Electrochemical energy storage system battery charging effect evaluation method

Abstract

The invention relates to the technical field of electrochemical energy storage systems, and discloses an electrochemical energy storage system battery charging effect evaluation method, which comprises the following steps: acquiring energy storage system operation data and battery charging records, extracting time sequence characteristics, equipment state characteristics and charging parameter characteristics, and calculating the charging effect of an electrochemical energy storage system battery. Constructing a multi-dimensional feature vector and generating an initial charging effect feature library; the evaluation model is dynamically adjusted based on feature library feedback, potential effect features are identified through mode clustering analysis, feature priorities are optimized, and model parameters are updated; and in combination with system configuration information and target model parameters, determining an optimal power compensation effect evaluation result by adopting a hierarchical disassembly and classification statistical method. According to the method, through multi-dimensional feature fusion, dynamic model optimization and refined statistical analysis, the accuracy and adaptability of electricity supplement effect evaluation are remarkably improved, and a scientific basis is provided for efficient operation and maintenance of the electrochemical energy storage system.

Description

Technical Field

[0001] The present invention relates to the technical field of electrochemical energy storage systems, and in particular to a method for evaluating the charging effect of a battery in an electrochemical energy storage system. Background Art

[0002] With the rapid development of renewable energy, electrochemical energy storage systems are increasingly being used in power systems. As a core component of energy storage systems, battery performance is directly related to the overall system's operating efficiency and stability. In actual operation, battery recharging is one of the key means of maintaining battery performance. However, traditional recharging effectiveness evaluation methods often rely on single indicators or static models, making it difficult to fully reflect the actual effectiveness of recharging operations. For example, existing technologies typically focus only on voltage or capacity recovery after recharging, while ignoring the combined effects of time series characteristics, changes in device status, and recharging parameters. This simplified evaluation approach can easily lead to insufficient optimization of recharging strategies and may even cause system operational risks due to evaluation bias.

[0003] Existing methods have significant shortcomings in dynamic adjustment and model updating. Due to the complex and changing operating environment of energy storage systems, the effectiveness of charging is affected by a variety of factors, such as temperature, charge and discharge frequency, and battery aging. Traditional methods lack real-time monitoring and dynamic feedback mechanisms for these factors, making it difficult for evaluation models to adapt to actual operational needs. Especially in large-scale energy storage systems, the frequency and scale of charging operations have increased significantly, and the limitations of static models are even more prominent. For example, although some methods have introduced historical data analysis, they have failed to effectively integrate real-time operating data and cannot achieve dynamic optimization of the evaluation model.

[0004] Existing technologies lack the ability to conduct hierarchical analysis and statistical classification of recharging effectiveness. Recharging operations involve multiple dimensions, including time, equipment, and parameters. However, existing methods often simply overlay these features, failing to fully explore their inherent relationships. For example, there is a lack of systematic analysis of the relationship between the recharging cycle and key system monitoring phases, or the impact of the time decay coefficient on performance evaluation. This crude approach not only reduces the accuracy of the evaluation but also increases the complexity of operations and maintenance. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for evaluating the charging effect of a battery in an electrochemical energy storage system to solve the problems raised in the above background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for evaluating the battery charging effect of an electrochemical energy storage system, the method comprising:

[0007] Acquire energy storage system operating data and battery charging records, analyze the energy storage system operating data and battery charging records, construct an evaluation feature set, and determine and output an initial charging effect feature library;

[0008] Determine the updated information set of the evaluation model based on the dynamic adjustment operation of the initial power replenishment effect feature library feedback;

[0009] Analyze the evaluation model update information set, perform hierarchical decomposition and classification statistics on the evaluation model update information set, and determine target model parameters;

[0010] A system parameter configuration information set is obtained, and based on the system parameter configuration information set and the target model parameters, a charging effect evaluation result is determined and output.

[0011] Preferably, the analyzing the energy storage system operation data and battery charging records, constructing an evaluation feature set, and determining and outputting an initial charging effect feature library includes:

[0012] Analyze the energy storage system operation data and battery charging records to extract time series feature sets, device status features, and charging parameter features;

[0013] Analyze the time series feature set to determine the time interval weight and performance correlation score corresponding to each feature point;

[0014] Determining a key monitoring phase of the system according to a charging cycle in the battery charging record, and determining a time decay coefficient according to the charging cycle and the key monitoring phase of the system;

[0015] constructing a multidimensional feature vector according to the time interval weight, the performance association score, the time decay coefficient, and the device state feature;

[0016] According to the multi-dimensional feature vector, a matching search is performed on a preset historical power replenishment database to determine and output the initial power replenishment effect feature library.

[0017] Preferably, constructing a multidimensional feature vector according to the time interval weight, the performance correlation score, the time decay coefficient and the device state feature specifically includes:

[0018] The time weight parameter is determined according to the time interval length of the time series characteristics, the state weight parameter is determined according to the abnormality degree of the equipment state characteristics, and the parameter weight parameter is determined according to the cumulative amount of the power replenishment parameter characteristics;

[0019] The time weight parameter, state weight parameter, parameter weight parameter and time attenuation coefficient are weighted and fused to form a composite feature vector including time dimension, device dimension and parameter dimension.

[0020] Preferably, the dynamic adjustment operation based on the feedback from the initial power replenishment effect feature library to determine the evaluation model update information set includes:

[0021] Based on the initial power replenishment effect feature library, a real-time effect visualization interface is provided for the monitoring system, and the feature space mapping effect is synchronously updated according to the monitoring data trajectory fed back by the system;

[0022] Based on the monitoring data trajectory, identifying the potential effect characteristics of power replenishment through pattern cluster analysis, dynamically adjusting the sorting priority of the initial features, and generating a feature update plan;

[0023] Analyze the feature update plan, evaluate the stability of the evaluation model, determine model stability feedback information, and feedback the corresponding feature adjustment suggestions to the operation and maintenance end, and receive the operation and maintenance confirmation instructions;

[0024] According to the operation and maintenance confirmation instruction, an evaluation model update information set is determined.

