Block chain-based energy storage transaction and carbon credit management system and method

Abstract

The invention discloses an energy storage transaction and carbon credit management system and method based on a block chain, and relates to the technical field of block chains, and the system comprises four modules: a dynamic carbon factor accounting module obtains data in real time, calculates the carbon emission reduction amount, and generates a carbon credit value; the heterogeneous cross-chain interoperation transaction module is connected with an energy chain and a carbon chain, and binds carbon credit and transaction records; the multi-target reinforcement learning energy storage scheduling optimization module is combined with a multi-target deep network algorithm to generate a strategy and a proportion; the enhanced carbon account module stores data and generates a voucher. The system has the advantages that data are acquired in real time through the dynamic carbon factor accounting module, the equivalent carbon emission reduction amount is calculated, the dynamic carbon credit value is generated, the problem of carbon credit calculation lag is solved, and accurate calculation and timely issuing of the carbon credit are realized; and meanwhile, by utilizing a hash time lock and a zero-knowledge proof verification mechanism of the heterogeneous cross-chain interoperation transaction module, the carbon credit is bound with the transaction record, and the energy chain and the carbon chain are connected, so that the mobility of the transaction and the integration of carbon emission reduction are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of blockchain, and specifically to a blockchain-based energy storage trading and carbon credit management system and method. Background Art

[0002] With the increasing global attention to sustainable development, carbon emission reduction and energy efficiency improvement have become the common goals of governments and societies around the world. Especially in the context of addressing climate change, promoting the development of a low-carbon economy, the application of green energy, and the innovation of energy storage technologies have become key measures. As an important means to achieve the consumption of clean energy, energy storage technologies are widely used in the power system to help balance energy supply and demand and reduce carbon emissions. However, the operation and optimization of energy storage systems still face challenges. Especially in the process of effectively managing carbon emission reduction, how to ensure the effective linkage between energy storage and carbon credits has become an important topic in technological development. In order to achieve more transparent, efficient, and compliant carbon credit management, while promoting the optimization of energy trading, blockchain technology, due to its characteristics such as decentralization, immutability, and data transparency, has become an important technical means to solve this problem. The energy storage trading and carbon credit management system combined with blockchain can achieve the accurate calculation and management of energy trading and carbon emission reduction certificates, while protecting user privacy, ensuring the auditability of regulatory agencies, and providing innovative support for the future development of the carbon trading market and smart grid.

[0003] The existing energy storage trading and carbon credit management methods fail to achieve real-time synchronization in the actual operation of carbon credit accounting and energy storage scheduling, resulting in a lag in the calculation of carbon emission reduction, which affects the timely issuance of carbon credit certificates; the energy trading and carbon credit management systems often operate independently, and cross-system trading and data sharing face high technical thresholds and lack effective cross-chain mechanisms, restricting the liquidity of transactions and the integration of carbon emission reduction; due to the centralized storage method of carbon credit trading data, the transparency and credibility of data are threatened, prone to tampering or abuse, and the audit ability of regulatory agencies is limited, unable to fully guarantee the fairness and compliance of the market. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a blockchain-based energy storage trading and carbon credit management system and method, which solve the problems in the above background art.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A blockchain-based energy storage trading and carbon credit management system and method, including the following steps: the following modules: a dynamic carbon factor accounting module, a heterogeneous cross-chain interoperable trading module, a multi-objective reinforcement learning energy storage scheduling optimization module, and an enhanced carbon account module; the dynamic carbon factor accounting module is used to obtain the grid marginal carbon emission factor and the energy storage device operation data in real time, calculate the equivalent carbon emission reduction amount generated by the energy storage charging and discharging behavior through a dynamic carbon factor model, and generate a dynamic carbon credit value; the heterogeneous cross-chain interoperable trading module is used to connect the energy chain and the carbon chain, and through a hash time lock and a zero-knowledge proof verification mechanism, atomically bind the dynamic carbon credit value to the energy storage trading record; the multi-objective reinforcement learning energy storage scheduling optimization module is used to receive the real-time electricity price, carbon price, and grid marginal carbon emission factor of the energy chain, combine the state of charge of the energy storage device, and dynamically generate an energy storage trading strategy and a carbon credit allocation ratio through a multi-objective deep network algorithm, and feedback the decision instruction to the energy storage device control end; the enhanced carbon account module is used to store carbon credit trading data and generate a verification carbon emission reduction certificate through zero-knowledge proof.

