Flow management and control data determination method and device based on electric power project

Abstract

The invention relates to a flow management and control data determination method and device based on an electric power project. The method comprises the following steps: acquiring storage battery real-time electrical data corresponding to an electric power project, each preset charge state estimation sub-model, storage battery real-time scene data and an initial charge state estimation model; deploying each preset charge state estimation sub-model at each neuron node of the model according to the real-time scene data of the storage battery; inputting real-time electric data of the storage battery into the deployed model to obtain real-time estimation data of the state of charge; adjusting the deployed model parameters by adopting a random state-of-charge algorithm and state-of-charge real-time estimation data to obtain an optimized state-of-charge estimation model; inputting real-time electric data of the storage battery into the optimized model to obtain charge state optimization estimation data; and fusing the charge state real-time estimation data and the charge state optimization estimation data to obtain the target charge state estimation data, so that the operation and maintenance efficiency of the electric power project can be improved.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology, and in particular to a method and apparatus for determining process control data based on power projects. Background Technology

[0002] In traditional technologies, the process management of battery-powered power projects mainly relies on real-time monitoring of basic battery parameters, such as voltage, current, and temperature. Data is collected through sensors, and judgments and controls are made based on preset thresholds. For example, when the battery voltage is lower than a set value, the system will automatically start the charging process or issue a warning. However, traditional technologies rely on simple threshold judgments and rule-based control, lacking accurate prediction and real-time optimization of the power project's status, resulting in low efficiency in the operation and maintenance of power projects. Summary of the Invention

[0003] Therefore, it is necessary to provide a method, device, computer equipment, computer-readable storage medium, and computer program product for determining process control data of power projects that can effectively improve the operation and maintenance efficiency of power projects, addressing the aforementioned technical problems.

[0004] Firstly, this application provides a method for determining process control data based on power projects, including:

[0005] The system acquires real-time electrical data of the battery corresponding to the power project, various preset state-of-charge estimation sub-models, real-time battery scenario data, and an initial state-of-charge estimation model; the initial state-of-charge estimation model is constructed using the biological neural network corresponding to the power project.

[0006] Based on the real-time scenario data of the battery, each of the preset state of charge estimation sub-models is deployed to each neuron node of the initial state of charge estimation model to obtain the deployed state of charge estimation model.

[0007] The real-time electrical data of the battery is input into the deployed state of charge estimation model to obtain real-time state of charge estimation data.

[0008] The model parameters of the deployed state of charge estimation model are adjusted using the stochastic state of charge algorithm and the real-time state of charge estimation data to obtain an optimized state of charge estimation model.

[0009] The real-time electrical data of the battery is input into the optimized state of charge estimation model to obtain optimized state of charge estimation data.

[0010] By integrating the real-time state of charge estimation data and the optimized state of charge estimation data, the target state of charge estimation data corresponding to the power project is obtained; the target state of charge estimation data is used to generate the process control data of the power project.

[0011] Secondly, this application also provides a device for determining process control data based on power projects, comprising:

[0012] The data acquisition module is used to acquire real-time electrical data of the battery corresponding to the power project, various preset state of charge estimation sub-models, real-time scenario data of the battery, and the initial state of charge estimation model; the initial state of charge estimation model is constructed through the biological neural network corresponding to the power project.

[0013] The model deployment module is used to deploy each of the preset state of charge estimation sub-models to each neuron node of the initial state of charge estimation model based on the real-time scene data of the battery, so as to obtain the deployed state of charge estimation model.

[0014] The data estimation module is used to input the real-time electrical data of the battery into the deployed state of charge estimation model to obtain real-time state of charge estimation data;

[0015] The model optimization module is used to adjust the model parameters of the deployed state of charge estimation model using a stochastic state of charge algorithm and the real-time state of charge estimation data, so as to obtain an optimized state of charge estimation model.

[0016] The data optimization module is used to input the real-time electrical data of the battery into the optimized state of charge estimation model to obtain optimized state of charge estimation data;

[0017] The data fusion module is used to fuse the real-time state of charge estimation data and the optimized state of charge estimation data to obtain the target state of charge estimation data corresponding to the power project; the target state of charge estimation data is used to generate the process control data of the power project.

[0018] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0019] The system acquires real-time electrical data of the battery corresponding to the power project, various preset state-of-charge estimation sub-models, real-time battery scenario data, and an initial state-of-charge estimation model; the initial state-of-charge estimation model is constructed using the biological neural network corresponding to the power project.

[0020] Based on the real-time scenario data of the battery, each of the preset state of charge estimation sub-models is deployed to each neuron node of the initial state of charge estimation model to obtain the deployed state of charge estimation model.

[0021] The real-time electrical data of the battery is input into the deployed state of charge estimation model to obtain real-time state of charge estimation data.

[0022] The model parameters of the deployed state of charge estimation model are adjusted using the stochastic state of charge algorithm and the real-time state of charge estimation data to obtain an optimized state of charge estimation model.

[0023] The real-time electrical data of the battery is input into the optimized state of charge estimation model to obtain optimized state of charge estimation data.

[0024] By integrating the real-time state of charge estimation data and the optimized state of charge estimation data, the target state of charge estimation data corresponding to the power project is obtained; the target state of charge estimation data is used to generate the process control data of the power project.

[0025] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0026] The system acquires real-time electrical data of the battery corresponding to the power project, various preset state-of-charge estimation sub-models, real-time battery scenario data, and an initial state-of-charge estimation model; the initial state-of-charge estimation model is constructed using the biological neural network corresponding to the power project.

[0027] Based on the real-time scenario data of the battery, each of the preset state of charge estimation sub-models is deployed to each neuron node of the initial state of charge estimation model to obtain the deployed state of charge estimation model.

[0028] The real-time electrical data of the battery is input into the deployed state of charge estimation model to obtain real-time state of charge estimation data.

[0029] The model parameters of the deployed state of charge estimation model are adjusted using the stochastic state of charge algorithm and the real-time state of charge estimation data to obtain an optimized state of charge estimation model.

[0030] The real-time electrical data of the battery is input into the optimized state of charge estimation model to obtain optimized state of charge estimation data.

[0031] By integrating the real-time state of charge estimation data and the optimized state of charge estimation data, the target state of charge estimation data corresponding to the power project is obtained; the target state of charge estimation data is used to generate the process control data of the power project.

[0032] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0033] The system acquires real-time electrical data of the battery corresponding to the power project, various preset state-of-charge estimation sub-models, real-time battery scenario data, and an initial state-of-charge estimation model; the initial state-of-charge estimation model is constructed using the biological neural network corresponding to the power project.

[0034] Based on the real-time scenario data of the battery, each of the preset state of charge estimation sub-models is deployed to each neuron node of the initial state of charge estimation model to obtain the deployed state of charge estimation model.

[0035] The real-time electrical data of the battery is input into the deployed state of charge estimation model to obtain real-time state of charge estimation data.

[0036] The model parameters of the deployed state of charge estimation model are adjusted using the stochastic state of charge algorithm and the real-time state of charge estimation data to obtain an optimized state of charge estimation model.

[0037] The real-time electrical data of the battery is input into the optimized state of charge estimation model to obtain optimized state of charge estimation data.

