Distributed energy storage method and system with highest local consumption rate

Abstract

The invention belongs to the technical field of distributed energy storage, and discloses a distributed energy storage method and system with the highest on-site consumption rate. The method comprises the following steps: collecting real-time operation data of the distributed energy storage system, and performing distributed energy storage analysis on the real-time operation data by using a distributed energy storage analysis model to obtain a real-time distributed energy storage analysis result; by taking maximization of the local consumption rate as an optimization target, performing distributed energy storage planning on the real-time distributed energy storage analysis result by using a distributed energy storage planning model to obtain a real-time distributed energy storage planning scheme; and performing distributed energy storage control on the real-time distributed energy storage planning scheme by using the distributed energy storage control model to obtain a real-time distributed energy storage control strategy, and sending the real-time distributed energy storage control strategy to the distributed energy storage system. According to the invention, the problems of low intelligent degree and inflexible energy storage planning in the prior art are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of distributed energy storage, and in particular relates to a distributed energy storage method and system with the highest local consumption rate. Background Art

[0002] With the rapid development of renewable energy, distributed energy storage systems play an important role in balancing energy supply and demand and improving grid stability. Distributed energy storage systems refer to energy storage units installed at various nodes of the power system (such as the grid side, user side or power supply side). Through coordinated control, they jointly provide energy storage and management services for the power system. Distributed energy storage systems play an important role in promoting the transformation of energy structure, improving the efficiency of grid operation and promoting the consumption of renewable energy. With the advancement of technology and the reduction of costs, distributed energy storage systems have developed rapidly around the world. However, how to efficiently manage distributed energy storage systems and maximize the local consumption of energy is an urgent problem to be solved.

[0003] The existing distributed energy storage technology has the following defects:

[0004] 1) Low level of intelligence: Most existing distributed energy storage technologies rely on manual analysis, planning, and control, which is costly and has limited capabilities in big data processing and analysis, making them unsuitable for distributed energy storage in large systems.

[0005] 2) Inflexible energy storage planning: Existing energy storage planning methods for distributed energy storage technologies are often based on fixed rules or simple logical procedures, and lack the ability to adapt to dynamically changing environments and load demands, resulting in the inability to adjust the configuration and operation strategies of distributed energy storage systems in real time to maximize local consumption. Summary of the invention

[0006] In order to solve the problems of low intelligence and inflexible energy storage planning in the prior art, the present invention aims to provide a distributed energy storage method and system with the highest local absorption rate.

[0007] The technical solution adopted by the present invention is:

[0008] A distributed energy storage method with the highest local consumption rate comprises the following steps:

[0009] Collect real-time operation data of the distributed energy storage system, and use the distributed energy storage analysis model to perform distributed energy storage analysis on the real-time operation data to obtain real-time distributed energy storage analysis results;

[0010] Taking maximizing the local consumption rate as the optimization goal, a distributed energy storage planning model is used to perform distributed energy storage planning on the real-time distributed energy storage analysis results to obtain a real-time distributed energy storage planning solution;

[0011] The distributed energy storage control model is used to perform distributed energy storage control on the real-time distributed energy storage planning scheme, and the real-time distributed energy storage control strategy is obtained and sent to the distributed energy storage system.

[0012] Furthermore, the real-time operation data includes the real-time distribution network operation data of the distribution network of the distributed energy storage system, the real-time renewable energy power generation actual data and real-time equipment operation data of the renewable energy power generation equipment, the real-time power consumption data and real-time load operation data of the local load, and the real-time device operation data of the distributed energy storage device.

[0013] Furthermore, the real-time distributed energy storage analysis results include real-time renewable energy generation prediction results of the distributed energy storage system, real-time local load demand prediction results, and real-time energy storage device stored energy prediction results;

[0014] The real-time distributed energy storage planning scheme includes the real-time distributed energy storage device operation location of the distributed energy storage system, the real-time distributed energy storage device capacity configuration, the real-time distributed energy storage device charging and discharging scheme, the real-time distributed energy storage system grid interaction scheme and the real-time distributed energy storage system investment and operation cost;

[0015] The real-time distributed energy storage control strategy includes the real-time distributed energy storage device charging and discharging control decision of the distributed energy storage system, the real-time distributed energy storage system power scheduling decision, the real-time distributed energy storage system energy management decision, the real-time distributed energy storage device demand response decision, the real-time distributed energy storage device fault response decision and the real-time distributed energy storage system economic optimization decision.