[0006] Further, in the dynamic carbon factor accounting module, the energy storage device operation data includes the rated capacity of the energy storage device, the state of charge at the start of charging, the state of charge at the end of charging, the charging efficiency, the state of charge at the start of discharging, the state of charge at the end of discharging, and the discharging efficiency. Data cleaning and normalization processing are performed on the grid marginal carbon emission factor and the energy storage device operation data, and the equivalent carbon emission reduction amount is calculated through the dynamic carbon factor model for the grid marginal carbon emission factor and the energy storage device operation data after data cleaning and normalization processing. The specific method is as follows: Among them, DP represents the equivalent carbon emission reduction amount, TY represents the grid marginal carbon emission factor, ED represents the rated capacity of the energy storage device, CQ represents the state of charge at the start of charging, CJ represents the state of charge at the end of charging, CX represents the charging efficiency, FX represents the discharging efficiency, FQ represents the state of charge at the start of discharging, and FJ represents the state of charge at the end of discharging.

[0007] Further, in the heterogeneous cross-chain interoperable trading module, the specific steps of the hash time lock are as follows: A first randomly generates a dynamic value, then converts the dynamic value into a hash value through a hash function, A sends the hash value to B and sets a time lock T, B provides a dynamic value that matches the hash value within the time lock T, and finally A sends the dynamic carbon credit value to B to complete the energy storage trading record.

[0008] Further, the specific steps of converting the dynamic value into a hash value through the hash function are as follows: First, convert the dynamic value into binary according to UTF-8 encoding to obtain binary data, and expand the length of the binary data to meet the requirement of taking the remainder of 448 when divided by 512 to obtain the expanded data; then use 8 32-bit constants preset by the SHA-256 algorithm as the initial hash values; divide the expanded data into 512-bit message blocks, expand each message block into 64 32-bit words first, then initialize the working variables with 8 initial hash values, and then perform 64 rounds of loop compression operations, update the working variables according to the logical function and constants, and finally update the hash values; finally, connect the 8 32-bit hash values obtained after processing all the message blocks, convert the 256-bit binary number into a hexadecimal string, which is the final hash value.

[0009] Further, the specific steps of using 8 32-bit constants preset by the SHA-256 algorithm as the initial hash values are as follows: S1: Set the golden ratio constant, and obtain a new number by calculating the product of the square roots of irrational numbers related to the golden ratio constant; S2: Take the fractional part of the new number and convert it into binary, and intercept the first 32 bits to form a 32-bit binary word; S3: Convert the 32-bit binary word into hexadecimal representation according to the rule that 4-bit binary corresponds to 1-bit hexadecimal to obtain the initial hash value.

[0010] Furthermore, in the multi-objective reinforcement learning energy storage scheduling optimization module, the real-time electricity price, carbon price, grid marginal carbon emission factor, and the state of charge of the energy storage device in the energy chain are normalized to obtain processed data. The specific steps of calculating the energy storage trading strategy and carbon credit allocation ratio through the multi-objective deep network algorithm are as follows: S1: Select a multi-layer perceptron (MLP) architecture. According to the four features of the real-time electricity price, carbon price, grid marginal carbon emission factor, and the state of charge of the energy storage device in the input data, determine that the number of neurons in the input layer is 4; the number of hidden layers and neurons is determined through experiments and cross-validation. Set 2 - 3 hidden layers, with 32 - 128 neurons in each layer; the number of neurons in the output layer is 2, which respectively output the energy storage trading strategy and the carbon credit allocation ratio; use the Xavier initialization method to determine the initial value range of the weights according to the number of input and output neurons; S2: Use the weighted mean square error as the loss function, comprehensively consider the prediction errors of the energy storage trading strategy and the carbon credit allocation ratio, and balance the importance of the two objectives by adjusting the weights; use the Adam optimization algorithm to adaptively adjust the learning rate to update the network weights, set an appropriate initial learning rate, and divide the data set into a training set, a validation set, and a test set according to the ratio of 70%, 15%, and 15%; during the training process, judge whether the model converges according to the loss of the validation set. If the loss of the validation set does not decrease for several consecutive rounds, stop training; after training, use the test set to evaluate the generalization ability of the model; S3: Input the new normalized real-time electricity price, carbon price, grid marginal carbon emission factor, and the state of charge of the energy storage device data into the trained model; the two values output by the model respectively correspond to the energy storage trading strategy and the carbon credit allocation ratio; judge the charge and discharge operations of the energy storage device according to the positive or negative value of the output value of the energy storage trading strategy, and the numerical size represents the charge and discharge power; according to the output value of the carbon credit allocation ratio, determine the specific carbon credit allocation plan in combination with the actual situation to achieve the collaborative decision-making of energy storage trading and carbon credit management.

