Mobile energy storage online decision-making method and device based on porous electrode aging model
By combining multilayer perceptron and deep reinforcement learning models, the problem that existing electrochemical energy storage aging assessment models are not applicable to mobile energy storage is solved, thereby improving the accuracy of mobile energy storage scheduling decisions and arbitrage profits.
Patent Information
- Application Number
- CN202410089917.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-22
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-01-22
AI Technical Summary
Existing electrochemical energy storage aging assessment models are not applicable to mobile energy storage, leading to biased estimation of battery aging conditions, reducing the decision-making accuracy and arbitrage profits of mobile energy storage, and existing models are either too simple or too complex to be applied, lacking physical interpretation.
A combined model based on multilayer perceptron and deep reinforcement learning is adopted. By acquiring the initial health state of the battery, the initial and operational parameters of the mobile energy storage are determined using a pre-trained model. The battery aging is updated in combination with different working states, so as to achieve an accurate estimate of the battery aging status.
It improves the accuracy of mobile energy storage scheduling decisions and arbitrage profits. By combining multilayer perceptrons and deep reinforcement learning models, it accurately determines battery aging conditions, thereby improving decision-making accuracy and economic efficiency.
Smart Images

Figure CN118131045B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to a mobile energy storage online decision-making method and device based on a porous electrode aging model. Background Art
[0002] Mobile energy storage is a new energy storage technology with more flexible deployment methods. It can perform energy arbitrage and alleviate grid congestion by charging or discharging between grid nodes. However, the battery aging of mobile energy storage is affected by factors such as charging and discharging rate and temperature. Battery aging can cause functional degradation of mobile energy storage, such as energy capacity and power capacity attenuation, which in turn affects the operational decision of mobile energy storage.
[0003] Current energy storage technology is mainly aimed at fixed energy storage. As time goes by, the capacity of energy storage batteries may gradually decrease. By comparing the capacity of new and old batteries, the battery capacity decay rate can be calculated to evaluate the degree of battery aging.
[0004] However, the existing aging assessment models for electrochemical energy storage are not suitable for mobile energy storage. First, the existing aging assessment models for electrochemical energy storage are mainly for fixed energy storage. The environment (such as temperature, etc.) and working conditions (such as charging and discharging power, etc.) faced by mobile energy storage are more complex, and the aging assessment of batteries is also more complex. Directly applying the existing fixed energy storage aging assessment model in the mobile energy storage scheduling decision-making process will cause estimation bias in the battery aging situation, thereby reducing the decision-making accuracy and arbitrage benefits of mobile energy storage. Second, most of the existing battery aging assessment models are single linear aging models, mechanism models and data-driven models. Among them, the linear aging model is too simple and cannot accurately describe the nonlinear factors affecting energy storage aging. The mechanism model has high assessment accuracy and good interpretability for energy storage aging, but its modeling is complex and online calculation is difficult. The data-driven model does not require complex modeling, but lacks physical interpretability and has high data requirements. Third, battery aging will cause functional degradation of mobile energy storage (such as the attenuation of energy capacity, power capacity and charging and discharging efficiency), which will affect the decision-making accuracy of mobile energy storage. However, existing studies rarely integrate battery aging assessment into mobile energy storage decision optimization. Battery aging assessment or mobile energy storage optimization scheduling needs to be integrated into a unified decision-making framework to improve the accuracy of battery aging assessment and mobile energy storage decision-making. Summary of the invention
[0005] In view of this, the present invention provides a mobile energy storage online decision-making method and device based on a porous electrode aging model to solve the problem that directly applying the existing fixed energy storage aging assessment model in the mobile energy storage scheduling decision-making process will cause estimation deviation of the battery aging condition, thereby reducing the decision-making accuracy and arbitrage benefits of mobile energy storage.
[0006] In a first aspect, the present invention provides an online decision-making method for mobile energy storage based on a porous electrode aging model, the method comprising: obtaining an initial health state of a battery; based on the initial health state of the battery, using a pre-trained online decision-making model for mobile energy storage based on deep reinforcement learning to determine the initial state parameters of the mobile energy storage at the first moment; based on the initial state parameters, using a pre-trained online decision-making model for mobile energy storage based on deep reinforcement learning to determine the action parameters of the mobile energy storage at the first moment; wherein the action parameters include: power, working state and next site location corresponding to the first moment; if the working state is a charging state, based on the power corresponding to the first moment, the initial state parameters and a pre-trained mobile energy storage aging assessment model of a multi-layer perceptron, determine the battery aging amount of the mobile energy storage at the next site location at the second moment, and update the initial state parameters based on the pre-trained online decision-making model for mobile energy storage based on deep reinforcement learning to obtain a first target state parameter; if the working state is a discharge state or a holding state, update the initial state parameters using a pre-trained online decision-making model for mobile energy storage based on deep reinforcement learning to obtain a second target state parameter.
[0007] In an optional embodiment, the initial state parameters include: the remaining energy of the mobile energy storage at the first moment, the inventory energy cost of the mobile energy storage at the first moment, the marginal electricity price of the site at the first moment, the temperature of the site at the first moment, and the health state of the mobile energy storage at the first moment; wherein, based on the power corresponding to the first moment and a pre-trained mobile energy storage aging assessment model of a multi-layer perceptron, the battery aging amount of the mobile energy storage at the next site location at the second moment is determined, including: determining the termination remaining power based on the power corresponding to the first moment and the current remaining power; based on the power corresponding to the first moment, the current remaining power, the termination remaining power, the temperature of the site at the first moment, and the health state of the mobile energy storage at the first moment, the battery aging amount of the mobile energy storage at the next site location at the second moment is determined using a pre-trained mobile energy storage aging assessment model of a multi-layer perceptron.
[0008] In an optional embodiment, the above method also includes: detecting whether the second moment satisfies the termination moment parameters; if the second moment does not satisfy the termination moment parameters, repeatedly executing the steps of determining the initial state parameters of the mobile energy storage at the first moment based on the initial health state of the battery, using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning, and if the working state is a discharge state or a holding state, using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning to update the initial state parameters to obtain the second target state parameters, until the second moment satisfies the termination moment parameters.
[0009] In an optional embodiment, the above method also includes: based on the target state parameters, using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning to determine the target reward function.
