Carbon emission data monitoring and management method, device and medium based on photovoltaic panels

Abstract

The present application relates to the field of data processing technology, and discloses a method, device and medium for monitoring and managing carbon emission data based on photovoltaic panels. The method includes: removing outliers and filling missing values ​​from the original data set of the photovoltaic system to obtain photovoltaic power generation time series data; adjusting the back propagation parameters of the photovoltaic power generation time series data to obtain the photovoltaic system carbon emission reduction prediction function; performing sliding window calculation on the real-time photovoltaic power generation data to obtain a carbon emission reduction estimation sequence; performing multi-scale analysis on the carbon emission reduction estimation sequence to obtain a long-term trend component and a periodic fluctuation component, constructing a carbon emission reduction efficiency evaluation index, comparing the carbon emission reduction efficiency evaluation index with a preset threshold, and when the carbon emission reduction efficiency evaluation index exceeds the threshold, triggering a smart contract to generate a carbon credit token for the corresponding period, and obtaining a carbon credit token sequence. The present application improves the accuracy of carbon emission data monitoring and management based on photovoltaic panels.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a method, device and medium for monitoring and managing carbon emission data based on photovoltaic panels. Background Art

[0002] As the global climate change problem becomes increasingly serious, reducing carbon emissions has become a common goal for countries around the world. Photovoltaic power generation, as a clean energy technology, plays an important role in reducing carbon emissions. At present, many photovoltaic power stations have established basic monitoring systems to collect data such as power generation, voltage, current, etc., and estimate carbon emission reduction through simple calculation methods. Some advanced systems have also introduced artificial intelligence technologies, such as machine learning algorithms, to improve the accuracy of carbon emission reduction predictions.

[0003] However, the existing carbon emission monitoring system for photovoltaic power plants still has many shortcomings. First, the real-time and accuracy of data collection needs to be improved, especially under abnormal weather conditions. Second, the calculation method of carbon emission reduction is often oversimplified and fails to fully consider the changes in photovoltaic system performance over time and the impact of environmental factors. Furthermore, the credibility and transparency of carbon emission reduction data are insufficient to meet the strict requirements of the carbon trading market. Finally, the existing system lacks an effective mechanism to convert carbon emission reduction data into tradable carbon credits, which limits the participation of photovoltaic power plants in the carbon trading market. Summary of the invention

[0004] The present application provides a method, device and medium for monitoring and managing carbon emission data based on photovoltaic panels, which are used to improve the accuracy of monitoring and management of carbon emission data based on photovoltaic panels.

[0005] In a first aspect, the present application provides a method for monitoring and managing carbon emission data based on photovoltaic panels, the method for monitoring and managing carbon emission data based on photovoltaic panels comprising:

[0006] The voltage, current, temperature and light intensity of the photovoltaic panel array are sampled in real time to obtain the original data set of the photovoltaic system including timestamp;

[0007] Using a sliding window median filtering algorithm, outliers are removed from the original data set of the photovoltaic system and missing values ​​are filled to obtain photovoltaic power generation time series data;

[0008] The photovoltaic power generation time series data is back-propagated and parameterized by a long short-term memory network to obtain a photovoltaic system carbon emission reduction prediction function;

[0009] The photovoltaic system carbon emission reduction prediction function is used to perform sliding window calculation on the real-time photovoltaic power generation data collected in real time to obtain a carbon emission reduction estimation sequence with an interval of 5 minutes;

[0010] Performing a multi-scale analysis on the carbon emission reduction estimation sequence to obtain a long-term trend component and a periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system, wherein the long-term trend component reflects the slow change of system performance over time, and the periodic fluctuation component reflects diurnal and seasonal fluctuations;

[0011] Based on the long-term trend component and the cyclical fluctuation component, a carbon emission reduction efficiency evaluation index is constructed, and the carbon emission reduction efficiency evaluation index is compared with a preset threshold. When the carbon emission reduction efficiency evaluation index exceeds the threshold, the smart contract is triggered to generate a carbon credit token for the corresponding period, and a carbon credit token sequence that can be circulated on a decentralized trading platform is obtained.

[0012] In a second aspect, the present application provides a carbon emission data monitoring and management device based on a photovoltaic panel, the carbon emission data monitoring and management device based on a photovoltaic panel comprising:

[0013] The acquisition module is used to sample the voltage, current, temperature and light intensity of the photovoltaic panel array in real time to obtain the original data set of the photovoltaic system including the timestamp;

[0014] A filling module is used to remove outliers and fill missing values ​​from the original data set of the photovoltaic system by a sliding window median filtering algorithm to obtain photovoltaic power generation time series data;

[0015] An adjustment module is used to adjust the back propagation parameters of the photovoltaic power generation time series data through a long short-term memory network to obtain a photovoltaic system carbon emission reduction prediction function;

[0016] A calculation module, used to perform sliding window calculation on the real-time photovoltaic power generation data collected in real time through the photovoltaic system carbon emission reduction prediction function to obtain a carbon emission reduction estimation sequence with an interval of 5 minutes;

[0017] An analysis module is used to perform multi-scale analysis on the carbon emission reduction estimation sequence to obtain a long-term trend component and a periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system, wherein the long-term trend component reflects the slow change of system performance over time, and the periodic fluctuation component reflects diurnal and seasonal fluctuations;

[0018] A comparison module is used to construct a carbon emission reduction efficiency evaluation index based on the long-term trend component and the cyclical fluctuation component, and compare the carbon emission reduction efficiency evaluation index with a preset threshold. When the carbon emission reduction efficiency evaluation index exceeds the threshold, the smart contract is triggered to generate a carbon credit token for the corresponding period, and a carbon credit token sequence that can be circulated on a decentralized trading platform is obtained.

[0019] The third aspect of the present application provides a carbon emission data monitoring and management device based on photovoltaic panels, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the carbon emission data monitoring and management device based on photovoltaic panels executes the above-mentioned carbon emission data monitoring and management method based on photovoltaic panels.

[0020] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned carbon emission data monitoring and management method based on photovoltaic panels.

[0021] In the technical solution provided by the present application, by installing micro sensors on each photovoltaic panel and performing real-time high-frequency sampling, the accuracy and timeliness of data collection are significantly improved, laying a solid data foundation for the subsequent calculation of carbon emission reduction; secondly, the sliding window median filtering algorithm is used to remove outliers and fill missing values ​​in the original data, which effectively improves the data quality and reduces the impact of environmental interference and equipment failure on the calculation results; the photovoltaic power generation time series data is deeply learned using long short-term memory networks, which not only improves the accuracy of carbon emission reduction prediction, but also can capture the trend of system performance changes over time, making the prediction results closer to the actual situation; and by performing multi-scale analysis on the carbon emission reduction estimation sequence, the long-term trend component and the periodic fluctuation component are obtained, and the carbon emission reduction efficiency evaluation index constructed based on the long-term trend and periodic fluctuation greatly improves the credibility and transparency of the carbon emission reduction calculation; the generation of carbon credit tokens is automatically triggered by smart contracts and deployed to the blockchain network, which not only simplifies the generation and certification process of carbon credits, but also significantly improves the efficiency and security of the entire process, providing convenient conditions for photovoltaic power stations to participate in the carbon trading market. By converting carbon emission reduction data into carbon credit tokens that can be circulated on decentralized trading platforms, new value growth points have been created for photovoltaic power stations, which will help promote the development of the clean energy industry. At the same time, this carbon credit token trading mechanism based on blockchain technology has greatly enhanced the traceability and non-tamperability of carbon emission reduction data. The implementation of this method also helps to improve the operational efficiency of photovoltaic power stations. Through real-time monitoring and analysis of carbon emission reduction data, operators can promptly discover and solve problems in the system, such as component performance degradation, local shading, etc., thereby maximizing the power generation efficiency and carbon emission reduction effect of the photovoltaic system. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0023] Figure 1 This is a schematic diagram of an embodiment of a method for monitoring and managing carbon emission data based on photovoltaic panels in an embodiment of the present application;

[0024] Figure 2 This is a schematic diagram of an embodiment of a carbon emission data monitoring and management device based on photovoltaic panels in an embodiment of the present application;

[0025] Figure 3 It is a structural schematic diagram of a photovoltaic panel-based carbon emission data monitoring and management device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The embodiments of the present application provide a method, device and medium for monitoring and managing carbon emission data based on photovoltaic panels. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of the carbon emission data monitoring and management method based on photovoltaic panels includes:

[0028] Step S101, sampling the voltage, current, temperature and light intensity of the photovoltaic panel array in real time to obtain a photovoltaic system original data set including a timestamp;

[0029] Step S102, using a sliding window median filtering algorithm to remove outliers and fill missing values ​​from the original data set of the photovoltaic system to obtain photovoltaic power generation time series data;

[0030] Step S103, adjusting the back propagation parameters of the photovoltaic power generation time series data through the long short-term memory network to obtain a photovoltaic system carbon emission reduction prediction function;

[0031] Step S104: Perform sliding window calculation on the real-time photovoltaic power generation data collected in real time through the photovoltaic system carbon emission reduction prediction function to obtain a carbon emission reduction estimation sequence with an interval of 5 minutes;

[0032] Step S105, performing multi-scale analysis on the carbon emission reduction estimation sequence to obtain a long-term trend component and a periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system, wherein the long-term trend component reflects the slow change of system performance over time, and the periodic fluctuation component reflects diurnal and seasonal fluctuations;

[0033] Step S106: Based on the long-term trend component and the periodic fluctuation component, a carbon emission reduction efficiency evaluation index is constructed, and the carbon emission reduction efficiency evaluation index is compared with a preset threshold. When the carbon emission reduction efficiency evaluation index exceeds the threshold, the smart contract is triggered to generate a carbon credit token for the corresponding period, and a carbon credit token sequence that can be circulated on a decentralized trading platform is obtained.

