 representing current data.

[0078] The data body contains specific power data values, with voltage data accurate to 0.01V and current data accurate to 0.01A, stored in floating-point data format to ensure data accuracy meets subsequent application requirements. The data tail contains data verification information, with a 16-bit check code generated by the cyclic redundancy check algorithm to detect data errors during data storage and transmission.

[0079] After completing the data format encapsulation, the data encapsulation subunit encrypts the encapsulated data using a symmetric encryption algorithm. Each data chaining unit is equipped with a dedicated encryption key, and the encrypted data can effectively prevent theft or tampering during transmission, ensuring data security.

[0080] The on-chain storage subunit is responsible for uploading and storing the encapsulated and encrypted power data to the blockchain network for distributed storage. First, the encapsulated data is fragmented, and the data is divided into multiple data fragments according to its size, with each data fragment set to 1MB to avoid reduced storage efficiency due to large single data.

[0081] After completing the data fragmentation, the on-chain storage subunit uses distributed hash table technology to distribute each data fragment to different nodes in the blockchain network for storage. Each data fragment is stored on 5 different nodes, achieving multi-copy storage of data.

[0082] The selection of the node is based on the storage remaining space and the network connection state of the node, and the node with a storage remaining space greater than 1 GB and a stable network connection is preferentially selected to ensure that the data can be stored for a long time. After the data shard storage is completed, the on-chain storage subunit records the storage location information of each data shard to generate a data index table, which includes data shard number, storage node address, storage time and other information. Through the data index table, the storage location of each data shard can be quickly located, which is convenient for subsequent data query and calling.

[0083] The on-chain storage subunit regularly checks the integrity of the stored data. Every 24 hours, a data verification request is sent to the storage node, and the storage node feeds back the verification code of the data. The on-chain storage subunit compares the feedback verification code with the original verification code. If they are consistent, it is determined that the data storage is complete. If they are inconsistent, the data recovery mechanism is started to obtain data from other nodes storing the data shard and store it again to ensure that data is not lost due to node failure.

[0084] Embodiment three, please refer to Figure 4 The green power consumption and storage data processing system based on the blockchain is shown in FIG. 1. The energy storage health management module is used to monitor the health status of the green power energy storage device and assess the residual value, providing data support for energy storage asset accounting, including a health data acquisition unit and a residual value calculation unit.

[0085] The health data acquisition unit of the energy storage health management module includes a sensor group and a data encryption subunit.

[0086] The sensor group is used to acquire index data of the energy storage device, and the data encryption subunit uploads the encrypted index data to the blockchain after encryption processing by an encryption algorithm.

[0087] The residual value calculation unit includes a coefficient determination subunit and a calculation subunit.

[0088] The coefficient determination subunit determines the health degree coefficient, the use time coefficient and the environmental adaptation coefficient of the energy storage device, and the calculation subunit calculates the residual value of the energy storage device based on the determined coefficients.

[0089] The sensor group includes a voltage sensor, a temperature sensor, a current sensor and a capacity detection sensor.

[0090] The voltage sensor is used to acquire output voltage data of the energy storage device, the temperature sensor is used to acquire internal temperature data of the energy storage device, the current sensor is used to acquire charge and discharge current data of the energy storage device, and the capacity detection sensor is used to acquire residual capacity data of the energy storage device to calculate the capacity attenuation rate.

[0091] Further, the sensor group in the health data acquisition unit includes four types of voltage sensor, temperature sensor, current sensor and capacity detection sensor. The voltage sensor is installed at the positive and negative output terminals of the energy storage device, adopts high-precision direct current voltage sensing technology, the acquisition range covers 0V to 1500V, the sampling frequency is set to 1 time / s, and the dynamic change of the output voltage of the energy storage device can be captured in real time. When the voltage fluctuation exceeds ±2% of the rated voltage, a warning signal is triggered in time.