[0011] Furthermore, the specific steps of using the Xavier initialization method to determine the initial value range of the weights according to the number of input and output neurons are as follows: For a layer with x input neurons and y output neurons, the initial value of the weights is set within a uniform distribution where U represents the distribution range of the initial value of the weights, c represents a fixed constant, x represents the number of input neuron layers, and y represents the number of output neuron layers; randomly sample a value for each connection weight of this layer as the initial value to complete the Xavier initialization of the weights.

[0012] Further, the specific steps of adaptively adjusting the learning rate to update the network weights by using the Adam optimization algorithm are as follows: S1: Input the real-time electricity price, carbon price, grid marginal carbon emission factor, and the state of charge of the energy storage device into the neural network. First, perform forward propagation calculation to obtain the output, then calculate the loss value according to the loss function, and then use the backpropagation algorithm to calculate the gradient of the loss with respect to the network weights; S2: Update the first-order matrix estimate and the second-order matrix estimate based on the gradient of the network weights obtained in S1 and correct the first-order matrix estimate and the second-order matrix estimate; S3: According to the corrected first-order matrix estimate and second-order matrix estimate, use the formula where, DI t+1 represents the network weight at the (t + 1)-th time, DI t represents the network weight at the t-th time, XL represents the learning rate, EG represents the second-order matrix estimate, CS represents a constant, YG represents the first-order matrix estimate. The network weights are adaptively adjusted through this formula, and then repeat the steps from S1 to S3, continuously iteratively updating the network weights until the training stop condition is met.

[0013] Further, in the enhanced carbon account module, the distributed ledger technology is combined with the traditional database storage, and the data is stored in the form of blocks in the blockchain and hash connections are established.

[0014] Further, Step 1: Obtain the grid marginal carbon emission factor and the operation data of the energy storage device in real time, calculate the equivalent carbon emission reduction amount generated by the charging and discharging behavior of the energy storage through the dynamic carbon factor model, and generate a dynamic carbon credit value; Step 2: Connect the energy chain and the carbon chain, and through the hash time lock and zero-knowledge proof verification mechanism, atomically bind the dynamic carbon credit value to the energy storage transaction record; Step 3: Receive the real-time electricity price, carbon price, and grid marginal carbon emission factor of the energy chain, combine with the state of charge of the energy storage device, dynamically generate the energy storage transaction strategy and the carbon credit allocation ratio through the multi-objective deep network algorithm, and feedback the decision instruction to the control end of the energy storage device; Step 4: Store the carbon credit transaction data and generate a verification carbon emission reduction certificate through zero-knowledge proof.

[0015] The present invention has the following beneficial effects:

[0016] (1) The energy storage transaction and carbon credit management system and method based on the blockchain achieve the accurate calculation and timely issuance of carbon credits: The dynamic carbon factor accounting module obtains data in real time and calculates the equivalent carbon emission reduction amount to generate a dynamic carbon credit value, solving the problem of calculation lag caused by the asynchrony between carbon credit accounting and energy storage scheduling, and ensuring the timely issuance of carbon credit certificates.

[0017] (2) The energy storage trading and carbon credit management system and method based on blockchain improve the liquidity of trading and the integration of carbon emission reduction: By using the heterogeneous cross-chain interoperable trading module and the verification mechanisms of hash time lock and zero-knowledge proof, the energy chain and the carbon chain are connected, and the dynamic carbon credit value is atomically bound to the energy storage trading record, solving the problems that the energy trading and carbon credit management systems operate independently and lack a cross-chain mechanism, and promoting the liquidity of trading and the integration of carbon emission reduction.

[0018] Of course, it is not necessary for any product implementing the present invention to achieve all the above advantages simultaneously. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a structural diagram of the energy storage trading and carbon credit management system based on blockchain of the present invention.

[0020] Figure 2 It is a flowchart of the energy storage trading and carbon credit management method based on blockchain of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] In the embodiments of the present application, through the energy storage trading and carbon credit management system and method based on blockchain, the problems that the carbon credit accounting and energy storage scheduling are out of sync, resulting in a lag in the calculation of carbon emission reduction and a delay in the issuance of carbon credit certificates, the energy trading and carbon credit management systems operate independently and lack a cross-chain mechanism, restricting the trading liquidity and the integration of carbon emission reduction, and the carbon credit trading data is centrally stored, resulting in limited transparency and credibility, are solved.