[0025] Preferably, analyzing the feature update scheme, evaluating the stability of the evaluation model, and determining model stability feedback information includes:

[0026] According to the feature update scheme, the number of feature dimensions, evaluation confidence interval, historical evaluation result distribution and feature importance ranking of the model are extracted;

[0027] According to the system significance of each feature, the model's evaluation error tolerance threshold, historical evaluation accuracy ratio and continuous evaluation consistency rate are determined;

[0028] Based on the number of feature dimensions, the evaluation confidence interval, and the distribution of historical evaluation results, and according to the feature importance ranking and the evaluation error tolerance threshold, a statistical stability determination condition is established;

[0029] Determining an effective time range for model evaluation based on the power replenishment cycle and the key monitoring phase of the system;

[0030] Based on the historical evaluation accuracy ratio and the effective time range, and according to the continuous evaluation consistency rate and the feature importance ranking, a temporal stability determination condition is established;

[0031] According to the statistical stability judgment condition and the temporal stability judgment condition, it is judged whether the statistical stability or temporal stability corresponding to the current feature update scheme meets the requirements, and a judgment conclusion is determined, which is used as the model stability feedback information.

[0032] Preferably, the statistical stability judgment condition is constructed based on the number of feature dimensions, the evaluation confidence interval and the distribution of historical evaluation results, according to the feature importance ranking and the evaluation error tolerance threshold, and specifically includes:

[0033] Calculate the correlation between the number of feature dimensions and the evaluation confidence interval, analyze the matching degree between the distribution of historical evaluation results and the feature importance ranking, and set the critical range of statistical stability based on the evaluation error tolerance threshold;

[0034] The time series stability determination conditions are constructed based on the historical evaluation accuracy ratio and the effective time range, according to the continuous evaluation consistency rate and the feature importance ranking, and specifically include:

[0035] Calculate the fluctuation range of the historical evaluation accuracy ratio within the effective time range, analyze the temporal correlation between the continuous evaluation consistency rate and the feature importance ranking, and set the critical range of temporal stability.

[0036] Preferably, analyzing the evaluation model update information set, performing hierarchical decomposition and classification statistics on the evaluation model update information set, and determining target model parameters include:

[0037] Update the information set according to the evaluation model, and count the key feature adjustment set and the auxiliary parameter correction set;

[0038] According to the model update cycle set by the system, the battery charging cycles corresponding to different batteries in the evaluation model update information set are divided into several update batches to determine the update batch information set;

[0039] Analyzing the key feature adjustment set based on the update batch information set, and determining the parameter redundancy corresponding to each feature in each update batch according to the parameter drift ratio corresponding to each feature in the key feature adjustment set in the historical update records;

[0040] The target model parameters are constructed according to the auxiliary parameter correction set, the update batch information set and the parameter redundancy.

[0041] Preferably, the determining and outputting the supplementary power effect evaluation result based on the system parameter configuration information set and according to the target model parameters includes:

[0042] Analyzing the target model parameters based on the system parameter configuration information set to determine a number of target system parameters corresponding to each characteristic adjustment item in the target model parameters;

[0043] Extracting historical collaboration records corresponding to each target system parameter based on the system parameter configuration information set, thereby determining the historical application count, historical adjustment cost, historical evaluation accuracy ratio, and historical execution risk of each feature adjustment item in the current system under the corresponding target system parameter;

[0044] Based on the historical application times, historical adjustment costs, historical evaluation accuracy ratio and historical execution risks of each feature adjustment item under the corresponding target system parameters, comprehensive cost optimization processing is performed on each feature adjustment item to determine the optimal system parameters corresponding to each feature adjustment item, thereby constructing and outputting the power replenishment effect evaluation results.

[0045] Preferably, performing comprehensive cost optimization processing on each feature adjustment item based on the historical application times, the historical adjustment cost, the historical evaluation accuracy ratio, and the historical execution risk of each feature adjustment item under the corresponding target system parameters specifically includes:

[0046] Calculate the weighted sum of the number of historical applications and the historical adjustment costs as the resource use cost indicator, and calculate the weighted sum of the historical evaluation accuracy ratio and the historical execution risk as the effect risk cost indicator;

[0047] The resource use cost index is comprehensively compared with the effect risk cost index, and the system parameter with the lowest comprehensive cost is selected as the optimal parameter.

[0048] Preferably, identifying the potential effect characteristics of power replenishment through pattern cluster analysis includes:

[0049] Dividing the monitoring data trajectory into time windows and extracting characteristic fluctuation amplitude and frequency within each time window;

[0050] The initial classification standard of the clustering algorithm is set as the historical high-effect power replenishment feature library, and clustering iteration is performed based on the matching degree between the feature fluctuation amplitude and frequency and the initial classification standard;

[0051] The reliability of the clustering results is evaluated by classification purity, and cluster groups whose matching degree with historical high-effect features exceeds the preset critical value are screened out, and their corresponding feature combinations are determined as potential power replenishment effect features.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] The present invention provides a method for evaluating the recharging effect of batteries in electrochemical energy storage systems. This method achieves a comprehensive evaluation of the recharging effect by constructing a multidimensional feature vector. A composite feature vector is formed by extracting time series features, device status features, and recharging parameter features, and combining them with a time decay coefficient and a weight parameter. This multidimensional analysis method can more accurately reflect the actual effect of the recharging operation, avoiding the limitations of traditional single indicator evaluation. For example, the introduction of time interval weights and performance correlation scores enables the evaluation results to dynamically reflect the time dependence of the recharging operation, thereby providing a more scientific basis for operation and maintenance decisions.

[0054] Based on feedback from the initial recharging effect feature library, the system can adjust the feature space mapping effect in real time and identify potential effect features through pattern clustering analysis. This process not only optimizes the feature sorting priority but also ensures the reliability of the model through stability assessment. For example, by establishing statistical stability judgment conditions and temporal stability judgment conditions, the system can dynamically determine the feasibility of feature update solutions, thereby avoiding evaluation bias caused by model drift. This dynamic adjustment capability enables the evaluation model to adapt to complex and changing operating environments, making it particularly suitable for the needs of large-scale energy storage systems.