[0092] The temperature sensor adopts a distributed arrangement, and one is installed at each of the key parts of the battery module, heat dissipation system and control module of the energy storage device. The high-precision thermocouple sensing technology is adopted, the measurement range covers-20℃ to 80℃, the measurement accuracy is controlled within ±0.5℃, the sampling frequency is set to 1 time / 30 seconds, and the temperature change of each part of the device is monitored in real time to avoid performance degradation or safety accidents caused by local overheating.

[0093] The current sensor is installed in the charge-discharge circuit of the energy storage device, adopts Hall current sensing technology, the acquisition range covers-500A to 500A, the negative sign represents the discharge current, and the positive sign represents the charging current. The sampling frequency is set to 1 time / s, which can accurately record the current change during the charging and discharging process of the device, and provide data basis for analyzing the charging and discharging efficiency and battery loss of the device.

[0094] The capacity detection sensor is integrated in the battery management system of the energy storage device, adopts the detection technology combining constant current discharge method and alternating current impedance method, detects the remaining capacity of the energy storage device once every 24 hours, and detects it once after each charge-discharge cycle. The detection accuracy is controlled within ±1%, the capacity attenuation rate is calculated by comparing the initial rated capacity and the current remaining capacity of the device, and the aging degree of the battery is intuitively reflected.

[0095] The data encryption subunit adopts a hybrid encryption mode combining asymmetric encryption algorithm and symmetric encryption algorithm. First, the original index data collected is encrypted by the symmetric encryption algorithm to generate an encrypted data block. The symmetric encryption key is randomly generated by the subunit and automatically updated once every 24 hours to ensure the security of the key.

[0096] Then, the symmetric encryption key is encrypted by the asymmetric encryption algorithm to generate a key ciphertext. The public key used for encryption is stored in the shared nodes of the blockchain network, and the private key used for decryption is saved by the exclusive key management unit of the energy storage health management module. The private key is stored in hardware encryption mode to prevent private key leakage.

[0097] After the encryption of data and keys is completed, the data encryption subunit packs the encrypted data block and key ciphertext into a data transmission package, uploads it to the blockchain storage verification module through a secure communication protocol, and stores it in the blockchain after the data verification is completed by the blockchain storage verification module, realizing the safe storage and traceability of data.

[0098] The coefficient determination subunit in the residual value calculation unit is responsible for determining the health degree coefficient, use time length coefficient and environment adaptation coefficient of the energy storage device. These three coefficients reflect the value loss of the device from three dimensions of device state, use time and external environment. The value range of each coefficient is 0 to 1. The closer the coefficient value is to 1, the smaller the value loss of the device and the higher the residual value.

[0099] The health degree coefficient is mainly calculated based on the device index data collected by the sensor group. The calculation process comprehensively considers four indexes: capacity attenuation rate, voltage stability, temperature fluctuation range and charge-discharge efficiency. The weight proportion of the capacity attenuation rate is 40%. When the capacity attenuation rate is ≤10%, the health degree coefficient score is 1.0. When the capacity attenuation rate is between 10% and 20%, the score linearly decreases to 0.8 with the increase of the attenuation rate. When the capacity attenuation rate exceeds 20%, the score linearly decreases to 0.5 with the increase of the attenuation rate.

[0100] The weight proportion of voltage stability is 20%. The proportion of the number of times that the voltage fluctuation exceeds the normal range within the past 72 hours to the total sampling number is calculated. When the proportion is ≤5%, the score is 1.0. When the proportion is between 5% and 10%, the score linearly decreases to 0.8. When the proportion exceeds 10%, the score linearly decreases to 0.6.

[0101] The weight proportion of temperature fluctuation range is 20%. The difference between the maximum temperature and the minimum temperature of each part of the device within the past 72 hours is calculated. When the difference is ≤10℃, the score is 1.0. When the difference is between 10℃ and 20℃, the score linearly decreases to 0.8. When the difference exceeds 20℃, the score linearly decreases to 0.6.