[0022] For the solution in the embodiments of the present application, please refer to Figure 2 , and the general idea is as follows:

[0023] Step 1: Obtain the grid marginal carbon emission factor and the operation data of the energy storage device in real time, calculate the equivalent carbon emission reduction amount generated by the charging and discharging behavior of the energy storage through the dynamic carbon factor model, and generate a dynamic carbon credit value;

[0024] Step 2: Connect the energy chain and the carbon chain, and atomically bind the dynamic carbon credit value to the energy storage trading record through the verification mechanisms of hash time lock and zero-knowledge proof;

[0025] Step 3: Receive the real-time electricity price, carbon price and grid marginal carbon emission factor of the energy chain, combine with the state of charge of the energy storage device, dynamically generate an energy storage trading strategy and a carbon credit allocation ratio through the multi-objective deep network algorithm, and feedback the decision instruction to the control end of the energy storage device;

[0026] Step 4: Store the carbon credit trading data, and generate a verified carbon emission reduction certificate through zero-knowledge proof.

[0027] Please refer to Figure 1, an embodiment of the present invention provides a technical solution: a blockchain-based energy storage trading and carbon credit management system, including the following modules: The dynamic carbon factor accounting module is used to obtain the grid marginal carbon emission factor and energy storage device operation data in real time, calculate the equivalent carbon emission reduction amount generated by the energy storage charging and discharging behavior through the dynamic carbon factor model, and generate a dynamic carbon credit value;

[0028] The heterogeneous cross-chain interoperability trading module is used to connect the energy chain and the carbon chain, and through the hash time lock and zero-knowledge proof verification mechanism, atomically bind the dynamic carbon credit value with the energy storage trading record;

[0029] The multi-objective reinforcement learning energy storage scheduling optimization module is used to receive the real-time electricity price, carbon price and grid marginal carbon emission factor of the energy chain, combine the state of charge of the energy storage device, and dynamically generate an energy storage trading strategy and a carbon credit allocation ratio through the multi-objective deep network algorithm, and feedback the decision-making instruction to the control end of the energy storage device;

[0030] The enhanced carbon account module is used to store carbon credit trading data and generate a verified carbon emission reduction certificate through zero-knowledge proof.

[0031] In the dynamic carbon factor accounting module, the energy storage device operation data includes the rated capacity of the energy storage device, the state of charge at the start of charging, the state of charge at the end of charging, the charging efficiency, the state of charge at the start of discharging, the state of charge at the end of discharging, and the discharging efficiency. Data cleaning and normalization processing of the grid marginal carbon emission factor and the energy storage device operation data helps to remove noise, errors and outliers in the data, correct inconsistent data, fill in missing values, and enable different types of energy storage device operation data and grid marginal carbon emission factors to be compared and comprehensively analyzed under a unified standard, avoiding calculation errors caused by data dimension differences. The specific method for calculating the equivalent carbon emission reduction amount by calculating the grid marginal carbon emission factor and the energy storage device operation data after data cleaning and normalization processing through the dynamic carbon factor model is as follows:

[0032]

[0033] Among them, DP represents the equivalent carbon emission reduction, which is an important indicator to measure the contribution of energy storage equipment in carbon emission reduction. TY represents the grid marginal carbon emission factor, which reflects the carbon emission intensity of the grid. ED represents the rated capacity of the energy storage equipment, which is a key parameter to measure the scale and capacity of the energy storage equipment. CQ represents the state of charge at the start of charging, which reflects the proportion of the remaining power in the battery. CJ represents the state of charge at the end of charging, which is used to calculate the change in power during the charging process. CX represents the charging efficiency, which reflects the conversion efficiency of electrical energy during the charging process. FX represents the discharging efficiency, which reflects the conversion efficiency of electrical energy during the discharging process. FQ represents the state of charge at the start of discharging, which reflects the conversion efficiency of electrical energy during the discharging process. FJ represents the state of charge at the end of discharging, which is used to calculate the change in power during the discharging process.