[0055] By hierarchically processing the evaluation model update information set, the system can divide update batches and determine parameter redundancy, thereby optimizing the generation of target model parameters. This process combines historical collaboration records with comprehensive cost optimization to ensure the optimal system parameter selection for each feature adjustment. For example, by calculating resource utilization cost indicators and effect risk cost indicators, the system can achieve a balance between cost and effect, providing cost-effective power replenishment strategies for operations and maintenance.

[0056] The monitoring system can provide operators with intuitive display of results and dynamic data traces, helping them quickly identify problems and make decisions. At the same time, the feedback and confirmation mechanism for feature adjustment suggestions further enhances the system's interactivity and practicality. These features not only reduce the complexity of operation and maintenance, but also provide strong support for the long-term stable operation of the energy storage system. The present invention has outstanding advantages in comprehensiveness, dynamism, accuracy, and practicality, and can effectively improve the battery management level and overall performance of electrochemical energy storage systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a working principle diagram of the method for evaluating the battery charging effect of an electrochemical energy storage system according to the present invention;

[0058] Figure 2 Flowchart for constructing the initial charging effect feature library;

[0059] Figure 3 Flowchart constructed for multidimensional feature vectors;

[0060] Figure 4 Flowchart for determining updated information sets for evaluation models;

[0061] Figure 5 Flowchart constructed for statistical and temporal stability criteria. DETAILED DESCRIPTION

[0062] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0063] See also Figure 1-Figure 5 The present invention relates to a method for evaluating the battery charging effect of an electrochemical energy storage system, and the specific implementation steps are as follows:

[0064] The energy storage system's operating data and battery charging records are acquired and analyzed to construct an evaluation feature set. The initial charging effect feature library is then determined and output. The acquired operating data and charging records are then analyzed in depth to extract a time series feature set, device status features, and charging parameter features. The time series feature set is analyzed to determine the time interval weight and performance correlation score corresponding to each feature point. Based on the charging cycle in the battery charging records, the system's key monitoring phase is determined. The time decay coefficient is then determined based on the charging cycle and the system's key monitoring phase. A multidimensional feature vector is then constructed based on the time interval weight, performance correlation score, time decay coefficient, and device status features. Finally, the multidimensional feature vector is used to perform a match search against a preset historical charging database to determine and output the initial charging effect feature library.

[0065] Based on the dynamic adjustment operations fed back by the initial power replenishment effect feature library, the evaluation model update information set is determined. Specifically, based on the initial power replenishment effect feature library, a real-time effect visualization interface is provided for the monitoring system. Based on the monitoring data trajectory fed back by the system, the feature space mapping effect is synchronously updated. By analyzing the monitoring data trajectory, pattern clustering analysis is used to identify the potential power replenishment effect characteristics, and then the sorting priority of the initial features is dynamically adjusted to generate a feature update plan. The feature update plan is analyzed to evaluate the stability of the evaluation model, determine the model stability feedback information, and feedback the corresponding feature adjustment suggestions to the operation and maintenance end. The operation and maintenance confirmation instructions are received, and finally, the evaluation model update information set is determined based on the operation and maintenance confirmation instructions.

[0066] Analyze the evaluation model update information set, perform hierarchical decomposition and classification of the evaluation model update information set, and determine the target model parameters. The specific operation is to count the key feature adjustment set and the auxiliary parameter correction set based on the evaluation model update information set. According to the model update cycle set by the system, combined with the recharging cycles corresponding to different batteries in the evaluation model update information set, several update batches are divided to determine the update batch information set. Based on the update batch information set, the key feature adjustment set is analyzed, and the parameter redundancy corresponding to each feature in each update batch is determined according to the parameter drift ratio corresponding to each feature in the key feature adjustment set in the historical update records. Finally, the target model parameters are constructed based on the auxiliary parameter correction set, the update batch information set, and the parameter redundancy.

[0067] Obtain a system parameter configuration information set, and based on the system parameter configuration information set and the target model parameters, determine and output the supplementary power effect evaluation result. Specifically, based on the system parameter configuration information set, analyze the target model parameters to determine several target system parameters corresponding to each feature adjustment item in the target model parameters. Extract the historical collaboration records corresponding to each target system parameter from the system parameter configuration information set to determine the historical application times, historical adjustment costs, historical evaluation accuracy ratios, and historical execution risks of each feature adjustment item in the current system under the corresponding target system parameters. Based on these historical data of each feature adjustment item under the corresponding target system parameters, perform comprehensive cost optimization processing on each feature adjustment item to determine the optimal system parameters corresponding to each feature adjustment item, thereby constructing and outputting the supplementary power effect evaluation result.

[0068] Example 1:

[0069] This example describes in detail the specific implementation method for constructing an evaluation feature set and determining an initial charging effect feature library. During implementation, energy storage system operating data and battery charging records must be acquired, forming the foundation of the entire example. Once this data is acquired, it is analyzed to extract a time series feature set, device status features, and charging parameter features.

[0070] Analyzing a time series feature set requires determining the time interval weight and performance correlation score for each feature point. The time interval weight is related to the length of the time series feature interval; different time interval lengths affect the weight. The performance correlation score is determined by analyzing the correlation between the feature point and the energy storage system's performance.

[0071] When analyzing battery recharge records, the recharge cycles can be used to identify the critical system monitoring phases. This identification is crucial for subsequent evaluations, as it helps clarify which phases require focused attention. The time decay coefficient is further determined by combining the recharge cycles and critical system monitoring phases. This factor accounts for the impact of time on battery recharge effectiveness. As battery performance changes over time, the recharge effect may also diminish.

[0072] A multidimensional feature vector needs to be constructed. When constructing a multidimensional feature vector, the time weight parameter is first determined based on the time interval length of the time series feature. Different time interval lengths result in different time weight parameters. The state weight parameter is then determined based on the degree of abnormality in the device state feature. Higher abnormalities correspond to higher potential state weight parameters. Furthermore, the parameter weight parameters are determined based on the cumulative amount of the recharging parameter features. This cumulative amount reflects the accumulation of key parameters during the recharging process and influences the determination of the parameter weight parameters.