[0016] Furthermore, the distributed energy storage analysis model is constructed based on the RF-LSTM algorithm, and the distributed energy storage analysis model includes a key feature screening module constructed based on the RF algorithm and a distributed energy storage analysis module constructed based on the LSTM algorithm, which are connected in sequence.

[0017] Furthermore, real-time operation data of the distributed energy storage system is collected, and a distributed energy storage analysis model is used to perform distributed energy storage analysis on the real-time operation data to obtain real-time distributed energy storage analysis results, including the following steps:

[0018] Collecting real-time operation data of the distributed energy storage system, and preprocessing the real-time operation data to obtain preprocessed real-time operation data;

[0019] Use the key feature screening module of the distributed energy storage analysis model to extract several real-time key features of the pre-processed real-time operation data;

[0020] The distributed energy storage analysis module of the distributed energy storage analysis model is used to perform distributed energy storage analysis according to a number of real-time key features to obtain real-time distributed energy storage analysis results.

[0021] Furthermore, the distributed energy storage planning model is constructed based on the FSSA algorithm, and the distributed energy storage planning model includes an initialization module, an iterative optimization module and a vector decoding module which are connected in sequence.

[0022] Furthermore, with maximizing the local consumption rate as the optimization goal, a distributed energy storage planning model is used to perform distributed energy storage planning on the real-time distributed energy storage analysis results to obtain a real-time distributed energy storage planning solution, including the following steps:

[0023] Taking maximizing the local consumption rate as the optimization goal, the fitness function and algorithm parameters of the distributed energy storage planning model are set according to the real-time distributed energy storage analysis results;

[0024] Use the initialization module of the distributed energy storage planning model to initialize and obtain several initial solutions;

[0025] According to the fitness function, algorithm parameters and several initial solutions, the iterative optimization module of the distributed energy storage planning model is used to perform iterative optimization to obtain the optimal individual;

[0026] The vector decoding module of the distributed energy storage planning model is used to decode the solution vector of the optimal individual to obtain the optimal real-time distributed energy storage planning solution with the highest local consumption rate.

[0027] Furthermore, the distributed energy storage control model is constructed based on the MPO-DQN algorithm, and the distributed energy storage control model includes a meta-strategy optimization module constructed based on the MPO algorithm and a reinforcement learning module constructed based on the DQN algorithm. The reinforcement learning module is provided with an intelligent agent, a deep Q network and an experience replay pool. The intelligent agent is respectively connected to the deep Q network, the experience replay pool and the meta-strategy optimization module, and the meta-strategy optimization module is connected to the experience replay pool.

[0028] Furthermore, using the distributed energy storage control model, the real-time distributed energy storage planning scheme is subjected to distributed energy storage control, a real-time distributed energy storage control strategy is obtained, and the strategy is sent to the distributed energy storage system, including the following steps:

[0029] According to the real-time distributed energy storage analysis results, the real-time meta-strategy optimization experience with the highest similarity is extracted from several historical meta-strategy optimization experiences in the experience playback pool of the distributed energy storage control model;

[0030] Based on the experience of real-time meta-strategy optimization, the meta-strategy optimization module of the distributed energy storage control model is used to generate the initial deep Q network of the reinforcement learning module;

[0031] According to the real-time distributed energy storage analysis results, the real-time reinforcement learning experience with the highest similarity is extracted from several historical reinforcement learning experiences in the experience replay pool of the distributed energy storage control model;

[0032] According to the real-time distributed energy storage planning scheme and real-time reinforcement learning experience, based on the initial deep Q network, the distributed energy storage control is carried out using the intelligent agent to obtain the real-time distributed energy storage control strategy;

[0033] The real-time distributed energy storage control strategy is sent to the distributed energy storage system, and the real-time distributed energy storage control strategy is executed based on the distributed energy storage system.

[0034] A distributed energy storage system with the highest local consumption rate is used to implement a distributed energy storage method with the highest local consumption rate. The system includes a distributed energy storage analysis unit, a distributed energy storage planning unit and a distributed energy storage control unit connected in sequence.