[0038] By integrating the real-time state of charge estimation data and the optimized state of charge estimation data, the target state of charge estimation data corresponding to the power project is obtained; the target state of charge estimation data is used to generate the process control data of the power project.

[0039] The aforementioned method, apparatus, computer equipment, storage medium, and computer program product for determining process control data for power projects acquire real-time electrical data of batteries corresponding to the power project, various preset state-of-charge (POC) estimation sub-models, real-time battery scenario data, and an initial POC estimation model. The initial POC estimation model is constructed using a biological neural network corresponding to the power project. Based on the real-time battery scenario data, each preset POC estimation sub-model is deployed at each neuron node of the initial POC estimation model to obtain a deployed POC estimation model. Real-time battery electrical data is input into the deployed POC estimation model to obtain real-time POC estimation data. The model parameters of the deployed POC estimation model are adjusted using a stochastic POC algorithm and the real-time POC estimation data to obtain an optimized POC estimation model. Real-time battery electrical data is input into the optimized POC estimation model to obtain optimized POC estimation data. The real-time POC estimation data and the optimized POC estimation data are fused to obtain target POC estimation data corresponding to the power project. The target POC estimation data is used to generate process control data for the power project.

[0040] An initial state-of-charge (POC) estimation model is constructed using a biological neural network corresponding to the power project, ensuring the model can adapt to the characteristics of different batteries. Then, based on real-time scenario data from the batteries, various pre-defined POC estimation sub-models are deployed to the neuron nodes of the initial model, forming a deployed POC estimation model. This deployment process allows the model to estimate the POC from multiple angles and levels. Next, real-time electrical data is input into the deployed model to obtain preliminary real-time estimation data. The model parameters are further adjusted and optimized using a stochastic POC algorithm to improve the accuracy and stability of the estimation. Finally, by fusing the real-time and optimized estimation data, more accurate target POC estimation data is obtained, and reasonable process control data for the power project is further established. This approach enables real-time and high-precision estimation of battery POC in complex and ever-changing real-world usage scenarios, significantly improving the operation and maintenance efficiency of power projects, extending battery life, and optimizing energy utilization efficiency. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is an application environment diagram of a method for determining process control data based on power projects in one embodiment.

[0043] Figure 2 This is a flowchart illustrating a method for determining process control data based on a power project in one embodiment;

[0044] Figure 3 This is a flowchart illustrating a method for deploying a state of charge estimation model in one embodiment.

[0045] Figure 4 This is a flowchart illustrating the method for deploying a state of charge estimation model in another embodiment;

[0046] Figure 5 This is a flowchart illustrating a method for obtaining real-time state of charge estimation data in one embodiment;

[0047] Figure 6 This is a flowchart illustrating a method for obtaining a synaptic state of charge estimation model in one embodiment;

[0048] Figure 7 This is a flowchart illustrating the method for optimizing the state of charge estimation model in one embodiment;

[0049] Figure 8 This is a flowchart illustrating the method for optimizing the state of charge estimation model in another embodiment;

[0050] Figure 9 This is a structural block diagram of a process control data determination device for a power project in one embodiment;

[0051] Figure 10 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0053] This application provides a method for determining process control data based on power projects, which can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be located in the cloud or on another network server. Server 104 acquires real-time electrical data of the battery corresponding to the power project, various preset state of charge (SCC) estimation sub-models, real-time battery scene data, and an initial SCC estimation model through terminal 102. The initial SCC estimation model is constructed using a biological neural network corresponding to the power project. Based on the real-time battery scene data, each preset SCC estimation sub-model is deployed at each neuron node of the initial SCC estimation model to obtain a deployed SCC estimation model. The real-time electrical data of the battery is input into the deployed SCC estimation model to obtain real-time SCC estimation data. The model parameters of the deployed SCC estimation model are adjusted using a stochastic SCC algorithm and the real-time SCC estimation data to obtain an optimized SCC estimation model. The real-time electrical data of the battery is input into the optimized SCC estimation model to obtain optimized SCC estimation data. The real-time SCC estimation data and the optimized SCC estimation data are fused to obtain the target SCC estimation data corresponding to the power project. The target SCC estimation data is used to generate process control data for the power project. The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle systems. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. The server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0054] In one exemplary embodiment, such as Figure 2 As shown, a method for determining process control data based on power projects is provided, which can be applied to... Figure 1 Taking the server in the example, the explanation includes the following steps 202 to 212. Wherein:

[0055] Step 202: Obtain the real-time electrical data of the battery corresponding to the power project, each preset state of charge estimation sub-model, real-time scenario data of the battery, and the initial state of charge estimation model.

[0056] Among them, real-time electrical data of the battery can be collected and recorded in real time by monitoring equipment to collect parameters such as battery voltage, current, temperature and charging status, so as to understand the battery's operating status and performance in real time and ensure its reliability and safety under various working conditions.

[0057] The preset state-of-charge (SOC) estimation sub-model can be used to predict and estimate the SOC of a battery at a specific point in time. This model, by inputting real-time parameters such as voltage, current, and temperature, combined with preset algorithms and correction coefficients, quickly calculates the current charge level of the battery, helping to optimize charging and discharging strategies and the performance of the battery management system. Generally, the preset SOC estimation sub-model can be based on the Coulomb counting method, open circuit voltage method, internal resistance method, model-based methods, or data-driven methods.

[0058] Among these, real-time battery scenario data can reflect the battery's actual usage conditions. For example, specific scenario parameters of the battery under different usage scenarios (such as vehicle operation, backup power use, etc.).

[0059] The initial state of charge estimation model can be constructed by simulating a biological neural network, and the state of charge (SOC) of the power project can be predicted through this model.

[0060] Specifically, server 104 acquires real-time electrical data of the power project, such as voltage, current, and temperature, through various terminals 102 deployed around the battery. This data reflects the battery's operating status in real time. Secondly, it acquires various preset state-of-charge estimation sub-models. These models are pre-established based on the battery's performance characteristics under different states of charge, through experimental or theoretical methods. Next, it collects real-time scenario data of the battery, which is used to adjust the estimation model to more accurately reflect the specific conditions of the actual use scenario. Finally, it constructs an initial state-of-charge estimation model. This model utilizes the biological neural network corresponding to the power project. Through the self-learning and adaptive capabilities of the neural network, the model's accuracy and robustness are enhanced, ensuring that the model can adapt to the characteristics and usage environment of different batteries.

[0061] Step 204: Based on the real-time scenario data of the battery, deploy each preset state of charge estimation sub-model on each neuron node of the initial state of charge estimation model to obtain the deployed state of charge estimation model.

[0062] Deploying the state of charge (SOC) estimation model can be achieved in a real-world application environment by integrating and implementing the necessary pre-defined SOC estimation sub-models into the various neuron nodes of the initial SOC estimation model. This process involves configuring, adjusting, and validating the model to ensure it can accurately and in real-time predict the SOC of the battery under actual operating conditions.

[0063] Specifically, the current usage scenario is determined based on real-time battery scenario data. Then, each preset state of charge estimation sub-model is mapped to the neuron nodes of the initial state of charge estimation model. The mapping method can be a one-to-one mapping between the preset state of charge estimation sub-model and the neuron node, multiple preset state of charge estimation sub-models can be mapped to the same neuron node simultaneously, or one preset state of charge estimation sub-model can be mapped to multiple neuron nodes simultaneously. This process requires embedding the parameters, weights, and feature information of each sub-model into the neural network structure of the initial model, so that each neuron node can independently run the corresponding sub-model. By adjusting the connection weights and activation functions of the neural network, the entire network can work collaboratively, thereby forming a unified deployment state of charge estimation model that can dynamically adapt to the real-time environment.