[0010] In an optional implementation, the initial state parameter is updated based on a pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain a first target state parameter, including:
[0011] The first target state parameter is determined by the following formula:
[0012] Among them, E h is the remaining energy of h mobile energy storage, E h' h' is the remaining energy of the mobile energy storage, n is the current site location, n' is the next site location, h is the first moment, h' is the second moment, COST E,h is the storage energy cost of mobile energy storage at the first moment, COST E,h' h' is the storage energy cost of mobile energy storage, λ n,h h is the marginal electricity price of the current site n, λ n',h' is the site marginal electricity price of h'next site n', TEMP n,h is the temperature of the current site n, TEMP n',h' h' is the temperature of the next station n', SOH h The health status and SOH of mobile energy storage h' The health status of h'mobile energy storage, h' is the charging power of the mobile energy storage at the next station, is the charging power of the mobile energy storage at the next site at the second moment, LMP(·) is the site marginal electricity price update function, Temperature(·) is the ambient temperature update function, Soh(·) is the health status update function of the mobile energy storage mobile battery, duration n,n' is the time distance between sites n and n'.
[0013] In an optional implementation, the initial state parameter is updated based on a pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain a second target state parameter, including: if the working state is a discharge state, the second target state parameter is determined by the following formula:
[0014] in, is the discharge power of the mobile energy storage at the next station at the second moment, E h is the remaining energy of h mobile energy storage, E h'h' is the remaining energy of the mobile energy storage, n is the current site location, n' is the next site location, h is the first moment, h' is the second moment, COST E,h is the storage energy cost of mobile energy storage at the first moment, COST E,h' h' is the storage energy cost of mobile energy storage, λ n,h h is the marginal electricity price of the current site n, λ n',h' is the site marginal electricity price of h'next site n', TEMP n,h is the temperature of the current site n, TEMP n',h' h' is the temperature of the next station n', SOH h The health status and SOH of mobile energy storage h' The health status of h'mobile energy storage, h' is the charging power of the mobile energy storage at the next site, LMP(·) is the site marginal price update function, Temperature(·) is the ambient temperature update function, Soh(·) is the health status update function of the mobile energy storage mobile battery, duration n,n' is the time distance between sites n and n';
[0015] If the working state is the hold state, the second target state parameter is determined by the following formula:
[0016] Among them, Action(n) is the action selected by the second site.
[0017] In an optional implementation, based on the target state parameters, a pre-trained mobile energy storage online decision model based on deep reinforcement learning is used to determine the target reward function, including:
[0018] If the working state is charging state, the target reward function is determined by the following formula:
[0019] r h =-C TRA -C DEG Among them, C TRA is the transportation cost of mobile energy storage, C DEG The aging cost of mobile energy storage;
[0020] If the working state is the discharge state, the target reward function is determined using the following formula:
[0021] Among them, λ n',h' is the node marginal electricity price of node n' at time h', is the discharge power of the mobile energy storage at node n' at time h', Δh is the time interval, COST E,h'is the inventory energy cost of mobile energy storage at time h' and E h is the remaining energy of the mobile energy storage at time h.
[0022] If the working state is the hold state, the target reward function is determined using the following formula:
[0023] r h =-C TRA -C DEG .
[0024] In a second aspect, the present invention provides a mobile energy storage online decision-making device based on a porous electrode aging model, the device comprising an acquisition module for acquiring an initial health state of a battery; a first determination module for determining the initial state parameters of the mobile energy storage at a first moment based on the initial health state of the battery using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning; a second determination module for determining the action parameters of the mobile energy storage at the first moment based on the initial state parameters using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning; wherein the action parameters include: power, working state and next site location corresponding to the first moment; a third determination module for determining the battery aging amount of the mobile energy storage at the next site location at the second moment based on the power corresponding to the first moment and a pre-trained mobile energy storage aging assessment model of a multi-layer perceptron if the working state is a charging state, and updating the initial state parameters based on the pre-trained mobile energy storage online decision-making model based on deep reinforcement learning to obtain a first target state parameter; a fourth determination module for updating the initial state parameters using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning to obtain a second target state parameter if the working state is a discharge state or a holding state.
[0025] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the online decision-making method for mobile energy storage based on the porous electrode aging model of the above-mentioned first aspect or any corresponding embodiment thereof by executing the computer instructions.
[0026] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the online decision-making method for mobile energy storage based on a porous electrode aging model of the above-mentioned first aspect or any corresponding embodiment thereof.
[0027] The present invention provides a mobile energy storage online decision-making method and device based on a porous electrode aging model, the method comprising: obtaining an initial health state of a battery; based on the initial health state of the battery, using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning to determine the initial state parameters of the mobile energy storage at the first moment; based on the initial state parameters, using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning to determine the action parameters of the mobile energy storage at the first moment; wherein the action parameters include: power, working state and next site location corresponding to the first moment; if the working state is a charging state, based on the power corresponding to the first moment and a pre-trained mobile energy storage aging assessment model of a multi-layer perceptron, the battery aging amount of the mobile energy storage at the next site location at the second moment is determined, and based on the pre-trained mobile energy storage online decision-making model based on deep reinforcement learning, the initial state parameters are updated to obtain a first target state parameter; if the working state is a discharge state or a holding state, the initial state parameters are updated to obtain a second target state parameter using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning.
[0028] The present invention can determine the initial state parameters of the mobile energy storage at the first moment and the action parameters of the mobile energy storage at the first moment through the initial health state of the battery and the pre-trained mobile energy storage online decision model based on deep reinforcement learning, and then determine the corresponding processing method through the different working states of the mobile energy storage, that is, if the working state is the charging state, first determine the battery aging amount of the mobile energy storage through the pre-trained multi-layer perceptron mobile energy storage aging assessment model, and then update the initial state parameters through the pre-trained mobile energy storage online decision model based on deep reinforcement learning. If the working state is the discharge state or the holding state, the pre-trained mobile energy storage online decision model based on deep reinforcement learning can be directly used to update the initial state parameters to obtain the second target state parameters. Therefore, the present invention combines the mobile energy storage aging assessment model based on the multi-layer perceptron with the mobile energy storage online decision model based on deep reinforcement learning, and can accurately determine the battery aging in the mobile energy storage scheduling decision process, thereby improving the decision accuracy and arbitrage benefits of mobile energy storage. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0030] Figure 1It is a flow chart of a mobile energy storage online decision-making method based on a porous electrode aging model according to an embodiment of the present invention;
[0031] Figure 2 is a schematic diagram of an online decision-making method for mobile energy storage based on a porous electrode aging model according to an embodiment of the present invention;
[0032] Figure 3 is a flow chart of another online decision-making method for mobile energy storage based on a porous electrode aging model according to an embodiment of the present invention;
[0033] Figure 4 is a structural block diagram of a mobile energy storage online decision-making device based on a porous electrode aging model according to an embodiment of the present invention;
[0034] Figure 5 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0035] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0036] Based on relevant technologies, mobile energy storage, as a new type of energy storage technology with more flexible deployment methods, can conduct energy arbitrage and alleviate grid congestion by charging or discharging between grid nodes. However, the battery aging of mobile energy storage is affected by factors such as charging and discharging rate and temperature. Battery aging can cause functional degradation of mobile energy storage, such as energy capacity and power capacity attenuation, which in turn affects the operational decision of mobile energy storage.