[0034] In the heterogeneous cross-chain interoperability transaction module, the specific steps of the hash time lock are as follows: A first randomly generates a dynamic value, which is the key information for subsequent transaction verification, providing a unique identifier and security guarantee for the entire transaction process. Then, the dynamic value is converted into a hash value through a hash function. The hash function has one-wayness, that is, it is almost impossible to reverse the original dynamic value from the hash value, which provides an additional layer of security for the transaction, preventing the dynamic value from being tampered with or leaked during transmission. A sends the hash value to B and sets a time lock T to prevent the transaction from being delayed indefinitely. B provides a dynamic value that matches the hash value within the time lock T, ensuring the security and conditional nature of the transaction. Finally, A sends the dynamic carbon credit value to B to complete the energy storage transaction record, realizing the transfer and confirmation of carbon credits between the two parties.

[0035] The specific steps for converting the dynamic value into a hash value through the hash function are as follows:

[0036] First, convert the dynamic value to binary according to UTF-8 encoding to obtain binary data, ensuring the integrity of the original information of the dynamic value during the conversion process. Expand the length of the binary data to meet the requirement of taking the remainder of 448 when divided by 512 to obtain the expanded data, ensuring that the data can be correctly block-divided and processed by the algorithm, and guaranteeing the standardization and consistency of the hash calculation. Then, use the 8 32-bit constants preset by the SHA-256 algorithm as the initial hash values to provide a fixed starting point for subsequent hash calculations. Divide the expanded data into 512-bit message blocks. For each message block, first expand it into 64 32-bit words, then initialize the working variables with the 8 initial hash values, and then perform 64 rounds of loop compression operations. Update the working variables according to the logical function and constants, and finally update the hash values, which can fully mix and compress the information in the data, making the finally generated hash values highly random and unpredictable. Finally, connect the 8 32-bit hash values obtained after processing all message blocks to get a 256-bit binary number and convert it into a hexadecimal string, which is the final hash value. The final hash value is used as the unique identifier of the dynamic value, with a fixed length, high randomness, and uniqueness, and can be used to verify the integrity and authenticity of the data.

[0037] The specific steps for using the 8 32-bit constants preset by the SHA-256 algorithm as the initial hash values are as follows:

[0038] S1: Set the golden ratio constant. The selection of the golden ratio constant helps to ensure that the generated initial hash values have good statistical characteristics and unpredictability. By calculating the product of the square roots of irrational numbers related to the golden ratio constant, a new number is obtained, introducing a value with high randomness and good distribution characteristics, providing a basis for subsequent generation of initial hash values.

[0039] S2: Take the fractional part of the new number and convert it to binary, and intercept the first 32 bits to form a 32-bit binary word. Convert the irrational number obtained by mathematical calculation into a binary data form that can be processed by a computer, and ensure the fixed length and accuracy of the data by intercepting the first 32 bits, providing convenience for subsequent conversion and use.

[0040] S3: Convert the 32-bit binary word to hexadecimal representation according to the rule that 4-bit binary corresponds to 1-bit hexadecimal to obtain the initial hash value.

[0041] In the multi-objective reinforcement learning energy storage scheduling optimization module, the real-time electricity price, carbon price, grid marginal carbon emission factor, and the state of charge of the energy storage device in the energy chain are normalized to obtain processed data, making the data distribution more uniform and reducing the model bias caused by the difference in feature scales. The specific steps for calculating the energy storage trading strategy and the carbon credit allocation ratio by using the multi-objective deep network algorithm for the processed data are as follows:

[0042] S1: Select a multi-layer perceptron (MLP) architecture, which is suitable for complex multi-objective optimization tasks. Determine the number of input layer neurons to be 4 based on the four characteristics of the input data: real-time electricity price, carbon price, grid marginal carbon emission factor, and the state of charge of the energy storage device. Using them as inputs can comprehensively capture the dynamic characteristics of the system. The number of hidden layers and neurons is determined through experiments and cross-validation. Set 2 - 3 hidden layers, with 32 - 128 neurons in each layer, which can achieve a balance between computational complexity and model performance. The number of output layer neurons is 2, which respectively output the energy storage trading strategy and the carbon credit allocation ratio, enabling the output of the model to be directly used for decision-making and optimization. Use the Xavier initialization method to determine the initial value range of the weights according to the number of input and output neurons, avoiding the problems of gradient disappearance or gradient explosion.

[0043] S2: Adopt the weighted mean square error as the loss function, comprehensively considering the prediction errors of the energy storage trading strategy and the carbon credit allocation ratio. By adjusting the weights, the importance of the two objectives can be balanced. Use the Adam optimization algorithm to adaptively adjust the learning rate to update the network weights, which helps to find the optimal solution during the network weight update process. Set an appropriate initial learning rate, and divide the dataset into a training set, a validation set, and a test set at a ratio of 70%, 15%, and 15% to adjust the model parameters and prevent overfitting. During the training process, judge whether the model converges based on the loss of the validation set. If the loss of the validation set does not decrease for several consecutive rounds, stop the training. Judging whether the model converges through the loss of the validation set can effectively prevent overfitting. After the training is completed, use the test set to evaluate the generalization ability of the model to ensure that the model has good generalization ability.