[0064] Step 206: Input the real-time electrical data of the battery into the deployed state of charge estimation model to obtain real-time state of charge estimation data.

[0065] The real-time state of charge (SOC) estimation data can be the result of the SOC estimation model being deployed to estimate the SOC based on the real-time electrical data of the battery. However, this result has not been corrected and may contain anomalies.

[0066] Specifically, the preprocessed real-time electrical data of the battery is input into a deployed state-of-charge (POC) estimation model, which integrates the parameters and structures of multiple preset POC sub-models. The biological neural network in the POC model begins to process the input real-time electrical data of the battery layer by layer. The neurons perform calculations according to preset weights and activation functions, gradually extracting and analyzing features in the electrical data. Each neuron node corresponds to one or more preset POC sub-models, which obtain corresponding processed data during the processing. By integrating the calculation results of different preset POC sub-models, the biological neural network continuously adjusts its internal connections and weights, ultimately generating a comprehensive output as the real-time POC estimation data.

[0067] Step 208: Using the stochastic state of charge algorithm and real-time state of charge estimation data, the model parameters of the deployed state of charge estimation model are adjusted to obtain an optimized state of charge estimation model.

[0068] Among them, the stochastic state of charge (SOC) algorithm can be an algorithm based on stochastic processes and probabilistic statistical methods, used to estimate the SOC of a battery after deploying the model parameters of the SOC estimation model. By introducing random variables and distribution models, this algorithm considers the uncertainty and randomness of the battery in actual operation when deploying the SOC estimation model, thus providing more flexible and robust SOC estimation results. This method is suitable for handling battery management needs under complex and variable operating conditions.

[0069] Specifically, using real-time state of charge (SOC) estimation data as a benchmark, the stochastic SOC algorithm is applied to optimize the model parameters of the deployed SOC estimation model. The stochastic SOC algorithm generates a series of random SOC samples and inputs these samples into the deployed SOC estimation model for simulation calculation. Based on the simulation results, the error is calculated by comparing them with the real-time estimation data. The model parameters and weights are adjusted based on error backpropagation, and the model is continuously iterated and optimized. Through multiple iterations and adjustments, the error is gradually reduced, and the accuracy and stability of the model are improved. Finally, an optimized SOC estimation model is obtained.

[0070] In one specific embodiment, when the number of batteries is very large, resulting in a massive amount of data to be processed later, traditional computers encounter certain computational bottlenecks in data processing. Therefore, quantum computing methods are considered, fixing the random state of charge algorithm to quantum Monte Carlo and quantum Fourier transform. The quantum Monte Carlo algorithm is used to generate multiple random state of charge samples, representing the battery's state of charge under different conditions. These samples are then input into a deployed state of charge estimation model for simulation calculations, obtaining preliminary estimation results. Next, quantum Fourier transform is applied to perform frequency domain analysis on these preliminary estimation results to extract key frequency feature information. This feature information is compared with actual real-time state of charge estimation data to identify computational errors. Using an error backpropagation mechanism, the model's parameters and weights are adjusted, and through multiple iterations of optimization, the error is gradually reduced. Finally, an optimized state of charge estimation model is obtained.

[0071] Step 210: Input the real-time electrical data of the battery into the optimized state of charge estimation model to obtain optimized state of charge estimation data.

[0072] Among them, the optimized state of charge estimation data can be the estimation results obtained by using the optimized model to estimate the state of charge.

[0073] Specifically, similar to step 206, the preprocessed real-time electrical data of the battery is input into the optimized state of charge estimation model, which integrates the parameters and structures of multiple preset state of charge estimation sub-models. The biological neural network in the optimized state of charge estimation model begins to process the input real-time electrical data of the battery layer by layer. The neuron nodes perform calculations according to preset weights and activation functions, and gradually extract and analyze the features in the electrical data. Each neuron node corresponds to one or more preset state of charge estimation sub-models, which obtain corresponding processed data during the processing. By integrating the calculation results of different preset state of charge estimation sub-models, the biological neural network continuously adjusts its internal connections and weights, and finally generates a comprehensive output as the optimized state of charge estimation data.

[0074] Step 212: Integrate the real-time state of charge estimation data and the optimized state of charge estimation data to obtain the target state of charge estimation data for the power project.

[0075] The target state of charge estimation data can be obtained by integrating different data and eliminating accidental cases.

[0076] Specifically, the real-time state of charge (SOC) estimation data and the optimized SOC estimation data are aligned and preprocessed to ensure consistency in data format and time series. The dynamic response characteristics of the real-time SOC estimation data and the accuracy characteristics of the optimized SOC estimation data are analyzed to determine the importance of each data set at different time points. The weight of each data set is calculated, and the two sets of data are weighted and averaged or filtered using a filtering algorithm (such as Kalman filtering) based on these weights. This process combines the rapid response capability of real-time data with the high accuracy of optimized data to obtain more accurate, smooth, and stable target SOC estimation data.

[0077] Furthermore, based on the target state of charge (SBC) estimation data, safety control objectives and constraints are established for each stage (e.g., charging, discharging, energy management) within the safety regulations of the power project. Then, the target SBC estimation data, safety control objectives, and safety constraints are combined to generate specific process control data, such as charging / discharging power settings, load adjustments, and temperature control strategies. To ensure the stability and efficiency of the project during operation, this control data needs to be dynamically adjusted periodically based on real-time data to adapt to different needs arising from changes in battery status and the environment.

[0078] In the aforementioned method for determining process control data for power projects, the following steps are taken: real-time electrical data of the battery corresponding to the power project, various preset state of charge (SOC) estimation sub-models, real-time battery scenario data, and an initial SOC estimation model are acquired. The initial SOC estimation model is constructed using a biological neural network corresponding to the power project. Based on the real-time battery scenario data, each preset SOC estimation sub-model is deployed at each neuron node of the initial SOC estimation model to obtain a deployed SOC estimation model. The real-time electrical data of the battery is input into the deployed SOC estimation model to obtain real-time SOC estimation data. The model parameters of the deployed SOC estimation model are adjusted using a stochastic SOC algorithm and the real-time SOC estimation data to obtain an optimized SOC estimation model. The real-time electrical data of the battery is input into the optimized SOC estimation model to obtain optimized SOC estimation data. Finally, the real-time SOC estimation data and the optimized SOC estimation data are fused to obtain the target SOC estimation data corresponding to the power project.

[0079] An initial state-of-charge (POC) estimation model is constructed using a biological neural network corresponding to the power project, ensuring the model can adapt to the characteristics of different batteries. Then, based on real-time scenario data from the batteries, various pre-defined POC estimation sub-models are deployed to the neuron nodes of the initial model, forming a deployed POC estimation model. This deployment process allows the model to estimate the POC from multiple angles and levels. Next, real-time electrical data is input into the deployed model to obtain preliminary real-time estimation data. The model parameters are further adjusted and optimized using a stochastic POC algorithm to improve the accuracy and stability of the estimation. Finally, by fusing the real-time and optimized estimation data, more accurate target POC estimation data is obtained, and reasonable process control data for the power project is further established. This approach enables real-time and high-precision estimation of battery POC in complex and ever-changing real-world usage scenarios, significantly improving the operation and maintenance efficiency of power projects, extending battery life, and optimizing energy utilization efficiency.