[0073] The determined time weight parameter, state weight parameter, parameter weight parameter, and time decay coefficient are weighted and fused to form a composite feature vector encompassing the time, device, and parameter dimensions—a multidimensional feature vector. This multidimensional feature vector comprehensively considers factors from multiple dimensions and can more comprehensively reflect the characteristics of battery charging.

[0074] Using the constructed multidimensional feature vector, a matching search is performed against a pre-set historical power replenishment database. This database contains a large amount of historical power replenishment data and corresponding feature information. Through matching search, historical data matching the current multidimensional feature vector is found, thereby determining and outputting an initial power replenishment effect feature library. This initial power replenishment effect feature library contains power replenishment effect feature information relevant to the current situation, providing an important reference for subsequent evaluation work.

[0075] Throughout the implementation process, every step must be rigorously executed. For example, when acquiring data, its accuracy and completeness must be ensured, as inaccurate or incomplete data can affect subsequent analysis and feature extraction. When analyzing data and extracting features, scientific and rational methods must be employed to ensure that the extracted features truly reflect the operating status and recharging conditions of the energy storage system. When determining various weighting parameters and attenuation coefficients, the actual situation and relevant factors must be fully considered to ensure the rationality of these parameters.

[0076] When constructing the multi-dimensional feature vector, attention should be paid to the weight distribution of each parameter in the process of weighted fusion, so that it can reasonably reflect the importance of each dimension factor. When performing matching retrieval on the pre-set historical power compensation database, the accuracy and efficiency of the retrieval algorithm should be ensured to quickly and accurately find the matching historical data, thereby generating an accurate initial power compensation effect feature library.

[0077] Embodiment 2:

[0078] In this embodiment, the process of determining the evaluation model update information set based on the dynamic adjustment operation feedback from the initial power compensation effect feature library is described. After obtaining the initial power compensation effect feature library, a real-time effect visualization interface needs to be built for the monitoring system. Based on the initial power compensation effect feature library, the interface presents the power compensation effect related data in a visual form. When the monitoring data trajectory is generated during system operation and fed back to the interface, the feature space mapping effect is updated synchronously, so that the visualization content can reflect the current system state in real time.

[0079] The monitoring data trajectory is processed, and the specific operation is to divide it into time windows. The division of time windows needs to be combined with data characteristics and evaluation requirements, for example, it can be divided according to fixed time intervals or according to different stages of the power compensation period. After division, the fluctuation amplitude and frequency of the features in each time window are extracted. The feature fluctuation amplitude reflects the change range of the feature within the window, and the frequency reflects the frequency of the change. These information is an important basis for subsequent analysis.

[0080] After completing feature extraction, set the initial classification standard of the clustering algorithm as the historical high-effect power compensation feature library. The historical high-effect power compensation feature library stores good feature combinations in the past, which can improve the pertinence of clustering as the initial standard. According to the matching degree of feature fluctuation amplitude and frequency and the initial classification standard, iterative clustering is performed. In the iteration process, the categories and feature attribution of clustering are constantly adjusted, so that the features in the same category have more similarity in fluctuation amplitude and frequency, and have higher matching degree with historical high-effect features.

[0081] After clustering, the reliability of the clustering result is evaluated by classification purity. Classification purity refers to the proportion of the number of features belonging to the class in each cluster to the total number of features. The higher the purity, the more reliable the clustering result. Filter out the cluster groups with a matching degree exceeding the preset threshold with historical high-effect features. The preset threshold needs to be determined according to the actual application scenario and historical data experience, and is generally set to a reasonable percentage, such as 70% or 80%. The feature combinations corresponding to these cluster groups are determined as potential power compensation effect features. These potential features may be new features that have an important impact on power compensation effect under the current running state.

[0082] Based on the identified potential recharge effect features, the ranking priority of the initial features is dynamically adjusted. Initial features already have a certain ranking in the initial recharge effect feature library, and the addition of potential effect features may change the importance of each feature. For example, if a potential feature frequently appears in multiple clusters and has a high match with historically high-performance features, its ranking priority may be increased. After the adjustment, a feature update plan is generated, which includes changes in feature ranking and possible feature additions or deletions.

[0083] Next, we move on to the model stability assessment phase. The feature update plan extracts the number of feature dimensions, the evaluation confidence interval, the distribution of historical evaluation results, and the feature importance ranking. The number of feature dimensions reflects the number of features considered by the model, the evaluation confidence interval reflects the reliability range of the evaluation results, the distribution of historical evaluation results shows the central tendency and dispersion of previous evaluation results, and the feature importance ranking clarifies the impact of each feature on the evaluation results.

[0084] Based on the system significance of each feature, determine the model's assessment error tolerance threshold, historical assessment accuracy ratio, and continuous assessment consistency rate. The assessment error tolerance threshold represents the allowable error range for assessment results, the historical assessment accuracy ratio represents the proportion of accurate results in previous assessments, and the continuous assessment consistency rate represents the probability that multiple consecutive assessment results are consistent. The determination of these indicators should be based on the actual system operational requirements and historical data statistics.

[0085] Based on the number of feature dimensions, the evaluation confidence interval, and the distribution of historical evaluation results, combined with the feature importance ranking and the evaluation error tolerance threshold, a statistical stability judgment condition is constructed. For example, if the number of feature dimensions increases, whether the evaluation confidence interval is within the error tolerance threshold, whether the distribution of historical evaluation results remains stable, and whether the changes in the feature importance ranking are reasonable. At the same time, based on the historical evaluation accuracy ratio and the effective time range, combined with the continuous evaluation consistency rate and feature importance ranking, a temporal stability judgment condition is constructed. The effective time range refers to the time period during which the model evaluation results are valid. It is necessary to analyze whether the fluctuations in the historical evaluation accuracy ratio within this time range are within an acceptable range, whether the continuous evaluation consistency rate meets the requirements, and whether the changes in the feature importance ranking over time are reasonable.

[0086] The statistical stability and temporal stability criteria are used to determine whether the current feature update scheme meets the statistical stability or temporal stability requirements. If so, the feature update scheme will not significantly negatively impact model stability. If not, the scheme needs to be adjusted. This ultimately generates model stability feedback information, which includes the evaluation conclusions and possible adjustment suggestions for the feature update scheme.