[0037] Current energy storage technology is mainly aimed at fixed energy storage. As time goes by, the capacity of energy storage batteries may gradually decrease. By comparing the capacity of new and old batteries, the battery capacity decay rate can be calculated to evaluate the degree of battery aging.
[0038] However, the existing aging assessment models for electrochemical energy storage are not suitable for mobile energy storage. First, the existing aging assessment models for electrochemical energy storage are mainly for fixed energy storage. The environment (such as temperature, etc.) and working conditions (such as charging and discharging power, etc.) faced by mobile energy storage are more complex, and the aging assessment of batteries is also more complex. Directly applying the existing fixed energy storage aging assessment model in the mobile energy storage scheduling decision-making process will cause estimation bias in the battery aging situation, thereby reducing the decision-making accuracy and arbitrage benefits of mobile energy storage. Second, most of the existing battery aging assessment models are single linear aging models, mechanism models and data-driven models. Among them, the linear aging model is too simple and cannot accurately describe the nonlinear factors affecting energy storage aging. The mechanism model has high assessment accuracy and good interpretability for energy storage aging, but its modeling is complex and online calculation is difficult. The data-driven model does not require complex modeling, but lacks physical interpretability and has high data requirements. Third, battery aging will cause functional degradation of mobile energy storage (such as the attenuation of energy capacity, power capacity and charging and discharging efficiency), which will affect the decision-making accuracy of mobile energy storage. However, existing studies rarely integrate battery aging assessment into mobile energy storage decision optimization. Battery aging assessment or mobile energy storage optimization scheduling needs to be integrated into a unified decision-making framework to improve the accuracy of battery aging assessment and mobile energy storage decision-making.
[0039] Based on this, the present invention can determine the initial state parameters of the mobile energy storage at the first moment and the action parameters of the mobile energy storage at the first moment through the initial health state of the battery and the pre-trained mobile energy storage online decision model based on deep reinforcement learning, and then determine the corresponding processing method through the different working states of the mobile energy storage, that is, if the working state is the charging state, first determine the battery aging amount of the mobile energy storage through the pre-trained multi-layer perceptron mobile energy storage aging assessment model, and then update the initial state parameters through the pre-trained mobile energy storage online decision model based on deep reinforcement learning. If the working state is the discharge state or the holding state, the pre-trained mobile energy storage online decision model based on deep reinforcement learning can be directly used to update the initial state parameters to obtain the second target state parameters. Therefore, the present invention combines the mobile energy storage aging assessment model based on the multi-layer perceptron with the mobile energy storage online decision model based on deep reinforcement learning, and can accurately determine the estimated deviation of the battery aging situation in the mobile energy storage scheduling decision process, thereby improving the decision accuracy and arbitrage benefits of mobile energy storage.
[0040] According to an embodiment of the present invention, an embodiment of an online decision-making method for mobile energy storage based on a porous electrode aging model is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.
[0041] In this embodiment, a mobile energy storage online decision-making method based on a porous electrode aging model is provided, which can be used for computer equipment, such as computers, servers, etc. Figure 1 is a flow chart of a mobile energy storage online decision-making method based on a porous electrode aging model according to an embodiment of the present invention, such as Figure 1 As shown, the process includes the following steps:
[0042] Step S101, obtaining the initial health status of the battery.
[0043] The initial health state of a battery refers to the performance and capacity state of a battery after it is manufactured, usually expressed as a percentage, reflecting the ratio of the current capacity of the battery to its original design capacity. The initial health state of a battery depends on the manufacturing process and quality control of the battery, as well as the material and design of the battery. Among them, the higher the initial health state of the battery, the better the performance and capacity of the battery, which can better meet the use requirements. Among them, the initial health state of the battery is not static, and the performance and capacity of the battery will gradually decrease over time. Therefore, during long-term use, it is necessary to regularly check and evaluate the health state of the battery to understand whether the performance and capacity of the battery meet the use requirements, and take timely measures to maintain the battery.
[0044] Mobile energy storage is a portable and movable energy storage device that can store electrical energy in mobile devices by using technologies such as batteries and fuel cells, thereby meeting the needs of emergency power supply and mobile device charging. The advantages of mobile energy storage devices lie in their flexibility, portability and mobility. Compared with traditional fixed energy storage systems, mobile energy storage devices can provide electricity anytime and anywhere, without being restricted by geographical location, and have a wider range of application scenarios.
[0045] Step S102, based on the initial health state of the battery, using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning, to determine the initial state parameters of the mobile energy storage at the first moment.
[0046] The pre-trained mobile energy storage online decision model based on deep reinforcement learning is a model that combines deep learning and reinforcement learning techniques and is used to make real-time decisions in mobile energy storage systems. This model can learn the optimal decision strategy through reinforcement learning algorithms based on the real-time status and historical data of the system (i.e., the initial health status of the battery) to achieve optimal management and control of the energy storage system. Among them, the initial health status of the battery can be used as input, and the initial state parameters at the first moment can be used as output.
[0047] Preferably, the initial state parameters can be in, is the initial state parameter at time h, E h is the remaining energy of mobile energy storage at time h, COST E,h is the inventory energy cost of mobile energy storage at time h, λ n,h is the marginal electricity price of the current site location n at time h, TEMP n,h is the temperature of the current station location n at time h, SOH h is the health status of mobile energy storage at time h, where h is the first moment. E,h , n,h It can be set to the value at the current station position n; E h and TEMP n,h It can be obtained by directly measuring with a measuring instrument or by calculation, and is not specifically limited here.