[0080] In one exemplary embodiment, such as Figure 3 As shown, based on real-time battery scenario data, each preset state of charge estimation sub-model is deployed at each neuron node of the initial state of charge estimation model to obtain the deployed state of charge estimation model, including steps 302 to 306. Wherein:

[0081] Step 302: Obtain the historical electrical data of the battery corresponding to the power project.

[0082] Historical electrical data of the battery can be electrical parameter data recorded during the battery's past operation, including voltage, current, temperature, and state of charge (SOC). This data is used to analyze battery performance trends, assess state of health (SOH), predict future behavior, and optimize battery management strategies.

[0083] Specifically, historical electrical data of the power project's batteries is retrieved from the battery management system or database. This historical electrical data typically includes parameters such as voltage, current, and temperature from historical records. The retrieved historical electrical data is then cleaned and preprocessed to remove outliers and noise, ensuring the integrity and consistency of the data.

[0084] Step 304: Based on the real-time scenario data of the battery, input the real-time electrical data and historical electrical data of the battery into each preset state of charge estimation sub-model to obtain the applicable analysis data and the limitation analysis data of each sub-model.

[0085] The sub-model applicability analysis data can be used to evaluate the applicability and performance of a specific sub-model in a particular application scenario. This data includes experimental results, test parameters, and operating conditions. By analyzing this data, the accuracy and reliability of the sub-model in a specific scenario can be determined, thereby confirming its suitability for practical applications and guiding model optimization and adjustment.

[0086] Sub-model limitation analysis data can be used to evaluate and identify the limitations and shortcomings of a specific sub-model in a particular application scenario. This data includes the model's performance under different conditions, error analysis, boundary condition testing, etc. By analyzing this data, we can understand the model's weaknesses and limitations, thus providing a basis for model improvement and optimization, and ensuring its reliability and effectiveness in practical applications.

[0087] Specifically, based on the real-time scenario data of the battery, the real-time electrical data and historical electrical data of the battery are input into each preset state of charge (SOC) estimation sub-model. Each preset SOC estimation sub-model performs calculations based on the input data to generate corresponding SOC estimation results. These results are analyzed to obtain the initial applicability data and initial limitation data of each preset SOC estimation sub-model in the current scenario. By comparing the output error and response characteristics of each sub-model, its performance under different conditions is evaluated. Multimodal mutual self-optimization is performed on the initial applicability data and initial limitation data of each preset SOC estimation sub-model in the current scenario to obtain the corresponding sub-model applicability analysis data and sub-model limitation analysis data of each preset SOC estimation sub-model.

[0088] Step 306: Based on the applicable analysis data and the limiting analysis data of each sub-model, determine the node deployment mapping relationship between each neuron node and at least one preset state of charge estimation sub-model.

[0089] The node deployment mapping relationship can be the correspondence between neuron nodes and preset state of charge estimation sub-models. One neuron node can establish a mapping relationship with one preset state of charge estimation sub-model, or one neuron node can establish a mapping relationship with multiple preset state of charge estimation sub-models.

[0090] Specifically, based on the applicable analysis data and the limitation analysis data of each sub-model, the performance and limitations of each preset state of charge estimation sub-model are evaluated in different scenarios. Based on the evaluation results, the optimal mapping relationship between each neuron node and at least one preset state of charge estimation sub-model is determined. This includes assigning the at least one preset state of charge estimation sub-model with the best performance to the at least one neuron node that best matches it, so as to ensure the estimation accuracy and reliability in various use scenarios.

[0091] Step 308: Based on the deployment mapping relationship of each node, each preset state of charge estimation sub-model is deployed to each neuron node to obtain the deployed state of charge estimation model.

[0092] Specifically, based on the node deployment mapping relationship, each preset state of charge estimation sub-model is assigned to the corresponding neuron node. At the same time, the model parameters and structure of the preset state of charge estimation sub-model matched by each neuron node are embedded into the neuron node, so that each neuron node can run the assigned preset state of charge estimation sub-model. The connection weights and activation functions of the state of charge estimation model are adjusted so that the entire state of charge estimation model can work together to form a complete deployed state of charge estimation model.

[0093] In this embodiment, historical electrical data from the power project is acquired and combined with real-time scenario data. The real-time and historical electrical data are then input into preset state-of-charge (SOC) estimation sub-models to obtain applicability and limitation analysis data for each sub-model. Based on this analysis data, the optimal mapping relationship between neuron nodes and corresponding sub-models is determined. Then, each sub-model is deployed to its corresponding neuron node, forming an optimized SOC estimation model. This process considers not only the dynamism of real-time data but also the comprehensiveness of historical data, enabling the model to more accurately estimate the battery's SOC under different usage scenarios. This improves the model's accuracy and applicability, thereby effectively enhancing the reliability and performance of the battery management system.

[0094] In one exemplary embodiment, such as Figure 4 As shown, according to the deployment mapping relationship of each node, each preset state of charge estimation sub-model is deployed to each neuron node to obtain the deployed state of charge estimation model, including steps 402 to 414.

[0095] in:

[0096] Step 402: Based on the real-time scenario data of the battery and the real-time environmental data corresponding to the power project, adjust the deployment mapping relationship of each node and each neuron node to obtain the deployment mapping relationship of each first adjustment node and each first adjustment neuron node.

[0097] Among them, adjusting the node deployment mapping relationship can be done by adjusting the correspondence between the neuron nodes and the preset state of charge estimation sub-model in the node deployment mapping relationship to obtain a new mapping relationship.

[0098] Among them, adjusting the neuron node can be a new neuron node obtained by adjusting the parameters of the original neuron node.

[0099] Among them, real-time environmental data can be the real-time changing environmental parameters of the battery under the current conditions. The difference between real-time scenario data and real-time scenario data of the battery is that real-time scenario data of the battery represents the usage scenario in which the battery is located, while real-time environmental data represents the environmental parameters of the battery under that scenario.

[0100] Specifically, real-time environmental data corresponding to the power project is acquired through sensors, and these data are analyzed based on the real-time scenario data and real-time environmental data of the battery to understand the usage conditions and environmental changes in the current scenario. Based on the usage conditions and environmental changes in the current scenario, the previously determined node deployment mapping relationship is evaluated and adjusted to ensure that the deployment model of each neuron node can adapt to the new environmental conditions. At the same time, the parameters and connection weights of the neuron nodes are adjusted so that they can still accurately estimate the state of charge under the new conditions, forming the first adjusted node deployment mapping relationship and the first adjusted neuron node.

[0101] Step 404: Based on the deployment mapping relationship of each first adjustment node, each preset state of charge estimation sub-model is deployed to each first adjustment neuron node to obtain the first intermediate state of charge estimation model.

[0102] Among them, the intermediate charge state estimation model can be a new model obtained by adjusting the node deployment mapping relationship and neuron nodes.

