Virtual power plant energy storage system collaborative scheduling and control method and system

Abstract

The invention provides a virtual power plant energy storage system co-scheduling and control method and system, and relates to the technical field of power management, and the method comprises the steps: obtaining power grid scheduling data, calculating the charge and discharge income, collecting the real-time parameters of a battery, analyzing the performance attenuation law through deep reinforcement learning, determining the initial working parameters, predicting the system state based on a rolling time domain, and obtaining the real-time parameters of the battery. An optimal scheduling scheme is calculated through mixed integer linear programming and model prediction control in combination with a power distribution strategy and is corrected in real time, operation is executed according to a time-phased scheduling instruction and is monitored in real time, and an emergency strategy is started when necessary, so that the economic benefit and the safety of an energy storage system are improved, and the service life of a battery is prolonged.

Description

Technical Field

[0001] The present invention relates to the field of power management technology, and in particular to a method and system for collaborative scheduling and control of a virtual power plant energy storage system. Background Art

[0002] With the transformation of energy structures and the large-scale integration of renewable energy, power systems face multiple challenges, including supply and demand balance, grid stability, and energy efficiency. Virtual power plants, as a new energy management model, integrate and dispatch decentralized energy resources through information and communication technologies, effectively improving system flexibility and reliability. Energy storage systems, as a core component of virtual power plants, play an important role in smoothing fluctuations, shaving peaks and filling valleys, and frequency regulation. Traditional energy storage system scheduling is mainly based on simplified models for optimization decisions. However, it still ignores the performance differences and attenuation characteristics between battery cluster cells, resulting in uneven utilization of battery components, difficulty in coping with the uncertainty of grid fluctuations and load changes, inability to achieve global optimal control effects, and difficulty in dynamically adjusting control strategies according to actual state changes during operation. The system safety is insufficient, especially in the inability to respond in a timely manner under extreme working conditions. Summary of the Invention

[0003] The embodiments of the present invention provide a method and system for collaborative scheduling and control of a virtual power plant energy storage system, which can at least solve some of the problems existing in the prior art.

[0004] A first aspect of an embodiment of the present invention provides a method for coordinated scheduling and control of a virtual power plant energy storage system, comprising: Collecting real-time operating parameters of battery cluster cells in the energy storage system, analyzing the real-time operating parameters based on a deep reinforcement learning method to obtain performance degradation patterns of the battery cluster cells, and determining initial operating parameters of each battery cluster cell based on the performance degradation patterns; Based on a rolling time domain, the system operating state of multiple future time periods is predicted to obtain a system state prediction value. A power allocation strategy is dynamically generated based on the system state prediction value and the initial operating parameters. The optimal scheduling plan for each time period is calculated by combining the power allocation strategy with mixed integer linear programming and model predictive control methods. The optimal scheduling plan is comprehensively considered, including grid fluctuations, load changes, and charging and discharging benefits, and the optimal scheduling plan is modified in real time to obtain a time-based scheduling instruction for the energy storage system. The charging and discharging operations are performed according to the time-division scheduling instructions and the operating parameters of the energy storage system are monitored in real time. When the operating parameters exceed a preset safety threshold, the charging and discharging power is adjusted according to a preset emergency control strategy.

[0005] In an optional embodiment, Obtain grid dispatch data within a preset time period and calculate the charging and discharging benefits of the energy storage system while meeting system constraints, including: Obtain grid load demand information and electricity price information within a preset time period; Performing a time series analysis on the load demand information and electricity price information of the power grid to obtain a charging and discharging period division result including a high-load and high-price period and a low-load and low-price period; According to the charging and discharging period division results, combined with the power constraint and capacity constraint of the energy storage system, charging operations are performed during low-load and low-electricity-price periods, and discharging operations are performed during high-load and high-electricity-price periods, and the charging and discharging benefits of the energy storage system are calculated.

[0006] In an optional embodiment, Collecting real-time operating parameters of battery cluster cells in the energy storage system, analyzing the real-time operating parameters based on a deep reinforcement learning method to obtain a performance degradation law of the battery cluster cells, and determining the initial operating parameters of each battery cluster cell based on the performance degradation law includes: Collecting real-time operating parameters of battery cluster cells, removing outliers from the real-time operating parameters using a triple standard deviation criterion, normalizing the real-time operating parameters after outlier removal to their maximum and minimum values, and extracting time-domain features from the normalized real-time operating parameters using a wavelet transform to obtain characteristic parameters of the battery cluster cells, wherein the real-time operating parameters include terminal voltage parameters, open-circuit voltage parameters, battery state of charge, cycle count, surface temperature parameters, and internal temperature distribution parameters; Analyzing the characteristic parameters based on a deep reinforcement learning algorithm to obtain a capacity attenuation law and an internal resistance growth law of the battery cluster monomers, and calculating a capacity retention rate and an internal resistance change rate of the battery cluster monomers, and determining a health status of the battery cluster monomers based on the capacity retention rate and the internal resistance change rate; Determine a charge and discharge depth limit according to the health state and the real-time operating parameters, allocate power to the battery cluster monomers based on the charge and discharge depth limit, and obtain initial operating parameters of each battery cluster monomer.

[0007] In an optional embodiment, The characteristic parameters are analyzed based on a deep reinforcement learning algorithm to obtain the capacity attenuation law and internal resistance growth law of the battery cluster monomer and calculate the capacity retention rate and internal resistance change rate of the battery cluster monomer. Establishing a temperature field gradient conduction network based on the characteristic parameters, calculating the spatial temperature gradient and time temperature change rate of the surface temperature parameter and the internal temperature distribution parameter in the characteristic parameters, and determining the temperature distribution characteristics of the battery cluster monomer according to the spatial temperature gradient and time temperature change rate; Establishing a heat conduction relationship matrix based on the physical position relationship of the battery cluster cells, calculating the thermal resistance between the cells and establishing a temperature field conduction equation to obtain the heat flow conduction characteristics of the battery cluster cells, calculating the thermal stress based on the heat flow conduction characteristics and determining the stress distribution characteristics of the electrode material, and coupling the stress distribution characteristics with the temperature distribution characteristics to obtain a temperature-stress coupling characteristic; The spatial temperature gradient, time temperature change rate, and temperature stress coupling characteristics are fused with the characteristic parameters to obtain a fused feature vector, and the fused feature vector is analyzed based on a deep reinforcement learning algorithm to establish a temperature field capacity decay coupling model and a temperature field internal resistance growth coupling model to obtain the capacity decay law and internal resistance growth law of the battery cluster monomer; The initial capacity retention rate and initial internal resistance change rate of the battery cluster monomer are calculated according to the capacity attenuation law and the internal resistance growth law, and the initial capacity retention rate and initial internal resistance change rate are corrected by using the spatial temperature gradient and temperature stress coupling characteristics to obtain the capacity retention rate and internal resistance change rate of the battery cluster monomer.

[0008] In an optional embodiment, Predicting the system operating state for multiple future time periods based on a rolling time domain to obtain a system state prediction value, and dynamically generating a power allocation strategy based on the system state prediction value and the initial operating parameters includes: Set a rolling prediction interval and a rolling update period, and use a deep learning algorithm to predict the load power, grid power, energy storage power, and state of charge within the rolling prediction interval to obtain the load power prediction value, grid power prediction value, energy storage power prediction value, and state of charge prediction value for each time period, and combine them to obtain the system state prediction value; Calculate the maximum allowable fluctuation range of grid power and energy storage power, the maximum allowable charging and discharging interval of energy storage power, and the maximum allowable switching interval of grid power according to the system state prediction value and the initial operating parameters, and generate an initial power allocation plan; The initial power allocation scheme is applied to the grid power forecast value and the energy storage power forecast value of each time period to ensure that the sum of the grid power and the energy storage power in each time period is equal to the load power forecast value of the corresponding time period, the energy storage power is within the maximum allowable charging and discharging range, and the grid power is within the maximum allowable exchange range, thereby obtaining the power allocation strategy.

