Method and system for optimizing working temperature of liquid cooling system of energy storage battery compartment and medium

Abstract

The invention relates to the technical field of energy storage system operation optimization, and discloses an energy storage battery tank liquid cooling system working temperature optimization method and system and a medium, and the method comprises the steps: obtaining a time sequence of battery operation condition characteristics at each moment in a future time period, and obtaining the working temperature of a liquid cooling system at each moment in the future time period through each iteration of an optimization algorithm; and iteratively solving by utilizing the constructed battery body temperature prediction model, battery aging acceleration factor evaluation model and working temperature optimization model in combination with an optimization algorithm, and outputting the working temperature optimal value of the liquid cooling system at each moment in the future time period. According to the method, the optimal value of the working temperature of the battery compartment liquid cooling system at each moment in the future time period is dynamically determined by predicting the battery operation condition characteristics and the body temperature change trend and taking the optimal battery aging acceleration factor as an optimization target, so that the working temperature of the battery compartment liquid cooling system is adjusted to quickly, accurately and dynamically respond to different operation conditions of the battery; and optimal management of battery aging of the energy storage power station is also realized.

Description

Technical Field

[0001] This invention relates to the field of energy storage system operation optimization technology, specifically to a method, system, and medium for optimizing the operating temperature of an energy storage battery compartment liquid cooling system. Background Technology

[0002] Currently, energy storage technologies in power systems are mainly divided into three categories: physical energy storage, electromagnetic energy storage, and battery energy storage. Among them, battery energy storage technology, with its significant cost advantages, fast response speed, and good recyclability, has been widely used in renewable energy grid integration and smart grid construction, providing an effective technical solution for grid frequency regulation, power fluctuation mitigation, and improved system reliability.

[0003] However, with the long-term operation of energy storage batteries under charge-discharge cycles and the accumulation of service years, their aging problems are becoming increasingly prominent. This issue not only significantly reduces the economics of energy storage systems but may also adversely affect the safe and stable operation of the entire power system. Battery temperature is the most critical factor affecting the battery aging process. Excessively high operating temperatures accelerate electrolyte decomposition and degradation of positive and negative electrode materials, leading to rapid capacity decay; while excessively low operating temperatures may induce lithium dendrite growth, increasing the risk of battery short circuits. In addition, frequent temperature fluctuations can exacerbate uneven stress distribution within the battery, further accelerating the aging process. Therefore, reasonable and effective battery temperature control is key to suppressing battery aging and improving the reliability of energy storage systems.

[0004] Currently, energy storage power stations generally use liquid-cooled air conditioning systems to regulate the ambient temperature inside the battery compartment, thereby indirectly controlling the battery temperature. However, in actual operation, most energy storage power stations still employ a crude control strategy of "fixed threshold start-up (e.g., battery temperature reaches 30°C) + temperature-guided adjustment" for the liquid cooling system's operating temperature. This means that the liquid cooling system is activated after the battery temperature reaches a preset threshold, and its operating temperature is simply adjusted based on the battery temperature change trend. This crude liquid cooling system operating temperature control strategy has the following significant technical limitations: The current operating temperature threshold of the battery compartment liquid cooling system is set empirically, failing to consider the complex and dynamically changing mapping relationship between battery aging and operating temperature. Furthermore, temperature-guided adjustment is based solely on the current battery temperature at the moment of operation, resulting in significant lag. Consequently, the existing crude liquid cooling system operating temperature control strategy cannot dynamically respond to the differentiated heat dissipation needs of the battery under various operating conditions, leading to inaccurate temperature regulation and causing accelerated battery aging or unsuitable use. For example, during high-rate charging and discharging, the battery's heat generation increases dramatically; under such conditions, a fixed liquid cooling system operating temperature setting often results in insufficient heat dissipation, easily triggering accelerated aging. Summary of the Invention

[0005] This invention aims to provide a method, system, and medium for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment. By predicting the battery's operating conditions and temperature change trends, and with the goal of optimizing the battery aging acceleration factor, the operating temperature of the liquid cooling system at various times in the future (e.g., 24 hours in the next day) is dynamically determined in advance. This enables rapid and precise dynamic response of the liquid cooling system to different battery operating conditions, while also providing precise control over battery aging. This provides an efficient technical solution for temperature management in energy storage power stations, significantly improving the economy and reliability of energy storage power stations.

[0006] The basic solution provided by this invention is: a method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment, comprising: A battery body temperature prediction model is constructed with battery operating condition characteristics and battery compartment liquid cooling system operating temperature as inputs and battery body temperature as output. Construct a battery aging acceleration factor assessment model that characterizes the mapping relationship between the bulk temperature of a battery cell and the aging acceleration factor; Construct an operating temperature optimization model with the operating temperature of the battery compartment liquid cooling system as the decision variable and the optimal battery aging acceleration factor as the objective; The time series of battery operating condition characteristics at each time point in the future period is obtained. The operating temperature of the liquid cooling system at each time point in the future period is obtained through each iteration of the optimization algorithm. The battery body temperature prediction model, battery aging acceleration factor evaluation model and operating temperature optimization model are constructed and iteratively solved by the optimization algorithm to output the optimal value of the operating temperature of the liquid cooling system at each time point in the future period.