[0034] It is understandable that the execution subject of the present application can be a carbon emission data monitoring and management device based on photovoltaic panels, or a terminal or a server, which is not limited here. The present application embodiment is described by taking the server as the execution subject as an example.

[0035] Specifically, the implementation process of the carbon emission data monitoring and management method based on photovoltaic panels begins with comprehensive real-time data collection of the photovoltaic panel array. It involves installing high-precision micro sensors on each photovoltaic panel to measure key parameters such as voltage, current, temperature and light intensity. These sensors transmit data in real time to a local data concentrator via a low-power wireless network to form a raw data set with precise timestamps. The frequency of data collection is usually set to once per second to ensure that small changes in the performance of the photovoltaic system are captured.

[0036] Preprocessing the original data set is crucial. The sliding window median filter algorithm plays a key role in this stage, which is used to remove outliers and fill missing data. The algorithm calculates the median of each parameter in the window by sliding a fixed-size window (such as 10 minutes) on the data sequence, and uses it as the representative value at that time point. This method effectively eliminates data anomalies caused by sensor failure, electromagnetic interference or temporary occlusion. For the detected missing values, the algorithm uses adjacent valid data points for interpolation filling to ensure the continuity of the data. After this processing, the original data is converted into high-quality photovoltaic power generation time series data. In order to accurately predict carbon emission reduction, the long short-term memory network (LSTM) is used to perform deep learning on the processed time series data. The special structure of the LSTM network enables it to capture long-term dependencies in the data, which is crucial for understanding the performance of photovoltaic systems at different time scales. During the network training process, the back propagation algorithm is used to optimize the model parameters to minimize the error between the predicted value and the actual carbon emission reduction. After the training is completed, a function model that can predict future carbon emission reduction based on the current input is obtained.

[0037] Using the trained carbon emission reduction prediction function, a sliding window calculation is performed on the real-time collected photovoltaic power generation data. This step sets the prediction window to 5 minutes and slides forward 5 minutes each time, thereby generating a carbon emission reduction estimation sequence with a 5-minute interval. This method not only provides carbon emission reduction estimates with high temporal resolution, but also reflects changes in system performance in a timely manner. Multi-scale analysis of the obtained carbon emission reduction estimation sequence is the key to understanding the long-term performance and short-term fluctuations of photovoltaic systems. This analysis process usually uses wavelet transform technology to decompose the time series into subsequences of different frequencies. The low-frequency part reflects the long-term trend of system performance, such as the slow degradation of photovoltaic panel efficiency over time; the high-frequency part captures diurnal and seasonal fluctuations, such as the impact of cloud cover or seasonal temperature changes on power generation efficiency.

[0038] Finally, the carbon reduction efficiency evaluation index is constructed based on the long-term trend component and the cyclical fluctuation component. This index comprehensively considers the stability and responsiveness of the system, and provides a scientific basis for the generation of carbon credit tokens. When the evaluation index exceeds the preset threshold, the smart contract is automatically triggered to generate carbon credit tokens for the corresponding period. These tokens are based on blockchain technology, which ensures their immutability and traceability, and lays the foundation for circulation on decentralized trading platforms.

[0039] For example, a photovoltaic power station installed 1,000 photovoltaic panels, each of which was equipped with a data acquisition sensor. On a clear summer day, the system collected data once a second, accumulating 86.4 million data points in a day. Through the sliding window median filter algorithm, about 0.1% of outliers were identified and about 0.05% of missing data were successfully filled. After LSTM network training, the model achieved 95% prediction accuracy on the test set. In the following week, the system generated carbon emission reduction estimates every 5 minutes, accumulating 2016 estimated data points. Multi-scale analysis showed that the long-term efficiency of the photovoltaic system showed a slow decline of about 0.5% per year, while the fluctuation range of daytime power generation efficiency was about ±15%. Based on these data, the system generated 168 carbon credit tokens, each representing a reduction of 1 ton of carbon dioxide equivalent. These tokens were successfully deployed to the blockchain network, creating additional economic value for the photovoltaic power station and making a substantial contribution to climate change mitigation.

[0040] In the embodiment of the present application, by installing micro sensors on each photovoltaic panel and performing real-time high-frequency sampling, the accuracy and timeliness of data collection are significantly improved, laying a solid data foundation for the subsequent calculation of carbon emission reduction; secondly, the sliding window median filtering algorithm is used to remove outliers and fill missing values ​​in the original data, which effectively improves the data quality and reduces the impact of environmental interference and equipment failure on the calculation results; the photovoltaic power generation time series data is deeply learned using a long short-term memory network, which not only improves the accuracy of carbon emission reduction prediction, but also can capture the trend of system performance changes over time, making the prediction results closer to the actual situation; and by performing multi-scale analysis on the carbon emission reduction estimation sequence, the long-term trend component and the periodic fluctuation component are obtained, and the carbon emission reduction efficiency evaluation index constructed based on the long-term trend and periodic fluctuation greatly improves the credibility and transparency of the carbon emission reduction calculation; the generation of carbon credit tokens is automatically triggered by smart contracts and deployed to the blockchain network, which not only simplifies the generation and certification process of carbon credits, but also significantly improves the efficiency and security of the entire process, providing convenient conditions for photovoltaic power stations to participate in the carbon trading market. By converting carbon emission reduction data into carbon credit tokens that can be circulated on decentralized trading platforms, new value growth points have been created for photovoltaic power stations, which will help promote the development of the clean energy industry. At the same time, this carbon credit token trading mechanism based on blockchain technology has greatly enhanced the traceability and non-tamperability of carbon emission reduction data. The implementation of this method also helps to improve the operational efficiency of photovoltaic power stations. Through real-time monitoring and analysis of carbon emission reduction data, operators can promptly discover and solve problems in the system, such as component performance degradation, local shading, etc., thereby maximizing the power generation efficiency and carbon emission reduction effect of the photovoltaic system.

[0041] In a specific embodiment, the expression of the photovoltaic system carbon emission reduction prediction function is as follows:

[0042]

[0043] in, for The predicted carbon emission reduction value at the time, in kilograms of carbon dioxide equivalent; is the Sigmoid activation function; Output weight matrix for carbon emission reduction; is the hidden state weight matrix; for The hidden state vector at time t; for t The input vector at the moment, including voltage, current, temperature and light intensity; and are the carbon emission reduction output bias vector and the hidden state bias vector respectively; is the hyperbolic tangent activation function.

[0044] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0045] (1) installing a micro voltage sensor, a current sensor, a temperature sensor, and a light intensity sensor on each photovoltaic panel of the photovoltaic panel array to obtain a sensor array, and calibrating the sensor array to obtain a calibrated sensor array;

[0046] (2) Performing high-frequency sampling on the photovoltaic panel array according to the calibrated sensor array to obtain an original sampling data stream, and performing a timestamp on the original sampling data stream to obtain an original sampling data stream with a timestamp;

[0047] (3) compressing the original sampling data stream with the timestamp to obtain a compressed sampling data stream, and encrypting the compressed sampling data stream to obtain an encrypted compressed sampling data stream;

[0048] (4) The encrypted compressed sampling data stream is packetized and transmitted through a low-power wireless transmission module to obtain multiple data packets, and the multiple data packets are reassembled in a local data concentrator to obtain a reassembled data stream;

[0049] (5) Decrypting and decompressing the reassembled data stream to obtain a restored sampled data stream, and performing data integrity verification on the restored sampled data stream to obtain a verified sampled data stream;

[0050] (6) Generate a structured photovoltaic system raw data set based on the verified sampled data stream, and create an index for the photovoltaic system raw data set to obtain an indexed photovoltaic system raw data set;

[0051] (7) The original data set of the photovoltaic system with an index is sharded to obtain multiple data shards, and the multiple data shards are distributedly stored to obtain a distributedly stored original data set of the photovoltaic system.