[0009] In an optional embodiment, By combining the power allocation strategy with mixed integer linear programming and model predictive control methods, the optimal scheduling plan for each time period is calculated. The optimal scheduling plan is modified in real time, taking into account grid fluctuations, load changes, and charging and discharging benefits. The time-segment scheduling instructions for the energy storage system are obtained, including: Extracting grid power time series data to calculate grid fluctuation sequence, extracting load power time series data to calculate load change sequence, constructing the current grid power and load power into a root node state vector, superimposing the grid fluctuation sequence and load change sequence onto the root node state vector at preset time intervals to obtain child node state vectors at different times, constructing the root node state vector and child node state vectors into a scenario tree structure, and calculating the occurrence probability of each node in the scenario tree structure and the transition probability between adjacent nodes; Based on the scenario tree structure, a continuous variable vector including an energy storage power variable and a state of charge variable is established; an integer variable vector including a charge state indicator variable and a discharge state indicator variable is established; a power balance constraint condition is established that the sum of the grid power and the energy storage power is equal to the load power; and an energy storage system operation constraint condition is established for the range of energy storage power and state of charge values; Linearizing the continuous variable vector and the integer variable vector, calculating the energy storage power and charge-discharge status corresponding to the minimum cost under the power balance constraint and the energy storage system operation constraint to obtain an initial optimal solution, setting a prediction time window and a control time window, calculating the grid power forecast value and load power forecast value for the future time period based on the current grid power and load power within the prediction time window, calculating the energy storage power and charge-discharge status for each time period within the control time window, using the power allocation strategy as a constraint condition, optimizing and iteratively calculating the energy storage power and charge-discharge status to obtain the optimal scheduling solution for each time period; Calculate the deviation between the current grid power and load power and the grid power and load power in the sub-node state vector, calculate the scenario weight coefficient and power correction based on the deviation, combine the grid fluctuation sequence, load change sequence and the charging and discharging benefits, superimpose the power correction with the energy storage power in the optimal scheduling scheme, and obtain the time-sharing scheduling instructions for the energy storage system.

[0010] In an optional embodiment, Executing charging and discharging operations according to the time-segment scheduling instructions and monitoring the operating parameters of the energy storage system in real time, and adjusting the charging and discharging power according to a preset emergency control strategy when the operating parameters exceed a preset safety threshold, includes: receiving a time-division scheduling instruction, and controlling the energy storage system to perform charging and discharging operations according to the energy storage power and charging and discharging status in the time-division scheduling instruction; Monitor the real-time operating parameters of the energy storage system and compare them with the corresponding preset safety thresholds; When any parameter exceeds the corresponding preset safety threshold, the charging power or discharging power is reduced according to the power adjustment rules in the preset emergency control strategy until the operating parameter returns to the preset safety threshold range.

[0011] A second aspect of an embodiment of the present invention provides a virtual power plant energy storage system coordinated dispatching and control system, including: The first unit is used to obtain grid dispatch data within a preset time period and calculate the charging and discharging benefits of the energy storage system while meeting system constraints; The second unit is used to collect real-time operating parameters of battery cluster cells in the energy storage system, analyze the real-time operating parameters based on a deep reinforcement learning method, obtain performance degradation patterns of the battery cluster cells, and determine initial operating parameters of each battery cluster cell based on the performance degradation patterns; The third unit is used to predict the system operating status of multiple time periods in the future based on a rolling time domain to obtain a system state prediction value, dynamically generate a power allocation strategy based on the system state prediction value and the initial operating parameters, calculate the optimal scheduling plan for each time period by combining the power allocation strategy through mixed integer linear programming and model predictive control methods, comprehensively consider power grid fluctuations, load changes, and charging and discharging benefits, and make real-time corrections to the optimal scheduling plan to obtain time-based scheduling instructions for the energy storage system; The fourth unit is used to perform charging and discharging operations according to the time-sharing scheduling instructions and monitor the operating parameters of the energy storage system in real time. When the operating parameters exceed the preset safety threshold, the charging and discharging power is adjusted according to the preset emergency control strategy.

[0012] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including: A processor and a memory for storing processor-executable instructions, wherein the processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0013] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0014] In the present invention, the performance degradation law of the battery cluster monomers of the energy storage system is analyzed through deep reinforcement learning, and the initial operating parameters are determined according to the degradation law, which effectively prolongs the battery life and improves the overall reliability of the system. The power allocation strategy is dynamically generated and the optimal scheduling plan is calculated. The grid fluctuations, load changes and charging and discharging benefits are comprehensively considered to achieve efficient scheduling and optimized operation of the energy storage system, significantly improving the economic benefits. When the operating parameters exceed the safety threshold, the charging and discharging power can be adjusted according to the preset emergency control strategy, ensuring the safe operation of the energy storage system, enhancing the system's ability to respond to emergencies, and improving the stability and reliability of the entire virtual power plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1Schematic diagram of the flow of a method for collaborative scheduling and control of a virtual power plant energy storage system according to an embodiment of the present invention; Figure 2 A thermal stress distribution diagram of battery electrode materials in a collaborative scheduling and control method for a virtual power plant energy storage system according to an embodiment of the present invention; Figure 3 A flow chart for generating a power allocation strategy for a method for collaborative scheduling and control of a virtual power plant energy storage system according to an embodiment of the present invention; Figure 4 This is a comparison chart of the power grid power balance performance of the collaborative scheduling and control method of the virtual power plant energy storage system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only 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 ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0017] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0018] Figure 1 FIG. 1 is a flow chart of a method for collaborative scheduling and control of a virtual power plant energy storage system according to an embodiment of the present invention. Figure 1 As shown, the method includes: Collecting real-time operating parameters of battery cluster cells in the energy storage system, analyzing the real-time operating parameters based on a deep reinforcement learning method to obtain performance degradation patterns of the battery cluster cells, and determining initial operating parameters of each battery cluster cell based on the performance degradation patterns; Based on a rolling time domain, the system operating state of multiple future time periods is predicted to obtain a system state prediction value. A power allocation strategy is dynamically generated based on the system state prediction value and the initial operating parameters. The optimal scheduling plan for each time period is calculated by combining the power allocation strategy with mixed integer linear programming and model predictive control methods. The optimal scheduling plan is comprehensively considered, including grid fluctuations, load changes, and charging and discharging benefits, and the optimal scheduling plan is modified in real time to obtain a time-based scheduling instruction for the energy storage system. The charging and discharging operations are performed according to the time-division scheduling instructions and the operating parameters of the energy storage system are monitored in real time. When the operating parameters exceed a preset safety threshold, the charging and discharging power is adjusted according to a preset emergency control strategy.

[0019] In an optional embodiment, Obtain grid dispatch data within a preset time period and calculate the charging and discharging benefits of the energy storage system while meeting system constraints, including: Obtain grid load demand information and electricity price information within a preset time period; Performing a time series analysis on the load demand information and electricity price information of the power grid to obtain a charging and discharging period division result including a high-load and high-price period and a low-load and low-price period; According to the charging and discharging period division results, combined with the power constraint and capacity constraint of the energy storage system, charging operations are performed during low-load and low-electricity-price periods, and discharging operations are performed during high-load and high-electricity-price periods, and the charging and discharging benefits of the energy storage system are calculated.

[0020] Obtain grid load demand information and electricity price information within a preset time period. In practical applications, data can be obtained through power trading centers, power grid companies, or energy management systems. For example, obtain the grid load curve and time-of-use electricity price data for a certain area for the next 24 hours. Load demand information is usually recorded in units of kilowatts or megawatts per hour or every 15 minutes, recording the power load of the grid in each time period; electricity price information records the electricity price in different time periods, in yuan / kilowatt-hour. Establish a data connection with the power grid dispatching system through the API interface, automatically obtain updated load and electricity price data on a regular basis, and store them in the local database for subsequent analysis.

[0021] After acquiring the data, we conduct a time series analysis of the grid's load demand and electricity price information. This involves three steps: data preprocessing, time period feature extraction, and time period segmentation. We also perform missing value processing, outlier detection, and correction on the raw data to ensure data quality. For example, we use a moving average method to smooth out abnormal fluctuations in the load curve, and we use the average value of historical data from the same time period to fill in missing electricity price data.

[0022] In the period feature extraction stage, the comprehensive index of each period is calculated. This index takes into account both the load level and the electricity price level. The load value and electricity value of each period are divided by their average values ​​to obtain normalized values, and the weighted sum is calculated as the comprehensive index of the period. For example, the 24-hour load data (in megawatts) for a certain day is: [80, 70, 65, 60, 65, 75, 90, 110, 120, 125, 130, 135, 140, 135, 130, 135, 140, 145, 140, 130, 120, 110, 100, 90], and the corresponding electricity price data (in yuan / kWh) is: [0.3, 0.3, 0.3, 0.3, 0.4, 0.5, 0.8, 0.8, 0.8, 1.0, 1.0, 1.0, 0.8, 0.8, 0.8, 1.2, 1.2, 1.2, 0.8, 0.8, 0.5, 0.4, 0.3]. After normalization and comprehensive calculation, the hourly comprehensive index value is generated.

[0023] In the period division stage, thresholds are set according to the comprehensive index values, and the periods above the high threshold are divided into high-load and high-electricity-price periods (discharging periods), and the periods below the low threshold are divided into low-load and low-electricity-price periods (charging periods). For example, 1 a.m. to 5 a.m. are divided into low-load and low-electricity-price periods (charging periods), and 10 a.m. to 2 p.m. and 5 p.m. to 7 p.m. are divided into high-load and high-electricity-price periods (discharging periods).

[0024] After the time period division is complete, the energy storage system's charging and discharging benefits are calculated based on the results of the charge and discharge period divisions, combined with the energy storage system's power and capacity constraints. Charging operations are performed during low-load, low-price periods, and discharging operations are performed during high-load, high-price periods. Power constraints refer to the energy storage system's maximum charge and discharge power limits, while capacity constraints refer to the energy storage system's available capacity range and charge and discharge depth limits.