[0102] The weight proportion of charge-discharge efficiency is 20%. When the charge-discharge efficiency is ≥90%, the score is 1.0. When the charge-discharge efficiency is between 80% and 90%, the score linearly decreases to 0.8. When the charge-discharge efficiency is less than 80%, the score linearly decreases to 0.6. The scores of the four indexes are weighted and summed according to the weights to obtain the health degree coefficient.

[0103] The use time length coefficient is determined according to the ratio of the used time length of the energy storage device to the designed service life. The environment adaptation coefficient is determined according to four indexes of temperature and humidity, dust concentration and vibration intensity of the operating environment of the energy storage device. The weight proportion of each index is 33.33%.

[0104] In terms of temperature, the corresponding score is 1.0 when the ambient temperature is between 15℃ and 25℃, 0.8 when the temperature is between 5℃ and 15℃ or 25℃ and 35℃, and 0.6 when the temperature is below 5℃ or above 35℃.

[0105] In terms of humidity, the corresponding score is 1.0 when the ambient relative humidity is between 40% and 60%, 0.8 when the humidity is between 20% and 40% or 60% and 80%, and 0.6 when the humidity is below 20% or above 80%.

[0106] In terms of dust concentration, the corresponding score is 1.0 when the ambient dust concentration is ≤0.1mg / m 3 , 0.8 when the dust concentration is between 0.1mg / m 3 and 0.5mg / m 3 , and 0.6 when the dust concentration exceeds 0.5mg / m 3 .

[0107] In terms of vibration intensity, the corresponding score is 1.0 when the ambient vibration acceleration is ≤0.1g, 0.8 when the vibration acceleration is between 0.1g and 0.5g, and 0.6 when the vibration acceleration exceeds 0.5g. The weighted sum of the scores of the four indicators according to the weights is the environmental adaptation coefficient.

[0108] The calculation sub-unit determines the health degree coefficient using the time length coefficient and the environmental adaptation coefficient determined by the coefficient determination sub-unit, and calculates the current residual value of the energy storage device in combination with the initial purchase cost of the device.

[0109] After obtaining the initial purchase cost of the device, a residual value calculation model is constructed, and the calculation formula is: device residual value = initial purchase cost × health degree coefficient × time length coefficient × environmental adaptation coefficient.

[0110] During the calculation process, the calculation sub-unit will call the three coefficient values uploaded by the coefficient determination sub-unit and the initial purchase cost data from the blockchain, automatically substitute them into the formula for calculation, and the calculation result is accurate to two decimal places. At the same time, the calculation sub-unit will package the residual value calculation result, the coefficient values and the initial purchase cost data for each calculation into a residual value evaluation report, and upload it to the blockchain for storage and verification. Users can access the evaluation report at any time through the blockchain query interface, realizing the transparency and traceability of the residual value evaluation process.

[0111] Example four, please refer to Figure 5 : Green power consumption and storage data processing system based on blockchain, multi-scenario prediction module is used for dynamic prediction of green power output and electricity load, including scene recognition unit and prediction model switching unit;

[0112] The scene recognition unit of the multi-scene prediction module comprises a scene data receiving subunit and a scene determination subunit;

[0113] The scene data receiving subunit receives green power historical load data, temperature and humidity data, meteorological warning data, and regional power consumption event information, and the scene determination subunit determines the current green power prediction scene based on the received data;

[0114] The prediction model switching unit comprises a model calling subunit and a model calibration subunit;

[0115] The model calling subunit calls the corresponding prediction model according to the determined scene, and the model calibration subunit compares the predicted data and the actual data periodically to calibrate the parameters of the prediction model.

[0116] Further, the scene data receiving subunit receives four types of key data. The first type is the past three years of regional green power historical load data stored in the blockchain, with a sampling interval of fifteen minutes, and after de-duplication and smoothing processing to eliminate accidental fluctuations.

[0117] The second type is the temperature and humidity data obtained by the regional meteorological monitoring station, with a temperature collection range of minus thirty to forty-five degrees Celsius and a humidity collection range of 0% to 100%, a sampling frequency of once an hour, and data containing real-time values and twenty-four-hour prediction values.