[0035] The beneficial effects of the present invention are:

[0036] The present invention provides a distributed energy storage method and system with the highest local absorption rate. By adopting a distributed energy storage analysis model constructed by a deep learning algorithm, it is possible to more accurately predict the renewable energy generation and load demand, thereby providing reliable data support for the distributed energy storage planning and distributed energy storage control of the distributed energy storage system, and significantly improving the local absorption rate. The distributed energy storage planning constructed by the swarm intelligence optimization algorithm takes maximizing the local absorption rate as the optimization goal, and can adjust the configuration and operation strategy of the distributed energy storage system in real time to adapt to the dynamically changing energy environment and load demand, ensure that the energy storage system always operates in the optimal state, and ensure the highest local absorption rate. The distributed energy storage control model constructed in combination with the reinforcement learning algorithm realizes a fast-response distributed energy storage control strategy, which can respond to emergencies in a timely manner and improve the stability and reliability of the system. By integrating deep learning, swarm intelligence optimization and reinforcement learning algorithms, the integration and intelligence of the distributed energy storage system are improved, the efficient circulation of information and the optimal configuration of resources are realized, and flexible energy storage analysis, energy storage planning and energy storage control are carried out to realize the efficient, economical and safe operation of the distributed energy storage system, which has significant application value.

[0037] Other beneficial effects of the present invention will be further described in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flow chart of the distributed energy storage method with the highest local consumption rate in the present invention.

[0039] Figure 2 It is a structural block diagram of the distributed energy storage system with the highest local consumption rate in the present invention. DETAILED DESCRIPTION

[0040] The present invention will be further explained below in conjunction with the accompanying drawings and specific embodiments.

[0041] Embodiment 1:

[0042] like Figure 1 As shown, this embodiment provides a distributed energy storage method with the highest local consumption rate, including the following steps:

[0043] S1: Collecting real-time operation data of the distributed energy storage system, and using the distributed energy storage analysis model to perform distributed energy storage analysis on the real-time operation data to obtain real-time distributed energy storage analysis results, including the following steps:

[0044] S1-1: collect real-time operation data of the distributed energy storage system, and pre-process the real-time operation data to obtain pre-processed real-time operation data;

[0045] The real-time operation data includes the real-time distribution network operation data of the distribution network of the distributed energy storage system, the real-time renewable energy power generation actual data and real-time equipment operation data of the renewable energy power generation equipment, the real-time power consumption data and real-time load operation data of the local load, and the real-time device operation data of the distributed energy storage device;

[0046] The distributed energy storage analysis model is constructed based on the Random Forest (RF)-Long Short-Term Memory (LSTM) algorithm, and the distributed energy storage analysis model includes a key feature screening module constructed based on the RF algorithm and a distributed energy storage analysis module constructed based on the LSTM algorithm, which are connected in sequence;

[0047] The key feature screening module screens the feature components of the input operation data through the internal Classification And Regression Tree (CART), which can process a large number of feature components, generate the key feature importance score of each feature component, and select the most stable and discriminative key feature components according to the key feature importance score. The trained key feature screening module can directly screen the newly input real-time operation data according to the selected key features to obtain the corresponding key features; the distributed energy storage analysis module learns the deep association of several key features through the gate mechanism to achieve time series prediction;

[0048] The method for constructing a distributed energy storage analysis model includes the following steps:

[0049] A-1: Collect historical operation data of a number of distributed energy storage systems, and pre-process the historical operation data to obtain a number of pre-processed historical operation data;

[0050] Preprocessing includes data format conversion, data cleaning, Gaussian denoising, and magnitude normalization in order to improve data quality and provide support for subsequent model training;

[0051] A-2: Use the RF-LSTM algorithm to build an initial distributed energy storage analysis model;

[0052] A-3: Use some pre-processed historical operation data to optimize and train the initial distributed energy storage analysis model, obtain the final distributed energy storage analysis model, and generate some historical distributed energy storage analysis results;

[0053] S1-2: Use the key feature screening module of the distributed energy storage analysis model to extract several real-time key features of the pre-processed real-time operation data;

[0054] S1-3: using the distributed energy storage analysis module of the distributed energy storage analysis model, performing distributed energy storage analysis according to a number of real-time key features, and obtaining real-time distributed energy storage analysis results;

[0055] The real-time distributed energy storage analysis results include the real-time renewable energy generation forecast results of the distributed energy storage system, the real-time local load demand forecast results, and the real-time energy storage device storage energy forecast results;

[0056] S2: Taking maximizing the local consumption rate as the optimization goal, using the distributed energy storage planning model, the distributed energy storage planning is performed on the real-time distributed energy storage analysis results to obtain a real-time distributed energy storage planning solution, including the following steps:

[0057] S2-1: Taking maximizing the local consumption rate as the optimization goal, the fitness function and algorithm parameters of the distributed energy storage planning model are set according to the real-time distributed energy storage analysis results;