[0052] Specifically, in the carbon emission data monitoring and management method based on photovoltaic panels, it is first necessary to install micro sensors on each panel of the photovoltaic panel array. These sensors include voltage sensors, current sensors, temperature sensors, and light intensity sensors, which together form a comprehensive sensor array. After installation, calibrating the sensor array is a key step to ensure data accuracy. The calibration process involves comparing the output of each sensor with a standard reference value and minimizing the measurement error by adjusting the sensor's parameters or compensation algorithm.

[0053] The calibrated sensor array begins high-frequency sampling of the photovoltaic panel array. The sampling frequency is usually set between 1Hz and 10Hz to capture subtle changes in the performance of the photovoltaic system. Each data point is assigned a precise timestamp, which is usually accurate to the millisecond level for subsequent timing analysis and data synchronization. This results in a raw sampled data stream with a timestamp. Considering the large amount of data generated by high-frequency sampling, data compression becomes a necessary processing step. Commonly used compression algorithms include lossless compression methods (such as Huffman coding or LZW algorithm) and lossy compression methods (such as wavelet transform). Compression not only reduces the burden of data transmission and storage, but also helps to improve the energy efficiency of the entire system. The compressed data stream is then encrypted, usually using symmetric encryption algorithms such as AES (Advanced Encryption Standard) to protect the security of data during transmission.

[0054] The encrypted compressed sampled data stream is transmitted in packets through a low-power wireless transmission module. The "low power" here refers to the use of communication protocols designed for the Internet of Things, such as LoRaWAN or NB-IoT, which minimize energy consumption while ensuring the transmission distance. The data is divided into multiple smaller data packets, each of which contains specific header information for subsequent data reassembly. In the local data concentrator, these data packets are received and reassembled according to the predetermined protocol to restore the complete data stream. The reassembled data stream needs to be decrypted and decompressed. The decryption process uses the same key as the encryption process to restore the original data through the reverse operation. Decompression is the inverse process of the compression algorithm, restoring the compressed data to its original form. Subsequently, the restored sampled data stream is checked for data integrity. Common methods include cyclic redundancy check (CRC) or hash function check to ensure that the data has not been tampered with or lost during transmission and processing.

[0055] The validated sampled data stream is used to generate a structured raw data set of the PV system. This process involves organizing the raw data into a predefined format, such as arranging parameters such as voltage, current, temperature, and light intensity in chronological order, and adding necessary metadata information. In order to improve data retrieval efficiency, an index is created for this structured data set. Common indexing methods include B-tree or hash indexes, which can significantly accelerate subsequent data query operations. Finally, in order to achieve efficient data management and utilization, the indexed raw data set of the PV system is sharded. Data sharding is the process of dividing a large data set into multiple smaller, manageable parts. The sharding strategy may be based on time range, data type, or other criteria. These data shards are distributed and stored on multiple servers or storage nodes, using technologies such as Hadoop Distributed File System (HDFS) to improve data availability and processing efficiency.

[0056] For example, a large photovoltaic power plant has 10,000 photovoltaic panels, each of which is equipped with the above four sensors. On a sunny summer day, the system samples at a frequency of 5Hz, generating 43,200,000 raw data points (10,000×4×5×60×60 = 43,200,000). The data is processed by the LZW compression algorithm, with a compression ratio of 3:1, and the data volume is reduced to 14,400,000 bytes. The compressed data is encrypted using 128-bit AES and then divided into 1,440 10KB data packets for transmission over the LoRaWAN network. In the data concentrator, these packets are reassembled, decrypted and decompressed to restore the original data. The data integrity check uses the SHA-256 hash algorithm to ensure 99.99% data integrity. The structured data set is indexed with a B+ tree, and the index size is about 5% of the original data. Finally, the data is divided into 24-hour time segments and distributed and stored on 8 storage nodes, with each node responsible for 3 hours of data.

[0057] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0058] (1) The original data set of the photovoltaic system is segmented into time series to obtain a data subset of 24 hours in length, and the data subset is divided into overlapping sliding windows with a width of 10 minutes and a step length of 5 minutes to obtain 144 overlapping data windows;

[0059] (2) Calculate the median of the voltage, current, temperature and light intensity data in each data window to obtain the median sequence of each parameter, and calculate the upper and lower limits of 3 times the interquartile range based on the median sequence to obtain the outlier identification threshold;

[0060] (3) The original data set of the photovoltaic system is compared point by point according to the outlier recognition threshold to obtain a binary outlier mark sequence, and the data points marked as abnormal are removed according to the binary outlier mark sequence to obtain a preliminary cleaned data set;

[0061] (4) Perform missing value detection on the data set after preliminary cleaning to obtain the timestamp index of the missing value, and perform linear regression analysis on the 48 adjacent non-missing data points based on the timestamp index of the missing value to obtain the candidate filling value for each missing position;

[0062] (5) Smoothing the candidate filling values ​​using a cubic spline interpolation algorithm to obtain an interpolated filling value sequence, and inserting the interpolated filling value sequence into the corresponding position of the data set after preliminary cleaning to obtain a filled data set;

[0063] (6) The filled data set is time aligned and resampled at 5-minute intervals to obtain equally spaced photovoltaic power generation time series data, and each parameter in the photovoltaic power generation time series data is normalized to obtain standardized photovoltaic power generation time series data.

[0064] Specifically, the original data set of the photovoltaic system is segmented into time series to obtain a data subset of 24 hours in length. This segmentation method takes into account the daily cycle characteristics of the photovoltaic system and is conducive to capturing the changes in power generation patterns within a day. Subsequently, each 24-hour data subset is divided into overlapping sliding windows with a width of 10 minutes and a step length of 5 minutes to obtain 144 overlapping data windows. The design of this overlapping sliding window allows for a more fine-grained analysis of the data while maintaining the continuity of the data. The median is calculated for the voltage, current, temperature and light intensity data in each data window to obtain the median sequence of each parameter. As a robust statistic, the median is not easily affected by extreme values ​​and is therefore suitable for the preliminary identification of outliers. Next, the upper and lower limits of 3 times the interquartile range are calculated based on the median sequence to obtain the outlier identification threshold. The interquartile range is a statistic that describes the degree of data dispersion. By calculating the upper and lower limits of 3 times the interquartile range, data points that deviate significantly from the normal range can be effectively identified.

[0065] Using the obtained outlier recognition threshold, the original data set of the photovoltaic system is compared point by point to generate a binary outlier mark sequence. In this sequence, 1 represents an outlier point and 0 represents a normal point. According to this mark sequence, the data points marked as outliers are removed to obtain a preliminary cleaned data set. This step effectively removes abnormal data that may be caused by sensor failure, temporary occlusion or other interference factors. Missing value detection is performed on the preliminary cleaned data set to obtain the timestamp index of the missing value. Missing values ​​may be caused by data transmission interruption, equipment failure or outlier removal. In order to fill these missing values, linear regression analysis is performed on the adjacent 48 non-missing data points to obtain candidate filling values ​​for each missing position. Linear regression analysis takes into account the local trend of the data and can provide relatively accurate estimates.

[0066] However, linear regression may produce non-smooth filling results in places where data changes drastically. Therefore, the candidate filling values ​​are smoothed by the cubic spline interpolation algorithm to obtain the interpolated filling value sequence. Cubic spline interpolation can provide smooth curve fitting while maintaining data continuity. These smoothed filling values ​​are inserted into the corresponding positions of the data set after preliminary cleaning to obtain the filled data set. Finally, the filled data set is time-aligned and resampled at 5-minute intervals to obtain photovoltaic power generation time series data with equal intervals. This step ensures the temporal consistency of the data and facilitates subsequent time series analysis. Each parameter in the photovoltaic power generation time series data is normalized to obtain standardized photovoltaic power generation time series data. Normalization unifies parameters of different dimensions to the same scale, which is conducive to subsequent multi-parameter analysis and model training.