[0118] The third type is the meteorological warning information pushed by the meteorological department, covering types such as heavy rain, gale, high temperature, and cold wave, including warning levels and warning duration, and the warning levels are divided into four levels: blue, yellow, orange, and red.

[0119] The fourth type is the regional power consumption event information reported by the regional power management department and related units, including large-scale industrial enterprise production plan adjustment, regional large-scale event holding, and residential power consumption policy change, etc., with the explanation of event influence period and influence load amplitude.

[0120] The scene determination subunit determines the current green power prediction scene using a hierarchical determination method, and divides it into six core scenes: regular weekday scene, regular weekend scene, holiday scene, extreme weather scene, large power consumption event scene, and policy adjustment scene. The determination process is divided into three levels. The first level preliminarily divides the scene according to the date type. If the date is a statutory holiday and a holiday including rest, it is preliminarily determined as a holiday scene. If it is a weekday or a weekend, it enters the second level determination.

[0121] The second level combines meteorological warning data to determine. If there is an orange or red meteorological warning and the warning duration covers more than 50% of the future twenty-four-hour prediction period, it is determined as an extreme weather scene. If there is only a blue or yellow warning or the warning duration is short, it enters the third level determination.

[0122] The third-level comprehensive regional power consumption event information and temperature and humidity data are determined. If there is a large power consumption event that is expected to affect the regional power consumption load by more than 10%, it is determined to be a large power consumption event scenario. The load impact threshold is derived from a preset scenario determination threshold library, and is linked with the correction coefficient of the subsequent large power consumption event prediction model. When the impact load amplitude is 10% to 20%, the model correction coefficient is set to 1.1. When the impact amplitude exceeds 20%, the correction coefficient is set to 1.2.

[0123] If there is a residential power consumption policy adjustment and the adjustment implementation time is within the prediction period, it is determined to be a policy adjustment scenario. If there is no special event as described above, and the temperature and humidity data are within 5 degrees Celsius of the normal temperature fluctuation and within 10% of the humidity fluctuation of the same period in the region, then according to the date type, it is finally determined to be a regular weekday scenario or a regular weekend scenario. The determination process relies on a preset scenario determination threshold library, which sets quantitative standards for each type of determination condition.

[0124] The model calling subunit of the prediction model switching unit has six prediction models built in, each corresponding to one of the six scenarios. The first regular weekday prediction model uses a long short-term memory network algorithm, focusing on learning the early peak, mid-peak, and late peak fluctuation characteristics of power consumption load during weekdays, as well as the linear law of photovoltaic output changes with sunlight. The model training data uses historical data from the same type of working days over the past two years, with a prediction step of fifteen minutes, and outputs the load and output prediction values every fifteen minutes in the next twenty-four hours.

[0125] The second regular weekend prediction model is based on a gradient boosting decision tree algorithm, taking into account the characteristics of delayed load fluctuation and gentle load fluctuation during the peak load of the weekend. During the training process, the weights of the afternoon and evening load data are strengthened to improve the prediction accuracy of these time periods.

[0126] The third holiday prediction model uses a time series decomposition algorithm to decompose the holiday load into trend, seasonal, and residual terms, separating the special power consumption mode of holidays from regular fluctuations. The training data includes load and output records of all holidays over the past five years.

[0127] The fourth extreme weather prediction model is based on a support vector regression algorithm, with the weights of meteorological warning data strengthened in the input data. For example, when a high temperature warning is issued, the air conditioning load prediction coefficient is increased, and when a strong wind warning is issued, the wind power output prediction threshold is adjusted. The built-in extreme weather load correction formula dynamically adjusts the prediction results according to the warning level. For example, when a red high temperature warning is issued, the basic load prediction value is increased by 15% to 20%.

[0128] The fifth large-scale electricity consumption event prediction model adopts a hybrid architecture of a basic model plus event correction. First, a basic load model is trained through historical data, and then the basic prediction result is corrected according to the electricity consumption scale and duration of the large-scale event. The correction coefficient is fitted through the historical impact data of similar events.