[0058] The distributed energy storage planning model is constructed based on the Fast Sparrow Search Algorithm (FSSA) algorithm, and the distributed energy storage planning model includes an initialization module, an iterative optimization module and a vector decoding module which are connected in sequence;

[0059] The initialization module is used to introduce the chaotic mapping sequence to initialize the population and obtain several initial solutions. The iterative optimization module is used to perform iterative optimization according to the fitness function, algorithm parameters and several initial solutions, using the iterative optimization module of the distributed energy storage planning model to obtain the optimal individual. The vector decoding module is used to decode the solution vector of the optimal individual to obtain the optimal real-time distributed energy storage planning solution with the highest local consumption rate.

[0060] The method for constructing a distributed energy storage planning model includes the following steps:

[0061] B-1: Use the FSSA algorithm to build an initial distributed energy storage planning model;

[0062] B-2: Input historical distributed energy storage analysis results, optimize the initial distributed energy storage planning model, and adjust the model hyperparameters of the distributed energy storage planning model during the optimization training, including learning rate, initial step size, etc.;

[0063] B-3: Traverse all historical distributed energy storage analysis results, retain the optimal model hyperparameters, obtain the final distributed energy storage planning model, and generate several historical distributed energy storage planning schemes;

[0064] The formula of the fitness function is:

[0065]

[0066] In the formula, f(X c ) is the FSSA individual X c The fitness value of L(X c ) is the FSSA individual X c The load balancing index of the distribution network; A(Xc) is the FSSA individual X c The local consumption rate; T(Xc) is the FSSA individual X c The total cost of All are weight coefficients; X c is the FSSA individual variable; c is the FSSA individual indicator; H is the optimization objective function;

[0067]

[0068] Where P gen,t' ,P load,t' The real-time renewable energy generation forecast result and the real-time local load demand forecast result of the distributed energy storage system at time t'; E stored,t' is the energy prediction result stored in the real-time energy storage device of the distributed energy storage system at time t'; t' is the time indicator; t' max is the total time;

[0069] Determine the parameters of the FSSA algorithm, including FSSA population parameters, maximum number of iterations, and search space limits;

[0070] In this embodiment, the FSSA population parameters include a search space of N×D dimensions, and a search space food of F=[F 1 ,F 2 ,...,F D ] T The sparrow position is X = [X h1 ,X h2 ,...,X hD ] T ; Where F is the search space food matrix, F 1 ,F 2 ,...,F D are all search space food matrix elements, X is the sparrow position matrix, X h1 ,X h2 ,...,X hD are all sparrow position matrix elements, N is the number of FSSA individuals, D is the dimension of the distributed energy storage planning problem; h is the FSSA individual indicator; the search space limit includes the search space upper limit ub and the search space lower limit lb;

[0071] S2-2: Use the initialization module of the distributed energy storage planning model to initialize and obtain several initial solutions. The specific method is as follows:

[0072] According to the FSSA population parameters, the Circle chaotic mapping sequence is used for initialization to obtain the initial FSSA population, which includes several initial solutions;

[0073] The formula is:

[0074]

[0075] Where X' c is the initial FSSA individual of Circle chaotic mapping; X c * is the randomly generated initial FSSA individual; c is the FSSA individual indicator;

[0076] S2-3: According to the fitness function, algorithm parameters and several initial solutions, the iterative optimization module of the distributed energy storage planning model is used to perform iterative optimization to obtain the optimal individual, including the following steps:

[0077] S2-3-1: Use the fitness function to obtain the fitness value of each initial FSSA individual (initial solution) in the initial FSSA population;

[0078] S2-3-2: Sort the initial FSSA individuals according to their fitness values ​​to obtain the initial discoverers, initial joiners and initial predators;

[0079] S2-3-3: updating the initial FSSA population to obtain an updated FSSA population; the updated FSSA population includes updated discoverers, updated joiners and updated predators;

[0080] The update formula of the discoverer is:

[0081]

[0082] In the formula, are the cth discoverer FSSA individuals in the t+1th and tth iterations respectively; t max is the maximum number of iterations; ξ is a random number between 0 and 1; Q is a normally distributed random number; L is a 1×D matrix whose elements are all 1; R 2 is the warning value; ST is the safety threshold;

[0083] The update formula for the joiner is:

[0084]

[0085] In the formula, are the cth joiner FSSA individuals in the t+1th and tth iterations respectively; The best position for the exposed person to occupy; is the current worst position; iter max is the maximum iteration threshold; ξ is a random number between 0 and 1; L is a 1×D matrix, whose elements are all 1 or -1; c is the sparrow indicator;