[0025] Initialize the energy storage system's initial state of charge (SOC), for example, to 30% capacity. For each time period, determine whether to charge or discharge based on the time period division. During the charging period, calculate the chargeable capacity, which depends on the energy storage system's remaining capacity and maximum charging power. During the discharging period, calculate the dischargeable capacity, which depends on the energy storage system's current charge and maximum discharge power, taking into account the energy storage system's charge and discharge efficiency.

[0026] For example, assume the energy storage system has a capacity of 1000 kWh, a maximum charge and discharge power of 200 kW, and a charge and discharge efficiency of 90%. At 2:00 AM (a low-load, low-price period, with an electricity price of 0.3 yuan / kWh), a decision is made to charge. The current stored energy capacity is 350 kWh. The available charge capacity is min(200 kW, (1000 - 350) / 0.9) = 200 kW. After one hour of charging, the stored energy capacity increases to 350 + 200 × 0.9 = 530 kWh, resulting in a charging cost of 200 × 0.3 = 60 yuan. At 6:00 PM (a high-load, high-price period, with an electricity price of 1.2 yuan / kWh), a decision is made to discharge the energy. The current energy storage capacity is 830 kWh, and the discharge capacity is min(200 kW, 830×0.9)=200 kW. After one hour of discharge, the energy storage capacity is reduced to 830-200 / 0.9=608 kWh, and the discharge profit is 200×1.2=240 yuan.

[0027] By accumulating the revenue and costs of all charging and discharging operations within a preset time period, the net revenue of the energy storage system within that time period is calculated. If the energy storage system charges a total of 600 kWh (cost 180 yuan) and discharges 550 kWh (revenue 605 yuan) in a day, the daily net revenue is 605-180=425 yuan.

[0028] In this embodiment, through precise time series analysis, it is possible to capture the characteristics of electricity price fluctuations and maximize the charge-discharge price gap. The decision-making mechanism based on time series analysis enables the system to adapt to the changing characteristics of grid load and electricity prices in different regions and seasons. It has strong versatility and adaptability. The charging and discharging operations of the energy storage system can effectively smooth out grid load fluctuations, reduce grid peak pressure, and improve grid stability and reliability. At the same time, it can improve energy utilization efficiency, providing strong support for the stable operation of the entire power system and the efficient use of renewable energy.

[0029] In an optional embodiment, Collecting real-time operating parameters of battery cluster cells in the energy storage system, analyzing the real-time operating parameters based on a deep reinforcement learning method to obtain a performance degradation law of the battery cluster cells, and determining the initial operating parameters of each battery cluster cell based on the performance degradation law includes: Collecting real-time operating parameters of battery cluster cells, removing outliers from the real-time operating parameters using a triple standard deviation criterion, normalizing the real-time operating parameters after outlier removal to their maximum and minimum values, and extracting time-domain features from the normalized real-time operating parameters using a wavelet transform to obtain characteristic parameters of the battery cluster cells, wherein the real-time operating parameters include terminal voltage parameters, open-circuit voltage parameters, battery state of charge, cycle count, surface temperature parameters, and internal temperature distribution parameters; Analyzing the characteristic parameters based on a deep reinforcement learning algorithm to obtain a capacity attenuation law and an internal resistance growth law of the battery cluster monomers, and calculating a capacity retention rate and an internal resistance change rate of the battery cluster monomers, and determining a health status of the battery cluster monomers based on the capacity retention rate and the internal resistance change rate; Determine a charge and discharge depth limit according to the health state and the real-time operating parameters, allocate power to the battery cluster monomers based on the charge and discharge depth limit, and obtain initial operating parameters of each battery cluster monomer.

[0030] During the operation of the energy storage system, the deployed battery management system collects real-time operating parameters of each battery cell at a frequency of once per second. Collected parameters include terminal voltage, open-circuit voltage, battery state of charge, cycle count, surface temperature, and internal temperature distribution. Terminal voltage is measured by voltage acquisition modules connected to the positive and negative terminals of the battery, with a range of 0-5V and an accuracy of ±0.01V. Open-circuit voltage is measured by leaving the battery at rest for at least 30 minutes. Battery state of charge is determined using coulomb counting combined with open-circuit voltage correction, with an accuracy of ±2%. Cycle count is accumulated and recorded by a counter in the battery management system. Surface temperature is collected using thermistors attached to the battery surface, with collection points located at the top, middle, and bottom of the battery, with an accuracy of ±0.5°C. Internal temperature distribution is acquired using an array of micro-temperature sensors embedded within the battery, with a total of eight measurement points, generating internal temperature field data.

[0031] The collected real-time operating parameters are processed for outlier removal using the triple standard deviation criterion. Taking the terminal voltage parameter as an example, for 1000 consecutive sets of data, the average value is calculated to be 3.65V, and the standard deviation is 0.08V. Data outside the range of [3.41V, 3.89V] is considered an outlier and is removed. The same method is used to remove outliers for open-circuit voltage, battery state of charge, surface temperature parameters, and internal temperature distribution parameters. For example, if the average value of a set of battery surface temperature data is 28.5°C and the standard deviation is 1.2°C, then data outside the range of [24.9°C, 32.1°C] is determined to be an outlier and is removed.

[0032] After removing outliers, the data is normalized to its maximum and minimum values. For example, the battery state of charge (SOC) ranges from 10% to 90%. Through normalization, the data is converted to a value between 0 and 1. The normalized value is equal to the original value minus the minimum value, divided by the difference between the maximum and minimum values. For example, if the original battery SOC is 65%, the normalized value is (65% - 10%) / (90% - 10%) = 0.6875.

[0033] After standardization, wavelet transform is used to extract time-domain features from the data. The db4 wavelet basis function is selected, and a five-layer wavelet decomposition is performed on the standardized voltage, temperature, and other parameters to extract energy, entropy, and statistical features. The mean, variance, kurtosis, and skewness of the battery terminal voltage parameters are extracted, and the energy proportion of the wavelet coefficients at different scales is calculated. For example, after wavelet transforming the terminal voltage curve of a battery cell during charging, the extracted mean characteristic is 0.72, the variance characteristic is 0.15, and the energy proportions of the first to fifth layers of wavelet decomposition are 0.08, 0.12, 0.25, 0.35, and 0.20, respectively. These features together constitute the characteristic parameter set of the battery cluster cell.

[0034] Based on a deep reinforcement learning algorithm, the feature parameters were analyzed and a deep reinforcement learning network architecture was constructed. The network consists of a feature extraction network and a policy network. The feature extraction network uses a four-layer convolutional neural network structure with convolution kernel sizes of 5×5, 3×3, 3×3, and 3×3, and the number of convolution kernels is 32, 64, 128, and 256, respectively. Each convolution layer is followed by a max pooling layer and a batch normalization layer. The policy network uses a three-layer fully connected network with 512, 256, and 128 neurons, respectively, and uses Reinforced Luminance (ReLU) as the activation function. The network input is the battery feature parameters, and the output is the predicted value of the battery capacity decay rate and internal resistance growth rate.

[0035] The network was trained using an experience replay mechanism, with an experience pool size of 10,000 and a batch size of 64. During training, the reward function was designed to be the negative of the mean squared error between the predicted and measured values, with the goal of maximizing the cumulative reward. The learning rate was initially set to 0.001, and the Adam optimizer was used with a learning rate decay strategy that decayed by a factor of 0.9 every 50 epochs. During training, the exploration rate ε was initially set to 0.9 and linearly decreased to 0.1 as training progressed. After 5,000 epochs of training, the model achieved a prediction accuracy of 92.3% on the validation set.

[0036] Using a trained deep reinforcement learning model, we analyze the capacity decay and internal resistance growth patterns of battery cluster cells. Based on the predicted results, we calculate the battery capacity retention rate, which is the ratio of the current capacity to the initial capacity. For example, if a battery cluster cell has an initial capacity of 100Ah and after 200 cycles, the capacity decays to 93Ah, with a capacity retention rate of 93%. We also calculate the internal resistance change rate, which is the ratio of the current internal resistance to the initial internal resistance. For example, if the initial internal resistance is 5mΩ and the current internal resistance is 5.8mΩ, the internal resistance change rate is 116%.

[0037] The health status of each battery cell is determined based on the capacity retention rate and internal resistance change rate. The health status assessment criteria are: when the capacity retention rate is greater than 90% and the internal resistance change rate is less than 120%, the health status is "good"; when the capacity retention rate is between 80% and 90% or the internal resistance change rate is between 120% and 150%, the health status is "fair"; when the capacity retention rate is less than 80% or the internal resistance change rate is greater than 150%, the health status is "poor".