[0129] The sixth policy adjustment prediction model is based on a logistic regression algorithm, which analyzes the guiding effect of policy changes on electricity consumption behavior. In the training process, load comparison data before and after the implementation of the policy are introduced. After the scene judgment result is received by the model calling subunit, the corresponding model is loaded and initialized within ten seconds, the historical data related to the scene in the blockchain are automatically retrieved as model input, and the prediction calculation is started.

[0130] The model calibration subunit is responsible for periodically verifying and optimizing the prediction model accuracy to avoid increasing prediction errors due to data drift or scene changes. First, data comparison is performed. The subunit retrieves the prediction data output by the prediction model within the past twenty-four hours at a fixed time when the electricity load is relatively stable at two o'clock in the morning every day, and compares it point by point with the actual power data stored in the blockchain at the same period. The output data in the actual power data is collected and recorded by the green power terminal, and the load data is derived from the regional power grid metering data. After comparison, the prediction error is calculated. The error indicators include the mean absolute error and the root mean square error. The mean absolute error reflects the average deviation of the predicted value from the actual value, and the root mean square error reflects the influence degree of the extreme deviation.

[0131] Threshold determination is then performed. The subunit presets two types of error thresholds. When the mean absolute error is not more than 3% and the root mean square error is not more than 5%, it is determined that the model accuracy meets the standard and no adjustment is needed. When the mean absolute error is between 3% and 5% or the root mean square error is between 5% and 8%, it is determined that the model needs to be slightly calibrated. When the mean absolute error exceeds 5% or the root mean square error exceeds 8%, it is determined that the model needs to be severely calibrated.

[0132] Finally, parameter adjustment is performed. When slight calibration is needed, the subunit fine-tunes the weight coefficients of the model, such as adjusting the historical data weight in the early peak period of the regular working day model, and the adjustment range is controlled within 5% to avoid large fluctuations in the model. When severe calibration is needed, the latest data of the past three months are selected to incrementally train the model, and the core parameters of the model are updated, such as the number of hidden layer nodes of the long short-term memory network model and the kernel function parameters of the support vector regression model. After training, the model is tested to ensure that the mean absolute error and the root mean square error after calibration fall within the standard range.

[0133] Embodiment five, please refer to Figure 6 The blockchain-based green power consumption and storage data processing system is shown in FIG. 1. The cross-chain permission management module is used to realize the permission control and traceability of green power cross-chain data, including a dynamic permission control unit and a cross-chain traceability unit.

[0134] The dynamic permission control unit of the cross-chain permission management module includes a permission division subunit and a permission management subunit;

[0135] The permission division subunit divides the data access permission into three levels of viewing permission, use permission and modification permission, the permission management subunit allocates corresponding permissions based on user types, and recycles the allocated permissions when a preset condition is triggered;

[0136] The cross-chain traceability unit includes an identity generation subunit and a traceability query subunit;

[0137] The identity generation subunit generates a unique identity for each cross-chain data, which contains source chain ID, original data hash, cross-chain time, receiving chain ID and use record, and the traceability query subunit provides identity query function to obtain data flow path.

[0138] Further, the permission division subunit in the dynamic permission control unit divides the data access permission into three levels of viewing permission, use permission and modification permission, and each level of permission range is clear and does not overlap, the viewing permission only allows users to browse the basic content of cross-chain data, such as green power output statistical data and electricity load summary information, cannot download or copy data, and sensitive fields in the data such as user privacy information and device core parameters will be desensitized.

[0139] The use permission allows users to use cross-chain data in specific scenarios based on the viewing permission, such as importing cross-chain power data into dispatching system to assist decision-making, and using it for carbon emission accounting report preparation, but needs to record data use process and synchronize to blockchain.

[0140] The modification permission is only open to system operation and maintenance personnel and core management users, allowing to correct error information in cross-chain data such as data entry deviation and format exception, and each modification needs to submit modification application explaining modification reason and basis, and can be executed after being approved, and modification record will be retained throughout the process.