[0086] The update formula of the predator is:

[0087]

[0088] In the formula, are the cth predator FSSA individuals of the t+1th and tth iterations respectively; δ is the step size control parameter, and δ=a"·γ", a" is the convergence factor, γ" is a non-zero positive real number for step size control; is the current best position; f c 、f g 、f w are the current, best and worst fitness of FSSA individuals respectively; γ is the minimum constant to prevent the denominator from being 0;

[0089]

[0090] In the formula, a" is the convergence factor; tanh(.) is the hyperbolic tangent function; a max 、a min are the maximum and minimum values ​​of the convergence factor, respectively; λ is the decreasing rate parameter, k" is the decreasing period parameter, λ = -2π, k" = π;

[0091] S2-3-4: Use the dynamic reverse learning algorithm to perform dynamic reverse learning on the updated FSSA population to generate a dynamic reverse FSSA population;

[0092] The formula is:

[0093]

[0094] In the formula, is the dynamically reversed FSSA individual; γ is the decreasing inertia coefficient; ub is the upper limit of the search space; lb is the lower limit of the search space; For updated FSSA individuals;

[0095] S2-3-5: According to the fitness function, calculate the fitness values ​​of all FSSA individuals in the updated FSSA population and the dynamically reversed FSSA population, and take the FSSA individual with the minimum fitness value as the optimal individual;

[0096] S2-3-6: If the number of iterations of the algorithm reaches the maximum number of iterations or the fitness value of the optimal individual meets the requirements, the optimal individual is output;

[0097] S2-4: Use the vector decoding module of the distributed energy storage planning model to decode the solution vector of the optimal individual to obtain the optimal real-time distributed energy storage planning solution with the highest local consumption rate;

[0098] The real-time distributed energy storage planning scheme includes the real-time distributed energy storage device operation location of the distributed energy storage system, the real-time distributed energy storage device capacity configuration, the real-time distributed energy storage device charging and discharging scheme, the real-time distributed energy storage system grid interaction scheme and the real-time distributed energy storage system investment and operation cost;

[0099] S3: Using the distributed energy storage control model, the real-time distributed energy storage planning scheme is subjected to distributed energy storage control, and the real-time distributed energy storage control strategy is obtained and sent to the distributed energy storage system, including the following steps:

[0100] S3-1: Based on the real-time distributed energy storage analysis results, extract the real-time meta-strategy optimization experience with the highest similarity from several historical meta-strategy optimization experiences in the experience playback pool of the distributed energy storage control model;

[0101] The distributed energy storage control model is constructed based on the Meta-Policy Optimization (MPO)-Deep Q Network (DQN) algorithm, and the distributed energy storage control model includes a meta-policy optimization module constructed based on the MPO algorithm and a reinforcement learning module constructed based on the DQN algorithm. The reinforcement learning module is provided with an intelligent agent, a deep Q network and an experience replay pool. The intelligent agent is respectively connected to the deep Q network, the experience replay pool and the meta-policy optimization module, and the meta-policy optimization module is connected to the experience replay pool.

[0102] The meta-strategy optimization module is used to generate the deep Q network parameters in the reinforcement learning module so that these parameters can quickly adapt to new distributed energy storage analysis results or new distributed energy storage planning schemes, improve the generalization ability of the model, and generate deep Q networks based on previous learning experience even under unseen data; through meta-strategy optimization, the DQN algorithm can converge faster, reduce the amount of data and time required for training, and significantly improve learning efficiency; the reinforcement learning module uses previous learning experience to output and select appropriate distributed energy storage control strategies based on the distributed energy storage planning scheme;

[0103] The method for constructing a distributed energy storage control model includes the following steps:

[0104] C-1: Use the historical distributed energy storage analysis results as optimization scenarios for meta-strategy optimization, obtain several optimization scenarios, and use the deep Q network generation problem of the DQN algorithm as the simulation environment for meta-strategy optimization;

[0105] C-2: Based on the simulation environment of meta-strategy optimization, use the MPO algorithm to build the initial meta-strategy optimization module, and train the initial meta-strategy optimization module under several optimization scenarios to obtain the final meta-strategy optimization module, and generate several historical meta-strategy optimization experiences;

[0106] C-3: Use the distributed energy storage control strategy generation problem as the simulation environment of the DQN algorithm, define the state space of the DQN algorithm based on several historical distributed energy storage planning schemes, and define the action space of the DQN algorithm based on several preset distributed energy storage control actions;