[0007] The present invention also provides a system for optimizing the operating temperature of an energy storage battery compartment liquid cooling system, so as to perform a method for optimizing the operating temperature of an energy storage battery compartment liquid cooling system; The system includes: a data processing unit, used to acquire the time series of battery operating condition characteristics at each moment in the future period and the liquid cooling system operating temperature at each moment in the future period obtained by each iteration of the optimization algorithm, and send them to each unit; it is also used to execute the optimization algorithm iterative solution; The battery body temperature prediction unit is used to output the predicted battery body temperature for each moment in the future period based on the time series of battery operating condition characteristics at each moment in the future period and the liquid cooling system operating temperature at each moment in the future period obtained by each iteration of the optimization algorithm, and send it to the battery aging acceleration factor evaluation unit. The battery aging acceleration factor evaluation unit is used to output the predicted values ​​of the battery aging acceleration factor at each time in the future period based on the predicted values ​​of the battery body temperature at each time in the future period, using a pre-built battery aging acceleration factor evaluation model, and send them to the operating temperature optimization unit. The operating temperature optimization unit is used to determine whether the operating temperature of the liquid cooling system at each time in the future period and the battery aging acceleration factor at each time in the future period are optimal based on the operating temperature of the liquid cooling system at each time in the future period obtained by each iteration of the optimization algorithm. If it is optimal, the optimal value of the operating temperature of the liquid cooling system at each time in the future period is output; otherwise, the feedback data processing unit executes the optimization algorithm iteratively to solve the problem.

[0008] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the above-described method for optimizing the operating temperature of an energy storage battery compartment liquid cooling system.

[0009] The working principle and advantages of this invention are as follows: Compared with existing technologies, this invention deeply analyzes the dynamic mapping relationship between battery operating conditions, battery body temperature, battery aging acceleration factor, and battery compartment liquid cooling system operating temperature. It constructs a battery body temperature prediction model, a battery aging acceleration factor evaluation model, and an operating temperature optimization model. Combined with optimization algorithms, it predicts battery operating condition characteristics and body temperature change trends, and aims to optimize the battery aging acceleration factor. It dynamically determines the operating temperature of the battery compartment liquid cooling system at various times in the future (e.g., 24 hours in the future), enabling rapid and accurate dynamic response of the liquid cooling system to different battery operating conditions. At the same time, it precisely controls battery aging, providing an efficient technical solution for temperature management of energy storage power stations, which can significantly improve the economy and reliability of energy storage power stations.

[0010] This invention deeply analyzes the operating characteristics of energy storage batteries, the relationship between the operating temperature of the battery compartment liquid cooling system and the battery body temperature, and establishes a multi-input single-output battery body temperature prediction model based on the Support Vector Regression (SVR) algorithm. This model can characterize the complex mapping relationship between the operating characteristics of energy storage batteries, the operating temperature of the battery compartment liquid cooling system, and the battery body temperature, achieving accurate prediction of the battery body temperature at various points in the future. Furthermore, SVR combined with five-fold cross-validation effectively avoids overfitting and improves the model's generalization ability. By optimizing parameters (such as the penalty factor C and kernel function parameters) using the five-fold method, sufficient learning of sample features can be achieved, enhancing prediction accuracy and stability. This model is suitable for modeling problems with moderate sample size and complex nonlinear relationships.

[0011] Based on the Arrhenius degradation equation, a battery aging acceleration factor assessment model is constructed using battery test analysis data, relating battery body temperature to the battery aging acceleration factor. Building upon this, the Block Hankel Tensor Autoregressive Integrated Moving Average (BHT-ARIMA) algorithm is employed to predict battery operating conditions in future time periods. Combined with preset liquid cooling system operating temperatures at various points in the future time period, the battery body temperature prediction model and the battery aging acceleration factor assessment model are used to achieve multi-step forward prediction of the battery aging acceleration factor. BHT-ARIMA can be used for short-term time series forecasting, integrating methods such as SDT, Tucker decomposition, and tensor ARIMA. Existing research has shown that BHT-ARIMA leverages the advantages of Hankel tensor quantization, significantly outperforming existing short-term time series forecasting methods in terms of effectiveness and efficiency (superior to ARIMA LSTM).