[0067] For example, a photovoltaic power station collected 86,400 data points (one per second) in one day. After dividing these data into 24-hour subsets, each subset was processed with a sliding window of 10 minutes width and 5 minutes step length, resulting in 144 overlapping windows. In one of the windows, the median of the voltage data was 220V and the interquartile range was 10V. Therefore, the outlier identification threshold was set to 190V (220V - 3 ×10V) to 250V (220V + 3 ×10V). With this threshold, 15 outliers were identified, accounting for about 2.5% of the data in the window. During the missing value filling process, a period of missing data lasting 2 minutes was found. Linear regression was performed using 48 adjacent data points (24 before and after), and a regression equation with a slope of 0.05V / min and an intercept of 215V was obtained. Based on this, 24 candidate filling values ​​were calculated. After cubic spline interpolation smoothing, these filling values ​​achieved a smooth transition with the original data. Finally, all parameters are normalized to the range of 0-1 to obtain a standardized time series dataset.

[0068] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0069] (1) Divide the photovoltaic power generation time series data into time windows to obtain an input sequence of fixed length, and normalize the input sequence to obtain standardized input data;

[0070] (2) Constructing a long short-term memory network structure, including an input layer, a hidden layer, and an output layer, and setting the number of neurons in the hidden layer to obtain an initial network structure;

[0071] (3) Randomly initialize the weight matrix and bias vector in the initial network structure to obtain the initial parameter set, and calculate the forward propagation output of the network based on the initial parameter set to obtain the prediction result;

[0072] (4) Compare the predicted results with the actual carbon emission reduction data, calculate the mean square error as the loss function, and calculate the gradient of the loss function with respect to each parameter through the back propagation algorithm to obtain the gradient matrix;

[0073] (5) The network parameters are updated according to the gradient matrix to obtain a new parameter set, and the forward propagation, loss calculation and back propagation steps are repeated until the preset number of iterations or convergence conditions are reached to obtain the trained network parameters, and the carbon emission reduction prediction function is constructed through the trained network parameters.

[0074] Specifically, the photovoltaic power generation time series data is divided into time windows to obtain a fixed-length input sequence. Usually 24 hours is selected as a window and sampled once an hour, so each input sequence contains 24 data points. These input sequences are normalized, and the minimum-maximum normalization method is used to scale the data to the [0,1] interval to obtain standardized input data, and a long short-term memory network (LSTM) structure is constructed, including an input layer, a hidden layer, and an output layer. The number of neurons in the input layer is equal to the number of features of the input sequence (such as voltage, current, temperature, light intensity, etc.), which is 4 in this case. The hidden layer adopts a two-layer LSTM structure, with 64 neurons in each layer. The output layer is a fully connected layer that outputs a single value, namely the predicted carbon emission reduction.

[0075] The weight matrix and bias vector in the initial network structure are randomly initialized to obtain the initial parameter set. The weight initialization uses the Xavier initialization method, and the bias is initialized to 0. Then, the forward propagation output of the network is calculated based on the initial parameter set to obtain the prediction result.

[0076] During the training process, the prediction results are compared with the actual carbon emission reduction data, and the mean square error (MSE) is calculated as the loss function. The gradient of the loss function to each parameter is calculated by the back propagation algorithm to obtain the gradient matrix. The Adam optimizer is used to update the network parameters, and the forward propagation, loss calculation and back propagation steps are repeated until the preset number of iterations or convergence conditions are reached to obtain the trained network parameters. Finally, the carbon emission reduction prediction function is constructed using the trained network parameters. Keeping consistent with the function provided initially, the carbon emission reduction prediction function expression is as follows:

[0077]

[0078] in, for The predicted carbon emission reduction value at the time, in kilograms of carbon dioxide equivalent; is the Sigmoid activation function; Output weight matrix for carbon emission reduction; is the hidden state weight matrix; for The hidden state vector at time t; for t The input vector at the moment, including voltage, current, temperature and light intensity; and are the carbon emission reduction output bias vector and the hidden state bias vector respectively; is the hyperbolic tangent activation function.

[0079] For example, a photovoltaic power station has 100 photovoltaic panels, and data is collected every hour, including voltage (180V-250V), current (0-10A), temperature (-10°C-50°C), and light intensity (0-1000W / m²). After collecting data for one year, a data set of 8760 time points is obtained. The data is divided into a window of 24 hours to obtain 365 input sequences. Each feature is normalized, such as normalizing the voltage from 180V-250V to between 0-1. Construct an LSTM network with 4 neurons in the input layer, 64 neurons in each of the two hidden layers, and 1 neuron in the output layer. Initialize the weights, such as the weight matrix dimension of the first layer of LSTM is [64, 4], ranging from [-0.25, 0.25]. During the training process, the MSE of the first epoch is 0.15, and after 1000 iterations, the MSE is reduced to 0.01. The final carbon emission reduction prediction function can predict the carbon emission reduction in the next hour based on 24 hours of input data, with a prediction error within 5%.

[0080] For example, given 24 hours of data on a certain day, the carbon emission reduction output by the prediction function is 15.8 kg CO2 equivalent, while the actual measured value is 16.2 kg CO2 equivalent, with a relative error of 2.47%. This method not only takes into account short-term fluctuations, but also captures long-term trends, providing an accurate prediction tool for carbon emission monitoring of photovoltaic power plants.

[0081] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0082] (1) Time synchronization calibration is performed on the real-time collected photovoltaic power generation data to obtain a calibrated real-time power data stream, and the calibrated real-time power data stream is divided into 5-minute time segments to obtain data segments of equal length;

[0083] (2) Perform maximum and minimum value scaling on the voltage, current, temperature, and light intensity data in data segments of equal length to obtain a normalized feature vector, and extract the state vector at the previous moment from the memory unit to obtain the function input data;

[0084] (3) Input the function input data into the photovoltaic system carbon emission reduction prediction function to obtain a preliminary estimate of the carbon emission reduction in the current time segment;

[0085] (4) Reversely scale the preliminary estimate of carbon emission reduction to obtain the actual unit carbon emission reduction value, and pair the carbon emission reduction value with the corresponding time stamp to obtain the time-stamped carbon emission reduction data point;

[0086] (5) Append the time-stamped carbon emission reduction data points to the carbon emission reduction data set to obtain the extended data set, and slide the time segment backward by 5 minutes to obtain the input data of the next time segment;

[0087] (6) The expanded data set is filled with time series to complete all 5-minute interval points in the monitoring period to obtain the carbon emission reduction time series. The carbon emission reduction time series is then verified for data continuity to obtain a carbon emission reduction estimation series with a 5-minute interval.

[0088] Specifically, the real-time collected photovoltaic power generation data is time-synchronized and calibrated. This process uses the Network Time Protocol (NTP) to align the timestamps of all sensors to a unified time standard, thereby obtaining a calibrated real-time power data stream. Next, the calibrated real-time power data stream is divided into 5-minute time segments to obtain data segments of equal length. This division method not only ensures the real-time nature of the data, but also provides sufficient information for short-term predictions. The voltage, current, temperature, and light intensity data in the equal-length data segments are scaled to the maximum and minimum values ​​to obtain a normalized feature vector. This step uses the following formula:

[0089]

[0090] in, is the normalized value, is the original data value, and are the minimum and maximum values ​​of the feature in the entire data set, respectively. At the same time, the state vector of the previous moment is extracted from the memory unit. This state vector contains the hidden state information of the previous time step in the LSTM network. Combined with the normalized feature vector, the complete function input data is obtained. The function input data is input into the photovoltaic system carbon emission reduction prediction function to obtain a preliminary estimate of the carbon emission reduction in the current time segment. The form of the prediction function is as follows:

[0091]

[0092] in, for The predicted carbon emission reduction value at the time, in kilograms of carbon dioxide equivalent; is the Sigmoid activation function; Output weight matrix for carbon emission reduction; is the hidden state weight matrix; for The hidden state vector at time t; for t The input vector at the moment, including voltage, current, temperature and light intensity; and are the carbon emission reduction output bias vector and the hidden state bias vector respectively; is the hyperbolic tangent activation function.

[0093] The preliminary estimate of carbon emission reduction is reversely scaled to obtain the actual unit carbon emission reduction value. The reverse scaling formula is:

[0094]

[0095] in, A is the actual carbon emission reduction value after reverse scaling, N is the normalized predicted value, and are the maximum and minimum values ​​of carbon emission reduction in the training data, respectively. The obtained carbon emission reduction values ​​are paired with the corresponding time tags to obtain the time-tagged carbon emission reduction data points. The time-tagged carbon emission reduction data points are appended to the carbon emission reduction data set to obtain the extended data set, and the time segment is slid back 5 minutes to obtain the input data of the next time segment. This sliding window method ensures the continuity and smoothness of the prediction.