[0107] C-4: Based on the impact of the preset distributed energy storage control action on the historical distributed energy storage control state, define the reward function of the DQN algorithm, use the final meta-strategy optimization module, generate the deep Q network of the reinforcement learning module, and obtain the initial reinforcement learning module;

[0108] C-5: Based on the state space, action space and reward function, use several historical distributed energy storage planning schemes to train the initial reinforcement learning module, obtain the final reinforcement learning module, and generate several historical reinforcement learning experiences;

[0109] C-6: Store several historical meta-strategy optimization experiences and several historical reinforcement learning experiences in the experience replay pool of the final reinforcement learning module, and integrate the final meta-strategy optimization module and the final reinforcement learning module to obtain a distributed energy storage control model;

[0110] S3-2: Based on the real-time meta-strategy optimization experience, the meta-strategy optimization module of the distributed energy storage control model is used to generate the initial deep Q network of the reinforcement learning module;

[0111] S3-3: Based on the real-time distributed energy storage analysis results, extract the real-time reinforcement learning experience with the highest similarity from several historical reinforcement learning experiences in the experience replay pool of the distributed energy storage control model;

[0112] S3-4: According to the real-time distributed energy storage planning scheme and real-time reinforcement learning experience, based on the initial deep Q network, use the intelligent agent to perform distributed energy storage control and obtain the real-time distributed energy storage control strategy, including the following steps:

[0113] S3-4-1: Analyze the real-time distributed energy storage planning scheme to obtain several real-time distributed energy storage states, and update the state space of the reinforcement learning module according to the several real-time distributed energy storage states to obtain the updated state space S'=[s' 1 ,...,s' i" ,...,s' I' ], where s' i" is the updated i'th state value, i' is the state indicator, and I' is the total number of state space dimensions;

[0114] S3-4-2: Extract several preset distributed energy storage control actions from the real-time reinforcement learning experience, update the action space of the reinforcement learning module, and obtain the updated action space A'=[a' 1 ,...,a' j" ,...,a' I ], where a' j" is the updated j-th "action value", j" is the action indicator, and I is the total number of action space dimensions;

[0115] S3-4-3: Using the intelligent agent of the updated reinforcement learning module to control the updated deep Q network, generating the Q value of each preset distributed energy storage control action in the updated action space for each real-time distributed energy storage state in the updated state space;

[0116] S3-4-4: According to the reward function, obtain the reward value of each preset distributed energy storage control action for each real-time distributed energy storage state, and update the corresponding Q value according to the reward value to obtain an updated Q value;

[0117] The formula is:

[0118] Q(s' p' ,a' p' )=(1-α")·Q(s p' ,a p' )+α"·(R(s p' ,a p' ,s' p' )+γ*·Q max (s p' ,a p' ))

[0119] In the formula, Q(s' p' ,a' p' ) is the updated state value s' p' and the updated action value a' p' The corresponding updated Q value; Q(s p' ,a p' ) is the state value s p' and action value a p' The corresponding Q value; α" is the learning rate; Q max (s p' ,a p' ) is the state value s p' and action value a p' The highest Q value corresponding to the value; p' is the comprehensive indicator; γ* is the update parameter; R(s p' ,a' p' ,s' p' ) is the updated action value a' p' Set the state value s p' Transform to the updated state value s' p' The reward value of

[0120] S3-4-5: Repeat the above Q value update steps until the number of iterations reaches the iteration number threshold, and obtain the final Q value of each preset distributed energy storage control action for each real-time distributed energy storage state;

[0121] S5-3-6: using a greedy strategy, selecting a preset distributed energy storage control action corresponding to the highest final Q value as the execution distributed energy storage control action corresponding to the real-time distributed energy storage state;

[0122] S5-3-7: Integrate all distributed energy storage control actions to obtain a real-time distributed energy storage control strategy;

[0123] Real-time distributed energy storage control strategies include real-time distributed energy storage device charging and discharging control decisions of distributed energy storage systems, real-time distributed energy storage system power scheduling decisions, real-time distributed energy storage system energy management decisions, real-time distributed energy storage device demand response decisions, real-time distributed energy storage device fault response decisions, and real-time distributed energy storage system economic optimization decisions;

[0124] S3-5: Send the real-time distributed energy storage control strategy to the distributed energy storage system, and execute the real-time distributed energy storage control strategy based on the distributed energy storage system.