[0087] The feature adjustment suggestions are fed back to the operations and maintenance team, who review and confirm them. Upon receiving the confirmation instruction, the evaluation model update information set is determined based on the instruction. This evaluation model update information set contains all confirmed feature adjustments and related parameters and serves as the basis for subsequent model updates. This ensures that the evaluation model can be updated in a timely manner as the system's operating status changes, maintaining the accuracy and effectiveness of the battery charging effect assessment.

[0088] Example 3:

[0089] This embodiment mainly involves the implementation method of analyzing the evaluation model update information set and determining the target model parameters. During implementation, it is necessary to perform statistical processing on the contents of the evaluation model update information set to obtain the key feature adjustment set and the auxiliary parameter correction set. The evaluation model update information set here is determined in the aforementioned steps based on the dynamic adjustment operation of the initial power replenishment effect feature library feedback, which contains various types of information related to the model update. The key feature adjustment set mainly involves the adjustment content of the key features in the model, while the auxiliary parameter correction set is the correction information about the auxiliary parameters. Accurate statistics of these sets are the basis for subsequent work.

[0090] The update batches should be divided according to the model update cycle set by the system, and in combination with the recharging cycles corresponding to different batteries in the evaluation model update information set. The model update cycle set by the system is determined according to the general rules and needs of system operation, while the recharging cycles of different batteries may vary due to factors such as the battery's usage status and type. By comprehensively considering these two cycle factors, the update work is divided into several update batches. Each update batch corresponds to a certain time range and update task, and then the update batch information set is determined. The update batch information set records the specific information of each batch in detail, such as the batch's time range, the number and type of batteries involved, etc., which helps to carry out subsequent work in an orderly manner.

[0091] After determining the update batch information set, the key feature adjustment set is analyzed based on this information set. Specifically, the parameter drift ratio corresponding to each feature in the key feature adjustment set in the historical update records needs to be examined. The parameter drift ratio reflects the degree to which the feature parameter has deviated from the initial value during previous updates. By analyzing this ratio, the changing trend and stability of the feature parameter can be understood. Based on the parameter drift ratio of each feature, the parameter redundancy corresponding to each feature in each update batch can be determined. The determination of parameter redundancy is to account for possible parameter drift during model updates and reserve a certain amount of adjustment space to ensure model stability and accuracy.

[0092] After determining the auxiliary parameter correction set, update batch information set, and parameter redundancy, this information needs to be integrated to construct the target model parameters. The auxiliary parameter correction set provides the direction and magnitude of auxiliary parameter corrections, the update batch information set clarifies the update batch schedule and task allocation, and the parameter redundancy provides a reference for adjusting the characteristic parameters. When constructing the target model parameters, the interrelationships and influences between these factors must be fully considered to ensure that the target model parameters accurately reflect the evaluation model's update requirements and provide reliable parameter support for subsequent recharging effectiveness evaluation.

[0093] Throughout the implementation process, data accuracy and completeness are crucial. When compiling the key feature adjustment sets and auxiliary parameter correction sets, it's crucial to ensure a comprehensive and accurate analysis of the evaluation model update information set to avoid missing important adjustments and corrections. When dividing update batches, it's important to consider the system's model update cycle and the recharge cycles of different batteries. This ensures that the batches meet the overall system operational requirements while allowing for flexible adjustments based on the characteristics of different batteries.

[0094] When analyzing key feature adjustment sets, the accuracy and completeness of historical update records directly impact the calculation of parameter drift ratios and the determination of parameter redundancy. Therefore, it's important to ensure the authenticity and reliability of historical update records and employ scientific and rational methods during analysis to accurately determine parameter drift ratios and parameter redundancy.

[0095] Constructing target model parameters requires comprehensive consideration of multiple factors, requiring the organic integration of auxiliary parameter correction sets, update batch information sets, and parameter redundancy. This process may require multiple analyses and adjustments to ensure the rationality and effectiveness of the target model parameters. For example, when considering auxiliary parameter corrections, the order and magnitude of auxiliary parameter corrections within each batch must be determined in conjunction with the update batch schedule. Furthermore, the parameter redundancy setting must also align with the update batch timeframe and the changing trends of the characteristic parameters.

[0096] Each step is closely logically linked. Statistical analysis of key feature adjustment sets and auxiliary parameter correction sets forms the basis for grouping update batches. This grouping, in turn, provides a timeframe and task schedule for analyzing key feature adjustment sets and determining parameter redundancy. Constructing target model parameters builds upon the previous steps by comprehensively applying and integrating all this information. Therefore, during implementation, it's crucial to strictly adhere to the sequential order of the steps, ensuring that each step lays a solid foundation for subsequent steps.

[0097] In practical applications, it is also necessary to flexibly adjust each link in the implementation process according to the actual operation of the energy storage system and changes in the external environment. For example, when the system operating state changes greatly, it may be necessary to reevaluate the model update period and the battery power compensation period, and then adjust the division of the update batch; when it is found that the parameter drift proportion of some features is abnormal, the parameter redundancy needs to be adjusted in a timely manner to ensure the accuracy and stability of the model.

[0098] Embodiment 4:

[0099] This embodiment is an implementation method of determining the power compensation effect evaluation result based on the system parameter configuration information set and the target model parameter. In implementation, the target model parameter is analyzed based on the system parameter configuration information set to determine a plurality of target system parameters corresponding to each feature adjustment item in the target model parameter. For example, if a feature adjustment item in the target model parameter involves adjustment of the battery voltage, the corresponding target system parameters may include the upper limit of the charging voltage, the voltage sampling frequency, etc.

[0100] The historical collaboration records corresponding to each target system parameter are extracted from the system parameter configuration information set. Assuming that the target system parameter is the upper limit of the charging voltage, its historical collaboration records may include the operation time, adjustment amplitude, and battery operating state during the same period when the upper limit of the charging voltage was adjusted in the past half year. Through analysis of these historical collaboration records, the historical application times, historical adjustment cost, historical evaluation accuracy proportion, and historical execution risk of the current system for each feature adjustment item under the corresponding target system parameter are determined.