[0048] Step S103, based on the initial state parameters, using a pre-trained mobile energy storage online decision model based on deep reinforcement learning, to determine the action parameters of the mobile energy storage at the first moment; wherein the action parameters include: the power, working state and next site location corresponding to the first moment.
[0049] After determining the above initial state parameters, the initial state parameters can be used as input, and the action parameters of the mobile energy storage at the first moment can be used as output. The action parameters include: the power corresponding to the first moment, the working state and the next station location.
[0050] Preferably, the action parameter can be a h =(n',cdh,pwr); where n' is the destination node of the mobile energy storage, cdh is the charging and discharging selection of the mobile energy storage, namely, the charging state, discharging state and holding state, and pwr is the charging and discharging power of the mobile energy storage; where the charging and discharging power can be calculated by the method of "power = energy / time"; the next site location can be used to characterize the next site of the current site location, that is, the next destination of the mobile energy storage.
[0051] Step S104, if the working state is the charging state, based on the power corresponding to the first moment, the initial state parameters and the pre-trained mobile energy storage aging assessment model of the multi-layer perceptron, the battery aging amount of the mobile energy storage at the next site location at the second moment is determined, and the initial state parameters are updated based on the pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the first target state parameters.
[0052] Combination Figure 2As shown, the pre-trained multi-layer perceptron mobile energy storage aging assessment model is a supervised learning model based on a feedforward neural network. The multi-layer perceptron includes an input layer, a hidden layer, and an output layer. The input layer is responsible for receiving data features, and the hidden layer is composed of multiple neurons, which are responsible for extracting features from the input layer and learning data representation. This patent adopts a regression model based on a multi-layer perceptron, with power charge and discharge rate, initial SOC, termination SOC, current battery health state (State of Health, SOH) and temperature as input features, and the data generated by PETLION as training data, and finally fits and outputs the attenuation of the battery health state ΔSOH.
[0053] Specifically, the final remaining capacity (State of Charge, SOC) of the mobile energy storage can be determined by the power corresponding to the first moment, and then the power corresponding to the first moment, the final remaining capacity, and the initial state parameters are used as input to obtain the attenuation ΔSOH of the battery health state.
[0054] After determining the attenuation of the battery health state, the initial state parameters can be updated through the state transfer equation in the pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the first target state parameters. How to update the initial state parameters through the state transfer equation is described in detail below.
[0055] Step S105, if the working state is the discharge state or the holding state, the initial state parameters are updated using the pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the second target state parameters.
[0056] If the working state is a discharge state or a holding state (i.e., neither discharge nor charge), the initial state parameters can be directly updated through the state transfer equation of the mobile energy storage online decision model based on deep reinforcement learning that has been pre-trained to obtain the second target state parameters. How to update the initial state parameters through the state transfer equation is described in detail below.
[0057] The present invention can determine the initial state parameters of the mobile energy storage at the first moment and the action parameters of the mobile energy storage at the first moment through the initial health state of the battery and the pre-trained mobile energy storage online decision model based on deep reinforcement learning, and then determine the corresponding processing method through the different working states of the mobile energy storage, that is, if the working state is the charging state, first determine the battery aging amount of the mobile energy storage through the pre-trained multi-layer perceptron mobile energy storage aging assessment model, and then update the initial state parameters through the pre-trained mobile energy storage online decision model based on deep reinforcement learning. If the working state is the discharge state or the holding state, the pre-trained mobile energy storage online decision model based on deep reinforcement learning can be directly used to update the initial state parameters to obtain the second target state parameters. Therefore, the present invention combines the mobile energy storage aging assessment model based on the multi-layer perceptron with the mobile energy storage online decision model based on deep reinforcement learning, and can accurately determine the battery aging in the mobile energy storage scheduling decision process, thereby improving the decision accuracy and arbitrage benefits of mobile energy storage.
[0058] In this embodiment, a mobile energy storage online decision-making method based on a porous electrode aging model is provided, which can be used for the above-mentioned computer equipment, such as computers, servers, etc. Figure 3 is a flow chart of a mobile energy storage online decision-making method based on a porous electrode aging model according to an embodiment of the present invention, such as Figure 3 As shown, the process includes the following steps:
[0059] Step S201, obtain the initial health status of the battery. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.
[0060] Step S202, based on the initial health state of the battery, using the pre-trained mobile energy storage online decision model based on deep reinforcement learning, determine the initial state parameters of the mobile energy storage at the first moment. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.
[0061] Step S203, based on the initial state parameters, using the pre-trained mobile energy storage online decision model based on deep reinforcement learning, determine the action parameters of the mobile energy storage at the first moment; wherein the action parameters include: the power, working state and next station location corresponding to the first moment. For details, please refer to Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.
[0062] Step S204, if the working state is the charging state, based on the power corresponding to the first moment, the initial state parameters and the pre-trained mobile energy storage aging assessment model of the multi-layer perceptron, the battery aging amount of the mobile energy storage at the next site location at the second moment is determined, and the initial state parameters are updated based on the pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the first target state parameters.
[0063] Specifically, the initial state parameters include: the remaining energy of the mobile energy storage at the first moment, the inventory energy cost of the mobile energy storage at the first moment, the site marginal electricity price of the site at the first moment, the temperature of the site at the first moment, and the health state of the mobile energy storage at the first moment; wherein the above step S204 includes:
[0064] Step S2041, determining the termination remaining power based on the power corresponding to the first moment and the current remaining power.
[0065] The termination remaining power can be used to characterize that when the remaining power of the battery increases to the threshold, it can be considered that the battery has been fully charged. Specifically, the power corresponding to the first moment is obtained, and the power represents the charging power of the battery at the first moment. The current remaining power is obtained, that is, the remaining power of the battery. According to the power at the first moment and the current remaining power, the charging current of the battery at the first moment is calculated. The calculation formula of the charging current is: charging current = power / voltage. According to the charging current and the current remaining power, the power change trend of the battery in the future period is predicted. The battery charging curve or other related models can be used for prediction. According to the predicted power change trend, the termination remaining power is determined.
[0066] Step S2042, based on the power corresponding to the first moment, the current remaining power, the ending remaining power, the temperature of the site at the first moment, and the health status of the mobile energy storage at the first moment, the mobile energy storage aging assessment model of the pre-trained multi-layer perceptron is used to determine the battery aging amount of the mobile energy storage at the next site location at the second moment.
[0067] After determining the termination remaining power, the initial state parameters and the termination remaining power can be used as inputs to determine the battery aging of the mobile energy storage at the next site at the second moment. h+1 =SOH h +ΔSOH; where SOH h+1 The battery aging amount.