[0125] Embodiment 2:

[0126] like Figure 2 As shown, this embodiment provides a distributed energy storage system with the highest local consumption rate, which is used to implement a distributed energy storage method with the highest local consumption rate. The system includes a distributed energy storage analysis unit, a distributed energy storage planning unit, and a distributed energy storage control unit connected in sequence;

[0127] A distributed energy storage analysis unit is used to collect real-time operation data of the distributed energy storage system, and use a distributed energy storage analysis model to perform distributed energy storage analysis on the real-time operation data to obtain real-time distributed energy storage analysis results;

[0128] A distributed energy storage planning unit is used to optimize the local consumption rate by maximizing it, and to use a distributed energy storage planning model to perform distributed energy storage planning on the real-time distributed energy storage analysis results to obtain a real-time distributed energy storage planning solution;

[0129] The distributed energy storage control unit is used to use the distributed energy storage control model to perform distributed energy storage control on the real-time distributed energy storage planning scheme, obtain the real-time distributed energy storage control strategy, and send it to the distributed energy storage system.

[0130] The present invention provides a distributed energy storage method and system with the highest local absorption rate. By adopting a distributed energy storage analysis model constructed by a deep learning algorithm, it is possible to more accurately predict the renewable energy generation and load demand, thereby providing reliable data support for the distributed energy storage planning and distributed energy storage control of the distributed energy storage system, and significantly improving the local absorption rate. The distributed energy storage planning constructed by the swarm intelligence optimization algorithm takes maximizing the local absorption rate as the optimization goal, and can adjust the configuration and operation strategy of the distributed energy storage system in real time to adapt to the dynamically changing energy environment and load demand, ensure that the energy storage system always operates in the optimal state, and ensure the highest local absorption rate. The distributed energy storage control model constructed in combination with the reinforcement learning algorithm realizes a fast-response distributed energy storage control strategy, which can respond to emergencies in a timely manner and improve the stability and reliability of the system. By integrating deep learning, swarm intelligence optimization and reinforcement learning algorithms, the integration and intelligence of the distributed energy storage system are improved, the efficient circulation of information and the optimal configuration of resources are realized, and flexible energy storage analysis, energy storage planning and energy storage control are carried out to realize the efficient, economical and safe operation of the distributed energy storage system, which has significant application value.

[0131] The present invention is not limited to the above optional implementations, and anyone can derive other various forms of products under the enlightenment of the present invention. The above specific implementations should not be understood as limiting the scope of protection of the present invention. The scope of protection of the present invention should be based on the definition in the claims, and the description can be used to interpret the claims.

Claims

1. A distributed energy storage method with the highest local consumption rate, characterized by: The steps include: Collect real-time operation data of the distributed energy storage system, and use the distributed energy storage analysis model to perform distributed energy storage analysis on the real-time operation data to obtain real-time distributed energy storage analysis results; Taking maximizing the local consumption rate as the optimization goal, the distributed energy storage planning model is used to perform distributed energy storage planning on the real-time distributed energy storage analysis results to obtain a real-time distributed energy storage planning solution; The distributed energy storage control model is used to perform distributed energy storage control on the real-time distributed energy storage planning scheme, and the real-time distributed energy storage control strategy is obtained and sent to the distributed energy storage system.

2. A distributed energy storage method with the highest local consumption rate according to claim 1, characterized in that: The real-time operation data includes the real-time distribution network operation data of the distributed energy storage system's distribution network, the real-time renewable energy power generation actual data and real-time equipment operation data of the renewable energy power generation equipment, the real-time power consumption data and real-time load operation data of the local load, and the real-time device operation data of the distributed energy storage device.

3. A distributed energy storage method with the highest local consumption rate according to claim 2, characterized in that: The real-time distributed energy storage analysis results include the real-time renewable energy power generation prediction results of the distributed energy storage system, the real-time local load demand prediction results, and the real-time energy storage device storage energy prediction results; The real-time distributed energy storage planning scheme includes the real-time distributed energy storage device operation location of the distributed energy storage system, the real-time distributed energy storage device capacity configuration, the real-time distributed energy storage device charging and discharging scheme, the real-time distributed energy storage system grid interaction scheme and the real-time distributed energy storage system investment and operation cost; The real-time distributed energy storage control strategy includes real-time distributed energy storage device charging and discharging control decisions of the distributed energy storage system, real-time distributed energy storage system power scheduling decisions, real-time distributed energy storage system energy management decisions, real-time distributed energy storage device demand response decisions, real-time distributed energy storage device fault response decisions and real-time distributed energy storage system economic optimization decisions.