[0012] An optimization model for the operating temperature of the liquid cooling system in the battery compartment is constructed, with the operating temperature of the liquid cooling system in the battery compartment as the decision variable and the optimal battery aging acceleration factor as the objective. An optimization algorithm is used to solve the model, dynamically determining the operating temperature of the liquid cooling system in the battery compartment at various points in the future and pre-setting corresponding instructions to achieve optimized management of battery aging in the energy storage power station. Furthermore, the Particle Swarm Optimization (PSO) algorithm is employed. PSO is a population-based parallel search algorithm derived from research on bird predation behavior. It has advantages such as simple approach, fast convergence speed, few dependent parameters, and good versatility, making it suitable for handling various types of objective functions and constraints. It demonstrates superior performance in solving the optimization model proposed in this scheme. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating a method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment, as provided in an embodiment of the present invention. Figure 2 A schematic diagram illustrating the process of constructing the battery body temperature prediction model provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the process for predicting the battery aging acceleration factor at various points in the future, as provided in an embodiment of the present invention. Figure 4 This is a flowchart illustrating the prediction of the operating temperature of the battery compartment liquid cooling system at various points in the future, as provided in an embodiment of the present invention. Figure 5 This is a comparative diagram of MSE evaluation values ​​corresponding to different kernel functions provided in an embodiment of the present invention. Figure 6 This is a comparative diagram showing the calculation results of the accelerated aging factor provided in the embodiments of the present invention; Figure 7 This is a comparative diagram of the prediction errors of different models provided in the embodiments of the present invention; Figure 8 This is a comparative diagram illustrating the effect of controlling battery aging provided in an embodiment of the present invention; Figure 9 This is a schematic diagram of the operating temperature optimization system for a liquid cooling system in an energy storage battery compartment, provided in an embodiment of the present invention. Detailed Implementation

[0014] The following detailed explanation illustrates the specific implementation methods: The basic implementation examples are as follows: Figure 1 As shown: A method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment, comprising: S1. Construct a battery body temperature prediction model with battery operating condition characteristics and battery compartment liquid cooling system operating temperature as inputs and battery body temperature as output.

[0015] Specifically, such as Figure 2 As shown, historical operational data of the energy storage battery is acquired through a combination of sensor deployment and system data export. This includes the operating temperature records of the liquid cooling system, the operating condition characteristics of each battery cell, and the temperature data of the battery cell itself. It should be noted that this approach focuses on the battery cell as the subject of study.

[0016] S11, Historical Operation Data Acquisition; The operating temperature record of the liquid cooling system can be exported from the daily inspection record. If the liquid cooling system is not turned on at the time of acquisition, the ambient temperature inside the battery compartment will be used instead; The operating condition characteristic data of individual battery cells can be exported from the battery management system (BMS). The operating condition characteristics of individual battery cells include the state of charge (SOC), state of health (SOH), charging / discharging current characteristics, and charging / discharging voltage characteristics; If the energy storage power station is equipped with a battery safety monitoring system, the battery body temperature data can be obtained by exporting data from the system's backend. If a battery safety monitoring system is not yet equipped, the battery body temperature data can be obtained by evenly distributing temperature sensors in the battery compartment.

[0017] S12, Historical Operation Data Preprocessing; To facilitate subsequent data analysis, a unified format for storing historical operation data of energy storage batteries is established, uniformly storing the liquid cooling system operating temperature, battery cell operating condition characteristic data, and battery cell body temperature data at each acquisition moment. n Taking a single battery cell as an example, its corresponding historical operation data storage matrix It can be represented as:

[0018] In the formula, T This represents the total number of data collection points. WT n,t， For the firstn The first battery cell t The operating temperature (°C) of the liquid cooling system at each moment; SOC n,t SOH n,t , CR n,t , VT n,t and BT n,t The first n The first battery cell t The state of charge (SOC) (%), state of health (SOH) (%), charge / discharge current characteristics (A), charge / discharge voltage characteristics (V), and body temperature (°C) at each time point.

[0019] Based on this N Historical operation data storage matrix constructed from individual battery cells It can be represented as:

[0020] in, Indicates the first n The historical operation data storage matrix corresponds to each individual battery cell, with a total of N individual battery cells.

[0021] S13, After shuffling the historical operating data of the energy storage battery obtained in the previous steps, randomly divide it into training sets at a ratio of 80% and 20%. and test set Each sample in the set represents the historical operating data for each battery cell. The battery cell temperature prediction model is constructed using the training set based on the support vector regression algorithm. The specific steps are as follows.

[0022] S14, Establish 5-dimensional input parameters based on the Support Vector Regression Machine (SVR) model. With 1D output y i The regression function between them can be expressed as follows:

[0023]

[0024] In the formula, For training set The total number of samples included; For the sample SVR model predicted cell temperature; This is a transformation function used to map the five input parameters to a high-dimensional feature space to achieve nonlinear fitting; Hyperplane weight vector transpose, b This is the offset. and b These are all unknown parameters in the regression function, obtained by minimizing the regression model error.

[0025] Because the support vector regression model allows for the prediction of the cell's internal temperature. Compared with the actual battery body temperature value y (i.e., variables) BT There exists a maximum value between them. The error. Therefore, the error function corresponding to the aforementioned regression function. f error (·) can be represented as:

[0026] In the formula, For the sample Predicted temperature of individual battery cells Compared with actual value The error; For the sample Predicted battery body temperature; For the sample The actual temperature of the battery body. The maximum permissible error (i.e., the maximum tolerable error) between the predicted and actual temperatures of each sample battery cell.

[0027] It should be noted that the above formula means that only when and The absolute value of the difference between them is greater than Losses are only calculated at that time.