[0038] Depth of charge and discharge limits are determined based on the battery's health status and real-time operating parameters. For batteries with a "good" health status, the depth of charge and discharge limit is set at 10%-90%; for batteries with a "fair" health status, the depth of charge and discharge limit is adjusted to 20%-80%; and for batteries with a "poor" health status, the depth of charge and discharge limit is further narrowed to 30%-70%.

[0039] Power is allocated to each battery cell in a cluster based on the depth of charge and discharge limits. This power allocation is weighted based on the battery's health status, with batteries in better health receiving a higher power share. For example, in a system consisting of three battery clusters, the corresponding power allocation ratios for health statuses are 50%, 30%, and 20%, respectively. If the total system power is 100kW, each battery cluster will receive 50kW, 30kW, and 20kW, respectively. The initial operating parameters of each battery cell in this calculation, including operating voltage, operating current, and operating power, ensure optimal overall system performance while extending battery life.

[0040] In this embodiment, outliers are eliminated by using the triple standard deviation criterion and combined with maximum and minimum value normalization processing, which effectively improves the data quality of battery operating parameters and reduces the interference of abnormal data on subsequent analysis. The wavelet transform is used to extract time domain features, which can fully capture the dynamic change characteristics of battery operating parameters. Compared with traditional time domain analysis methods, it can more accurately reflect the changing laws of the battery's working state. Based on the deep reinforcement learning algorithm, the capacity attenuation law and internal resistance growth law are analyzed, and the adaptive evaluation of the battery health status is realized, avoiding the limitations of the traditional fixed threshold method and improving the evaluation accuracy.

[0041] In an optional embodiment, The characteristic parameters are analyzed based on a deep reinforcement learning algorithm to obtain the capacity attenuation law and internal resistance growth law of the battery cluster monomer and calculate the capacity retention rate and internal resistance change rate of the battery cluster monomer. Establishing a temperature field gradient conduction network based on the characteristic parameters, calculating the spatial temperature gradient and time temperature change rate of the surface temperature parameter and the internal temperature distribution parameter in the characteristic parameters, and determining the temperature distribution characteristics of the battery cluster monomer according to the spatial temperature gradient and time temperature change rate; Establishing a heat conduction relationship matrix based on the physical position relationship of the battery cluster cells, calculating the thermal resistance between the cells and establishing a temperature field conduction equation to obtain the heat flow conduction characteristics of the battery cluster cells, calculating the thermal stress based on the heat flow conduction characteristics and determining the stress distribution characteristics of the electrode material, and coupling the stress distribution characteristics with the temperature distribution characteristics to obtain a temperature-stress coupling characteristic; The spatial temperature gradient, time temperature change rate, and temperature stress coupling characteristics are fused with the characteristic parameters to obtain a fused feature vector, and the fused feature vector is analyzed based on a deep reinforcement learning algorithm to establish a temperature field capacity decay coupling model and a temperature field internal resistance growth coupling model to obtain the capacity decay law and internal resistance growth law of the battery cluster monomer; The initial capacity retention rate and initial internal resistance change rate of the battery cluster monomer are calculated according to the capacity attenuation law and the internal resistance growth law, and the initial capacity retention rate and initial internal resistance change rate are corrected by using the spatial temperature gradient and temperature stress coupling characteristics to obtain the capacity retention rate and internal resistance change rate of the battery cluster monomer.

[0042] The battery cluster is divided into n×m×l grid cells, each representing a tiny region in space. Each grid cell is assigned a surface temperature value extracted from the characteristic parameters, and a three-dimensional convolutional neural network is used to model the temperature field. This 3D convolutional neural network consists of five convolutional layers, each using 32 3×3×3 convolution kernels, with a Reinforced Luminance (ReLU) activation function. This network calculates the temperature difference between adjacent grid cells to obtain a spatial temperature gradient, ranging from 0.5°C / cm to 3.5°C / cm. The time-temperature rate of change is calculated from temperature data at consecutive time points, typically ranging from 0.02°C / s to 0.5°C / s under normal operating conditions. For example, in one set of test data, the spatial temperature gradient reached 2.8°C / cm in the center of the battery cluster and 1.2°C / cm at the edges. The maximum time-temperature rate of change during charge and discharge was 0.35°C / s.

[0043] Based on the physical position relationship of the battery cluster cells, a heat conduction relationship matrix is ​​established. Each cell in the battery cluster is regarded as a node, and the physical connection between the nodes is regarded as an edge to construct an undirected weighted graph. The weight of the edge in the graph represents the heat conduction coefficient between the two cells, and the heat conduction coefficient is determined by the material properties and the contact area. For a closely arranged lithium battery pack, the typical value of the heat conduction coefficient between adjacent cells is 0.8W / (m·K) to 1.5W / (m·K). The thermal resistance is calculated using a thermal network model, and the thermal resistance between each cell is expressed as the inverse of the heat conduction coefficient multiplied by the length of the conduction path, and then divided by the contact area. For example, the thermal resistance value between adjacent cells ranges from 0.5K / W to 2.0K / W. The temperature field conduction equation is solved by the finite difference method, with a time step of 10 seconds and a spatial step of 5mm. The heat flux density distribution obtained by the solution shows that during fast charging, the heat flux density in the central area of ​​the battery cluster can reach 120W / m 2 , while the edge area is about 60W / m 2 .

[0044] Thermal stress calculation is based on the thermal expansion coefficient and temperature difference. The linear thermal expansion coefficient of the electrode material is about 1.6×10 -5 / °C. Finite element analysis (FEM) revealed that the maximum stress within the battery is concentrated in the contact area between the positive and negative electrodes, reaching values ​​as high as 18 MPa. The temperature and stress distribution characteristics are coupled using a weight coefficient method to construct a temperature-stress coupling eigenvector. This vector consists of three components: the temperature gradient, the stress gradient, and their product. The weight coefficients are determined through training with historical battery performance data, with typical values ​​being [0.4, 0.3, 0.3].

[0045] The fused feature vector is constructed using feature concatenation and dimensionality reduction. The spatial temperature gradient (3 × n × m × l dimensions), the temporal temperature change rate (n × m × l dimensions), and the temperature-stress coupling feature (3 × n × m × l dimensions) are concatenated with the original feature parameters (including voltage, current, SOC, etc., with dimension k) to produce a high-dimensional feature vector. Principal component analysis is then used to reduce the dimensionality of the vector to 128 dimensions, retaining over 95% of the information.

[0046] The deep reinforcement learning algorithm is implemented using a dual deep Q-network framework. The state space is a fused feature vector, and the action space is the predicted adjustment values ​​for the capacity decay rate and internal resistance growth rate, ranging from [-0.5%, 0.5%] and [-1%, 1%], respectively, with 21 discrete values. The reward function is designed based on the prediction error, with smaller prediction errors resulting in higher rewards. The network structure consists of four fully connected layers with 256, 128, 64, and 32 neurons, respectively, and the activation function is Leaky ReLU. During training, the batch size is set to 64, the learning rate is 0.001, the discount factor is 0.95, and the target network update frequency is 100 steps. The training data consists of 200 sets of cycle test data for battery clusters under different operating conditions, each set containing at least 500 charge and discharge cycles.

[0047] Modeling results of capacity decay show that under standard operating conditions (25°C ambient temperature, 1C charge / discharge rate), the capacity decay of the battery cell cluster follows a bi-exponential decay pattern, with a decay rate of approximately 0.15% / cycle over the first 100 cycles, decreasing to 0.08% / cycle over the later stages. The internal resistance growth pattern exhibits a near-linear growth trend, with a growth rate of approximately 0.2% / cycle. Temperature gradient significantly influences capacity decay, with the capacity decay rate increasing by approximately 0.03% / cycle for every 1°C / cm increase in temperature gradient.

[0048] The calculations for the initial capacity retention rate and initial internal resistance change rate are based on the baseline values ​​of new batteries, with the new battery capacity defined as 100% and the internal resistance as 100%. Using a capacity decay model, the capacity after n cycles is predicted as Cn, and the capacity retention rate is Cn / C0 × 100%. Similarly, the internal resistance change rate is Rn / R0 × 100%, where Rn is the internal resistance after n cycles. Spatial temperature gradients and temperature stress coupling characteristics are corrected using a linear adjustment method, with the correction coefficient determined through regression analysis. For example, in one battery cluster test, the initially predicted capacity retention rate after 100 cycles was 92%. After correction for temperature gradients (correction coefficient -0.5% / (°C / cm)) and temperature stress coupling characteristics (correction coefficient -0.3% / MPa), the final capacity retention rate was corrected to 89.6%, close to the measured value of 90.1%.

[0049] In this embodiment, both the spatial temperature gradient and the temporal temperature change rate are considered simultaneously, achieving accurate modeling of the temperature distribution of the battery cluster cells. By coupling the temperature distribution characteristics with the stress distribution characteristics, the limitations of traditional separate temperature and stress analysis are overcome, and a collaborative analysis of the temperature field and stress field is achieved. Through multi-physics field coupling analysis, a more reliable battery health status assessment method is provided, providing strong technical support for the safe and stable operation of the energy storage system, significantly improving the accuracy and reliability of the battery health status assessment, and having important significance for extending the service life of the energy storage system.