[0068] Step S2043, based on the pre-trained mobile energy storage online decision-making model based on deep reinforcement learning, the initial state parameters are updated to obtain the first target state parameters.
[0069] Step S205: If the working state is the discharge state or the holding state, the initial state parameters are updated using the pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the second target state parameters. Figure 1 Step S105 of the illustrated embodiment will not be described in detail here.
[0070] In an optional embodiment, in the above step S204, the initial state parameters are updated based on the pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the first target state parameters, including:
[0071] The first target state parameter is determined by the following formula:
[0072] Among them, E h is the remaining energy of h mobile energy storage, E h' h' is the remaining energy of the mobile energy storage, n is the current site location, n' is the next site location, h is the first moment, h' is the second moment, COST E,h is the storage energy cost of mobile energy storage at the first moment, COST E,h' h' is the storage energy cost of mobile energy storage, λ n,h h is the marginal electricity price of the current site n, λ n',h' is the site marginal electricity price of h'next site n', TEMP n,h is the temperature of the current site n, TEMP n',h' h' is the temperature of the next station n', SOH h The health status and SOH of mobile energy storage h' The health status of h'mobile energy storage, h' is the charging power of the mobile energy storage at the next station, is the charging power of the mobile energy storage at the next site at the second moment, LMP(·) is the site marginal electricity price update function, Temperature(·) is the ambient temperature update function, Soh(·) is the health status update function of the mobile energy storage mobile battery, duration n,n' is the time distance between sites n and n'.
[0073] In an optional implementation, in the above step S205, the initial state parameters are updated based on the pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the second target state parameters, including:
[0074] If the working state is the discharge state, the second target state parameter is determined by the following formula:
[0075] in, is the discharge power of the mobile energy storage at the next station at the second moment, E h is the remaining energy of h mobile energy storage, E h' h' is the remaining energy of the mobile energy storage, n is the current site location, n' is the next site location, h is the first moment, h' is the second moment, COST E,h is the storage energy cost of mobile energy storage at the first moment, COST E,h' h' is the storage energy cost of mobile energy storage, λ n,h h is the marginal electricity price of the current site n, λ n',h' is the site marginal electricity price of h'next site n', TEMP n,h is the temperature of the current site n, TEMP n',h' h' is the temperature of the next station n', SOH h The health status and SOH of mobile energy storage h' The health status of h'mobile energy storage, h' is the charging power of the mobile energy storage at the next site, LMP(·) is the site marginal price update function, Temperature(·) is the ambient temperature update function, Soh(·) is the health status update function of the mobile energy storage mobile battery, duration n,n' is the time distance between sites n and n';
[0076] If the working state is the hold state, the second target state parameter is determined by the following formula:
[0077] Among them, Action(n) is the action selected by the second site.
[0078] In an optional embodiment, the above method further includes:
[0079] Step a1, detecting whether the second moment satisfies the termination moment parameter.
[0080] The termination time parameter can be used to characterize the threshold of the termination cycle. Among them, the termination state parameter can be 24 moments, 25 moments, etc., which are not specifically limited here. Detecting whether the second moment meets the termination time parameter can include two situations, situation 1: the second moment meets the termination time parameter; situation 2: the second moment does not meet the termination time parameter.
[0081] Step a2: If the termination time parameters are not met at the second moment, the steps of determining the initial state parameters of the mobile energy storage at the first moment based on the initial health state of the battery and using the pre-trained online decision-making model for mobile energy storage based on deep reinforcement learning to update the initial state parameters and obtain the second target state parameters if the working state is a discharge state or a holding state are repeatedly executed, until the termination time parameters are met at the second moment.
[0082] For the above situation 1: when the second moment does not meet the termination time parameter, the above steps S202 to S205 can be repeatedly performed until the second moment meets the termination time parameter. Please refer to the above steps S202 to S205 for the specific processing method, which will not be described in detail here.
[0083] The mobile energy storage online decision-making method based on the porous electrode aging model provided in this embodiment can determine the results corresponding to the locations of each site, that is, whether the mobile energy storage can go to the next site, by determining the first target state parameter or the second target state parameter corresponding to each moment.
[0084] In an optional embodiment, the above method also includes: based on the target state parameters, using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning to determine the target reward function.
[0085] The reward function is a core concept in reinforcement learning. It determines the feedback from the environment after the agent takes an action. This feedback can be positive (positive reward), negative (negative reward), or both (positive and negative rewards). The design of the reward function is crucial to the learning efficiency and effect of the agent. By using the target state parameter as input, the target reward function can be determined, thereby determining whether the mobile energy storage can go to the next station.
[0086] Specifically, based on the target state parameters, the target reward function is determined by using a pre-trained mobile energy storage online decision model based on deep reinforcement learning, including:
[0087] If the working state is charging state, the target reward function is determined by the following formula:
[0088] r h =-C TRA -C DEG Among them, C TRA is the transportation cost of mobile energy storage, C DEG The aging cost of mobile energy storage;
[0089] If the working state is the discharge state, the target reward function is determined using the following formula:
[0090] Among them, λ n',h' is the node marginal electricity price of node n' at time h', is the discharge power of the mobile energy storage at node n' at time h', Δh is the time interval, COST E,h' is the inventory energy cost of mobile energy storage at time h' and E h is the remaining energy of the mobile energy storage at time h.
[0091] If the working state is the hold state, the target reward function is determined using the following formula:
[0092] r h =-C TRA -C DEG .
[0093] The mobile energy storage online decision-making method based on the porous electrode aging model provided in this embodiment integrates a battery aging assessment method into the inventory path decision of the mobile energy storage, monitors the battery health of the mobile energy storage system in real time according to the decision information, and makes real-time decisions based on the battery aging conditions, so that the mobile energy storage can be more adaptable to complex operating environments.