[0028] S15, based on the aforementioned error function, by introducing slack variables and The problem of minimizing the regression error is transformed into a constrained convex quadratic programming optimization problem as shown in the following equation. The regression parameters are estimated by solving the optimization problem:

[0029] In the formula, It is a regularization parameter greater than 0; This represents the deviation between the fitted historical data and the actual data.

[0030] S16, to facilitate the solution, uses the Lagrange multiplier method and kernel function. The dual problem of the above problem is formed, and the optimal Lagrange multiplier is obtained by solving the dual problem through the Karush-Kuhn-Tucker (KKT) conditions. and Therefore, the optimal regression parameters are determined. and :

[0031] It should be noted that the formula obtained in S16 is a transformed form of the formula in S14, and the kernel function is... It has been implicitly defined In higher-dimensional space, there is no need to explicitly obtain... The specific value.

[0032] As can be seen from the above derivation, selecting a suitable kernel function... And accordingly determine the optimal parameter set, i.e., the kernel function parameters. Punishment factor Maximum tolerable error This is crucial for ensuring the accuracy of the temperature prediction model for the energy storage battery itself. The next step will involve a more detailed discussion of the optimal kernel function selection method.

[0033] S17. The four commonly used kernel functions in Support Vector Regression (SVR) include the linear kernel, polynomial kernel, Gaussian radial basis function (RBF) kernel, and sigmoid kernel. Therefore, five-fold cross-validation is first used to find the parameter set that minimizes the fit index of each candidate kernel function on the training set data.

[0034] In the formula, To adopt the first r The optimal parameter set obtained by regression analysis using kernel functions includes , and ; The number of samples contained in the z-th training subset in the five-fold cross-validation; For the sample Based on the r The predicted values ​​of the SVR regression model obtained by the kernel function.

[0035] S18, After determining the optimal parameters for each kernel function, the regression model is trained using the battery running data contained in the entire training set to determine the optimal regression parameters. and Then, the trained model is used to predict the test set. That is, the SVR prediction model corresponding to each kernel function is trained using the training set, and the optimal parameter set corresponding to each prediction model is obtained through five-fold cross-validation. Then, the MPAE index of the SVR prediction model corresponding to the r-th kernel function on the training set is calculated.

[0036] Among them, ther The MAPE index, representing the regression model prediction error corresponding to each kernel function, can be calculated using the following formula:

[0037] In the formula: Let r be the prediction error index of the SVR regression model corresponding to the r-th kernel function. For the test set The number of samples included is taken from D ALL The remaining 20% ​​of the samples; For the sample Based on the r The SVR regression model prediction values ​​obtained from the kernel functions For the sample Actual temperature of the battery itself.

[0038] S19. Based on this, select the kernel function that minimizes the prediction error. And a regression model to construct a multi-input single-output battery body temperature prediction function model. .

[0039]

[0040] In the formula, q in the table below represents the kernel function with the minimum prediction error. The corresponding parameters; and It is the optimal Lagrange multiplier; and These are the optimal regression parameters; For training set The samples included Total number. It should be noted that x represents any new x value, which needs to be calculated using the K function along with the historical Ntrain x values.

[0041] S2. Construct a battery aging acceleration factor evaluation model that characterizes the mapping relationship between battery body temperature and aging acceleration factor, and use it to evaluate the accelerated aging factor of each individual battery cell under different operating conditions.

[0042] S21. The aging process of an energy storage battery is a unidirectional and irreversible process, and its aging rate has a significant thermodynamic correlation with the battery's internal temperature. This step, based on the Arrhenius reaction rate theory, derives a mathematical model that describes the nonlinear mapping relationship between the battery's internal temperature and its aging acceleration factor. The specific expression is as follows:

[0043] In the formula, and For the first n The battery cell in the first h Accelerating aging factors and body temperature (°C) over a period of time. Reference temperature (°C); It is the free gas constant, with a value of 8.314 J / K / mol; This is the activation energy parameter. S22, the unknown parameters in the above model can be estimated by fitting the data from the battery accelerated aging test. .

[0044] S23, Based on the results obtained from the above steps, the first... n The equivalent accelerated aging factor of a single battery cell within a given time period If analyzed according to the 24 time periods of a day, it can be expressed as:

[0045] S3, based on the aforementioned battery body temperature prediction model and aging acceleration factor evaluation model, proposes a method for predicting battery aging acceleration factors based on time-series analysis. For example... Figure 3 As shown.

[0046] This method generates time series of operating condition characteristics (SOH, SOC, CR, VT) of each individual battery cell in the future period based on the Block Hankel Tensor AutoRegressive Integrated Moving Average (BHT-ARIMA) algorithm. Then, it combines the liquid cooling system operating temperature at each moment in the future period obtained by each optimization iteration of the optimization algorithm in S4, and inputs them sequentially into the battery body temperature prediction model and the aging acceleration factor evaluation model, and outputs the battery aging acceleration factor at each moment in the future period, realizing multi-step advance prediction of the battery aging acceleration factor.