[0103] Specifically, similar to step 308, according to the deployment mapping relationship of the first adjustment node, each preset state of charge estimation sub-model is assigned to the corresponding first adjustment neuron node. At the same time, the model parameters and structure of the preset state of charge estimation sub-model matched by each first adjustment neuron node are embedded into the first adjustment neuron node, so that each first adjustment neuron node can run the assigned preset state of charge estimation sub-model. The connection weights and activation functions of the state of charge estimation model are adjusted so that the entire state of charge estimation model can work together to form a complete first intermediate state of charge estimation model.

[0104] Step 406: Based on the real-time scenario data of the battery and the real-time demand data corresponding to the power project, adjust the deployment mapping relationship of each node and each neuron node to obtain the deployment mapping relationship of each second adjustment node and each second adjustment neuron node.

[0105] Among them, real-time demand data can be the demand data of power projects that change in real time due to changes in external conditions in the current scenario.

[0106] Specifically, the system acquires real-time demand data corresponding to power projects through sensors, and analyzes this data based on real-time scenario data and real-time demand data of batteries to understand the usage conditions and specific needs under the current scenario. Based on the usage conditions and specific needs under the previous scenario, the system evaluates and adjusts the previous node deployment mapping relationship and neuron nodes to ensure that the model can meet the new needs and usage environment. The system adjusts the parameters and connection weights of each neuron node to enable it to perform optimal calculations under the new mapping relationship, forming a second adjusted node deployment mapping relationship and a second adjusted neuron node.

[0107] Step 408: Based on the deployment mapping relationship of each second adjustment node, each preset state of charge estimation sub-model is deployed to each second adjustment neuron node to obtain the second intermediate state of charge estimation model.

[0108] Specifically, similar to step 308, according to the second adjustment node deployment mapping relationship, each preset state of charge estimation sub-model is assigned to the corresponding second adjustment neuron node. At the same time, the model parameters and structure of the preset state of charge estimation sub-model matched by each second adjustment neuron node are embedded into the second adjustment neuron node, so that each second adjustment neuron node can run the assigned preset state of charge estimation sub-model. The connection weights and activation functions of the state of charge estimation model are adjusted so that the entire state of charge estimation model can work together to form a complete second intermediate state of charge estimation model.

[0109] Step 410: Based on the deployment mapping relationship of each node, each preset state of charge estimation sub-model is deployed to each neuron node to obtain the third intermediate state of charge estimation model.

[0110] Specifically, similar to step 308, based on the node deployment mapping relationship, each preset state of charge estimation sub-model is assigned to the corresponding neuron node. At the same time, the model parameters and structure of the preset state of charge estimation sub-model matched to each neuron node are embedded into the neuron node, so that each neuron node can run the assigned preset state of charge estimation sub-model. The connection weights and activation functions of the state of charge estimation model are adjusted so that the entire state of charge estimation model can work together to form a complete third intermediate state of charge estimation model.

[0111] Step 412: Based on the first intermediate state of charge estimation model, the second intermediate state of charge estimation model, and the third intermediate state of charge estimation model, determine the deployment mapping relationship of each target node and each target neuron node.

[0112] Among them, the target node deployment mapping relationship and the target neuron node are the optimized mapping relationship and neuron node, respectively.

[0113] Specifically, real-time electrical data of the battery is input into the first, second, and third intermediate state of charge (SOC) estimation models, respectively, to obtain the output data of the three different intermediate SOC estimation models. Based on the output data of the three different intermediate SOC estimation models, a stacking ensemble method is used to fuse the three intermediate SOC estimation models. First, the output of each intermediate SOC estimation model is used as the input feature to train a meta-model (such as linear regression, decision tree, etc.). This meta-model is used to learn how to optimally combine the outputs of each intermediate model. Then, through the calculation of the meta-model, the output data of each intermediate model is integrated and fed back to the node deployment mapping relationship and the original neuron nodes. The parameters of the node deployment mapping relationship and the original neuron nodes are adjusted to obtain the target node deployment mapping relationship and the target neuron nodes.

[0114] Step 414: Based on the deployment mapping relationship of each target node, each preset state of charge estimation sub-model is deployed to each target neuron node to obtain the deployed state of charge estimation model.

[0115] Specifically, similar to step 308, based on the target node deployment mapping relationship, each preset state of charge estimation sub-model is assigned to the corresponding target neuron node. At the same time, the model parameters and structure of the preset state of charge estimation sub-model matched to each target neuron node are embedded into the target neuron node, so that each target neuron node can run the assigned preset state of charge estimation sub-model. The connection weights and activation functions of the state of charge estimation model are adjusted so that the entire state of charge estimation model can work together to form a complete deployment state of charge estimation model.

[0116] In this embodiment, by utilizing real-time scene data and real-time environmental data of the battery, the node deployment mapping relationship and neuron nodes are dynamically adjusted to obtain a series of optimized intermediate state of charge estimation models. Based on real-time demand data, these mapping relationships and node deployments are further adjusted to generate more accurate first, second, and third intermediate state of charge estimation models. Finally, by comprehensively analyzing and fusing these intermediate models, the optimal target node deployment mapping relationship and target neuron nodes are determined, and the preset state of charge estimation sub-models are deployed onto these nodes to form the final deployed state of charge estimation model. This process ensures the adaptability and accuracy of the model under different environments and demands, enabling it to more accurately reflect the actual state of charge of the battery.

[0117] In one exemplary embodiment, such as Figure 5As shown, real-time electrical data of the battery is input into the deployed state-of-charge estimation model to obtain real-time state-of-charge estimation data, including steps 502 to 508. Wherein:

[0118] Step 502: Based on the real-time electrical data of the battery, adjust the synaptic weights of each neuron node in the deployed state of charge estimation model to obtain the synaptic state of charge estimation model.

[0119] Among them, the synaptic charge state estimation model can be a new estimation model after adjusting the weights of the synapses of the neuron nodes in the original estimation model.

[0120] Specifically, real-time electrical data of the battery (such as voltage, current, temperature, etc.) is input into the deployed state of charge estimation model. Using the real-time electrical data of the battery, the error of each neuron node in the deployed state of charge estimation model is calculated through the backpropagation algorithm. The synaptic weights of each neuron node are adjusted according to these errors to optimize the performance of the model. Through multiple iterations, the weights are continuously adjusted and the errors are reduced so that the model can more accurately reflect the actual state of charge of the battery, forming a synaptic state of charge estimation model.

[0121] Step 504: Input the real-time electrical data of the battery into the synaptic state of charge estimation model to obtain the initial state of charge estimation data.

[0122] The initial state of charge estimation data is based on the preliminary state of charge data obtained from the synaptic state of charge estimation model.

[0123] Specifically, the pre-processed real-time electrical data of the battery (such as voltage, current, temperature, etc.) is input into the optimized synaptic state of charge estimation model. The biological neural network inside the synaptic state of charge estimation model uses the optimized synaptic weights to process and calculate the input data layer by layer, extracting and analyzing the feature information in the data. Finally, the synaptic state of charge estimation model outputs the initial state of charge estimation data through comprehensive calculation.

[0124] Step 506: If the initial state of charge estimation data representation has not converged, the model parameters of the synaptic state of charge estimation model are adjusted using a state of charge genetic algorithm to obtain an adjusted state of charge estimation model.