[0094] In an optional embodiment, the porous electrode theory is a model that describes the physical and electrochemical properties of battery electrodes. It assumes that the electrode is a porous medium containing electrolytes and active substances (such as lithium salts and electrode materials in lithium-ion batteries), and is suitable for describing different types of batteries, such as lithium-ion batteries, fuel cells, and supercapacitors. The model is based on the principles of conservation of mass, conservation of energy, and conservation of charge, and uses differential equations at the electrode scale to describe the physical and chemical processes within the electrode. The model mainly considers the growth and lithium precipitation mechanism of the battery solid electrolyte interface. In the battery aging model based on porous electrode theory, the transport process of lithium ions in the electrolyte at the electrode scale is calculated by the charge conservation equation and the ion conservation equation. The charge conservation equations in solid electrodes and electrolytes are as follows:
[0095]
[0096]
[0097] in, is the equivalent electronic conductivity at the electrode scale, φ s Solid phase potential, j tot is the total volume current density, is the equivalent conductivity of lithium ions in the electrolyte, is the equivalent diffusion rate of lithium ions in the electrolyte, φ e is the electrolyte potential, c e is the lithium ion concentration in the electrolyte.
[0098] The ion conservation equations in the electrolyte and active material particles are as follows
[0099]
[0100]
[0101] Where ε is the electrode porosity, t is the time, is the equivalent diffusion coefficient of lithium ions in the electrolyte, t + is the migration number of lithium ions in the electrolyte, F is the Faraday constant, c s is the solid phase lithium ion concentration, r is the particle radius, D s is the ion diffusion coefficient inside the particle.
[0102] Solid electrolyte interface growth and lithium precipitation are two important phenomena in the charging and discharging process of lithium-ion batteries. They have a significant impact on the performance and life of the battery. The reaction mainly occurs at the anode. Therefore, a total of three chemical reactions can occur at the anode, namely, the total volume current density j tot for
[0103] j tot =j int +j SEI +j lpl ;
[0104] Among them, j int is the transfer current density of lithium insertion reaction, j SEI is the current density of the solid electrolyte interface growth process, j lpl is the current density of the lithium deposition process.
[0105] Transfer current density j of lithium insertion reaction int It is calculated by the Butler-Volmer equation, which is an empirical equation that relates the reaction rate at an electrode to the overpotential at the electrode (the difference between the actual potential of the electrode and the equilibrium potential):
[0106]
[0107] Where a is the specific surface area, i 0,int is the exchange current density, a a,int 、a c,int are the transfer coefficients of the anode reaction and the cathode reaction, R is the ideal gas constant, T is the temperature, η int For overvoltage.
[0108] Current density j during the solid electrolyte interface growth process SEICalculated by the cathode Tafel expression, it is derived from the Butler-Volmer equation in the limit case and is applicable to the area where the current density is far away from the exchange current density. It is a simplified equation that describes the relationship between current density and overpotential during the electrochemical reaction:
[0109]
[0110] Among them, k 0,SEI is the kinetic rate constant, is the concentration of electrolyte on the graphite surface, a c,SEI is the transfer coefficient of the cathode reaction, φ e is the potential in the electrolyte, R film is the radius of the electrolyte particle, U SEI is the equilibrium potential of the SEI growth reaction.
[0111] The current density of lithium deposition process lpl The cathode Tafel expression is calculated as follows:
[0112]
[0113] Among them, i 0,lpl is the exchange current density of the lithium deposition process, a c,lpl is the transfer coefficient of the cathode reaction.
[0114] The material balance between SEI and metallic lithium can be expressed as
[0115]
[0116]
[0117] Among them, c SEI 、c Li are the molar concentrations of the solid electrolyte interface and lithium per unit volume of electrode, respectively, and β is the proportion of the precipitated lithium forming the solid electrolyte interface.
[0118] The battery aging model based on the above porous electrode theory can simulate the charge and discharge cycle process of the battery under different operating conditions such as temperature, SOC and power. By tracking key variables in the model, such as lithium ion concentration, electrolyte consumption and solid electrolyte interface thickness, the degree of battery aging can be evaluated. Specifically, the degree of battery aging and real-time battery SOH can be evaluated by analyzing indicators such as battery capacity loss and internal resistance increase.
[0119] In this embodiment, a mobile energy storage online decision-making device based on a porous electrode aging model is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and will not be repeated hereafter. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0120] This embodiment provides a mobile energy storage online decision-making device based on a porous electrode aging model. Figure 4 As shown, it includes: an acquisition module 401, which is used to acquire the initial health state of the battery; a first determination module 402, which is used to determine the initial state parameters of the mobile energy storage at the first moment based on the initial health state of the battery using a pre-trained mobile energy storage online decision model based on deep reinforcement learning; a second determination module 403, which is used to determine the action parameters of the mobile energy storage at the first moment based on the initial state parameters using a pre-trained mobile energy storage online decision model based on deep reinforcement learning; wherein the action parameters include: the power, working state and next site location corresponding to the first moment; a third determination module 404, which is used to determine the battery aging amount of the mobile energy storage at the next site location at the second moment based on the power corresponding to the first moment and a pre-trained mobile energy storage aging assessment model of a multi-layer perceptron if the working state is a charging state, and update the initial state parameters based on the pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the first target state parameters; a fourth determination module 405, which is used to update the initial state parameters using a pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the second target state parameters if the working state is a discharging state or a holding state.
[0121] In an optional embodiment, the initial state parameters include: the remaining energy of the mobile energy storage at the first moment, the inventory energy cost of the mobile energy storage at the first moment, the marginal electricity price of the site at the first moment, the temperature of the site at the first moment, and the health status of the mobile energy storage at the first moment; wherein the third determination module 404 includes: a first determination unit, used to determine the termination remaining power based on the power corresponding to the first moment and the current remaining power; a second determination unit, used to determine the battery aging amount of the mobile energy storage at the next site location at the second moment based on the power corresponding to the first moment, the current remaining power, the termination remaining power, the temperature of the site at the first moment, and the health status of the mobile energy storage at the first moment, using a pre-trained mobile energy storage aging assessment model of a multi-layer perceptron.
[0122] In an optional embodiment, the above-mentioned device also includes: a detection module, which is used to detect whether the second moment meets the termination moment parameters; a repeated execution module, which is used to repeatedly execute the steps of determining the initial state parameters of the mobile energy storage at the first moment based on the initial health state of the battery, using a pre-trained mobile energy storage online decision model based on deep reinforcement learning, and if the working state is a discharge state or a holding state, using a pre-trained mobile energy storage online decision model based on deep reinforcement learning to update the initial state parameters to obtain the second target state parameters, if the second moment does not meet the termination moment parameters, until the second moment meets the termination moment parameters.
[0123] In an optional embodiment, the above-mentioned device also includes: a fifth determination module, which is used to determine the target reward function based on the target state parameters and using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning.