[0141] The permission management subunit allocates corresponding permissions based on user types, and recycles the allocated permissions when a preset condition is triggered, the user types are divided into four categories of ordinary users, operation and maintenance personnel, management users and third-party institutions, ordinary users are mostly power users and are only allocated viewing permission, operation and maintenance personnel are responsible for data maintenance and are allocated viewing and modification permissions, management users are allocated all three types of permissions for overall data management, and third-party institutions such as carbon emission verification institutions are allocated viewing or use permissions according to cooperation needs.

[0142] The user identity information and the use period are bound when the authority is allocated. The use period is set according to business requirements. The authority period of an ordinary user is usually 30 days. The authority period of a third-party institution is consistent with the cooperation period and is no more than 180 days. The preset recovery conditions include three types of authority expiration, user identity change and data use scenario termination.

[0143] The identification generation subunit in the cross-chain traceability unit generates a unique identification for each piece of cross-chain data. The identification contains five core information: source chain ID, original data hash, cross-chain time, receiving chain ID and use record. The source chain ID is used to identify the initial blockchain where the data is located. It uses 8-character encoding. Different blockchains correspond to unique codes. The original data hash is calculated by the SHA-256 algorithm on the original data before cross-chain to ensure that the data has not been tampered with during the cross-chain process. The cross-chain time is accurate to the second and records the specific time when the data is transferred out of the source chain. The receiving chain ID identifies the target blockchain where the data is transferred into. The encoding rules are consistent with the source chain ID. The use record updates the use of the data in the receiving chain in real time, such as which users view it and which scenarios it is used for.

[0144] After the identification is generated, it will be bound with the cross-chain data and stored in the source chain, the receiving chain and the system main chain to form a triple backup to prevent the identification from being lost.

[0145] The traceability query subunit provides identification query function. Users can obtain the data flow path by inputting the unique identification. During the query process, the traceability query subunit will retrieve the corresponding information of the identification from the main chain and display the data flow nodes in chronological order, including the time when the data is transferred out of the source chain, the time when it is transferred into the receiving chain, the users in each link and the data use scenarios, etc. If there is an abnormality in the data flow process, such as mismatch between the identification and the data hash or missing flow nodes, the subunit will issue a warning prompt and locate the abnormal link.

[0146] Embodiment six, please refer to Figure 7 The federated collaborative processing module is used to realize collaborative analysis of multi-agent green power encrypted data, including a local model training unit and a global model aggregation unit.

[0147] The local model training unit of the federated collaborative processing module includes a local encrypted data acquisition subunit and a sub-model training subunit.

[0148] The local encrypted data acquisition subunit acquires the locally stored green power encrypted data. The sub-model training subunit trains a green power prediction sub-model locally using the acquired encrypted data.

[0149] The global model aggregation unit of the federated collaborative processing module includes a parameter receiving subunit, a model aggregation subunit and a contribution evaluation subunit.

[0150] The parameter receiving subunit receives the sub-model parameters uploaded by each subject;

[0151] The model aggregation subunit aggregates the received sub-model parameters by weighting according to the proportion of each subject's data amount, to generate a global prediction model.

[0152] The contribution degree evaluation subunit calculates the data contribution degree of each subject based on the influence of the uploaded sub-model parameters on the accuracy of the global prediction model, and uploads the contribution degree result to the blockchain.

[0153] Further, the local encrypted data acquisition subunit in the local model training unit is responsible for acquiring the green power encrypted data stored locally by each subject, supporting the local databases of different subjects such as photovoltaic power stations, wind farms, and regional power grids. The local encrypted data acquisition subunit uses secure multi-party computation technology in privacy computing to establish a connection with the local database through a pre-set encrypted communication protocol, and can read the feature dimension and data amount information of the encrypted data without decryption, avoiding data theft during reading.