[0044] S3: Input the new normalized real-time electricity price, carbon price, grid marginal carbon emission factor, and the state of charge of the energy storage device data into the trained model, enabling the model to correctly process new data. The two values output by the model respectively correspond to the energy storage trading strategy and the carbon credit allocation ratio, making the decision-making process more efficient. Judge the charging and discharging operations of the energy storage device according to the positive or negative value of the output value of the energy storage trading strategy, and the numerical size represents the charging and discharging power. Based on the output value of the carbon credit allocation ratio, determine the specific carbon credit allocation plan in combination with the actual situation, making the carbon credit management more scientific and reasonable, which can effectively support the carbon emission reduction goal and realize the collaborative decision-making of energy storage trading and carbon credit management. This not only improves the overall efficiency of the system but also better supports the goals of carbon emission reduction and energy management.

[0045] The specific steps to determine the initial value range of the weights according to the number of input and output neurons using the Xavier initialization method are as follows:

[0046] For a layer with x input neurons and y output neurons using the Xavier initialization method, where x and y represent the number of neurons in the input layer and output layer respectively, ensure that the initialization range is adapted to the scale of the layer, and set the initial weight values within a uniform distribution where U represents the distribution range of the initial weight values, c represents a fixed constant, x represents the number of layers of input neurons, y represents the number of layers of output neurons, and the interval range here is calculated based on the number of input and output neurons. The purpose is to make the initial weight values neither too large nor too small, avoiding the problems of gradient vanishing or gradient explosion in the initial stage of training; randomly sample a value for each connection weight of this layer as the initial value to complete the Xavier initialization of the weights. This step ensures that the initial weight values not only conform to the theoretical guidance of Xavier initialization but also have practical operability, laying a good foundation for subsequent model training.

[0047] The specific steps to adaptively adjust the learning rate and update the network weights using the Adam optimization algorithm are as follows:

[0048] S1: Input the real-time electricity price, carbon price, grid marginal carbon emission factor, and the state of charge of the energy storage device into the neural network. First, perform forward propagation calculation to obtain the predicted values of the energy storage trading strategy and the carbon credit allocation ratio. Then, calculate the loss value according to the loss function. By calculating the loss value, the performance and accuracy of the model can be evaluated, providing a basis for subsequent weight updates. Then, use the backpropagation algorithm to calculate the gradient of the loss with respect to the network weights. These gradient values indicate the direction and magnitude of weight adjustment to minimize the loss function;

[0049] S2: Update and correct the first-order matrix estimate and the second-order matrix estimate based on the gradients of the network weights obtained in S1. These estimated values are updated through exponential weighted averaging. The first-order moment estimate is used to capture the average change trend of the gradients, and the second-order moment estimate is used to capture the fluctuation of the gradients;

[0050] S3: According to the corrected first-order matrix estimate and second-order matrix estimate, the learning rate can be automatically adjusted according to the change of the gradients, improving the training efficiency and convergence speed of the model. Use the formula where DI t+1 represents the network weights at the (t + 1)-th time, DI t represents the network weights at the t-th time, XL represents the learning rate, EG represents the second-order matrix estimate, CS represents a constant, YG represents the first-order matrix estimate. Adaptively adjust the network weights through this formula, and then repeat the steps from S1 to S3, continuously iteratively updating the network weights to improve the prediction accuracy until the training stop condition is met.

[0051] In the enhanced carbon accounting module, distributed ledger technology is combined with traditional database storage. Data is stored in the form of blocks in the blockchain and hash connections are established. Through the hash connections, the privacy and security of the data can be protected, preventing the data from being illegally accessed and tampered with.

[0052] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0053] The present invention is described with reference to the flowcharts and / or block diagrams of systems, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0054] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0055] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are performed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0056] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic creative concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present invention.