[0050] Figure 2 This is a thermal stress distribution diagram of the battery electrode material in the collaborative scheduling and control method of the virtual power plant energy storage system according to the embodiment of the present invention. Based on the precise temperature gradient and the linear thermal expansion coefficient of the electrode material (1.6×10 -5 / °C) by finite element analysis to calculate the thermal stress distribution, and the temperature-stress coupling eigenvector (weight coefficient [0.4, 0.3, 0.3]) was used to achieve an effective fusion of temperature distribution characteristics and stress distribution characteristics; like Figure 2 As shown, the maximum thermal stress in the contact area of ​​the positive and negative electrodes reaches 17.8MPa (red area in the figure), the thermal stress in the middle area of ​​the electrode is 8.4MPa (yellow area in the figure), and the lowest in the edge area is only 3.2MPa (blue area in the figure). The non-uniform distribution is highly consistent with the actual internal structure and temperature distribution of the battery, and the prediction accuracy reaches 94.6%. The traditional model assumes uniform thermal expansion and predicts a maximum thermal stress of only 12.5MPa, which is far lower than the measured value of 18MPa, and cannot capture local high stress areas. The prediction accuracy is only 69.4%. This technical solution can accurately predict the stress concentration phenomenon in hot spot areas during rapid charging and discharging, providing a key basis for evaluating battery performance degradation and potential safety risks.

[0051] In an optional embodiment, Predicting the system operating state for multiple future time periods based on a rolling time domain to obtain a system state prediction value, and dynamically generating a power allocation strategy based on the system state prediction value and the initial operating parameters includes: Set a rolling prediction interval and a rolling update period, and use a deep learning algorithm to predict the load power, grid power, energy storage power, and state of charge within the rolling prediction interval to obtain the load power prediction value, grid power prediction value, energy storage power prediction value, and state of charge prediction value for each time period, and combine them to obtain the system state prediction value; Calculate the maximum allowable fluctuation range of grid power and energy storage power, the maximum allowable charging and discharging interval of energy storage power, and the maximum allowable switching interval of grid power according to the system state prediction value and the initial operating parameters, and generate an initial power allocation plan; The initial power allocation scheme is applied to the grid power forecast value and the energy storage power forecast value of each time period to ensure that the sum of the grid power and the energy storage power in each time period is equal to the load power forecast value of the corresponding time period, the energy storage power is within the maximum allowable charging and discharging range, and the grid power is within the maximum allowable exchange range, thereby obtaining the power allocation strategy.

[0052] The rolling forecast interval is set to 4 hours, with a rolling update cycle of 15 minutes. Within each rolling forecast interval, a deep learning algorithm based on a long short-term memory (LSTM) network is used to forecast load power, grid power, energy storage power, and state of charge for future time periods. The forecast process uses historical data as input, including load power data, grid power data, energy storage power data, and state of charge data recorded every 15 minutes for the past 24 hours. Through LSTM network training, load power, grid power, energy storage power, and state of charge are predicted every 15 minutes for the next 4 hours, resulting in forecast values ​​for a total of 16 time periods. For example, in one forecast, the load power forecast for the first 15-minute period was 120 kW, the grid power forecast was 80 kW, the energy storage power forecast was 40 kW, and the state of charge forecast was 75%. These forecast values ​​are combined to form the system state forecast.

[0053] After obtaining the predicted system state, the maximum allowable fluctuation range of grid power and energy storage power, the maximum allowable charging and discharging range of energy storage power, and the maximum allowable switching range of grid power are calculated based on the predicted system state and initial operating parameters. This generates an initial power allocation plan. The initial operating parameters include a maximum allowable grid power of 150kW, a minimum allowable power of -50kW (a negative value indicates feeding power into the grid), a maximum energy storage system charging power of 60kW, a maximum discharge power of 60kW, an upper state of charge of 95%, a lower state of charge of 15%, a current state of charge of 75%, and an energy storage system capacity of 200kWh.

[0054] Calculate the maximum allowable fluctuation ranges for grid power and energy storage power. Taking into account grid stability requirements, set the maximum change rate for grid power between two adjacent time periods to 20kW / 15 minutes, and the maximum change rate for energy storage power between two adjacent time periods to 30kW / 15 minutes. For example, if the current grid power is 80kW, the grid power in the next 15-minute period should be between 60kW and 100kW; if the current energy storage power is 40kW, the energy storage power in the next 15-minute period should be between 10kW and 70kW.

[0055] The maximum allowable charge and discharge range of energy storage power is determined by the state of charge (SOC) and physical limitations of the energy storage system. For example, at a SOC of 75%, there's still 20% of capacity remaining to charge from the upper limit of 95%, corresponding to 40 kWh. The maximum charge power allowed within 15 minutes is 60 kW (considering a charging efficiency of approximately 90%), resulting in a maximum actual charge power of 54 kW. The current dischargeable capacity is 60% (75% - 15%), corresponding to 120 kWh. The maximum discharge power allowed within 15 minutes is 60 kW (considering a discharge efficiency of approximately 95%), resulting in a maximum actual discharge power of 57 kW. Taking all these factors into consideration, the maximum allowable charge and discharge range for energy storage power is [-57 kW, 54 kW], where negative values ​​indicate discharge and positive values ​​indicate charge.

[0056] The maximum allowable exchange range of the grid power is determined according to the technical limitations of the grid, and the maximum allowable exchange range of the grid power is [-50kW, 150kW].

[0057] Based on the above calculation results, an initial power allocation plan is generated. For the first 15-minute period, the load power forecast is 120 kW, and the initial allocation plan is: 80 kW grid power and 40 kW energy storage power (discharging). For the second 15-minute period, the load power forecast is 125 kW, and the initial allocation plan is: 85 kW grid power and 40 kW energy storage power (discharging). For the third 15-minute period, the load power forecast is 130 kW, and the initial allocation plan is: 90 kW grid power and 40 kW energy storage power (discharging). Initial power allocation plans are generated for each 15-minute period over the next four hours.

[0058] The initial power allocation scheme is applied to the grid power forecast and energy storage power forecast for each time period, and adjustments are made to ensure that three conditions are met: the sum of the grid power and energy storage power in each time period is equal to the load power forecast for the corresponding time period; the energy storage power is within the maximum allowable charging and discharging range; and the grid power is within the maximum allowable switching range.

[0059] For example, if the load power forecast value for the fourth 15-minute period is 140kW, the initial allocation plan is 95kW grid power and 45kW energy storage power (discharging). However, since the grid power in the previous period was 90kW, the maximum change rate limit means that the maximum grid power in the fourth period can only reach 110kW. Therefore, it is adjusted to 110kW grid power and 30kW energy storage power (discharging).

[0060] After adjustment, the final power allocation strategy is obtained. After each rolling update cycle (15 minutes), the prediction and power allocation process is re-executed to achieve dynamic rolling optimization.

[0061] In this embodiment, a rolling prediction of the system status is achieved through a deep learning algorithm. This not only accurately grasps the dynamic change trends of load power, grid power, and energy storage system operating status, but also ensures the real-time and accuracy of the prediction by setting the rolling prediction interval and update cycle. The multi-dimensional constraint design not only ensures the safety of system operation, but also improves the feasibility of the power allocation plan. Through the organic combination of prediction and constraint, both the economy of the system and the stability of operation are guaranteed, providing effective technical support for the optimal scheduling of the energy storage system.

[0062] Figure 3 A flow chart for generating a power allocation strategy for a collaborative scheduling and control method for a virtual power plant energy storage system according to an embodiment of the present invention.

[0063] In an optional embodiment, By combining the power allocation strategy with mixed integer linear programming and model predictive control methods, the optimal scheduling plan for each time period is calculated. The optimal scheduling plan is modified in real time, taking into account grid fluctuations, load changes, and charging and discharging benefits. The time-segment scheduling instructions for the energy storage system are obtained, including: Extracting grid power time series data to calculate grid fluctuation sequence, extracting load power time series data to calculate load change sequence, constructing the current grid power and load power into a root node state vector, superimposing the grid fluctuation sequence and load change sequence onto the root node state vector at preset time intervals to obtain child node state vectors at different times, constructing the root node state vector and child node state vectors into a scenario tree structure, and calculating the occurrence probability of each node in the scenario tree structure and the transition probability between adjacent nodes; Based on the scenario tree structure, a continuous variable vector including an energy storage power variable and a state of charge variable is established; an integer variable vector including a charge state indicator variable and a discharge state indicator variable is established; a power balance constraint condition is established that the sum of the grid power and the energy storage power is equal to the load power; and an energy storage system operation constraint condition is established for the range of energy storage power and state of charge values; Linearizing the continuous variable vector and the integer variable vector, calculating the energy storage power and charge-discharge status corresponding to the minimum cost under the power balance constraint and the energy storage system operation constraint to obtain an initial optimal solution, setting a prediction time window and a control time window, calculating the grid power forecast value and load power forecast value for the future time period based on the current grid power and load power within the prediction time window, calculating the energy storage power and charge-discharge status for each time period within the control time window, using the power allocation strategy as a constraint condition, optimizing and iteratively calculating the energy storage power and charge-discharge status to obtain the optimal scheduling solution for each time period; Calculate the deviation between the current grid power and load power and the grid power and load power in the sub-node state vector, calculate the scenario weight coefficient and power correction based on the deviation, combine the grid fluctuation sequence, load change sequence and the charging and discharging benefits, superimpose the power correction with the energy storage power in the optimal scheduling scheme, and obtain the time-sharing scheduling instructions for the energy storage system.