[0125] Among them, the charge state genetic algorithm can optimize the model parameters of the synaptic charge state estimation model and find the optimal network structure and model parameters through operations such as selection, crossover and mutation.

[0126] Among them, adjusting the state of charge estimation model can be to create a state of charge estimation model with a better network structure and model parameters.

[0127] Specifically, the algorithm determines whether the initial state of charge (SCC) estimation data has converged. If the SCC estimation data has not converged, a genetic algorithm for SCC is used for adjustment. The specific steps include initializing a population of model parameters, performing selection, crossover, and mutation operations on these parameters to generate new parameter combinations. Then, these new parameter combinations are applied to the synaptic SCC estimation model, and the fitness of each combination, i.e., the estimation accuracy of the model, is calculated. Through multiple generations of iteration, parameter combinations with high fitness are continuously selected, while combinations with low fitness are eliminated, gradually optimizing the model parameters to obtain the adjusted SCC estimation model.

[0128] Step 508: Adjust the state of charge estimation model as the deployment state of charge estimation model, return to execute the step of adjusting the synaptic weights of each neuron node in the deployment state of charge estimation model according to the real-time electrical data of the battery, and obtain the synaptic state of charge estimation model, until the initial state of charge estimation data table converges and the real-time state of charge estimation data is obtained.

[0129] Specifically, the adjusted state of charge (SCC) estimation model is used as the new deployment SCC estimation model. Then, the synaptic weights of each neuron node in the new deployment SCC estimation model are adjusted using real-time battery electrical data. Error calculation and weight updates are performed repeatedly to obtain a new synaptic SCC estimation model. Real-time electrical data is input into the new synaptic SCC estimation model to generate new initial SCC estimation data. If the new initial SCC estimation data representation does not converge, the model parameters are adjusted again using the SCC genetic algorithm to generate a new adjusted SCC estimation model, and the above process is repeated. Through multiple iterations, the initial SCC estimation data representation converges, and finally, accurate real-time SCC estimation data is obtained.

[0130] In this embodiment, the synaptic weights of the neuron nodes in the deployed state of charge (SOC) estimation model are dynamically adjusted using real-time electrical data from the battery to form a synaptic SOC estimation model. Then, the real-time electrical data is input into this model to obtain initial SOC estimation data. If the initial estimation data fails to converge, a genetic algorithm for SOC further optimizes the model parameters, and the adjusted model is re-deployed as the model for iterative optimization until data convergence, ultimately obtaining accurate real-time SOC estimation data. This process, through continuous adjustment and optimization of model parameters, ensures high accuracy and stability of the estimation data, thereby significantly improving the accuracy of the battery management system, extending battery life, and optimizing its performance.

[0131] In one exemplary embodiment, such as Figure 6As shown, based on real-time electrical data of the battery, the synaptic weights of each neuron node in the deployed state of charge estimation model are adjusted to obtain the synaptic state of charge estimation model, including steps 602 to 606. Wherein:

[0132] Step 602: Based on the real-time electrical data of the battery, activate each neuron node in the deployed state of charge estimation model to obtain neuron node activation data.

[0133] Among these, neuron activation data can be the activation state and output values ​​of each neuron node during the operation of the neural network. This data reflects the internal reactions and processing mechanisms of the neural network when processing input information.

[0134] Specifically, real-time electrical data of the battery (such as voltage, current, temperature, etc.) is input into the deployed state of charge estimation model. The biological neural network inside the deployed state of charge estimation model processes the input data layer by layer. Each neuron node calculates the input signal according to its activation function (such as ReLU, Sigmoid, Tanh, etc.) to generate an activation value. These activation values ​​reflect the degree of response of each neuron node to the input data, and finally output the neuron node activation data.

[0135] Step 604: Based on the neuron node activation data, adjust the synaptic weights of each neuron node to obtain the adjusted neuron synaptic weights.

[0136] Among them, adjusting the neuron synaptic weights can be the adjusted synaptic weights of the neuron nodes.

[0137] Specifically, based on the activation data of neuron nodes, the error of each neuron node under the current input is calculated, which is usually achieved by comparing it with the target value of historical data under similar conditions. Then, using the backpropagation algorithm, the synaptic weights of each neuron node are adjusted according to the error, the gradient of each weight is calculated, and the weight values ​​are updated proportionally to reduce the error. This process is carried out layer by layer, from the output layer to the input layer, gradually adjusting the synaptic weights of all neuron nodes, and finally obtaining the adjusted neuron synaptic weights.

[0138] Step 606: The synaptic fitness algorithm is used to optimize the synaptic weights of each adjusted neuron to obtain neuron node activation data.

[0139] Among them, the synaptic fitness algorithm can be an algorithm (function) for evaluating the quality of synaptic connections in a neural network. It determines the fitness of each synapse by quantifying its contribution to signal transmission and learning, thereby guiding the adjustment and optimization of weights.

[0140] Specifically, a synaptic fitness algorithm is used to evaluate the synaptic weights of each adjusted neuron and calculate their fitness value. This fitness value reflects the contribution of each weight to the model's output accuracy in the current network state. Then, based on the fitness value, the weights with higher fitness are further optimized and adjusted, for example, by improving their performance through local search or fine-tuning. This updates the synaptic weights of each neuron node in the neural network, thereby improving the overall network performance. Finally, the optimized weights are applied to the model, and real-time electrical data from the battery is input to calculate and obtain new neuron node activation data.

[0141] In this embodiment, by activating the neuron nodes of the deployed state of charge estimation model, generating neuron node activation data using real-time electrical data of the battery, adjusting synaptic weights based on this activation data, and then optimizing using a synaptic fitness algorithm, the model's adaptability and accuracy can be significantly improved. The resulting optimized neuron node activation data enables the model to respond more accurately to real-time changes, reduce error accumulation, extend battery life, and improve energy utilization efficiency.

[0142] In one exemplary embodiment, such as Figure 7 As shown, the model parameters of the deployed state of charge estimation model are adjusted using a stochastic state of charge algorithm and real-time state of charge estimation data to obtain an optimized state of charge estimation model, including steps 702 to 706. Wherein:

[0143] Step 702: Input the real-time electrical data of the battery into the random state of charge algorithm to obtain the random state of charge dataset.

[0144] Among them, the random state of charge dataset can be a set of electrical parameter data of the battery under different states of charge (SOC) generated or collected by calculating the real-time electrical data of the battery through the random state of charge algorithm.

[0145] Specifically, real-time electrical data of the battery (such as voltage, current, temperature, etc.) is directly input into the random state of charge (RBC) algorithm. The RBC algorithm generates multiple RBC data under different conditions through random sampling and simulation. In each iteration, the RBC algorithm randomly generates multiple other possible RBC data based on the input data. This process is repeated to form a random RBC dataset covering various random conditions.

[0146] In one specific embodiment, when the random state of charge algorithm is quantum Monte Carlo, real-time electrical data of the battery (such as voltage, current, temperature, etc.) is input into the quantum Monte Carlo algorithm. The algorithm utilizes the parallel processing capability of quantum computing to generate state of charge data under multiple conditions through random sampling and quantum state superposition. In each iteration, the algorithm randomly samples the quantum states based on the input data and obtains a set of possible state of charge data through measurement. By repeating this process, the algorithm generates a random state of charge dataset containing a large number of possible states of charge.