[0154] Before data acquisition, the identity and authority of the subject will be verified, and only subjects that have been authenticated by the cross-chain authority management module are allowed to participate in data calling. Each data acquisition operation generates an operation log, recording information such as acquisition time, data type, and data amount, and uploading it to the blockchain. The acquired encrypted data includes green power output data, equipment operation data, and electricity load data, with a data time span of no less than 6 months. The sampling interval is set according to the data type, with a 15-minute sampling interval for output data and load data, and a 1-hour sampling interval for equipment operation data, ensuring the time continuity and integrity of the data and providing sufficient samples for sub-model training.

[0155] The sub-model training subunit uses the acquired encrypted data to train green power prediction sub-models locally, using the local training framework in federated learning, supporting multiple algorithms such as linear regression, random forest, and long short-term memory network. Each subject can choose an adaptive algorithm according to its own data characteristics.

[0156] Before training, the encrypted data is preprocessed, including missing value filling and outlier processing. Missing values are filled with the mean value of adjacent time data, and outliers are identified by the 3σ principle and replaced with the historical mean value of the period. The preprocessing process is completed locally and does not involve data transmission. The training process uses homomorphic encryption technology to directly encrypt the gradient parameters in the model training process, preventing the original data from being recovered through gradient backpropagation. At the same time, gradient calculation can be completed in an encrypted state without decryption, enabling parameter updating and ensuring data privacy and security.

[0157] The parameter receiving subunit in the global model aggregation unit is responsible for receiving the sub-model parameters uploaded by each subject, adopts a distributed receiving architecture, supports receiving more than 100 subject parameter upload requests at the same time, the upload link adopts the TLS1.3 encryption protocol, prevents the parameters from being tampered with or intercepted during transmission, and checks the integrity of the parameter ciphertext before receiving, compares the hash value of the parameter ciphertext with the hash value uploaded by the subject, and receives if consistent, and requires re-uploading if inconsistent.

[0158] The received parameters are temporarily stored in an encrypted cache area, the cache area adopts a hardware encryption storage mode, only allows the model aggregation subunit to access, and the parameter storage time in the cache area is not more than 24 hours, and the parameters are deleted immediately after aggregation is completed, so that the security risk caused by long-term retention of the parameters is avoided.

[0159] The model aggregation subunit performs weighted aggregation on the received sub-model parameters according to the proportion of the data amount of each subject, generates a global prediction model, the contribution degree evaluation subunit calculates the data contribution degree of each subject based on the influence of the uploaded sub-model parameters on the accuracy of the global prediction model, adopts the accuracy improvement method to evaluate the contribution degree, first calculates the accuracy of the temporary model generated by removing the sub-model parameters of a subject, and then compares it with the original global model accuracy, the greater the accuracy difference, the greater the contribution of the subject parameters to the global model.