[0096] Finally, the expanded data set is filled with time series to complete all 5-minute interval points in the monitoring period to obtain the carbon emission reduction time series. The filling method uses linear interpolation:

[0097]

[0098] in, is the value of the point to be filled, and is the value of the two known points before and after, , , are the time indexes of the points to be filled and the known points respectively. The data continuity of the carbon emission reduction time series is verified to check whether there are abnormal jumps or missing data, and the carbon emission reduction estimation series with an interval of 5 minutes is obtained.

[0099] For example, a photovoltaic power station starts collecting data at 6:00 in the morning on a sunny summer day. The average value of the raw data collected in the 5 minutes from 6:00 to 6:05 is: voltage 230V, current 8A, temperature 28°C, and light intensity 800W / m². After time synchronization calibration, ensure that the timestamps of all data points are accurate to the millisecond level. This set of data is normalized. Assuming that the voltage range is 180-250V, the current range is 0-10A, the temperature range is 10-40°C, and the light intensity range is 0-1000W / m², the normalized feature vector is [0.71, 0.8, 0.6, 0.8]. Extract the state vector of the previous moment (5:55-6:00) from the memory unit, assuming it is [0.6, 0.5, 0.7, 0.4]. Inputting these data into the prediction function, the preliminary estimate of carbon emission reduction for the period of 6:00-6:05 is 0.85 (normalized value). Assuming that the range of carbon emission reduction in the training data is 0-20kg CO2 equivalent / 5min, the actual carbon emission reduction after reverse scaling is 0.85×20 = 17kg CO2 equivalent.

[0100] This data point is added to the dataset and labeled "2023-07-15 06:05:00". Next, the time window slides back 5 minutes to start processing the data from 6:05-6:10. After one day of processing, 288 carbon emission reduction data points are obtained. During the continuity verification, it was found that there was missing data between 11:30-11:35. The linear interpolation method was used to fill in the data from 11:25-11:30 and 11:35-11:40. The final carbon emission reduction estimation sequence not only reflects the carbon emission reduction situation on that day, but also provides basic data for long-term carbon emission trend analysis.

[0101] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0102] (1) Perform wavelet decomposition on the carbon emission reduction estimation series to obtain wavelet coefficients of different scales, and then group the wavelet coefficients to obtain coefficient groups representing different time scales;

[0103] (2) Threshold processing is performed on the coefficient groups representing different time scales to obtain a denoised coefficient group, and an inverse wavelet transform is performed on the denoised coefficient group to obtain multiple components of different time scales;

[0104] (3) Clustering the components of multiple time scales to obtain long-term trend clusters and periodic fluctuation clusters, and reconstructing the long-term trend clusters and periodic fluctuation clusters to obtain the initial long-term trend component and the initial periodic fluctuation component;

[0105] (4) Perform polynomial fitting on the initial long-term trend component to obtain a smoothed long-term trend curve, and perform slope analysis on the smoothed long-term trend curve to obtain a system performance change rate indicator;

[0106] (5) Perform Fourier transform on the initial periodic fluctuation component to obtain a spectrum diagram, and perform peak detection on the spectrum diagram to obtain the main periodic components;

[0107] (6) According to the system performance change rate index and the main periodic components, the smoothed long-term trend curve and the initial periodic fluctuation component are weighted combined to obtain the long-term trend component and periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system.

[0108] Specifically, in the carbon emission data monitoring and management method based on photovoltaic panels, multi-scale analysis of the carbon emission reduction estimation sequence is a key step in understanding the long-term performance and periodic changes of the system. First, the carbon emission reduction estimation sequence is subjected to wavelet decomposition, which uses discrete wavelet transform (DWT) to decompose the time series into subsequences of different frequencies. The mathematical expression of wavelet decomposition is:

[0109]

[0110] in, are the wavelet coefficients, is the original signal, is the wavelet function, is the scale parameter, is the translation parameter. By adjusting and , we can get wavelet coefficients of different scales, which represent the characteristics of the signal at different times and frequencies. Group the obtained wavelet coefficients to form coefficient groups representing different time scales. For example, the coefficients can be divided into three groups: high frequency (intraday fluctuations), medium frequency (weekly fluctuations) and low frequency (monthly and above fluctuations). Then, threshold processing is performed on these coefficient groups to remove the influence of noise. The threshold processing method can use the soft threshold method:

[0111]

[0112] in, is the coefficient after threshold processing, are the original coefficients, is the threshold value, is a sign function. The noise component in the wavelet coefficients is effectively removed. The denoised coefficient group is subjected to inverse wavelet transform to obtain multiple components of different time scales. These components represent the changing characteristics of carbon emission reduction at different time scales. Then, these components of different time scales are clustered by time series, and the K-means algorithm is used to divide them into long-term trend clusters and periodic fluctuation clusters. The objective function of the K-means algorithm is:

[0113]

[0114] in, is the objective function, k is the number of clusters, n is the number of data points, It is The first i data points, It is j The center of a cluster.

[0115] The long-term trend cluster and the periodic fluctuation cluster are reconstructed respectively to obtain the initial long-term trend component and the initial periodic fluctuation component. In order to further optimize the long-term trend component, a polynomial fitting is performed on it to obtain a smoothed long-term trend curve. The general form of polynomial fitting is:

[0116]

[0117] in, is the fitting polynomial, is the time variable, , , ..., is the polynomial coefficient. Perform slope analysis on the smoothed long-term trend curve, calculate the derivative value at each time point, and obtain the system performance change rate index. Perform Fourier transform on the initial periodic fluctuation component to obtain the spectrum diagram. The formula for Fourier transform is:

[0118]

[0119] in, is the spectrum, is the time domain signal, is the angular frequency. Peak detection is performed on the spectrum to identify the main periodic components, which reflect the periodic changes in carbon emission reduction. Finally, according to the system performance change rate index and the main periodic components, the smoothed long-term trend curve and the initial periodic fluctuation component are weighted and combined to obtain the long-term trend component and periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system.

[0120] For example, a photovoltaic power station continuously monitors carbon emission reduction data for one year, for a total of 365 days, with 48 data points per day (one every 30 minutes). Wavelet decomposition is performed on these 17,520 data points, using Daubechies 4 wavelets to obtain wavelet coefficients of 5 scales. These coefficients are divided into three groups: high frequency (1-2 days), medium frequency (3-7 days), and low frequency (more than 8 days). The soft threshold method is applied to each group of coefficients for denoising, and the threshold is set to 1.5 times the standard deviation of the group of coefficients. After inverse wavelet transformation, three time scale components are obtained. The three components are clustered using the K-means algorithm (k=2) to obtain a long-term trend cluster and a periodic fluctuation cluster. A fifth-order polynomial is used to fit the long-term trend cluster to obtain a smooth long-term trend curve. The first-order derivative of the curve is calculated, and it is found that the average rate of change of system performance in one year is -0.02% / day, indicating that the system efficiency has decreased slightly.

[0121] The Fourier transform of the cyclical fluctuation component was performed, and three main peaks were detected in the spectrum diagram, corresponding to periods of 1 day, 7 days and 365 days, reflecting the cyclical changes between days, weeks and years. Based on these analysis results, the long-term trend and cyclical fluctuation were weighted and combined in a ratio of 6:4 to obtain the final long-term trend component and cyclical fluctuation component.

[0122] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0123] (1) Normalize the long-term trend component to obtain the standardized long-term trend value, and calculate the amplitude of the periodic fluctuation component to obtain the fluctuation intensity index;

[0124] (2) Based on the standardized long-term trend value and fluctuation intensity index, a comprehensive carbon emission reduction efficiency evaluation index is constructed, and the comprehensive carbon emission reduction efficiency evaluation index is divided into time windows to obtain an evaluation index sequence for multiple time periods;

[0125] (3) Conduct statistical analysis on the evaluation indicator sequences of multiple time periods to obtain the mean and standard deviation of the evaluation indicators, and set the carbon credit token generation threshold based on the mean and standard deviation to obtain a dynamic threshold sequence;

[0126] (4) Compare the evaluation index sequence with the dynamic threshold sequence point by point to obtain a set of super-threshold time points, and perform time clustering on the super-threshold time points to obtain a continuous super-threshold time period;

[0127] (5) Integrate and calculate the carbon emission reduction in the continuous exceeding threshold time period to obtain the cumulative carbon emission reduction in each time period, and design the smart contract triggering conditions based on the cumulative carbon emission reduction to obtain the contract triggering rules;

[0128] (6) Encode the contract triggering rules into a smart contract and deploy it to the blockchain network to obtain an executable smart contract, and generate a carbon credit token for the corresponding period through the executable smart contract to obtain an initial carbon credit token set;

[0129] (7) Digitally sign and encrypt the initial carbon credit token set to obtain a secure carbon credit token data package, and upload the secure carbon credit token data package to the decentralized storage network to obtain a carbon credit token with a unique identifier;

[0130] (8) Construct a smart contract interface for the decentralized trading platform, and associate the carbon credit token with a unique identifier with the smart contract interface to obtain a carbon credit token sequence that can be circulated on the decentralized trading platform.