[0064] Extract the last seven days of grid power data from the historical database, perform differential processing on it, and calculate the grid power change value for each period to form a grid fluctuation sequence. For example, if the power of a certain grid is 100kW, 102kW, 105kW, etc. over 24 consecutive hours, the corresponding grid fluctuation sequence is 2kW, 3kW, etc. Extract the last seven days of load power data and calculate the load power change value for each period to form a load change sequence. If the load power is 95kW, 98kW, 103kW, etc., the load change sequence is 3kW, 5kW, etc.

[0065] The grid power Pg(t0) and load power Pl(t0) at the current time t0 are combined into the root node state vector S0 = [Pg(t0), Pl(t0)]. For example, if the current grid power is 100 kW and the load power is 95 kW, the root node state vector S0 = [100 kW, 95 kW]. Based on a preset time interval Δt (e.g., 15 minutes), the grid fluctuation sequence and the load change sequence are superimposed on the root node state vector to obtain the child node state vectors. For example, the first child node state vector S1 = [100 kW + 2 kW, 95 kW + 3 kW] = [102 kW, 98 kW], and the second child node state vector S2 = [102 kW + 3 kW, 98 kW + 5 kW] = [105 kW, 103 kW]. The root node and all child nodes are constructed into a scenario tree structure, with each node representing the system state at a specific moment. Based on statistical analysis of historical data, the occurrence probability of each node and the transition probability between adjacent nodes are calculated. For example, according to historical data analysis, the probability that the grid power changes from 100kW to 102kW is 0.7, and the probability that it changes to 98kW is 0.3.

[0066] Based on the scenario tree structure, a mixed integer linear programming model is established. Define the continuous variable vector X = [Ps(1), Ps(2), ..., Ps(N), SOC(1), SOC(2), ..., SOC(N)], where Ps(i) represents the energy storage power in the i-th period and SOC(i) represents the state of charge in the i-th period. Define the integer variable vector Y = [δc(1), δc(2), ..., δc(N), δd(1), δd(2), ..., δd(N)], where δc(i) and δd(i) are the charge state indicator variable and discharge state indicator variable of the i-th period, respectively, and take the value of 0 or 1.

[0067] Establish a power balance constraint: For each time period i, Pg(i) + Ps(i) = Pl(i). Here, Pg(i) is the grid power, Ps(i) is the energy storage power, and Pl(i) is the load power. For example, if the grid power is 100kW and the load power is 95kW at a certain moment, the energy storage power Ps should be -5kW (a negative value indicates charging).

[0068] Establish the energy storage system's operating constraints: For each time period i, the energy storage power Ps(i) ranges from [-Ps_max, Ps_max], where Ps_max is the maximum power of the energy storage system, such as 100kW. The state of charge (SOC(i)) ranges from [SOC_min, SOC_max], where SOC_min and SOC_max represent the minimum and maximum states of charge, such as 20% and 90%, respectively. The charge and discharge state indicator variable satisfies δc(i) + δd(i) ≤ 1, indicating that the energy storage system cannot be charged and discharged simultaneously at the same time.

[0069] Linearize continuous and integer variables to avoid nonlinear terms in the model. For example, express the charge and discharge power as Ps(i) = Pc(i) - Pd(i), where Pc(i) = δc(i) × Ps(i) represents the charge power and Pd(i) = δd(i) × Ps(i) represents the discharge power. Linearize the product terms.

[0070] Under power balance and energy storage system operating constraints, calculate the energy storage power and charge / discharge status that minimizes total operating costs. This total cost includes the grid's electricity purchase costs and energy storage system losses. For example, assuming an electricity price of 0.8 yuan / kWh and a 90% energy storage system charge / discharge efficiency, the total cost of charging 1 kWh of energy storage during a specific time period is 0.8 × 1 ÷ 0.9 = 0.89 yuan. By solving the optimization problem, an initial optimal solution is obtained, including the energy storage power and charge / discharge status for each time period.

[0071] Set the prediction time window \(T_p\) (e.g., 24 hours) and the control time window \(T_c\) (e.g., 4 hours), where \(T_c < T_p\). Based on the current grid power and load power, use the autoregressive integrated moving average model to calculate the predicted values of grid power and load power within the future \(T_p\) period. For example, if the current grid power is 100 kW, the predicted values of grid power per hour within the next 24 hours calculated according to the prediction model are 102 kW, 105 kW, etc.

[0072] Within the control time window \(T_c\), combined with the power distribution strategy, calculate the energy storage power and charge / discharge status for each period. The power distribution strategy takes into account the characteristics of different types of energy storage devices. For example, battery energy storage is suitable for handling medium- and long-term fluctuations, while flywheel energy storage is suitable for handling short-term fluctuations. For a hybrid energy storage system composed of 10% battery energy storage and 90% flywheel energy storage, the fluctuating power is distributed according to the frequency characteristics, with the high-frequency part handled by the flywheel energy storage and the low-frequency part handled by the battery energy storage.

[0073] Through iterative optimization calculations, obtain the optimal scheduling plans for each period. Represent these plans as \([P_s^*(1), P_s^*(2),..., P_s^*(T_c)]\) and \([\delta_c^*(1), \delta_c^*(2),..., \delta_c^*(T_c), \delta_d^*(1), \delta_d^*(2),..., \delta_d^*(T_c)]\).

[0074] Real-time monitor the current grid power \(P_{g\_real}\) and load power \(P_{l\_real}\), and calculate the deviations from the predicted values \(\Delta P_g = P_{g\_real}-P_{g\_pred}\) and \(\Delta P_l = P_{l\_real}-P_{l\_pred}\). For example, if the predicted grid power is 100 kW and the actual value is 105 kW, then the deviation \(\Delta P_g = 5\) kW. Based on the deviation values, calculate the scenario weight coefficient \(w_i\) and the power correction amount \(\Delta P_s\). For example, if the deviation is large, increase the weight of the corresponding scenario and adjust the power correction amount accordingly.

[0075] Combined with the grid fluctuation sequence, load change sequence, and charge / discharge benefits, superimpose the power correction amount \(\Delta P_s\) on the energy storage power \(P_s^*\) in the optimal scheduling plan to obtain the corrected energy storage power \(P_{s\_adj}=P_s^*+\Delta P_s\). For example, if the optimal energy storage power for a certain period is -10 kW and the power correction amount is 2 kW, then the corrected energy storage power is -8 kW.

[0076] According to the corrected energy storage power and charge / discharge status, generate the time-segmented scheduling instructions for the energy storage system, including the charge / discharge power values and operating modes for each period. The time-segmented scheduling instructions are sent to the energy storage device actuator through the control system to achieve the optimal operation of the energy storage system.

[0077] In this embodiment, by extracting the time-series variation characteristics of grid power and load power and establishing a dynamic evolution path from the root node to the child node, not only the random characteristics of system operation are captured, but also scientific predictions are achieved through the calculation of occurrence probability and transition probability. By establishing power balance constraints and operation constraints, the feasibility of the optimization results is ensured. A rolling optimization strategy with two time windows is adopted. The long-term variation trend of the system is captured through the prediction time window, and the energy storage system is precisely regulated through the control time window. This not only improves the operating efficiency of the energy storage system, but also enhances the system's adaptability to changes in the external environment, providing a strong guarantee for the economical and efficient operation of the energy storage system.

[0078] Figure 4 This is a comparison chart of the grid power balance performance of the collaborative scheduling and control method for a virtual power plant energy storage system according to an embodiment of the present invention, showing the performance differences in power fluctuation control between this technical solution and the traditional PID control algorithm and fixed threshold control method. The technical solution achieves the smallest grid power fluctuation, with a fluctuation range of 7-15kW; the fluctuation range of the traditional PID control algorithm is between 18-25kW; and the fluctuation range of the fixed threshold control method is between 30-38kW. At the beginning of the test (0 hours), the power fluctuations of the three methods were 15kW, 25kW and 35kW respectively; by the 8th hour, the power fluctuation values ​​of the three methods dropped to 7kW, 18kW and 30kW respectively; and at the 16th hour, the power fluctuation values ​​were 11kW, 23kW and 38kW respectively; by the 24th hour, the fluctuation values ​​of the three methods were 10kW, 21kW and 37kW respectively; finally, at the 32nd hour (the end of the test), the fluctuation values ​​of the three methods were 12kW, 22kW and 38kW respectively.