[0101] Taking the feature adjustment item "optimizing the charging current curve" as an example, one of its corresponding target system parameters is the charging current change rate. The historical application times refer to the number of times this charging current change rate has been used in past power compensation operations; the historical adjustment cost includes the cost of human input, equipment debugging time, etc. required when adjusting this parameter; the historical evaluation accuracy proportion refers to the proportion of the number of times the evaluation result is accurate to the total number of evaluations after using this charging current change rate for power compensation; and the historical execution risk is determined according to the frequency and severity of problems such as battery overheating and charging efficiency decline that have occurred in previous operations.

[0102] After obtaining this historical data, comprehensive cost optimization is performed on each feature adjustment item. The weighted sum of the historical application count and the historical adjustment cost is calculated as the resource usage cost indicator. Assuming a weight of 0.4 for the historical application count and 0.6 for the historical adjustment cost, if the target system parameter A corresponding to a feature adjustment item has been historically applied 100 times, with a cost of 50 yuan per adjustment, then the resource usage cost indicator is 100 × 50 × 0.6 + 100 × 0.4 × 1 (here assuming the quantized value of the application count is 1) = 3000 + 40 = 3040. The weighted sum of the historical assessment accuracy rate and the historical execution risk is calculated as the effect risk cost indicator. If the historical assessment accuracy rate is 80% and the historical execution risk is 20% (the higher the risk, the greater the value), with weights of 0.7 and 0.3, respectively, then the effect risk cost indicator is 80% × 0.7 + 20% × 0.3 = 0.56 + 0.06 = 0.62.

[0103] A comprehensive comparison is performed between the resource cost and the performance-risk cost, and the system parameter with the lowest overall cost is selected as the optimal parameter. For example, the feature adjustment item "Optimize Charging Current Curve" corresponds to two target system parameters, A and B. Parameter A has a resource cost of 3040 and an performance-risk cost of 0.62; parameter B has a resource cost of 2800 and an performance-risk cost of 0.65. For this comprehensive comparison, the two indicators can be normalized and added together. Assuming a normalization coefficient of 0.001 for the resource cost and 1 for the performance-risk cost, the overall cost of parameter A is 3040 × 0.001 + 0.62 = 3.04 + 0.62 = 3.66, and the overall cost of parameter B is 2800 × 0.001 + 0.65 = 2.8 + 0.65 = 3.45. In this case, parameter B is selected as the optimal parameter.

[0104] Based on the optimal system parameters determined for each feature adjustment item, the battery charging effect evaluation results are constructed and output. For example, for the feature adjustment item "Optimize Charging Current Curve," the optimal parameter is determined to be a charging current change rate of 0.5A / min. The battery charging effect evaluation results will clearly indicate the optimal parameter setting for this feature adjustment item, and combined with the optimal parameters of other feature adjustment items, a complete evaluation report will be formed to provide a reference for subsequent battery charging operations.

[0105] Throughout the implementation process, the completeness and accuracy of the system parameter configuration information set are crucial. Missing certain historical collaboration records in the information set can lead to inaccurate data such as the number of historical applications and adjustment costs, which in turn affects the calculation of comprehensive costs and the selection of optimal parameters. When analyzing historical collaboration records, it's important to comprehensively consider various factors to avoid missing important information. For example, when calculating historical execution risk, it's important to consider not only the frequency of problems but also their severity; a single factor shouldn't be used alone.

[0106] When determining weighting factors, it's important to set them appropriately based on actual conditions and experience. Different feature adjustments and target system parameters may have different historical application times, adjustment costs, assessment accuracy, and execution risk, so the weighting factors should also vary. For example, for parameters that have a greater impact on battery life, the historical execution risk weight may need to be set higher.

[0107] Comprehensive cost optimization is a complex process that requires comprehensive consideration of multiple indicators. We cannot focus on a single metric while ignoring others. For example, while a parameter may have a low resource usage cost, if its effect risk is high, it may result in poor charging performance or even damage the battery. Therefore, a comprehensive trade-off is necessary.

[0108] The output of the battery charging effect evaluation must clearly and explicitly present the optimal parameters for each feature adjustment item, as well as the entire evaluation process, so that relevant personnel can understand and use it. In actual application, the evaluation results must be promptly updated and optimized based on new operating data and feedback to ensure that they always accurately reflect the actual battery charging effect and optimal parameter settings.

[0109] Example 5:

[0110] This embodiment focuses on the implementation method of identifying the potential effect characteristics of power replenishment through pattern cluster analysis. In the specific implementation, the monitoring data trajectory needs to be divided into time windows. For example, when the energy storage system is running, the collected battery voltage, current, temperature and other data will form a continuous monitoring data trajectory, which can be divided into a time window every 15 minutes. The length of the time window can also be dynamically adjusted according to the different stages of the power replenishment cycle, such as the charging stage, the static stage, and the discharging stage. The purpose of dividing the time window is to more carefully analyze the characteristic changes of the data in different time periods.

[0111] After the time window division, the fluctuation amplitude and frequency of the features in each time window are extracted. Taking the battery voltage data in a certain time window as an example, the fluctuation amplitude refers to the difference between the maximum and minimum values of the voltage in that time period, and the frequency refers to the number of times the voltage fluctuates in a unit of time. Assuming that in a 15-minute time window, the battery voltage fluctuates from 3.8V to 4.2V and then falls back to 3.9V, the fluctuation amplitude is 0.4V, and if the voltage fluctuates back and forth 5 times during this period, the frequency is 5 times / 15 minutes. The extraction of these feature fluctuation amplitudes and frequencies can reflect the running state of the battery in that time period and the parameter changes during the power compensation process.

[0112] The initial classification standard of the clustering algorithm is set as the historical high-effect power compensation feature library. The historical high-effect power compensation feature library stores information such as the fluctuation amplitude and frequency of each feature under the condition of good power compensation effect in the past. For example, when the power compensation effect was good in the past, the fluctuation amplitude of the battery voltage was between 0.2-0.3V and the fluctuation frequency was 3-4 times / 15 minutes, which will be included in the historical high-effect power compensation feature library. Using the features in this library as the initial classification standard can make the clustering analysis more targeted in finding feature combinations related to high-effect power compensation.