[0131] Specifically, the long-term trend component is normalized, and the data is scaled to the [0,1] interval using the minimum-maximum normalization method to obtain the standardized long-term trend value. This step makes the data of different time periods comparable. At the same time, the amplitude of the periodic fluctuation component is calculated by calculating the peak-to-valley difference of the fluctuation in each cycle to obtain the fluctuation intensity index. This indicator reflects the degree of fluctuation of carbon emission reduction on different time scales. Based on the standardized long-term trend value and the fluctuation intensity index, a comprehensive carbon emission reduction efficiency evaluation index is constructed. This comprehensive index is usually the weighted sum of two components, and the weight can be adjusted according to actual needs. For example, a higher weight can be given to the long-term trend to reflect the overall performance of the system, while retaining a certain proportion of the periodic fluctuation weight to capture short-term changes. After the construction is completed, the comprehensive carbon emission reduction efficiency evaluation index is divided into time windows, such as one week, to obtain a series of evaluation indicators for multiple time periods.

[0132] Next, the evaluation indicator sequence of multiple time periods is statistically analyzed to calculate the mean and standard deviation of the evaluation indicator in each time window. These statistics reflect the overall level and fluctuation of carbon emission reduction efficiency in different time periods. According to these means and standard deviations, the threshold for generating carbon credit tokens is set. A common method is to set the threshold to the mean plus a standard deviation, which ensures that only time periods that perform better than the average can generate carbon credit tokens. In this way, a dynamically changing threshold sequence is obtained, which can adapt to long-term changes and short-term fluctuations in system performance. The evaluation indicator sequence is compared with the dynamic threshold sequence point by point, and all time points exceeding the threshold are identified to form a set of super-threshold time points. In order to avoid frequent triggering caused by short-term fluctuations, the super-threshold time point set is temporally clustered. Here, a density-based clustering algorithm, such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise), can be used to aggregate points close in time to obtain continuous super-threshold time periods.

[0133] The carbon emission reduction in each continuous super-threshold time period is integrated and calculated to obtain the cumulative carbon emission reduction in each time period. This calculation process takes into account the length of the time period and the intensity of carbon emission reduction, and more accurately reflects the carbon emission reduction contribution of the time period. Design smart contract trigger conditions based on the cumulative carbon emission reduction. For example, a minimum carbon emission reduction threshold can be set, and only time periods exceeding this threshold can trigger the generation of carbon credit tokens. These conditions constitute the contract trigger rules. Encode the contract trigger rules into smart contracts, and use smart contract programming languages ​​such as Solidity to convert the rules into executable code. Deploy the encoded smart contract to a blockchain network, such as Ethereum, to obtain an executable smart contract. Through this executable smart contract, the carbon credit token for the corresponding time period is automatically generated to form an initial carbon credit token set.

[0134] The initial carbon credit token set is digitally signed and encrypted to ensure the authenticity and security of the token. Digital signatures use asymmetric encryption algorithms, such as RSA or elliptic curve encryption, while encryption can use symmetric encryption algorithms, such as AES. After processing, a secure carbon credit token data package is obtained, and these data packages are uploaded to a decentralized storage network, such as IPFS (InterPlanetary File System). Each token obtains a unique content identifier (CID) and becomes a carbon credit token with a unique identifier. Finally, a smart contract interface for a decentralized trading platform is constructed, which defines operations such as trading, transfer, and destruction of tokens. The carbon credit token with a unique identifier is associated with this smart contract interface so that the token can circulate freely on the decentralized trading platform. This association is usually achieved by recording the CID of the token in the state variable of the smart contract, thereby obtaining a sequence of carbon credit tokens that can be circulated on the decentralized trading platform.

[0135] For example, after processing the carbon emission reduction data of a photovoltaic power station within one year, the standardized long-term trend value and fluctuation intensity index were obtained. Assume that the long-term trend value fluctuates between 0.6 and 0.9, and the intensity index of periodic fluctuation varies between 0.05 and 0.2. When constructing the comprehensive index, a 70% weight is given to the long-term trend and a 30% weight is given to the periodic fluctuation. The evaluation index sequence of 52 time periods is obtained by dividing the window into one week. Statistical analysis shows that the mean of these 52 indicators is 0.75 and the standard deviation is 0.08. The threshold is set to the mean plus one standard deviation, that is, 0.83. After comparison, it is found that 15 time periods exceed the threshold. After time clustering, 8 consecutive over-threshold time periods are finally obtained. The integral calculation of these 8 time periods shows that the cumulative carbon emission reduction is 5 tons, 7 tons, 6 tons, 9 tons, 8 tons, 11 tons, 10 tons and 12 tons of CO2 equivalent respectively. The trigger condition of the smart contract is set to the cumulative carbon emission reduction exceeding 8 tons, and finally there are 5 time periods that trigger the generation of carbon credit tokens. These 5 tokens are uploaded to the IPFS network after encryption and signature processing to obtain a unique CID. The smart contract interface associates these 5 CIDs with the corresponding carbon emission reductions to form a tradable carbon credit token, which provides a quantitative and tradable certificate for the carbon emission reduction effect of the photovoltaic power station.

[0136] The above describes the carbon emission data monitoring and management method based on photovoltaic panels in the embodiment of the present application. The following describes the carbon emission data monitoring and management device based on photovoltaic panels in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of a carbon emission data monitoring and management device based on photovoltaic panels includes:

[0137] The acquisition module 201 is used to sample the voltage, current, temperature and light intensity of the photovoltaic panel array in real time to obtain the original data set of the photovoltaic system including the timestamp;

[0138] Filling module 202, used to remove abnormal values ​​and fill missing values ​​from the original data set of the photovoltaic system by using a sliding window median filtering algorithm to obtain photovoltaic power generation time series data;

[0139] An adjustment module 203 is used to adjust the back propagation parameters of the photovoltaic power generation time series data through a long short-term memory network to obtain a photovoltaic system carbon emission reduction prediction function;

[0140] The calculation module 204 is used to perform sliding window calculation on the real-time photovoltaic power generation data collected in real time by using the photovoltaic system carbon emission reduction prediction function to obtain a carbon emission reduction estimation sequence with an interval of 5 minutes;

[0141] An analysis module 205 is used to perform multi-scale analysis on the carbon emission reduction estimation sequence to obtain a long-term trend component and a periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system, wherein the long-term trend component reflects the slow change of system performance over time, and the periodic fluctuation component reflects diurnal and seasonal fluctuations;

[0142] The comparison module 206 is used to construct a carbon emission reduction efficiency evaluation index based on the long-term trend component and the periodic fluctuation component, and compare the carbon emission reduction efficiency evaluation index with a preset threshold. When the carbon emission reduction efficiency evaluation index exceeds the threshold, the smart contract is triggered to generate a carbon credit token for the corresponding period, and a carbon credit token sequence that can be circulated on a decentralized trading platform is obtained.

[0143] Through the coordinated cooperation of the above components, by installing micro sensors on each photovoltaic panel and performing real-time high-frequency sampling, the accuracy and timeliness of data collection are significantly improved, laying a solid data foundation for the subsequent calculation of carbon emission reduction; secondly, the sliding window median filtering algorithm is used to remove outliers and fill missing values ​​in the original data, which effectively improves the data quality and reduces the impact of environmental interference and equipment failure on the calculation results; the use of long short-term memory networks for deep learning of photovoltaic power generation time series data not only improves the accuracy of carbon emission reduction prediction, but also can capture the trend of system performance over time, making the prediction results closer to the actual situation; and through multi-scale analysis of the carbon emission reduction estimation sequence, the long-term trend component and cyclical fluctuation component are obtained, and the carbon emission reduction efficiency evaluation index constructed based on the long-term trend and cyclical fluctuation greatly improves the credibility and transparency of carbon emission reduction calculation; the generation of carbon credit tokens is automatically triggered by smart contracts and deployed to the blockchain network, which not only simplifies the generation and certification process of carbon credits, but also significantly improves the efficiency and security of the entire process, providing convenient conditions for photovoltaic power stations to participate in the carbon trading market. By converting carbon emission reduction data into carbon credit tokens that can be circulated on decentralized trading platforms, new value growth points have been created for photovoltaic power stations, which will help promote the development of the clean energy industry. At the same time, this carbon credit token trading mechanism based on blockchain technology has greatly enhanced the traceability and non-tamperability of carbon emission reduction data. The implementation of this method also helps to improve the operational efficiency of photovoltaic power stations. Through real-time monitoring and analysis of carbon emission reduction data, operators can promptly discover and solve problems in the system, such as component performance degradation, local shading, etc., thereby maximizing the power generation efficiency and carbon emission reduction effect of the photovoltaic system.