[0057] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. The energy storage transaction and carbon credit management system based on blockchain is characterized by: It includes the following modules: dynamic carbon factor accounting module, heterogeneous cross-chain interoperability transaction module, multi-objective reinforcement learning energy storage scheduling optimization module, and enhanced carbon account module; The dynamic carbon factor accounting module is used to obtain the marginal carbon emission factor of the power grid and the operation data of the energy storage equipment in real time, calculate the equivalent carbon emission reduction generated by the energy storage charging and discharging behavior through the dynamic carbon factor model, and generate a dynamic carbon credit value; The heterogeneous cross-chain interoperable transaction module is used to connect the energy chain and the carbon chain, and atomically bind the dynamic carbon credit value to the energy storage transaction record through the hash time lock and zero-knowledge proof verification mechanism; The multi-objective reinforcement learning energy storage scheduling optimization module is used to receive the real-time electricity price, carbon price and marginal carbon emission factor of the energy chain, and dynamically generate energy storage trading strategies and carbon credit allocation ratios through a multi-objective deep network algorithm in combination with the charge state of the energy storage equipment, and feed back the decision instructions to the energy storage equipment control terminal; The enhanced carbon account module is used to store carbon credit transaction data and generate and verify carbon emission reduction certificates through zero-knowledge proof.

2. The blockchain-based energy storage transaction and carbon credit management system according to claim 1 is characterized in that: In the dynamic carbon factor calculation module, the energy storage device operation data includes the rated capacity of the energy storage device, the state of charge at the start of charging, the state of charge at the end of charging, the charging efficiency, the state of charge at the start of discharging, the state of charge at the end of discharging, and the discharge efficiency. The marginal carbon emission factor of the power grid and the energy storage device operation data are cleaned and normalized. The marginal carbon emission factor of the power grid and the energy storage device operation data after data cleaning and normalization are calculated by the dynamic carbon factor model to obtain the equivalent carbon emission reduction. The specific method is as follows: Among them, DP represents the equivalent carbon emission reduction, TY represents the marginal carbon emission factor of the power grid, ED represents the rated capacity of the energy storage device, CQ represents the state of charge at the start of charging, CJ represents the state of charge at the end of charging, CX represents the charging efficiency, FX represents the discharge efficiency, FQ represents the state of charge at the start of discharging, and FJ represents the state of charge at the end of discharging.

3. The blockchain-based energy storage transaction and carbon credit management system according to claim 1 is characterized in that: In the heterogeneous cross-chain interoperability transaction module, the specific steps of the hash time lock are: A first randomly generates a dynamic value, then converts the dynamic value into a hash value through a hash function, A sends the hash value to B and sets the time lock T, B provides a dynamic value matching the hash value within the time lock T, and finally A sends the dynamic carbon credit value to B to complete the energy storage transaction record.

4. The blockchain-based energy storage transaction and carbon credit management system according to claim 3 is characterized in that: The specific steps of converting the dynamic value into a hash value through a hash function are: First, convert the dynamic value into binary according to UTF-8 encoding to obtain binary data, and expand the length of the binary data to meet the requirement of 512 modulo 448 to obtain the extended data; then use the 8 32-bit constants preset by the SHA-256 algorithm as the initial hash value; divide the extended data into 512-bit message blocks, expand each message block into 64 32-bit words, and then initialize the working variables with 8 initial hash values, and then perform 64 rounds of cyclic compression operations, update the working variables according to the logic function and constants, and finally update the hash value; finally, connect the 8 32-bit hash values ​​obtained after processing all message blocks to obtain a 256-bit binary number converted into a hexadecimal string, which is the final hash value.

5. The blockchain-based energy storage transaction and carbon credit management system according to claim 4 is characterized in that: The specific steps of using the 8 32-bit constants preset by the SHA-256 algorithm as the initial hash value are: S1: Set the golden section constant, and get the new number by calculating the square root product of the irrational numbers related to the golden section constant; S2: convert the decimal part of the new number into binary, and cut off the first 32 bits to form a 32-bit binary word; S3: Convert the 32-bit binary word into hexadecimal representation according to the rule that 4 binary bits correspond to 1 hexadecimal bit to obtain an initial hash value.