[0113] According to the matching degree of the feature fluctuation amplitude and frequency and the initial classification standard, the clustering iteration is performed. In the first clustering, the feature fluctuation amplitude and frequency of each time window are compared with the features in the initial classification standard to calculate the matching degree. The matching degree can be calculated using methods such as Euclidean distance, and the closer the distance, the higher the matching degree. Time window features with high matching degree are classified into a class, and with the iteration, the center position of each class is adjusted, the matching degree is recalculated and classified, until the clustering result no longer changes significantly or reaches the preset number of iterations.

[0114] After clustering is completed, the reliability of the clustering result is evaluated by classification purity. The calculation method of classification purity is to count the proportion of the number of time windows belonging to the historical high-effect power compensation features in the total number of time windows in each cluster. For example, there are 100 time window features in a certain cluster, of which 80 have a matching degree with the historical high-effect power compensation features exceeding a preset threshold, so the classification purity of this cluster is 80%. By setting a purity threshold, such as 70%, the cluster group with a classification purity higher than the threshold is selected.

[0115] After screening the cluster groups that meet the purity requirements, further screen the cluster groups that have a matching degree with the historical high-effect features exceeding a preset threshold. The preset threshold also needs to be determined according to historical data and actual needs, such as setting it to 75%. Assuming that the classification purity of a certain cluster group is 80%, and the average matching degree of each time window feature with the historical high-effect feature is 85%, which exceeds the threshold of 75%, then the feature combination corresponding to the cluster group will be determined as a power supply potential effect feature.

[0116] For example, after the above steps, it is found that the feature combination of a certain cluster group is: battery voltage fluctuation amplitude 0.25-0.35V, fluctuation frequency 3-5 times / 15 minutes, battery temperature fluctuation amplitude 2-3℃, fluctuation frequency 1-2 times / 15 minutes. These feature combinations frequently appear in historical high-effect power supply, and the current clustering result shows that there are a large number of time windows that meet the feature combination in recent monitoring data, so it is determined as a power supply potential effect feature.

[0117] The determination of these power supply potential effect features provides an important basis for subsequent feature updating and evaluation model optimization. For example, in the subsequent evaluation model, these potential effect features can be used as new input dimensions, or the weights of existing features can be adjusted, to more accurately evaluate the power supply effect. In practical applications, it is also necessary to regularly update the historical high-effect power supply feature library, and include the newly identified power supply potential effect features, to adapt to the changes in the operation state of the energy storage system and the development of power supply technology.

[0118] During the entire implementation process, the division method of the time window will directly affect the effect of feature extraction. If the time window is divided too short, it may lead to inaccurate calculation of feature fluctuation amplitude and frequency; if it is divided too long, it may not be able to capture short-term change features of the data. Therefore, the length of the time window needs to be reasonably selected according to the sampling frequency of the data and the actual operation situation.

[0119] The selection of the clustering algorithm and the setting of the initial classification standard are also crucial. Different clustering algorithms have different adaptability to data, and the appropriate algorithm needs to be selected according to the type and distribution characteristics of the features. The accuracy of the initial classification standard depends on the quality of the historical high-effect power supply feature library, so it is necessary to ensure the accuracy and completeness of the historical data and update the data in the library in a timely manner.

[0120] The setting of the preset threshold and the purity threshold needs to be tested and verified through a large amount of historical data and actual application, to find the most suitable numerical value for the current system. If the threshold is set too high, it may miss some valuable potential effect features; if it is set too low, it may introduce some irrelevant features, affecting the accuracy of the evaluation model.

[0121] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting; it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and illustrative figures, it should be apparent that the scope of the present application is not limited to these specific embodiments.

[0122] While the embodiments of the application have been shown and described herein, it is to be understood that the scope of the application, jointly pointed out in the appended claims, is not to be limited to the above-described embodiments but can be otherwise variously changed, modified, replaced, and altered within the principles and spirit of the present application.

Claims

1. A method for evaluating the battery charging effect of an electrochemical energy storage system, characterized in that: include: Acquire energy storage system operating data and battery charging records, analyze the energy storage system operating data and battery charging records, construct an evaluation feature set, and determine and output an initial charging effect feature library; Determine the updated information set of the evaluation model based on the dynamic adjustment operation of the initial power replenishment effect feature library feedback; Analyze the evaluation model update information set, perform hierarchical decomposition and classification statistics on the evaluation model update information set, and determine target model parameters; A system parameter configuration information set is obtained, and based on the system parameter configuration information set and the target model parameters, a charging effect evaluation result is determined and output.

2. The method for evaluating the battery charging effect of an electrochemical energy storage system according to claim 1, characterized in that: The analyzing the energy storage system operation data and battery charging records, constructing an evaluation feature set, and determining and outputting an initial charging effect feature library includes: Analyze the energy storage system operation data and battery charging records to extract time series feature sets, device status features, and charging parameter features; Analyze the time series feature set to determine the time interval weight and performance correlation score corresponding to each feature point; Determining a key monitoring phase of the system according to a charging cycle in the battery charging record, and determining a time decay coefficient according to the charging cycle and the key monitoring phase of the system; constructing a multidimensional feature vector according to the time interval weight, the performance association score, the time decay coefficient, and the device state feature; According to the multi-dimensional feature vector, a matching search is performed on a preset historical power replenishment database to determine and output the initial power replenishment effect feature library.

3. The method for evaluating the battery charging effect of an electrochemical energy storage system according to claim 2, characterized in that: The constructing of a multidimensional feature vector according to the time interval weight, the performance correlation score, the time decay coefficient, and the device state feature specifically includes: The time weight parameter is determined according to the time interval length of the time series characteristics, the state weight parameter is determined according to the abnormality degree of the equipment state characteristics, and the parameter weight parameter is determined according to the cumulative amount of the power replenishment parameter characteristics; The time weight parameter, state weight parameter, parameter weight parameter and time attenuation coefficient are weighted and fused to form a composite feature vector including time dimension, device dimension and parameter dimension.