[0144] The present application also provides a carbon emission data monitoring and management device based on photovoltaic panels, and the carbon emission data monitoring and management device based on photovoltaic panels includes a memory and a processor, and the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes the steps of the carbon emission data monitoring and management method based on photovoltaic panels in the above-mentioned embodiments.

[0145] Figure 3It is a structural schematic diagram of a carbon emission data monitoring and management device based on a photovoltaic panel provided by an embodiment of the present invention. The carbon emission data monitoring and management device 300 based on a photovoltaic panel may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage devices) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 can be short-term storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the carbon emission data monitoring and management device 300 based on a photovoltaic panel. Furthermore, the processor 310 may be configured to communicate with the storage medium 330 to execute a series of instruction operations in the storage medium 330 on the carbon emission data monitoring and management device 300 based on a photovoltaic panel.

[0146] The photovoltaic panel-based carbon emission data monitoring and management device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will appreciate that Figure 3 The structure of the photovoltaic panel-based carbon emission data monitoring and management device shown does not constitute a limitation on the photovoltaic panel-based carbon emission data monitoring and management device, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0147] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the carbon emission data monitoring and management method based on photovoltaic panels.

[0148] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0149] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0150] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for monitoring and managing carbon emission data based on photovoltaic panels. It is characterized in that The carbon emission data monitoring and management method based on photovoltaic panels includes: The voltage, current, temperature and light intensity of the photovoltaic panel array are sampled in real time to obtain the original data set of the photovoltaic system including timestamp; Using a sliding window median filtering algorithm, outliers are removed from the original data set of the photovoltaic system and missing values ​​are filled to obtain photovoltaic power generation time series data; The photovoltaic power generation time series data is back-propagated and parameterized by a long short-term memory network to obtain a photovoltaic system carbon emission reduction prediction function; The photovoltaic system carbon emission reduction prediction function is used to perform sliding window calculation on the real-time photovoltaic power generation data collected in real time to obtain a carbon emission reduction estimation sequence with an interval of 5 minutes; A multi-scale analysis is performed on the carbon emission reduction estimation sequence to obtain a long-term trend component and a periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system, wherein the long-term trend component reflects the slow change of system performance over time, and the periodic fluctuation component reflects diurnal and seasonal fluctuations, specifically comprising: performing wavelet decomposition on the carbon emission reduction estimation sequence to obtain wavelet coefficients of different scales, and grouping the wavelet coefficients to obtain coefficient groups representing different time scales; performing threshold processing on the coefficient groups representing different time scales to obtain denoised coefficient groups, and performing inverse wavelet transform on the denoised coefficient groups to obtain multiple components of different time scales; performing time series clustering on the multiple components of different time scales to obtain long-term trend clusters and Periodic fluctuation cluster, and reconstruct the long-term trend cluster and the periodic fluctuation cluster respectively to obtain an initial long-term trend component and an initial periodic fluctuation component; perform polynomial fitting on the initial long-term trend component to obtain a smoothed long-term trend curve, and perform slope analysis on the smoothed long-term trend curve to obtain a system performance change rate index; perform Fourier transform on the initial periodic fluctuation component to obtain a spectrum diagram, and perform peak detection on the spectrum diagram to obtain a main periodic component; according to the system performance change rate index and the main periodic component, perform weighted combination on the smoothed long-term trend curve and the initial periodic fluctuation component to obtain a long-term trend component and a periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system; Based on the long-term trend component and the periodic fluctuation component, a carbon emission reduction efficiency evaluation index is constructed, and the carbon emission reduction efficiency evaluation index is compared with a preset threshold. When the carbon emission reduction efficiency evaluation index exceeds the threshold, the smart contract is triggered to generate a carbon credit token for the corresponding time period, and a carbon credit token sequence that can be circulated on a decentralized trading platform is obtained, which specifically includes: normalizing the long-term trend component to obtain a standardized long-term trend value, and calculating the amplitude of the periodic fluctuation component to obtain a fluctuation intensity index; constructing a comprehensive carbon emission reduction efficiency evaluation index based on the standardized long-term trend value and the fluctuation intensity index, and dividing the comprehensive carbon emission reduction efficiency evaluation index into time windows to obtain a sequence of evaluation indicators for multiple time periods; performing statistical analysis on the evaluation indicator sequences for the multiple time periods to obtain the mean and standard deviation of the evaluation indicators, and setting the carbon credit token generation threshold according to the mean and standard deviation to obtain a dynamic threshold sequence; comparing the evaluation indicator sequence with the dynamic threshold sequence point by point Compare and obtain a set of super-threshold time points, and perform time clustering on the super-threshold time point set to obtain a continuous super-threshold time period; integrate and calculate the carbon emission reduction in the continuous super-threshold time period to obtain the cumulative carbon emission reduction in each time period, and design a smart contract trigger condition based on the cumulative carbon emission reduction to obtain a contract trigger rule; encode the contract trigger rule into a smart contract and deploy it to the blockchain network to obtain an executable smart contract, and generate a carbon credit token for the corresponding time period through the executable smart contract to obtain an initial carbon credit token set; digitally sign and encrypt the initial carbon credit token set to obtain a secure carbon credit token data packet, and upload the secure carbon credit token data packet to a decentralized storage network to obtain a carbon credit token with a unique identifier; construct a smart contract interface for a decentralized trading platform, and associate the carbon credit token with a unique identifier with the smart contract interface to obtain a carbon credit token sequence that can be circulated on the decentralized trading platform.

2. According to claim 1, the method for monitoring and managing carbon emission data based on photovoltaic panels, It is characterized in that The expression of the photovoltaic system carbon emission reduction prediction function is as follows: ;in, for The predicted carbon emission reduction value at the time, in kilograms of carbon dioxide equivalent; is the Sigmoid activation function; Output weight matrix for carbon emission reduction; is the hidden state weight matrix; for The hidden state vector at time t; for t The input vector at the moment, including voltage, current, temperature and light intensity; and are the carbon emission reduction output bias vector and the hidden state bias vector respectively; is the hyperbolic tangent activation function.

3. According to claim 1, the method for monitoring and managing carbon emission data based on photovoltaic panels, It is characterized in that The voltage, current, temperature and light intensity of the photovoltaic panel array are sampled in real time to obtain a photovoltaic system raw data set containing a timestamp, including: Installing a micro voltage sensor, a current sensor, a temperature sensor and a light intensity sensor on each photovoltaic panel of the photovoltaic panel array to obtain a sensor array, and calibrating the sensor array to obtain a calibrated sensor array; Performing high-frequency sampling on the photovoltaic panel array according to the calibrated sensor array to obtain an original sampling data stream, and marking the original sampling data stream with a timestamp to obtain an original sampling data stream with a timestamp; Performing data compression on the original sampled data stream with the timestamp to obtain a compressed sampled data stream, and encrypting the compressed sampled data stream to obtain an encrypted compressed sampled data stream; The encrypted compressed sampled data stream is packetized and transmitted through a low-power wireless transmission module to obtain a plurality of data packets, and the plurality of data packets are reassembled in a local data concentrator to obtain a reassembled data stream; Decrypting and decompressing the reorganized data stream to obtain a restored sampled data stream, and performing data integrity verification on the restored sampled data stream to obtain a verified sampled data stream; Generate a structured photovoltaic system original data set according to the verified sampled data stream, and create an index for the photovoltaic system original data set to obtain an indexed photovoltaic system original data set; The indexed photovoltaic system original data set is sharded to obtain a plurality of data shards, and the plurality of data shards are distributedly stored to obtain a distributedly stored photovoltaic system original data set.