[0147] Step 704: Perform a Fourier transform on the real-time state of charge estimation data to obtain the frequency domain characteristics of the estimation data.

[0148] Among them, the frequency domain characteristics of the estimated data can be obtained by processing and analyzing the estimated data through Fourier transform to extract its frequency components and periodic characteristics.

[0149] Specifically, the real-time state of charge estimation data is used as input, and the Fourier transform converts the real-time state of charge estimation data in the time domain to the frequency domain, and the frequency components and amplitude information of the data are analyzed; further, the Fourier transform algorithm calculates and extracts the features of the real-time state of charge estimation data in the frequency domain, and obtains the frequency domain features of the estimation data.

[0150] In one specific embodiment, the real-time state of charge estimation data is processed using quantum Fourier transform. Quantum Fourier transform utilizes the superposition and interference properties of quantum states to convert the state of charge data in the time domain to the frequency domain. During the conversion process, quantum Fourier transform analyzes the frequency components and amplitude information of the data. Finally, the quantum computing result outputs the frequency domain characteristics of the estimated data.

[0151] Step 706: Based on the random state of charge dataset and the frequency domain characteristics of the estimated data, adjust the model parameters of the deployed state of charge estimation model to obtain the optimized state of charge estimation model.

[0152] Specifically, the random state of charge (SOC) dataset and the frequency domain features of the estimated data are input into the deployed SOC estimation model. Based on the output data of the deployed SOC estimation model, Bayesian optimization techniques are used to adjust the model parameters. A surrogate model is constructed to approximate the true objective function, and the surrogate model is updated in each iteration based on new observation data. Bayesian optimization selects the next sampling point in the parameter space by balancing exploration and utilization strategies, gradually finding the optimal parameter combination that minimizes the error. Through multiple iterations, the surrogate model gradually approaches the optimal parameter settings, significantly reducing the difference between the model output and the actual data, and finally obtaining the optimized SOC estimation model.

[0153] In this embodiment, a random state of charge (SOC) dataset is generated by inputting real-time electrical data of the battery into a random state of charge (SOC) algorithm. Frequency domain features are extracted from the real-time SOC estimation data using Fourier transform. Then, the parameters of the deployed SOC estimation model are adjusted and optimized by combining these two types of data, forming an optimized SOC estimation model. This process not only improves the model's prediction accuracy but also enhances its adaptability to different operating conditions and electrical environments. Ultimately, the optimized model can more accurately monitor and predict the battery's SOC, prevent over-discharge or over-charge, improve battery safety and reliability, and optimize energy management and usage efficiency.

[0154] In one exemplary embodiment, such as Figure 8 As shown, based on the random state of charge dataset and the frequency domain characteristics of the estimated data, the model parameters of the deployed state of charge estimation model are adjusted to obtain an optimized state of charge estimation model, including steps 802 to 806. Wherein:

[0155] Step 802: High-dimensional feature extraction is performed on the random state of charge dataset and the frequency domain features of the estimated data to obtain a high-dimensional feature dataset of the state of charge.

[0156] The high-dimensional feature dataset of the state of charge (POC) can be a collection of data containing key feature information obtained by processing the random POC dataset and estimating the frequency domain features of the data through high-dimensional feature extraction techniques. These high-dimensional features retain the most important trends and information of change in the original data, thereby simplifying the complexity of the data.

[0157] Specifically, the random state of charge dataset and the frequency domain features of the estimated data are used as input, and high-dimensional feature extraction techniques such as principal component analysis (PCA) or linear discriminant analysis (LDA) are used to process the data. These techniques are used to reduce the dimensionality and extract the most representative high-dimensional features from the dataset. These high-dimensional features can capture the main changing trends and important information of the original data, resulting in a high-dimensional feature dataset of the state of charge.

[0158] Step 804: Using particle swarm optimization algorithm and high-dimensional feature dataset of state of charge, the model parameters of the deployed state of charge estimation model are optimized to obtain the model parameters of the approximate global optimal solution.

[0159] The parameters of the approximate global optimal solution model can be a set of parameter combinations obtained by searching in a high-dimensional parameter space through optimization algorithms. Although these parameters may not be the absolute optimal solution, they are very close to the global optimal solution in practical applications.

[0160] Specifically, a high-dimensional feature dataset of charged states is input into the particle swarm optimization algorithm to initialize a swarm of particles, each representing a possible combination of model parameters. The particles then search in the parameter space, continuously updating their velocity and position, and adjusting the parameters based on their own experience and the experience of the swarm. In each iteration, the fitness of each particle is evaluated, i.e., the error between the model output and charged states under similar conditions in historical data. Through cooperation and competition, the particle swarm gradually approaches the region with the minimum error. Finally, after multiple iterations, the particle swarm converges to an approximate global optimum, yielding the approximate global optimum model parameters.

[0161] Step 806: The annealing algorithm is used to iteratively optimize the parameters of the near-global optimal solution model to obtain the optimized state of charge estimation model.

[0162] Specifically, the model parameters of the approximate global optimal solution model are used as initial parameters input into the annealing algorithm. The annealing algorithm searches in the parameter space by simulating the annealing process. In each iteration, the current parameters are randomly perturbed, and the fitness of the new parameter combination is calculated. If the new parameter combination has a better fitness, it is accepted; otherwise, it is accepted with a certain probability to avoid getting trapped in local optima. The system temperature is gradually reduced so that the algorithm gradually focuses on the global optimal region. After multiple iterations and optimizations, the optimized state of charge estimation model is finally obtained.

[0163] In this embodiment, high-dimensional features are extracted from the frequency domain features of the random state of charge (SOC) dataset and the estimated data to generate a high-dimensional SOC feature dataset. Then, a particle swarm optimization algorithm is used to optimize the parameters of the deployed SOC estimation model, obtaining near-globally optimal model parameters. Subsequently, an annealing algorithm is used for further iterative optimization of these parameters, forming the final optimized SOC estimation model. This process effectively captures and utilizes complex SOC features, combining global and local optimization methods to make the model more sensitive and accurate in responding to environmental changes and battery state, thereby significantly improving the reliability of the battery management system and battery life, while optimizing energy utilization efficiency and reducing energy waste.

[0164] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0165] Based on the same inventive concept, this application also provides a process control data determination device for power projects to implement the above-mentioned method for determining process control data based on power projects. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the process control data determination device for power projects provided below can be found in the limitations of the method for determining process control data based on power projects described above, and will not be repeated here.

[0166] In one exemplary embodiment, such as Figure 9 As shown, a process control data determination device for power projects is provided, including: a data acquisition module 902, a model deployment module 904, a data estimation module 906, a model optimization module 908, a data optimization module 910, and a data fusion module 912. Each module in the above-mentioned process control data determination device for power projects can be implemented in whole or in part through software, hardware, or a combination thereof.

[0167] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 10 As shown. This computer device includes a processor, memory, input / output interfaces (I / O), and communication interfaces.

[0168] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0169] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0170] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.

[0171] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods.