4. The method for evaluating the battery charging effect of an electrochemical energy storage system according to claim 2, wherein: The dynamic adjustment operation based on the feedback from the initial charging effect feature library to determine the evaluation model update information set includes: Based on the initial power replenishment effect feature library, a real-time effect visualization interface is provided for the monitoring system, and the feature space mapping effect is synchronously updated according to the monitoring data trajectory fed back by the system; Based on the monitoring data trajectory, identifying the potential effect characteristics of power replenishment through pattern cluster analysis, dynamically adjusting the sorting priority of the initial features, and generating a feature update plan; Analyze the feature update plan, evaluate the stability of the evaluation model, determine model stability feedback information, and feedback the corresponding feature adjustment suggestions to the operation and maintenance end, and receive the operation and maintenance confirmation instructions; According to the operation and maintenance confirmation instruction, an evaluation model update information set is determined.

5. The method for evaluating the battery charging effect of an electrochemical energy storage system according to claim 4, characterized in that: The analyzing the feature update scheme, evaluating the stability of the evaluation model, and determining model stability feedback information includes: According to the feature update scheme, the number of feature dimensions, evaluation confidence interval, historical evaluation result distribution and feature importance ranking of the model are extracted; According to the system significance of each feature, the model's evaluation error tolerance threshold, historical evaluation accuracy ratio and continuous evaluation consistency rate are determined; Based on the number of feature dimensions, the evaluation confidence interval, and the distribution of historical evaluation results, and according to the feature importance ranking and the evaluation error tolerance threshold, a statistical stability determination condition is established; Determining an effective time range for model evaluation based on the power replenishment cycle and the key monitoring phase of the system; Based on the historical evaluation accuracy ratio and the effective time range, and according to the continuous evaluation consistency rate and the feature importance ranking, a temporal stability determination condition is established; According to the statistical stability judgment condition and the temporal stability judgment condition, it is judged whether the statistical stability or temporal stability corresponding to the current feature update scheme meets the requirements, and a judgment conclusion is determined, which is used as the model stability feedback information.

6. The method for evaluating the battery charging effect of an electrochemical energy storage system according to claim 5, characterized in that: The statistical stability determination condition is constructed based on the number of feature dimensions, the evaluation confidence interval, and the distribution of historical evaluation results, according to the feature importance ranking and the evaluation error tolerance threshold, and specifically includes: Calculate the correlation between the number of feature dimensions and the evaluation confidence interval, analyze the matching degree between the distribution of historical evaluation results and the feature importance ranking, and set the critical range of statistical stability based on the evaluation error tolerance threshold; The time series stability determination conditions are constructed based on the historical evaluation accuracy ratio and the effective time range, according to the continuous evaluation consistency rate and the feature importance ranking, and specifically include: Calculate the fluctuation range of the historical evaluation accuracy ratio within the effective time range, analyze the temporal correlation between the continuous evaluation consistency rate and the feature importance ranking, and set the critical range of temporal stability.

7. The method for evaluating the battery charging effect of an electrochemical energy storage system according to claim 6, characterized in that: The analyzing the evaluation model update information set, performing hierarchical decomposition and classification statistics on the evaluation model update information set, and determining target model parameters includes: Update the information set according to the evaluation model, and count the key feature adjustment set and the auxiliary parameter correction set; According to the model update cycle set by the system, the battery charging cycles corresponding to different batteries in the evaluation model update information set are divided into several update batches to determine the update batch information set; Analyzing the key feature adjustment set based on the update batch information set, and determining the parameter redundancy corresponding to each feature in each update batch according to the parameter drift ratio corresponding to each feature in the key feature adjustment set in the historical update records; The target model parameters are constructed according to the auxiliary parameter correction set, the update batch information set and the parameter redundancy.

8. The method for evaluating the battery charging effect of an electrochemical energy storage system according to claim 7, characterized in that: The determining and outputting a supplementary power effect evaluation result based on the system parameter configuration information set and according to the target model parameters includes: Analyzing the target model parameters based on the system parameter configuration information set to determine a number of target system parameters corresponding to each characteristic adjustment item in the target model parameters; Extracting historical collaboration records corresponding to each target system parameter based on the system parameter configuration information set, thereby determining the historical application count, historical adjustment cost, historical evaluation accuracy ratio, and historical execution risk of each feature adjustment item in the current system under the corresponding target system parameter; Based on the historical application times, historical adjustment costs, historical evaluation accuracy ratio and historical execution risks of each feature adjustment item under the corresponding target system parameters, comprehensive cost optimization processing is performed on each feature adjustment item to determine the optimal system parameters corresponding to each feature adjustment item, thereby constructing and outputting the power replenishment effect evaluation results.

9. The method for evaluating the battery charging effect of an electrochemical energy storage system according to claim 8, characterized in that: The comprehensive cost optimization process is performed on each feature adjustment item based on the historical application times, the historical adjustment cost, the historical evaluation accuracy ratio, and the historical execution risk of each feature adjustment item under the corresponding target system parameters, specifically including: Calculate the weighted sum of the number of historical applications and the historical adjustment costs as the resource use cost indicator, and calculate the weighted sum of the historical evaluation accuracy ratio and the historical execution risk as the effect risk cost indicator; The resource use cost index is comprehensively compared with the effect risk cost index, and the system parameter with the lowest comprehensive cost is selected as the optimal parameter.

10. The method for evaluating the battery charging effect of an electrochemical energy storage system according to claim 4, characterized in that: The identification of potential power supplement effect characteristics through pattern cluster analysis includes: Dividing the monitoring data trajectory into time windows and extracting characteristic fluctuation amplitude and frequency within each time window; The initial classification standard of the clustering algorithm is set as the historical high-effect power replenishment feature library, and clustering iteration is performed based on the matching degree between the feature fluctuation amplitude and frequency and the initial classification standard; The reliability of the clustering results is evaluated by classification purity, and cluster groups whose matching degree with historical high-effect features exceeds the preset critical value are screened out, and their corresponding feature combinations are determined as potential power replenishment effect features.

Images