[0047] The BHT-ARIMA algorithm represents the time series data of single-cell battery operating conditions as tensors and combines single-way delay-embedding transform (SDT), Tucker decomposition, tensor decomposition and ARIMA modeling to effectively capture complex temporal dependencies in the time series.

[0048] Taking the generation of SOH time series as an example, the specific steps of the BHT-ARIMA algorithm are as follows: S31, Perform SDT along the time direction on the SOH time series of multiple individual cells to convert the SOH series into a tensor form with time-varying characteristics, resulting in a third-order Hankel tensor. :

[0049] In the formula, for N Each battery cell is in T Time series data at each moment; For SDT conversion functions, Let be the rank of SDT.

[0050] S32, the tensor obtained through S31 of d The order difference eliminates the nonstationarity of the sequence, resulting in ; and then mapped it to a low-rank core tensor using the projection matrix method. This leads to Tucker decomposition:

[0051] In the formula, For the first t The core tensor of time can be considered as the past. p Individual differences and past q A linear function of random error observations; and These are the coefficients to be estimated in the BHT-ARIMA model; Indicates the past q The actual error of this calculation; Indicates the first t The prediction error at any given time.

[0052] S33, based on historical core tensor data Estimating model coefficients and A prediction equation for the characteristic sequence of single-cell operating conditions based on the BHT-ARIMA algorithm was established.

[0053] S34, Predicting the first T Core Tensor at Time +1 And calculate the corresponding number of... T +1 time d Step difference Then on Inverse difference is obtained .

[0054] S35, perform inverse SDT, and convert the predicted value Mapping back to the original SOH sequence space, we obtain the values ​​of each individual cell in the first... T Predicted SOH operating condition characteristics at time +1 :

[0055] S36. Iterate through S34-S35 to obtain the predicted SOH values ​​of the battery at each time point in the future period.

[0056] S37, similarly, uses the above steps to predict future time-series data of battery operating condition characteristics SOC, CR, and VT.

[0057] S38. Based on the time series of operating conditions of each individual battery cell in the future period obtained by the above process, and combined with the liquid cooling system operating temperature at each moment of the future period obtained by each optimization iteration of the optimization algorithm in S4 below, the battery body temperature prediction model and aging acceleration factor evaluation model are input sequentially, and the battery aging acceleration factor at each moment of the future period is output to realize multi-step advance prediction of battery aging acceleration factor.

[0058] S4 constructs a nonlinear optimization model for operating temperature, with the operating temperature of the battery compartment liquid cooling system as the decision variable and the optimal aging acceleration factor of the energy storage power station batteries as the objective. This model is solved using a particle swarm optimization (PSO) algorithm, outputting the optimal operating temperature of the liquid cooling system at each future time period, and pre-setting corresponding instructions to achieve optimized management of battery aging in the energy storage power station. Figure 4 As shown.

[0059] S41, determine the objective function of the optimization model for the operating temperature of the liquid cooling system of the energy storage power station. (In the future time period) Z Energy storage power station within a certain time period N The goal is to optimize the average aging acceleration factor of each individual battery cell, which can be specifically expressed as:

[0060] In the formula, For the first n The single cell in the first z The aging acceleration factor value within a given time period.

[0061] S42, determine the constraint objectives of the optimization model for the operating temperature of the liquid cooling system of the energy storage power station, as follows: 1) Upper and lower limits of liquid cooling system operating temperature: The operating temperature of the liquid cooling system at all times should be controlled within the rated operating temperature range designed for the system, i.e., it must meet the following requirements:

[0062] In the formula, and These are the upper and lower limits of the rated operating temperature of the liquid cooling system in an energy storage power station.

[0063] 2) Operating temperature regulation parameters of the liquid cooling system, such as regulation rate constraints: To avoid instability or equipment wear caused by frequent adjustments, the operating temperature variation of the liquid cooling system at adjacent moments must meet the following requirements:

[0064] In the formula, and These represent the maximum range of change for lowering and raising the operating temperature of the liquid cooling system, respectively.

[0065] 3) Temperature limits for individual cells: The operating temperature of each individual cell should be maintained within an acceptable safe range to prevent overheating or efficiency degradation, i.e., it must meet the following requirements:

[0066] In the formula, and These are the upper and lower limits that the operating temperature of a single battery cell must meet.

[0067] S43, the PSO algorithm is used to solve the above optimization model, and finally the optimal operating temperature of the liquid cooling system at each time point in the future is output.

[0068] The PSO algorithm initializes a swarm of particles randomly, using fitness as the evaluation criterion. Through continuous iteration and collaboration among individuals, it dynamically adjusts the search strategy to seek the global optimum. In solving the above optimization model, the operating temperature of the liquid cooling system at each moment can be considered as a particle in the search space, with the fitness value corresponding to the model's objective function value. During iteration, the particles continuously update their velocity and position based on their individual optimal position and the swarm's optimal position, until the termination criterion is met. Since the optimization objective of the above optimization model is... Z The operating temperature of the liquid cooling system at any given moment, therefore the search space is... Z Dimension; Suppose the initial population has Y The nth particle, then the nth y The initial position of a particle can be represented as a Z dimensional vector :

[0069] In each iteration, each particle updates its velocity and position based on its individual optimal position and the global optimal position. The velocity and position update formulas are as follows:

[0070]

[0071] In the formula, , , and The first u +1、 u In the nth iteration i The first particle z Position and velocity values ​​in the dimension; and The first u Individual optimal and group optimal values ​​in the next iteration; ω c1 and c2 are inertial weights, representing the degree to which particles retain their historical update speeds; c1 and c2 are learning factors. r 1 and r 2 are two independent random numbers that follow a uniform distribution of [0 1].