[0160] The above shows and describes the basic principles and main features of the present application and the advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above-mentioned embodiments, the above-mentioned embodiments and descriptions in the specification are only to illustrate the principles of the present application, and various changes and improvements can be made to the present application without departing from the spirit and scope of the present application, and these changes and improvements all fall within the scope of the present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A blockchain-based green power consumption and storage data processing system, characterized in that, The edge preprocessing module, the blockchain storage verification module, the energy storage health management module, the multi-scenario prediction module, the cross-chain permission management module and the federal collaborative processing module are connected through a distributed node network to realize data interaction. The edge preprocessing module is configured to preprocess original power data generated by a green power terminal, including a data cleaning unit and a data compression unit. The blockchain storage verification module is configured to verify and store the preprocessed power data in a distributed manner, and generate a green power traceability certificate, including a lightweight consensus unit and a data chaining unit. The energy storage health management module is configured to monitor the health status of green power energy storage equipment and assess the residual value, providing data support for energy storage asset accounting, including a health data acquisition unit and a residual value calculation unit. The multi-scenario prediction module is configured to dynamically predict green power output and power load, including a scenario identification unit and a prediction model switching unit. The cross-chain permission management module is configured to realize permission control and traceability of green power cross-chain data, including a dynamic permission control unit and a cross-chain traceability unit. The federal collaborative processing module is configured to realize collaborative analysis of multi-agent green power encrypted data, including a local model training unit and a global model aggregation unit. The lightweight consensus unit of the blockchain storage verification module selects edge nodes from the blockchain network as verification nodes, and limits each verification node to verify only the power data output by adjacent edge nodes. The data chaining unit includes a data packaging subunit and a chain storage subunit. The data packaging subunit standardizes and packages the preprocessed power data, and the chain storage subunit uploads the packaged data to the blockchain network for distributed storage. The local model training unit of the federal collaborative processing module includes a local encrypted data acquisition subunit and a sub-model training subunit. The local encrypted data acquisition subunit acquires locally stored green power encrypted data, and the sub-model training subunit trains a green power prediction sub-model locally using the acquired encrypted data. The global model aggregation unit of the federal collaborative processing module includes a parameter receiving subunit, a model aggregation subunit, and a contribution evaluation subunit. The parameter receiving subunit receives sub-model parameters uploaded by each agent. The model aggregation subunit weights and aggregates the received sub-model parameters according to the data volume proportion of each agent to generate a global prediction model. The contribution evaluation subunit calculates the data contribution of each agent based on the influence of the uploaded sub-model parameters on the accuracy of the global prediction model, and uploads the contribution result to the blockchain. 2.The blockchain-based green power consumption and storage data processing system of claim 1, wherein, The data cleaning unit of the edge preprocessing module includes a power data receiving subunit and an outlier elimination subunit. The power data receiving subunit is configured to receive voltage and current raw data output by a green power terminal, and the outlier elimination subunit uses a sliding window algorithm to identify and eliminate instantaneous fault data that exceeds the normal range. The data compression unit uses differential encoding technology to compress the volume of the cleaned power data to obtain preprocessed data. 3.The blockchain-based green power consumption and storage data processing system of claim 1, wherein, The health data acquisition unit of the energy storage health management module includes a sensor group and a data encryption subunit; The sensor group is used to acquire index data of the energy storage device, and the data encryption subunit uploads the acquired index data to the blockchain after encryption processing by using an encryption algorithm; The residual value calculation unit includes a coefficient determination subunit and a calculation subunit; The coefficient determination subunit determines the health degree coefficient, the use time length coefficient and the environmental adaptation coefficient of the energy storage device, and the calculation subunit calculates the residual value of the energy storage device based on the determined coefficients.

4. The blockchain-based green power consumption and storage data processing system of claim 3, wherein, The sensor group includes a voltage sensor, a temperature sensor, a current sensor and a capacity detection sensor; The voltage sensor is used to acquire output voltage data of the energy storage device, the temperature sensor is used to acquire internal temperature data of the energy storage device, the current sensor is used to acquire charge and discharge current data of the energy storage device, and the capacity detection sensor is used to acquire residual capacity data of the energy storage device to calculate the capacity attenuation rate.

5. The blockchain-based green power consumption and storage data processing system of claim 1, wherein, The scene identification unit of the multi-scenario prediction module includes a scene data receiving subunit and a scene determination subunit; The scene data receiving subunit receives green power historical load data, temperature and humidity data, weather warning data and regional power consumption event information, and the scene determination subunit determines the current green power prediction scene based on the received data; The prediction model switching unit includes a model calling subunit and a model calibration subunit; The model calling subunit calls the corresponding prediction model according to the determined scene, and the model calibration subunit compares the prediction data and the actual data periodically, and calibrates the parameters of the prediction model.

6. The blockchain-based green power consumption and storage data processing system of claim 1, wherein, The dynamic permission control unit of the cross-chain permission management module includes a permission division subunit and a permission management subunit; The permission division subunit divides the data access permission into three levels of viewing permission, using permission and modifying permission, the permission management subunit assigns the corresponding permission based on the user type, and recycles the assigned permission when the preset condition is triggered; The cross-chain traceability unit includes an identification generation subunit and a traceability query subunit; The identification generation subunit generates a unique identification for each piece of cross-chain data, which includes source chain ID, original data hash, cross-chain time, receiving chain ID and use record, and the traceability query subunit provides an identification query function to obtain the data flow path.

Images