[0172] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for determining process control data based on power projects, characterized in that, The method includes: The system acquires real-time electrical data of the battery corresponding to the power project, various preset state-of-charge estimation sub-models, real-time battery scenario data, and an initial state-of-charge estimation model; the initial state-of-charge estimation model is constructed using the biological neural network corresponding to the power project. Based on the real-time scenario data of the battery, each of the preset state of charge estimation sub-models is deployed to each neuron node of the initial state of charge estimation model to obtain the deployed state of charge estimation model. The real-time electrical data of the battery is input into the deployed state of charge estimation model to obtain real-time state of charge estimation data. The model parameters of the deployed state of charge estimation model are adjusted using the stochastic state of charge algorithm and the real-time state of charge estimation data to obtain an optimized state of charge estimation model. The real-time electrical data of the battery is input into the optimized state of charge estimation model to obtain optimized state of charge estimation data. By integrating the real-time state of charge estimation data and the optimized state of charge estimation data, the target state of charge estimation data corresponding to the power project is obtained; the target state of charge estimation data is used to generate the process control data of the power project.

2. The method according to claim 1, characterized in that, The step of deploying each of the preset state of charge estimation sub-models to each neuron node of the initial state of charge estimation model based on the real-time scene data of the battery, to obtain the deployed state of charge estimation model, includes: Obtain historical electrical data of batteries corresponding to power projects; Based on the real-time scenario data of the battery, the real-time electrical data of the battery and the historical electrical data of the battery are respectively input into each of the preset state of charge estimation sub-models to obtain the applicable analysis data and the limitation analysis data of each sub-model. Based on the applicable analysis data and the limiting analysis data of each sub-model, determine the node deployment mapping relationship between each neuron node and at least one preset state of charge estimation sub-model; Based on the node deployment mapping relationship, each preset state of charge estimation sub-model is deployed to each neuron node to obtain the deployed state of charge estimation model.

3. The method according to claim 2, characterized in that, The step of deploying each preset state of charge estimation sub-model to each neuron node according to the node deployment mapping relationship, to obtain the deployed state of charge estimation model, includes: Based on the real-time scenario data of the battery and the real-time environmental data corresponding to the power project, the deployment mapping relationship of each node and each neuron node are adjusted to obtain the deployment mapping relationship of each first adjusted node and each first adjusted neuron node. Based on the deployment mapping relationship of each of the first adjustment nodes, each of the preset state of charge estimation sub-models is deployed to each of the first adjustment neuron nodes to obtain the first intermediate state of charge estimation model. Based on the real-time scenario data of the battery and the real-time demand data corresponding to the power project, the deployment mapping relationship of each node and each neuron node are adjusted to obtain the deployment mapping relationship of each second adjusted node and each second adjusted neuron node. According to the deployment mapping relationship of each of the second adjustment nodes, each of the preset state of charge estimation sub-models is deployed to each of the second adjustment neuron nodes to obtain the second intermediate state of charge estimation model. Based on the node deployment mapping relationship, each of the preset state of charge estimation sub-models is deployed to each of the neuron nodes to obtain the third intermediate state of charge estimation model. Based on the first intermediate state of charge estimation model, the second intermediate state of charge estimation model, and the third intermediate state of charge estimation model, the deployment mapping relationship of each target node and each target neuron node are determined. The preset state of charge estimation sub-models are deployed to each of the target neuron nodes according to the deployment mapping relationship of each target node, thereby obtaining the deployed state of charge estimation model.

4. The method according to claim 1, characterized in that, The step of inputting the real-time electrical data of the battery into the deployed state-of-charge estimation model to obtain real-time state-of-charge estimation data includes: Based on the real-time electrical data of the battery, the synaptic weights of each neuron node in the deployed state of charge estimation model are adjusted to obtain the synaptic state of charge estimation model. The real-time electrical data of the battery is input into the synaptic state of charge estimation model to obtain the initial state of charge estimation data. If the initial state of charge estimation data does not converge, the model parameters of the synaptic state of charge estimation model are adjusted using a state of charge genetic algorithm to obtain an adjusted state of charge estimation model. The adjusted state of charge estimation model is used as the deployed state of charge estimation model. The step of adjusting the synaptic weights of each neuron node in the deployed state of charge estimation model based on the real-time electrical data of the battery to obtain the synaptic state of charge estimation model is returned to the execution until the initial state of charge estimation data table converges to obtain the real-time state of charge estimation data.

5. The method according to claim 4, characterized in that, The step of adjusting the synaptic weights of each neuron node in the deployed state of charge estimation model based on the real-time electrical data of the battery to obtain the synaptic state of charge estimation model includes: Based on the real-time electrical data of the battery, each neuron node in the deployed state of charge estimation model is activated to obtain neuron node activation data. Based on the activation data of the neuron nodes, the synaptic weights of each neuron node are adjusted to obtain the adjusted synaptic weights of each neuron. The synaptic fitness algorithm is used to optimize the synaptic weights of each adjusted neuron to obtain the neuron node activation data; the neuron node activation data includes the optimized synaptic weights of each whole neuron.

6. The method according to claim 1, characterized in that, The step of adjusting the model parameters of the deployed state of charge estimation model using a stochastic state of charge algorithm and the real-time state of charge estimation data to obtain an optimized state of charge estimation model includes: The real-time electrical data of the battery is input into the random state of charge algorithm to obtain a random state of charge dataset; The frequency domain characteristics of the estimated state of charge are obtained by performing a Fourier transform on the real-time estimated data. Based on the random state of charge dataset and the frequency domain characteristics of the estimated data, the model parameters of the deployed state of charge estimation model are adjusted to obtain the optimized state of charge estimation model.

7. The method according to claim 6, characterized in that, The step of adjusting the model parameters of the deployed state of charge estimation model based on the random state of charge dataset and the frequency domain characteristics of the estimated data to obtain the optimized state of charge estimation model includes: High-dimensional feature extraction is performed on the random state of charge dataset and the frequency domain features of the estimated data to obtain a high-dimensional feature dataset of the state of charge; The particle swarm optimization algorithm and the high-dimensional feature dataset of the state of charge are used to optimize the model parameters of the deployed state of charge estimation model to obtain the model parameters of the approximate global optimal solution; The annealing algorithm is used to iteratively optimize the parameters of the approximate global optimal solution model to obtain the optimized state of charge estimation model.

8. A device for determining process control data based on power projects, characterized in that, The device includes: The data acquisition module is used to acquire real-time electrical data of the battery corresponding to the power project, various preset state of charge estimation sub-models, real-time scenario data of the battery, and the initial state of charge estimation model; the initial state of charge estimation model is constructed through the biological neural network corresponding to the power project. The model deployment module is used to deploy each of the preset state of charge estimation sub-models to each neuron node of the initial state of charge estimation model based on the real-time scene data of the battery, so as to obtain the deployed state of charge estimation model. The data estimation module is used to input the real-time electrical data of the battery into the deployed state of charge estimation model to obtain real-time state of charge estimation data; The model optimization module is used to adjust the model parameters of the deployed state of charge estimation model using a stochastic state of charge algorithm and the real-time state of charge estimation data, so as to obtain an optimized state of charge estimation model. The data optimization module is used to input the real-time electrical data of the battery into the optimized state of charge estimation model to obtain optimized state of charge estimation data; The data fusion module is used to fuse the real-time state of charge estimation data and the optimized state of charge estimation data to obtain the target state of charge estimation data corresponding to the power project; the target state of charge estimation data is used to generate the process control data of the power project.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

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

Images