4. A distributed energy storage method with the highest local consumption rate according to claim 1, characterized in that: The distributed energy storage analysis model is constructed based on the RF-LSTM algorithm, and the distributed energy storage analysis model includes a key feature screening module constructed based on the RF algorithm and a distributed energy storage analysis module constructed based on the LSTM algorithm, which are connected in sequence.

5. A distributed energy storage method with the highest local consumption rate according to claim 4, characterized in that: Collecting real-time operation data of the distributed energy storage system and using the distributed energy storage analysis model to perform distributed energy storage analysis on the real-time operation data to obtain real-time distributed energy storage analysis results includes the following steps: Collecting real-time operation data of the distributed energy storage system, and preprocessing the real-time operation data to obtain preprocessed real-time operation data; Use the key feature screening module of the distributed energy storage analysis model to extract several real-time key features of the pre-processed real-time operation data; The distributed energy storage analysis module of the distributed energy storage analysis model is used to perform distributed energy storage analysis according to a number of real-time key features to obtain real-time distributed energy storage analysis results.

6. A distributed energy storage method with the highest local consumption rate according to claim 1, characterized in that: The distributed energy storage planning model is constructed based on the FSSA algorithm, and the distributed energy storage planning model includes an initialization module, an iterative optimization module and a vector decoding module which are connected in sequence.

7. A distributed energy storage method with the highest local consumption rate according to claim 6, characterized in that: Taking maximizing the local consumption rate as the optimization goal, the distributed energy storage planning model is used to perform distributed energy storage planning on the real-time distributed energy storage analysis results to obtain a real-time distributed energy storage planning solution, which includes the following steps: Taking maximizing the local consumption rate as the optimization goal, the fitness function and algorithm parameters of the distributed energy storage planning model are set according to the real-time distributed energy storage analysis results; Use the initialization module of the distributed energy storage planning model to initialize and obtain several initial solutions; According to the fitness function, algorithm parameters and several initial solutions, the iterative optimization module of the distributed energy storage planning model is used to perform iterative optimization to obtain the optimal individual; The vector decoding module of the distributed energy storage planning model is used to decode the solution vector of the optimal individual to obtain the optimal real-time distributed energy storage planning solution with the highest local consumption rate.

8. The distributed energy storage method with the highest local consumption rate according to claim 1, characterized in that: The distributed energy storage control model is constructed based on the MPO-DQN algorithm, and the distributed energy storage control model includes a meta-strategy optimization module constructed based on the MPO algorithm and a reinforcement learning module constructed based on the DQN algorithm. The reinforcement learning module is provided with an intelligent agent, a deep Q network and an experience replay pool. The intelligent agent is respectively connected to the deep Q network, the experience replay pool and the meta-strategy optimization module, and the meta-strategy optimization module is connected to the experience replay pool.

9. A distributed energy storage method with the highest local consumption rate according to claim 8, characterized in that: Using the distributed energy storage control model, the real-time distributed energy storage planning scheme is subjected to distributed energy storage control, and the real-time distributed energy storage control strategy is obtained and sent to the distributed energy storage system, including the following steps: According to the real-time distributed energy storage analysis results, the real-time meta-strategy optimization experience with the highest similarity is extracted from several historical meta-strategy optimization experiences in the experience playback pool of the distributed energy storage control model; Based on the experience of real-time meta-strategy optimization, the meta-strategy optimization module of the distributed energy storage control model is used to generate the initial deep Q network of the reinforcement learning module; According to the real-time distributed energy storage analysis results, the real-time reinforcement learning experience with the highest similarity is extracted from several historical reinforcement learning experiences in the experience replay pool of the distributed energy storage control model; According to the real-time distributed energy storage planning scheme and real-time reinforcement learning experience, based on the initial deep Q network, the distributed energy storage control is carried out using the intelligent agent to obtain the real-time distributed energy storage control strategy; The real-time distributed energy storage control strategy is sent to the distributed energy storage system, and the real-time distributed energy storage control strategy is executed based on the distributed energy storage system.

10. A distributed energy storage system with the highest local consumption rate, used to implement the distributed energy storage method with the highest local consumption rate as claimed in any one of claims 1 to 9, characterized in that: The system comprises a distributed energy storage analysis unit, a distributed energy storage planning unit and a distributed energy storage control unit which are connected in sequence.

Images