[0079] During the entire test cycle, the average power fluctuation of this technical solution was about 11kW, which was about 50% lower than the average fluctuation value of 22kW of the traditional PID control algorithm and about 69% lower than the average fluctuation value of 35kW of the fixed threshold control method. The advantages of this technical solution were particularly obvious during the high-load period from the 10th to the 14th hour, with the power fluctuation remaining at around 9kW, while the traditional PID control and fixed threshold control reached 19kW and 34kW, respectively.

[0080] This technical solution simultaneously considers grid fluctuations, load changes, and charging and discharging benefits, and dynamically adjusts the charging and discharging strategy of the energy storage system through a scenario tree structure and real-time correction mechanism. After a long period of operation (28-32 hours), the power fluctuation of this technical solution remained stable at around 12kW, while the other two methods increased to 22kW and 38kW, respectively, demonstrating the advantages of this solution in long-term stable control.

[0081] In an optional embodiment, Executing charging and discharging operations according to the time-segment scheduling instructions and monitoring the operating parameters of the energy storage system in real time, and adjusting the charging and discharging power according to a preset emergency control strategy when the operating parameters exceed a preset safety threshold, includes: receiving a time-division scheduling instruction, and controlling the energy storage system to perform charging and discharging operations according to the energy storage power and charging and discharging status in the time-division scheduling instruction; Monitor the real-time operating parameters of the energy storage system and compare them with the corresponding preset safety thresholds; When any parameter exceeds the corresponding preset safety threshold, the charging power or discharging power is reduced according to the power adjustment rules in the preset emergency control strategy until the operating parameter returns to the preset safety threshold range.

[0082] The energy storage system receives time-segment dispatch instructions from the grid dispatch center via a communication interface. These instructions include the energy storage power value and charge / discharge status for each time period. For example, the dispatch instruction specifies discharging at 500 kW during peak grid load periods (e.g., 6:00 PM to 9:00 PM) and charging at 400 kW during off-peak grid load periods (e.g., 2:00 AM to 5:00 AM). The energy storage system controller interprets the instructions and determines the appropriate charge / discharge operation based on the current system clock. The controller then sends corresponding control signals to the energy storage unit via the power conversion system (PCS), controlling its charge / discharge status and power level.

[0083] While charging and discharging, the energy storage system continuously monitors various operating parameters, including but not limited to battery temperature, voltage, current, state of charge (SOC), internal impedance of the energy storage unit, and power conversion efficiency. The monitoring frequency is set based on the importance of the parameter. Key parameters like battery temperature can be monitored once per second, while more stable parameters like internal impedance can be monitored once per minute. Real-time parameter data collected is transmitted to the controller via the data acquisition system for processing and analysis.

[0084] The controller compares various parameters monitored in real time with preset safety thresholds. These thresholds are determined based on the energy storage system's specifications and actual operating experience. For example, for lithium-ion battery energy storage systems, the safety threshold for battery cell temperature can be set between -20°C and 55°C; exceeding this range triggers emergency control; the safety threshold for battery cell voltage can be set between 2.5V and 4.2V; the safe operating range for state of charge can be set between 10% and 90% to avoid overcharging and over-discharging; the safety threshold for battery charge current can be set to no more than 0.5C (0.5 times the rated capacity of the battery); and the safety threshold for discharge current can be set to no more than 1C.

[0085] When any monitored parameter exceeds the corresponding preset safety threshold, the controller will activate the emergency control strategy. The emergency control strategy includes power adjustment rules for different parameter abnormalities. The controller dynamically adjusts the charging and discharging power based on the type of abnormal parameter and the degree of deviation from the safety threshold. For example, when the battery temperature is detected to reach 50°C (close to the upper threshold of 55°C), the controller will linearly reduce the charging power from the original 400kW to 200kW according to the preset rules to reduce heat; if the temperature continues to rise to 52°C, the charging power will be further reduced to 100kW; if it reaches 54°C, the charging operation will be completely stopped.

[0086] Different emergency control strategies are set for different parameters. For abnormal SOC conditions, when the SOC reaches 85% (close to the upper limit of 90%), the charging power will automatically be reduced to 50% of the original power; when the SOC reaches 88%, the charging power is further reduced to 20% of the original power; when the SOC reaches 89%, the charging power is reduced to the minimum value or charging is completely stopped. Similarly, when the SOC drops to 15% (close to the lower limit of 10%), the discharge power will be reduced to 50% of the original power; when the SOC drops to 12%, the discharge power will be reduced to 20% of the original power; when the SOC drops to 11%, the discharge power is reduced to the minimum value or discharge is completely stopped.

[0087] For abnormal battery voltage, when the single cell voltage reaches 4.1V (close to the upper limit of 4.2V), the charging power will be reduced to 70% of the original power; when the voltage reaches 4.15V, the charging power will be reduced to 30% of the original power; when the voltage reaches 4.18V, the battery unit will stop charging, and when the single cell voltage drops to 2.7V (close to the lower limit of 2.5V), the discharge power will be reduced to 70% of the original power; when the voltage drops to 2.6V, the discharge power will be reduced to 30% of the original power; when the voltage drops to 2.55V, the battery unit will stop discharging.

[0088] The emergency control strategy also includes a recovery mechanism that gradually restores charging and discharging power once abnormal parameters return to a certain margin within the safety threshold. For example, when the battery temperature drops from an abnormally high 50°C to 45°C (10°C below the safety threshold of 55°C), charging power is restored to 50% of the original rated power. When the temperature further drops to 40°C, charging power is fully restored to the original rated power of 400kW.

[0089] In this embodiment, by executing the energy storage power and charge-discharge status control requirements in the time-slot scheduling instructions, precise scheduling and control of the energy storage system is achieved, laying the foundation for the economic operation of the system. Threshold-based real-time monitoring can promptly detect problems in the early stages of abnormal situations and avoid further expansion of faults. Flexible power adjustment strategies maintain continuous operation of the system to the greatest extent while ensuring system safety, thereby improving the reliability and utilization efficiency of the energy storage system.

[0090] A second aspect of an embodiment of the present invention provides a virtual power plant energy storage system coordinated dispatching and control system, including: The first unit is used to obtain grid dispatch data within a preset time period and calculate the charging and discharging benefits of the energy storage system while meeting system constraints; The second unit is used to collect real-time operating parameters of battery cluster cells in the energy storage system, analyze the real-time operating parameters based on a deep reinforcement learning method, obtain performance degradation patterns of the battery cluster cells, and determine initial operating parameters of each battery cluster cell based on the performance degradation patterns; The third unit is used to predict the system operating status of multiple time periods in the future based on a rolling time domain to obtain a system state prediction value, dynamically generate a power allocation strategy based on the system state prediction value and the initial operating parameters, calculate the optimal scheduling plan for each time period by combining the power allocation strategy through mixed integer linear programming and model predictive control methods, comprehensively consider power grid fluctuations, load changes, and charging and discharging benefits, and make real-time corrections to the optimal scheduling plan to obtain time-based scheduling instructions for the energy storage system; The fourth unit is used to perform charging and discharging operations according to the time-sharing scheduling instructions and monitor the operating parameters of the energy storage system in real time. When the operating parameters exceed the preset safety threshold, the charging and discharging power is adjusted according to the preset emergency control strategy.

[0091] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including: A processor and a memory for storing processor-executable instructions, wherein the processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0092] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0093] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for collaborative scheduling and control of a virtual power plant energy storage system, characterized in that: include: Obtain grid dispatch data within a preset time period and calculate the charging and discharging benefits of the energy storage system while meeting system constraints; Collecting real-time operating parameters of battery cluster cells in the energy storage system, analyzing the real-time operating parameters based on a deep reinforcement learning method to obtain performance degradation patterns of the battery cluster cells, and determining initial operating parameters of each battery cluster cell based on the performance degradation patterns; Based on a rolling time domain, the system operating state of multiple future time periods is predicted to obtain a system state prediction value. A power allocation strategy is dynamically generated based on the system state prediction value and the initial operating parameters. The optimal scheduling plan for each time period is calculated by combining the power allocation strategy with mixed integer linear programming and model predictive control methods. The optimal scheduling plan is comprehensively considered, including grid fluctuations, load changes, and charging and discharging benefits, and the optimal scheduling plan is modified in real time to obtain a time-based scheduling instruction for the energy storage system. The charging and discharging operations are performed according to the time-division scheduling instructions and the operating parameters of the energy storage system are monitored in real time. When the operating parameters exceed a preset safety threshold, the charging and discharging power is adjusted according to a preset emergency control strategy.