[0124] In an optional implementation, the third determination module 404 includes: determining the first target state parameter by the following formula:
[0125] Among them, E h is the remaining energy of h mobile energy storage, E h' h' is the remaining energy of the mobile energy storage, n is the current site location, n' is the next site location, h is the first moment, h' is the second moment, COST E,h is the storage energy cost of mobile energy storage at the first moment, COST E,h' h' is the storage energy cost of mobile energy storage, λ n,h h is the marginal electricity price of the current site n, λ n',h' is the site marginal electricity price of h'next site n', TEMP n,h is the temperature of the current site n, TEMP n',h' h' is the temperature of the next station n', SOH h The health status and SOH of mobile energy storage h' The health status of h'mobile energy storage, h' is the charging power of the mobile energy storage at the next station, is the charging power of the mobile energy storage at the next site at the second moment, LMP(·) is the site marginal electricity price update function, Temperature(·) is the ambient temperature update function, Soh(·) is the health status update function of the mobile energy storage mobile battery, duration n,n' is the time distance between n and n'.
[0126] In an optional implementation, the fourth determination module 405 includes: if the working state is the discharge state, determining the second target state parameter by the following formula:
[0127] in, is the discharge power of the mobile energy storage at the next station at the second moment, E h is the remaining energy of h mobile energy storage, E h' h' is the remaining energy of the mobile energy storage, n is the current site location, n' is the next site location, h is the first moment, h' is the second moment, COST E,h is the storage energy cost of mobile energy storage at the first moment, COST E,h' h' is the storage energy cost of mobile energy storage, λ n,h h is the marginal electricity price of the current site n, λ n',h' is the site marginal electricity price of h'next site n', TEMP n,h is the temperature of the current site n, TEMP n',h' h' is the temperature of the next station n', SOH h The health status and SOH of mobile energy storage h' The health status of h'mobile energy storage, h' is the charging power of the mobile energy storage at the next site, LMP(·) is the site marginal price update function, Temperature(·) is the ambient temperature update function, Soh(·) is the health status update function of the mobile energy storage mobile battery, duration n,n' is the time distance between sites n and n';
[0128] If the working state is the hold state, the second target state parameter is determined by the following formula:
[0129] Among them, Action(n) is the action selected by the second site.
[0130] In an optional embodiment, the fifth determination module includes: if the working state is the charging state, determining the target reward function using the following formula:
[0131] r h =-C TRA -C DEG Among them, C TRA is the transportation cost of mobile energy storage, C DEG The aging cost of mobile energy storage;
[0132] If the working state is the discharge state, the target reward function is determined using the following formula:
[0133] Among them, λ n',h' is the node marginal electricity price of node n' at time h', is the discharge power of the mobile energy storage at node n' at time h', Δh is the time interval, COST E,h' is the inventory energy cost of mobile energy storage at time h' and E h is the remaining energy of the mobile energy storage at time h.
[0134] If the working state is the hold state, the target reward function is determined using the following formula:
[0135] r h =-C TRA -C DEG .
[0136] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.
[0137] The mobile energy storage online decision-making device based on the porous electrode aging model provided in this embodiment can determine the initial state parameters of the mobile energy storage at the first moment and the action parameters of the mobile energy storage at the first moment through the initial health state of the battery and the pre-trained mobile energy storage online decision-making model based on deep reinforcement learning, and then determine the corresponding processing method through the different working states of the mobile energy storage, that is, if the working state is the charging state, first determine the battery aging amount of the mobile energy storage through the pre-trained multi-layer perceptron mobile energy storage aging assessment model, and then update the initial state parameters through the pre-trained mobile energy storage online decision-making model based on deep reinforcement learning. If the working state is the discharge state or the holding state, the pre-trained mobile energy storage online decision-making model based on deep reinforcement learning can be directly used to update the initial state parameters to obtain the second target state parameters. Therefore, the present invention combines the mobile energy storage aging assessment model based on the multi-layer perceptron with the mobile energy storage online decision-making model based on deep reinforcement learning, and can accurately determine the battery aging in the mobile energy storage scheduling decision process, thereby improving the decision accuracy and arbitrage benefits of mobile energy storage.
[0138] The mobile energy storage online decision-making device based on the porous electrode aging model in this embodiment is presented in the form of a functional unit, where the functional unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.
[0139] The embodiment of the present invention also provides a computer device having the above Figure 4 The mobile energy storage online decision-making device based on the porous electrode aging model is shown.
[0140] See also Figure 5 , Figure 5is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 5 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 5 A processor 10 is taken as an example.
[0141] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.
[0142] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.
[0143] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0144] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.
[0145] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.
[0146] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.
[0147] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.
Claims
1. A mobile energy storage online decision-making method based on a porous electrode aging model, characterized in that: include: Get the initial health status of the battery; Based on the initial health state of the battery, using a pre-trained mobile energy storage online decision model based on deep reinforcement learning, determine the initial state parameters of the mobile energy storage at the first moment; Based on the initial state parameters, the action parameters of the mobile energy storage at the first moment are determined using a pre-trained online decision model of the mobile energy storage based on deep reinforcement learning; wherein the action parameters include: the power, working state and next site location corresponding to the first moment; If the working state is a charging state, based on the power corresponding to the first moment, the initial state parameters and the pre-trained mobile energy storage aging assessment model of the multi-layer perceptron, the battery aging amount of the mobile energy storage at the next site location at the second moment is determined, and based on the pre-trained mobile energy storage online decision model based on deep reinforcement learning, the initial state parameters are updated to obtain the first target state parameters; If the working state is a discharge state or a holding state, the initial state parameter is updated using a pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain a second target state parameter; The initial state parameters include: the remaining energy of the mobile energy storage at the first moment, the inventory energy cost of the mobile energy storage at the first moment, the marginal electricity price of the site at the first moment, the temperature of the site at the first moment, and the health state of the mobile energy storage at the first moment; wherein, based on the power corresponding to the first moment, the initial state parameters and the pre-trained mobile energy storage aging assessment model of the multi-layer perceptron, the battery aging amount of the mobile energy storage at the next site location at the second moment is determined, including: Determine the termination remaining power based on the power corresponding to the first moment and the current remaining power; Based on the power corresponding to the first moment, the current remaining power, the final remaining power, the temperature of the site at the first moment, and the health status of the mobile energy storage at the first moment, the battery aging amount of the mobile energy storage at the next site location at the second moment is determined using a pre-trained multi-layer perceptron mobile energy storage aging assessment model.