[0072] S44, the overall solution process for this working temperature optimization model includes: Input S3 to obtain battery operating condition characteristic data at each time point in the future time period; Determine the PSO algorithm parameters, including population size, maximum number of iterations M (usually a fixed standard M is 500-10000; the more iterations, the better the optimization effect, but the longer the computation time. To balance computation time and optimization effect, a value of 500 can be used); initialize the number of iterations u=1; Calculate the initial particle fitness values ​​to determine the optimal values ​​for individuals and the population; Update the particle velocity and position, calculate the particle fitness value and update the individual and group optimal values. Repeat this step until the maximum number of iterations is reached, and output the optimal operating temperature of the liquid cooling system at each future time.

[0073] In the model solution, the operating temperature value of the liquid cooling system at each moment is regarded as a particle in the search space, and the fitness value corresponds to the objective function value of the operating temperature optimization model. Therefore, the initial particle is a set of liquid cooling system operating temperature values ​​at each moment in the future time period, randomly generated by the PSO algorithm. The fitness value of the initial particle is calculated by using the battery body temperature prediction model to output the predicted battery body temperature at each moment in the future time period based on the time series of battery operating condition characteristics at each moment in the future time period and the set of liquid cooling system operating temperature values ​​at each moment in the future time period, and then using the aging acceleration factor evaluation model to predict the aging acceleration factor of each individual battery at each moment in the future time period. Based on the set of liquid cooling system operating temperature values ​​at each moment in the future time period and the aging acceleration factor of each individual battery at each moment in the future time period, the working optimization model is used to determine whether it is optimal. If it is not optimal, the particle is updated and iterative solution is performed until the optimal value of the liquid cooling system operating temperature at each moment in the future time period is output. It should be noted that since the optimal operating temperature of the liquid cooling system at each future time point output at this time is optimized with the goal of optimizing the battery aging acceleration factor of the energy storage power station, the optimal operating temperature of the liquid cooling system at each future time point output by this scheme not only conforms to the battery operating conditions, but also corresponds to the optimal battery aging acceleration factor, thereby realizing the optimized management of battery aging in the energy storage power station.

[0074] To verify the effectiveness of the model and algorithm proposed in this invention, an energy storage unit from an energy storage power station in Southwest China is selected for analysis and illustration. The main parameter settings for the example are as follows: liquid cooling system parameters. , , and The values ​​are 15℃, 35℃, 5℃, and 5℃ respectively; upper and lower limits of battery cell operating temperature. and The values ​​are taken at 55℃ and 20℃ respectively; the activation energy parameter in the battery aging acceleration factor evaluation model. The value is 8519; in the BHT-ARIMA prediction algorithm ( p , d , q The order is (3, 1, 2). Assume the battery data sampling interval is 10 minutes, and the sampled value is the average value within that time period. Assume the liquid cooling system operates at a constant temperature every hour. The PSO solution algorithm uses a particle swarm size of 30 and a learning factor of... c 1 and c 2 takes the value 2, inertia weight ω The value is 0.8, and the maximum number of iterations is [value missing]. M The value is 100. For example... Figure 5 As shown, the MAPE index evaluation results for the SVR model corresponding to four alternative kernel functions are provided.

[0075] Depend on Figure 5 It can be seen that the four kernel functions exhibit certain differences in their performance on the MAPE index. Among them, the polynomial kernel function has the lowest MAPE value, at only 4.65%, indicating that it is superior to other kernel functions in terms of battery body temperature fitting ability, and therefore was selected as the optimal kernel function for constructing the SVR model.

[0076] To analyze the impact of battery temperature on its aging acceleration factor, based on historical operating data, a typical time-series temperature curve of a single battery cell in the energy storage unit was selected, and its aging acceleration factor at each time point was calculated. Specific results are as follows: Figure 6 As shown.

[0077] Depend on Figure 6 It can be seen that there is a significant positive correlation between the battery's internal temperature and the accelerated aging factor, with a PEN correlation coefficient greater than 0.95. This is because the accelerated aging factor is typically calculated based on the Arrhenius equation, which is exponentially sensitive to changes in battery temperature. Therefore, by controlling the battery's internal temperature, the rise of the aging factor can be effectively suppressed, thus delaying battery aging.

[0078] Based on this, to verify the effectiveness of the battery accelerated aging factor prediction model based on the BHT-ARIMA algorithm proposed in this invention, the SOC time-series data of a single battery cell within a given time period was selected as the analysis object for prediction. Furthermore, to enhance the comprehensiveness of the comparative analysis, prediction results based on the classic ARIMA and XGBoost models are also presented, and the widely used Normalized Root Mean Square Error (NRMSE) and Symmetric Mean Absolute Percentage Error (SMAPE) are used to measure specific performance. Figure 7 As shown.