6. The blockchain-based energy storage transaction and carbon credit management system according to claim 1 is characterized in that: In the multi-objective reinforcement learning energy storage scheduling optimization module, the real-time electricity price, carbon price, marginal carbon emission factor of the power grid and the state of charge of the energy storage equipment of the energy chain are normalized to obtain the processed data. The processed data is calculated by the multi-objective deep network algorithm to obtain the specific steps of the energy storage trading strategy and carbon credit allocation ratio: S1: A multi-layer perceptron (MLP) architecture is selected. The number of input layer neurons is determined to be 4 based on the four characteristics of the input data, namely, the real-time electricity price, carbon price, marginal carbon emission factor of the power grid, and the state of charge of the energy storage equipment. The number of hidden layers and neurons is determined through experiments and cross-validation, and 2-3 hidden layers are set, with 32-128 neurons in each layer. The number of neurons in the output layer is 2, which outputs the energy storage trading strategy and the carbon credit allocation ratio respectively. The Xavier initialization method is used to determine the initial value range of the weight according to the number of input and output neurons. S2: Use weighted mean square error as the loss function, comprehensively consider the prediction errors of energy storage trading strategy and carbon credit allocation ratio, and balance the importance of the two goals by adjusting the weights; use the Adam optimization algorithm to adaptively adjust the learning rate to update the network weights, set a suitable initial learning rate, and divide the data set into training set, validation set, and test set at a ratio of 70%, 15%, and 15%; During the training process, the model converges based on the validation set loss. If the validation set loss does not decrease after several consecutive rounds, the training is stopped. After the training is completed, the test set is used to evaluate the generalization ability of the model. S3: Input the new normalized real-time electricity price, carbon price, grid marginal carbon emission factor and energy storage equipment charge state data into the trained model; the two values ​​output by the model correspond to the energy storage trading strategy and the carbon credit allocation ratio respectively; the charging and discharging operation of the energy storage equipment is judged according to the positive and negative output value of the energy storage trading strategy, and the value represents the charging and discharging power; according to the output value of the carbon credit allocation ratio, the specific carbon credit allocation plan is determined in combination with the actual situation to achieve coordinated decision-making of energy storage trading and carbon credit management.

7. The blockchain-based energy storage transaction and carbon credit management system according to claim 6 is characterized in that: The specific steps of using the Xavier initialization method to determine the initial value range of the weight according to the number of input and output neurons are as follows: Using the Xavier initialization method, for a layer with x input neurons and y output neurons, the initial weight values ​​are set to a uniform distribution. In which U represents the distribution range of the initial value of the weight, c represents a fixed constant, x represents the number of layers of input neurons, and y represents the number of layers of output neurons; a value is randomly sampled as the initial value for each connection weight of the layer to complete the Xavier initialization of the weight.

8. The blockchain-based energy storage transaction and carbon credit management system according to claim 6 is characterized in that: The specific steps of using the Adam optimization algorithm to adaptively adjust the learning rate and update the network weights are: S1: Input the real-time electricity price, carbon price, marginal carbon emission factor of the power grid and the state of charge of the energy storage device into the neural network, first perform forward propagation calculation output, then calculate the loss value based on the loss function, and then use the back propagation algorithm to calculate the gradient of the loss with respect to the network weight; S2: Update the first-order matrix estimate and the second-order matrix estimate based on the gradient of the network weights obtained in S1 and correct the first-order matrix estimate and the second-order matrix estimate; S3: Based on the modified first-order matrix estimation and second-order matrix estimation, use the formula Among them, DI t+1 Represents the network weight at time t+1, DI t Represents the network weight at the tth time, XL represents the learning rate, EG represents the second-order matrix estimate, CS represents the constant, and YG represents the first-order matrix estimate. The network weight is adaptively adjusted through this formula, and then the steps from S1 to S3 are repeated to continuously iterate and update the network weight until the training stop condition is met.

9. The blockchain-based energy storage transaction and carbon credit management system according to claim 1, characterized in that: In the enhanced carbon account module, distributed ledger technology is used in combination with traditional database storage to store data in the form of blocks in the blockchain and establish hash connections.

10. The method for energy storage trading and carbon credit management based on blockchain according to claim 1, characterized in that: Step 1: Obtain the marginal carbon emission factor of the power grid and the operation data of energy storage equipment in real time, calculate the equivalent carbon emission reduction generated by the energy storage charging and discharging behavior through the dynamic carbon factor model, and generate a dynamic carbon credit value; Step 2: Connect the energy chain and the carbon chain, and atomically bind the dynamic carbon credit value to the energy storage transaction record through the hash time lock and zero-knowledge proof verification mechanism; Step 3: Receive the real-time electricity price, carbon price and marginal carbon emission factor of the energy chain, combine the state of charge of the energy storage equipment, dynamically generate energy storage trading strategies and carbon credit allocation ratios through a multi-objective deep network algorithm, and feed back decision instructions to the energy storage equipment control end; Step 4: Store carbon credit transaction data and generate and verify carbon emission reduction certificates through zero-knowledge proof.

Images