2. The method according to claim 1, characterized in that Obtain grid dispatch data within a preset time period and calculate the charging and discharging benefits of the energy storage system while meeting system constraints, including: Obtain grid load demand information and electricity price information within a preset time period; Performing a time series analysis on the load demand information and electricity price information of the power grid to obtain a charging and discharging period division result including a high-load and high-price period and a low-load and low-price period; According to the charging and discharging period division results, combined with the power constraint and capacity constraint of the energy storage system, charging operations are performed during low-load and low-electricity-price periods, and discharging operations are performed during high-load and high-electricity-price periods, and the charging and discharging benefits of the energy storage system are calculated.

3. The method according to claim 1, characterized in that Collecting real-time operating parameters of battery cluster cells in the energy storage system, analyzing the real-time operating parameters based on a deep reinforcement learning method to obtain a performance degradation law of the battery cluster cells, and determining the initial operating parameters of each battery cluster cell based on the performance degradation law includes: Collecting real-time operating parameters of battery cluster cells, removing outliers from the real-time operating parameters using a triple standard deviation criterion, normalizing the real-time operating parameters after outlier removal to their maximum and minimum values, and extracting time-domain features from the normalized real-time operating parameters using a wavelet transform to obtain characteristic parameters of the battery cluster cells, wherein the real-time operating parameters include terminal voltage parameters, open-circuit voltage parameters, battery state of charge, cycle count, surface temperature parameters, and internal temperature distribution parameters; Analyzing the characteristic parameters based on a deep reinforcement learning algorithm to obtain a capacity attenuation law and an internal resistance growth law of the battery cluster monomers, and calculating a capacity retention rate and an internal resistance change rate of the battery cluster monomers, and determining a health status of the battery cluster monomers based on the capacity retention rate and the internal resistance change rate; Determine a charge and discharge depth limit according to the health state and the real-time operating parameters, allocate power to the battery cluster monomers based on the charge and discharge depth limit, and obtain initial operating parameters of each battery cluster monomer.

4. The method according to claim 3, characterized in that The characteristic parameters are analyzed based on a deep reinforcement learning algorithm to obtain the capacity attenuation law and internal resistance growth law of the battery cluster monomer and calculate the capacity retention rate and internal resistance change rate of the battery cluster monomer. Establishing a temperature field gradient conduction network based on the characteristic parameters, calculating the spatial temperature gradient and time temperature change rate of the surface temperature parameter and the internal temperature distribution parameter in the characteristic parameters, and determining the temperature distribution characteristics of the battery cluster monomer according to the spatial temperature gradient and time temperature change rate; Establishing a heat conduction relationship matrix based on the physical position relationship of the battery cluster cells, calculating the thermal resistance between the cells and establishing a temperature field conduction equation to obtain the heat flow conduction characteristics of the battery cluster cells, calculating the thermal stress based on the heat flow conduction characteristics and determining the stress distribution characteristics of the electrode material, and coupling the stress distribution characteristics with the temperature distribution characteristics to obtain a temperature-stress coupling characteristic; The spatial temperature gradient, time temperature change rate, and temperature stress coupling characteristics are fused with the characteristic parameters to obtain a fused feature vector, and the fused feature vector is analyzed based on a deep reinforcement learning algorithm to establish a temperature field capacity decay coupling model and a temperature field internal resistance growth coupling model to obtain the capacity decay law and internal resistance growth law of the battery cluster monomer; The initial capacity retention rate and initial internal resistance change rate of the battery cluster monomer are calculated according to the capacity attenuation law and the internal resistance growth law, and the initial capacity retention rate and initial internal resistance change rate are corrected by using the spatial temperature gradient and temperature stress coupling characteristics to obtain the capacity retention rate and internal resistance change rate of the battery cluster monomer.

5. The method according to claim 1, wherein Predicting the system operating state for multiple future time periods based on a rolling time domain to obtain a system state prediction value, and dynamically generating a power allocation strategy based on the system state prediction value and the initial operating parameters includes: Set a rolling prediction interval and a rolling update period, and use a deep learning algorithm to predict the load power, grid power, energy storage power, and state of charge within the rolling prediction interval to obtain the load power prediction value, grid power prediction value, energy storage power prediction value, and state of charge prediction value for each time period, and combine them to obtain the system state prediction value; Calculate the maximum allowable fluctuation range of grid power and energy storage power, the maximum allowable charging and discharging interval of energy storage power, and the maximum allowable switching interval of grid power according to the system state prediction value and the initial operating parameters, and generate an initial power allocation plan; The initial power allocation scheme is applied to the grid power forecast value and the energy storage power forecast value of each time period to ensure that the sum of the grid power and the energy storage power in each time period is equal to the load power forecast value of the corresponding time period, the energy storage power is within the maximum allowable charging and discharging range, and the grid power is within the maximum allowable exchange range, thereby obtaining the power allocation strategy.

6. The method according to claim 1, wherein By combining the power allocation strategy with mixed integer linear programming and model predictive control methods, the optimal scheduling plan for each time period is calculated. The optimal scheduling plan is modified in real time, taking into account grid fluctuations, load changes, and charging and discharging benefits. The time-segment scheduling instructions for the energy storage system are obtained, including: Extracting grid power time series data to calculate grid fluctuation sequence, extracting load power time series data to calculate load change sequence, constructing the current grid power and load power into a root node state vector, superimposing the grid fluctuation sequence and load change sequence onto the root node state vector at preset time intervals to obtain child node state vectors at different times, constructing the root node state vector and child node state vectors into a scenario tree structure, and calculating the occurrence probability of each node in the scenario tree structure and the transition probability between adjacent nodes; Based on the scenario tree structure, a continuous variable vector including an energy storage power variable and a state of charge variable is established; an integer variable vector including a charge state indicator variable and a discharge state indicator variable is established; a power balance constraint condition is established that the sum of the grid power and the energy storage power is equal to the load power; and an energy storage system operation constraint condition is established for the range of energy storage power and state of charge values; Linearizing the continuous variable vector and the integer variable vector, calculating the energy storage power and charge-discharge status corresponding to the minimum cost under the power balance constraint and the energy storage system operation constraint to obtain an initial optimal solution, setting a prediction time window and a control time window, calculating the grid power forecast value and load power forecast value for the future time period based on the current grid power and load power within the prediction time window, calculating the energy storage power and charge-discharge status for each time period within the control time window, using the power allocation strategy as a constraint condition, optimizing and iteratively calculating the energy storage power and charge-discharge status to obtain the optimal scheduling solution for each time period; Calculate the deviation between the current grid power and load power and the grid power and load power in the sub-node state vector, calculate the scenario weight coefficient and power correction based on the deviation, combine the grid fluctuation sequence, load change sequence and the charging and discharging benefits, superimpose the power correction with the energy storage power in the optimal scheduling scheme, and obtain the time-sharing scheduling instructions for the energy storage system.

7. The method according to claim 1, characterized in that Executing charging and discharging operations according to the time-segment scheduling instructions and monitoring the operating parameters of the energy storage system in real time, and adjusting the charging and discharging power according to a preset emergency control strategy when the operating parameters exceed a preset safety threshold, includes: receiving a time-division scheduling instruction, and controlling the energy storage system to perform charging and discharging operations according to the energy storage power and charging and discharging status in the time-division scheduling instruction; Monitor the real-time operating parameters of the energy storage system and compare them with the corresponding preset safety thresholds; When any parameter exceeds the corresponding preset safety threshold, the charging power or discharging power is reduced according to the power adjustment rules in the preset emergency control strategy until the operating parameter returns to the preset safety threshold range.

8. A virtual power plant energy storage system coordinated dispatching and control system, used to implement the method according to any one of claims 1 to 7, characterized in that: include: The first unit is used to obtain grid dispatch data within a preset time period and calculate the charging and discharging benefits of the energy storage system while meeting system constraints; The second unit is used to collect real-time operating parameters of battery cluster cells in the energy storage system, analyze the real-time operating parameters based on a deep reinforcement learning method, obtain performance degradation patterns of the battery cluster cells, and determine initial operating parameters of each battery cluster cell based on the performance degradation patterns; The third unit is used to predict the system operating status of multiple time periods in the future based on a rolling time domain to obtain a system state prediction value, dynamically generate a power allocation strategy based on the system state prediction value and the initial operating parameters, calculate the optimal scheduling plan for each time period by combining the power allocation strategy through mixed integer linear programming and model predictive control methods, comprehensively consider power grid fluctuations, load changes, and charging and discharging benefits, and make real-time corrections to the optimal scheduling plan to obtain time-based scheduling instructions for the energy storage system; The fourth unit is used to perform charging and discharging operations according to the time-sharing scheduling instructions and monitor the operating parameters of the energy storage system in real time. When the operating parameters exceed the preset safety threshold, the charging and discharging power is adjusted according to the preset emergency control strategy.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images