[0079] Depend on Figure 7 The results show that the battery operating characteristic time-series prediction model based on BHT-ARIMA has a significant advantage in prediction accuracy. Its NRMSE and SMAPE indices are reduced by approximately 37.93% and 45.51% respectively compared to ARIMA, and by approximately 29.50% and 22.32% respectively compared to the XGBoost model. This result demonstrates that the method proposed in this invention can effectively predict the operating conditions of individual battery cells in the future, thus providing a reliable data foundation for accelerating the calculation of aging factors and effectively improving the credibility of the prediction results.

[0080] Finally, taking the daily operating data of the energy storage unit as an example, we analyze the effects of the traditional, extensive liquid cooling system operating temperature control strategy and the proposed control strategy on controlling battery aging. Specifically, as follows... Figure 8 As shown.

[0081] from Figure 8As can be seen, after adopting the liquid cooling system temperature control strategy proposed in this invention, the average accelerated aging factor of the battery is closer to 1.0, indicating that the battery is in a more reasonable operating state. In contrast, the accelerated aging factor under the traditional strategy exceeds 1.0, posing a certain risk of accelerated aging.

[0082] like Figure 9 As shown, this embodiment also provides a system for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment, which executes the above-described method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment; the system includes: The data processing unit is used to acquire the time series of battery operating condition characteristics at each time in the future period and the liquid cooling system operating temperature at each time in the future period obtained by each iteration of the optimization algorithm, and send them to each unit; it is also used to execute the optimization algorithm iterative solution; The battery body temperature prediction unit is used to output the predicted battery body temperature for each moment in the future period based on the time series of battery operating condition characteristics at each moment in the future period and the liquid cooling system operating temperature at each moment in the future period obtained by each iteration of the optimization algorithm, and send it to the battery aging acceleration factor evaluation unit. The battery aging acceleration factor evaluation unit is used to output the predicted values ​​of the battery aging acceleration factor at each time in the future period based on the predicted values ​​of the battery body temperature at each time in the future period, using a pre-built battery aging acceleration factor evaluation model, and send them to the operating temperature optimization unit. The operating temperature optimization unit is used to determine whether the operating temperature of the liquid cooling system at each time in the future period and the battery aging acceleration factor at each time in the future period are optimal based on the operating temperature of the liquid cooling system at each time in the future period obtained by each iteration of the optimization algorithm. If it is optimal, the optimal value of the operating temperature of the liquid cooling system at each time in the future period is output; otherwise, the feedback data processing unit executes the optimization algorithm iteratively to solve the problem.

[0083] Understandably, the above system can execute the above method completely with the same effect, so it will not be elaborated further here.

[0084] This embodiment also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the steps of the above-described method for optimizing the operating temperature of an energy storage battery compartment liquid cooling system.

[0085] This embodiment provides a method, system, and medium for optimizing the operating temperature of a liquid cooling system in an energy storage battery compartment. It deeply analyzes the dynamic mapping relationship between battery operating conditions, battery body temperature, battery aging acceleration factor, and the operating temperature of the liquid cooling system in the battery compartment. It constructs a battery body temperature prediction model, a battery aging acceleration factor evaluation model, and an operating temperature optimization model. Combined with an optimization algorithm, it predicts battery operating condition characteristics and body temperature change trends, and dynamically determines the operating temperature of the liquid cooling system in advance for various future time periods (e.g., 24 hours in the future) based on the goal of optimizing the battery aging acceleration factor. This enables rapid and accurate dynamic response of the liquid cooling system to different battery operating conditions, while also achieving precise control of battery aging. This provides an efficient technical solution for temperature management in energy storage power stations, significantly improving the economy and reliability of energy storage power stations.

[0086] The above descriptions are merely embodiments of the present invention. Commonly known structures and characteristics of the solutions are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent.

Claims

1. A method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment, characterized in that, include: A battery body temperature prediction model is constructed with battery operating condition characteristics and battery compartment liquid cooling system operating temperature as inputs and battery body temperature as output. Construct a battery aging acceleration factor assessment model that characterizes the mapping relationship between the bulk temperature of a battery cell and the aging acceleration factor; Construct an operating temperature optimization model with the operating temperature of the battery compartment liquid cooling system as the decision variable and the optimal battery aging acceleration factor as the objective; The time series of battery operating condition characteristics at each time point in the future period is obtained. The operating temperature of the liquid cooling system at each time point in the future period is obtained through each iteration of the optimization algorithm. The battery body temperature prediction model, battery aging acceleration factor evaluation model and operating temperature optimization model are constructed and iteratively solved by the optimization algorithm to output the optimal value of the operating temperature of the liquid cooling system at each time point in the future period.