4. The method for monitoring and managing carbon emission data based on photovoltaic panels according to claim 1, It is characterized in that The photovoltaic system original data set is cleared of outliers and filled with missing values ​​by a sliding window median filtering algorithm to obtain photovoltaic power generation time series data, including: Performing time series segmentation on the original data set of the photovoltaic system to obtain a data subset of 24 hours in length, and performing overlapping sliding window division on the data subset with a width of 10 minutes and a step length of 5 minutes to obtain 144 overlapping data windows; The median of the voltage, current, temperature and light intensity data in each data window is calculated respectively to obtain the median sequence of each parameter, and the upper and lower limits of 3 times the interquartile range are calculated according to the median sequence to obtain the outlier identification threshold; According to the outlier recognition threshold, the photovoltaic system original data set is compared point by point to obtain a binary outlier mark sequence, and data points marked as abnormal are removed according to the binary outlier mark sequence to obtain a preliminarily cleaned data set; Perform missing value detection on the data set after preliminary cleaning to obtain the timestamp index of the missing value, and perform linear regression analysis on 48 adjacent non-missing data points according to the timestamp index of the missing value to obtain candidate filling values ​​for each missing position; Smoothing the candidate filling values ​​by a cubic spline interpolation algorithm to obtain an interpolated filling value sequence, and inserting the interpolated filling value sequence into corresponding positions of the data set after preliminary cleaning to obtain a filled data set; The filled data set is time aligned and resampled at 5-minute intervals to obtain photovoltaic power generation time series data at equal intervals, and each parameter in the photovoltaic power generation time series data is normalized to obtain standardized photovoltaic power generation time series data.

5. The method for monitoring and managing carbon emission data based on photovoltaic panels according to claim 1, It is characterized in that The back propagation parameter adjustment of the photovoltaic power generation time series data through the long short-term memory network to obtain the photovoltaic system carbon emission reduction prediction function includes: Dividing the photovoltaic power generation time series data into time windows to obtain an input sequence of fixed length, and normalizing the input sequence to obtain standardized input data; Constructing a long short-term memory network structure, including an input layer, a hidden layer and an output layer, and setting the number of neurons in the hidden layer to obtain an initial network structure; Randomly initializing the weight matrix and bias vector in the initial network structure to obtain an initial parameter set, and calculating the forward propagation output of the network according to the initial parameter set to obtain a prediction result; The prediction result is compared with the actual carbon emission reduction data, the mean square error is calculated as the loss function, and the gradient of the loss function to each parameter is calculated by the back propagation algorithm to obtain a gradient matrix; The network parameters are updated according to the gradient matrix to obtain a new parameter set, and the forward propagation, loss calculation and back propagation steps are repeated until a preset number of iterations or convergence conditions are reached to obtain the trained network parameters, and a carbon emission reduction prediction function is constructed using the trained network parameters.

6. The method for monitoring and managing carbon emission data based on photovoltaic panels according to claim 1, It is characterized in that The photovoltaic system carbon emission reduction prediction function performs sliding window calculation on the real-time photovoltaic power generation data collected in real time to obtain a carbon emission reduction estimation sequence at intervals of 5 minutes, including: Performing time synchronization calibration on the real-time collected photovoltaic power generation data to obtain a calibrated real-time power data stream, and dividing the calibrated real-time power data stream into 5-minute time segments to obtain data segments of equal length; Performing maximum and minimum value scaling on the voltage, current, temperature and light intensity data in the data segments of equal length to obtain a normalized feature vector, and extracting a state vector at the previous moment from a memory unit to obtain function input data; Inputting the function input data into the photovoltaic system carbon emission reduction prediction function to obtain a preliminary estimate of the carbon emission reduction for the current time segment; Performing reverse scaling processing on the preliminary estimated value of carbon emission reduction to obtain a carbon emission reduction value in actual units, and pairing the carbon emission reduction value with a corresponding time stamp to obtain a time-stamped carbon emission reduction data point; Appending the time-stamped carbon emission reduction data point to the carbon emission reduction data set to obtain an extended data set, and sliding the time segment backward by 5 minutes to obtain input data for the next time segment; The expanded data set is time-series filled to complete all 5-minute interval points in the monitoring period to obtain a carbon emission reduction time series, and the carbon emission reduction time series is verified for data continuity to obtain a carbon emission reduction estimation series with 5-minute intervals.

7. A photovoltaic panel-based carbon emission data monitoring and management device, used to execute the photovoltaic panel-based carbon emission data monitoring and management method as described in any one of claims 1 to 6, It is characterized in that The photovoltaic panel-based carbon emission data monitoring and management device comprises: The acquisition module is used to sample the voltage, current, temperature and light intensity of the photovoltaic panel array in real time to obtain the original data set of the photovoltaic system including the timestamp; A filling module is used to remove outliers and fill missing values ​​from the original data set of the photovoltaic system by a sliding window median filtering algorithm to obtain photovoltaic power generation time series data; An adjustment module is used to adjust the back propagation parameters of the photovoltaic power generation time series data through a long short-term memory network to obtain a photovoltaic system carbon emission reduction prediction function; A calculation module, used to perform sliding window calculation on the real-time photovoltaic power generation data collected in real time through the photovoltaic system carbon emission reduction prediction function to obtain a carbon emission reduction estimation sequence with an interval of 5 minutes; The analysis module is used to perform multi-scale analysis on the carbon emission reduction estimation sequence to obtain a long-term trend component and a periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system, wherein the long-term trend component reflects the slow change of system performance over time, and the periodic fluctuation component reflects diurnal and seasonal fluctuations, specifically comprising: performing wavelet decomposition on the carbon emission reduction estimation sequence to obtain wavelet coefficients of different scales, and grouping the wavelet coefficients to obtain coefficient groups representing different time scales; performing threshold processing on the coefficient groups representing different time scales to obtain denoised coefficient groups, and performing inverse wavelet transform on the denoised coefficient groups to obtain multiple components of different time scales; performing time series clustering on the multiple components of different time scales to obtain a long-term trend component; potential cluster and periodic fluctuation cluster, and reconstruct the long-term trend cluster and periodic fluctuation cluster respectively to obtain an initial long-term trend component and an initial periodic fluctuation component; perform polynomial fitting on the initial long-term trend component to obtain a smoothed long-term trend curve, and perform slope analysis on the smoothed long-term trend curve to obtain a system performance change rate index; perform Fourier transform on the initial periodic fluctuation component to obtain a spectrum diagram, and perform peak detection on the spectrum diagram to obtain a main periodic component; according to the system performance change rate index and the main periodic component, perform weighted combination on the smoothed long-term trend curve and the initial periodic fluctuation component to obtain a long-term trend component and a periodic fluctuation component that characterize the carbon emission reduction efficiency of the photovoltaic system; A comparison module is used to construct a carbon emission reduction efficiency evaluation index based on the long-term trend component and the periodic fluctuation component, and compare the carbon emission reduction efficiency evaluation index with a preset threshold value. When the carbon emission reduction efficiency evaluation index exceeds the threshold value, the smart contract is triggered to generate a carbon credit token for the corresponding time period, and a carbon credit token sequence that can be circulated on a decentralized trading platform is obtained, specifically including: normalizing the long-term trend component to obtain a standardized long-term trend value, and calculating the amplitude of the periodic fluctuation component to obtain a fluctuation intensity index; constructing a comprehensive carbon emission reduction efficiency evaluation index based on the standardized long-term trend value and the fluctuation intensity index, and dividing the comprehensive carbon emission reduction efficiency evaluation index into time windows to obtain a sequence of evaluation indicators for multiple time periods; performing statistical analysis on the evaluation indicator sequences for the multiple time periods to obtain the mean and standard deviation of the evaluation indicator, and setting the carbon credit token generation threshold according to the mean and standard deviation to obtain a dynamic threshold sequence; comparing the evaluation indicator sequence with the dynamic threshold sequence A point-by-point comparison is performed to obtain a set of super-threshold time points, and the super-threshold time point set is time-clustered to obtain a continuous super-threshold time period; the carbon emission reduction in the continuous super-threshold time period is integrated and calculated to obtain the cumulative carbon emission reduction in each time period, and a smart contract trigger condition is designed according to the cumulative carbon emission reduction to obtain a contract trigger rule; the contract trigger rule is encoded into a smart contract and deployed to a blockchain network to obtain an executable smart contract, and a carbon credit token for the corresponding time period is generated through the executable smart contract to obtain an initial carbon credit token set; the initial carbon credit token set is digitally signed and encrypted to obtain a secure carbon credit token data packet, and the secure carbon credit token data packet is uploaded to a decentralized storage network to obtain a carbon credit token with a unique identifier; a smart contract interface of a decentralized trading platform is constructed, and the carbon credit token with a unique identifier is associated with the smart contract interface to obtain a carbon credit token sequence that can be circulated on the decentralized trading platform.

8. A computer-readable storage medium having instructions stored thereon, It is characterized in that When the instructions are executed by the processor, a carbon emission data monitoring and management method based on photovoltaic panels as described in any one of claims 1 to 6 is implemented.

Images