2. The mobile energy storage online decision-making method based on the porous electrode aging model according to claim 1 is characterized in that: Also includes: Detecting whether the second moment satisfies a termination moment parameter; If the second moment does not meet the termination moment parameters, the steps of determining the initial state parameters of the mobile energy storage at the first moment based on the initial health state of the battery and using a pre-trained online decision-making model for mobile energy storage based on deep reinforcement learning to update the initial state parameters and obtain the second target state parameters if the working state is a discharge state or a holding state are repeatedly performed until the second moment meets the termination moment parameters.
3. The mobile energy storage online decision-making method based on the porous electrode aging model according to claim 1 is characterized in that: Also includes: Based on the target state parameters, a target reward function is determined using a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning.
4. The mobile energy storage online decision-making method based on the porous electrode aging model according to claim 1 is characterized in that: The initial state parameter is updated based on a pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain a first target state parameter, including: The first target state parameter is determined by the following formula: Among them, E h is the remaining energy of h mobile energy storage, E h' h' is the remaining energy of the mobile energy storage, n is the current site location, n' is the next site location, h is the first moment, h' is the second moment, COST E,h is the storage energy cost of mobile energy storage at the first moment, COST E,h' h' is the storage energy cost of mobile energy storage, λ n,h h is the marginal electricity price of the current site n, λ n',h' is the site marginal electricity price of h'next site n', TEMP n,h is the temperature of the current site n, TEMP n',h' h' is the temperature of the next station n', SOH h The health status and SOH of mobile energy storage h' The health status of h'mobile energy storage, is the charging power of the mobile energy storage at the next site at the second moment, LMP(·) is the site marginal electricity price update function, Temperature(·) is the ambient temperature update function, Soh(·) is the health status update function of the mobile energy storage mobile battery, Δh is the time interval, Action(n) is the action selection of the second site, duration n,n' is the time distance between n and n'.
5. The mobile energy storage online decision-making method based on the porous electrode aging model according to claim 1 is characterized in that: The initial state parameters are updated based on a pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain a second target state parameter, including: If the working state is a discharge state, the second target state parameter is determined by the following formula: in, is the discharge power of the mobile energy storage at the next station at the second moment, E h is the remaining energy of h mobile energy storage, E h' h' is the remaining energy of the mobile energy storage, n is the current site location, n' is the next site location, h is the first moment, h' is the second moment, COST E,h is the storage energy cost of mobile energy storage at the first moment, COST E,h' h' is the storage energy cost of mobile energy storage, λ n,h h is the marginal electricity price of the current site n, λ n',h' is the site marginal electricity price of h'next site n', TEMP n,h is the temperature of the current site n, TEMP n',h' h' is the temperature of the next station n', SOH h The health status and SOH of mobile energy storage h' h' is the health status of the mobile energy storage, LMP(·) is the site marginal price update function, Temperature(·) is the ambient temperature update function, Soh(·) is the health status update function of the mobile energy storage mobile battery, duration n,n' is the time distance between sites n and n', Δh is the time interval; If the working state is the holding state, the second target state parameter is determined by the following formula: Among them, Action(n) is the action selected by the second site.
6. The mobile energy storage online decision-making method based on the porous electrode aging model according to claim 3 is characterized in that: Based on the target state parameters, a pre-trained mobile energy storage online decision-making model based on deep reinforcement learning is used to determine the target reward function, including: If the working state is the charging state, the target reward function is determined using the following formula: r h =-C TRA -C DEG Among them, C TRA is the transportation cost of mobile energy storage, C DEG The aging cost of mobile energy storage; If the working state is the discharging state, the target reward function is determined using the following formula: Among them, λ n',h' is the node marginal electricity price of node n' at time h', is the discharge power of the mobile energy storage at node n' at time h', Δh is the time interval, COST E,h' is the inventory energy cost of mobile energy storage at time h' and E h is the remaining energy of mobile energy storage at time h; If the working state is the holding state, the target reward function is determined using the following formula: r h =-C TRA -C DEG 。 7. A mobile energy storage online decision-making device based on a porous electrode aging model, characterized in that: The device comprises: The acquisition module is used to obtain the initial health status of the battery; A first determination module is used to determine the initial state parameters of the mobile energy storage at the first moment based on the initial health state of the battery and using a pre-trained mobile energy storage online decision model based on deep reinforcement learning; A second determination module is used to determine the action parameters of the mobile energy storage at the first moment based on the initial state parameters and using a pre-trained mobile energy storage online decision model based on deep reinforcement learning; wherein the action parameters include: power, working state and next site location corresponding to the first moment; A third determination module is used to determine the battery aging amount of the mobile energy storage at the next site location at the second moment if the working state is the charging state, based on the power corresponding to the first moment, the initial state parameters and the pre-trained mobile energy storage aging assessment model of the multi-layer perceptron, and update the initial state parameters based on the pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain the first target state parameters; A fourth determination module is used to update the initial state parameters using a pre-trained mobile energy storage online decision model based on deep reinforcement learning to obtain a second target state parameter if the working state is a discharge state or a holding state; The initial state parameters include: the remaining energy of the mobile energy storage at the first moment, the inventory energy cost of the mobile energy storage at the first moment, the marginal electricity price of the site at the first moment, the temperature of the site at the first moment, and the health status of the mobile energy storage at the first moment; wherein the third determination module includes: a first determination unit, used to determine the termination remaining power based on the power corresponding to the first moment and the current remaining power; a second determination unit, used to determine the battery aging amount of the mobile energy storage at the next site location at the second moment based on the power corresponding to the first moment, the current remaining power, the termination remaining power, the temperature of the site at the first moment, and the health status of the mobile energy storage at the first moment, using a pre-trained mobile energy storage aging assessment model of a multi-layer perceptron.
8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the mobile energy storage online decision-making method based on the porous electrode aging model according to any one of claims 1 to 6 by executing the computer instructions.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the mobile energy storage online decision-making method based on the porous electrode aging model according to any one of claims 1 to 6.
Citation Information
Patent Citations
Distributed mobile digital energy storage system for rapid charging of electric vehicle
CN114274818A
Operation decision-making method and system based on hybrid energy storage system
CN114725971A
Storage battery health state prediction method based on feedforward neural network
CN115372852A
Mobile energy storage charging and discharging space-time planning method based on deep reinforcement learning
CN115392143A
Multi-resource coordination optimization method and device and storage medium
CN116436019A