2. The method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment according to claim 1, characterized in that, A battery body temperature prediction model is constructed based on the support vector regression machine algorithm, including: Each kernel function of the support vector regression machine model is used to construct the corresponding SVR regression model; The optimal parameters for each kernel function were determined using a five-fold cross-validation method. The SVR regression model corresponding to each kernel function was trained and predicted using historical battery operating data, and the model prediction error index was calculated. The historical battery operating data included the liquid cooling system operating temperature, battery operating condition characteristics, and battery body temperature at several acquisition times. A battery body temperature prediction model is constructed using the kernel function corresponding to the minimum model prediction error index and its SVR regression model.

3. The method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment according to claim 2, characterized in that, The battery body temperature prediction model is as follows: In the formula, the subscript q represents the q-th kernel function. The corresponding relevant parameters; and It is the optimal Lagrange multiplier; and These are the optimal regression parameters; For training set The samples included total.

4. The method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment according to claim 1, characterized in that, Based on the Arrhenius reaction rate theory, a battery aging acceleration factor evaluation model is constructed: In the formula, and The first n The battery in the first h Battery aging acceleration factors and battery body temperature over time; Reference temperature; The free gas constant; The activation energy parameter is determined by data fitting based on accelerated aging test data of the battery. .

5. The method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment according to claim 1, characterized in that, The process of building an operating temperature optimization model includes: Determine the objective function of the operating temperature optimization model, in the future time period. Z Within a certain time period N The goal is to optimize the average aging acceleration factor of each battery. The constraints of the operating temperature optimization model are determined, including at least one of the following: operating temperature limit constraint of the battery compartment liquid cooling system, operating temperature regulation parameter constraint of the battery compartment liquid cooling system, and battery body temperature constraint.

6. The method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment according to claim 1, characterized in that, Based on the Block Hankel Tensor Autoregressive Integrated Moving Average algorithm, the time series data of battery operating condition characteristics are represented in tensor form, and a prediction equation for the battery operating condition characteristic sequence is constructed to generate the time series of battery operating condition characteristics for future periods.

7. The method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment according to claim 6, characterized in that, A one-way delay embedding transformation is performed on the battery operating condition feature time series along the time direction to convert the T time series data of the battery operating condition feature into a battery operating condition feature tensor with time series variation. The battery operating condition characteristic tensor is decomposed using tensor decomposition and Tucker decomposition, and combined with an autoregressive integrated moving average model to construct a predictive equation for the battery operating condition characteristic sequence. The core tensor of the battery operating condition characteristics at time T+1 is predicted using the prediction equation. Then, by combining tensor decomposition and one-way delay embedding transformation, the predicted value of the battery operating condition characteristics at time T+1 is obtained.

8. The method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment according to claim 1, characterized in that, The particle swarm optimization algorithm is used for optimization. The working temperature value of the liquid cooling system of the battery compartment at each time in the future period is regarded as a particle in the search space, and the fitness value corresponds to the objective function value of the working temperature optimization model. The time series of battery operating condition characteristics at each time in the future period and the liquid cooling system operating temperature at each time in the future period obtained by each iteration of the optimization algorithm are input into the battery body temperature prediction model, and the battery body temperature value at each time in the future period is output. Input the battery body temperature value at each time in the future into the battery aging acceleration factor evaluation model, and output the battery aging acceleration factor at each time in the future. The operating temperatures of the liquid cooling system at each moment in the future time period and the battery aging acceleration factor at each moment in the future time period, obtained through each iteration of the optimization algorithm, are input into the operating temperature optimization model to confirm whether it is optimal. The optimal operating temperature of the liquid cooling system at each time point in the future is output through iterative solution using the particle swarm optimization algorithm.

9. A system for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment, characterized in that, A method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment according to any one of claims 1-8; the system includes: The data processing unit is used to acquire the time series of battery operating condition characteristics at each time in the future period and the liquid cooling system operating temperature at each time in the future period obtained by each iteration of the optimization algorithm, and send them to each unit; it is also used to execute the optimization algorithm iterative solution; The battery body temperature prediction unit is used to output the predicted battery body temperature for each moment in the future period based on the time series of battery operating condition characteristics at each moment in the future period and the liquid cooling system operating temperature at each moment in the future period obtained by each iteration of the optimization algorithm, and send it to the battery aging acceleration factor evaluation unit. The battery aging acceleration factor evaluation unit is used to output the predicted values ​​of the battery aging acceleration factor at each time in the future period based on the predicted values ​​of the battery body temperature at each time in the future period, using a pre-built battery aging acceleration factor evaluation model, and send them to the operating temperature optimization unit. The operating temperature optimization unit is used to determine whether the operating temperature of the liquid cooling system at each time in the future period and the battery aging acceleration factor at each time in the future period are optimal based on the operating temperature of the liquid cooling system at each time in the future period obtained by each iteration of the optimization algorithm. If it is optimal, the optimal value of the operating temperature of the liquid cooling system at each time in the future period is output; otherwise, the feedback data processing unit executes the optimization algorithm iteratively to solve the problem.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of the method for optimizing the operating temperature of a liquid cooling system for an energy storage battery compartment as described in any one of claims 1-8.

Images