Method for evaluating temperature consistency of battery cells of energy storage power station

Abstract

The invention relates to the technical field of energy storage power stations, in particular to an energy storage power station cell temperature consistency evaluation method, which comprises the following steps of S1, screening out abnormal cell temperature from cell temperature data acquired in real time according to a preset threshold value, and recording an equipment fault position of the abnormal cell temperature through a positioning device; and S2, by correlating the operation state data of the refrigeration equipment, when the method is used, based on multi-dimensional and multi-working-condition temperature index fusion, the influence of the equipment state on temperature abnormity is fully considered, misjudgment caused by external refrigeration faults is avoided, and through the combination of the analytic hierarchy process and the fuzzy threshold value, the reliability of the system is improved. Accurate quantitative evaluation is carried out on the battery temperature consistency of the energy storage power station, the accuracy and reliability of anomaly detection are enhanced, equipment faults are alarmed through the abnormal temperatures, then the abnormal temperatures are filtered through a temperature and fault alarm linkage strategy, and the preprocessed temperatures can truly reflect the temperature consistency performance of the power station.

Description

Technical Field

[0001] This invention relates to the field of energy storage power station technology, and specifically to a method for evaluating the temperature consistency of battery cells in an energy storage power station. Background Technology

[0002] Energy storage technology is an effective way to solve the problem of efficient utilization of renewable energy. As a basic component of energy storage power stations, battery cells are connected in series and parallel to form a battery energy storage system. Inconsistency in the battery pack is a significant factor affecting the performance of energy storage power stations, reducing the usable capacity and cycle life of the battery pack. Temperature is a major factor causing inconsistency in battery performance and the battery pack itself. During high-power operation, due to internal chemical reactions and the Joule effect, a large amount of heat is generated inside the battery module and accumulates rapidly, causing localized temperature increases, increasing the temperature difference within the battery pack, deteriorating battery pack performance, and potentially inducing thermal runaway, leading to fires and explosions. Conversely, excessively low battery temperatures increase internal resistance and reduce the electrochemical reaction rate, resulting in a significant decrease in the discharge capacity of the battery pack. Therefore, to ensure the safe and stable operation of energy storage power stations, it is necessary to assess the temperature consistency of the battery cells, quantifying the degree of temperature consistency of individual cells, and providing data support for improving the safety performance of energy storage power stations.

[0003] Reference 1: A Consistency Evaluation Method for Lithium-ion Battery Packs Based on Cloud Charging Data. Taking lithium-ion battery packs as the research object, this paper proposes a method to evaluate temperature consistency by calculating the root mean square error and coefficient of variation of the extreme temperature curves during charging. The root mean square error between the highest and lowest temperature curves during the charging stage is defined as a parameter for evaluating temperature consistency, which quantitatively describes the temperature consistency of each battery pack and provides data support for analyzing battery health status. However, this technology does not further analyze the causes of extreme temperatures. Cell temperature is not only affected by charging and discharging conditions, but also by cooling devices such as fans and air conditioners. When cooling equipment fails, the temperature control strategy will fail, resulting in abnormally high cell temperatures. At this time, the root mean square error of the extreme values ​​is large, leading to a large deviation in the temperature consistency evaluation results, which cannot truly reflect the performance of the power station.

[0004] Reference 2: Correlation Analysis of Battery Pack Temperature and Voltage Consistency Based on Cloud Data. This paper proposes a method for analyzing battery pack temperature consistency based on hierarchical clustering by effectively segmenting the charge and discharge segments in the cell data. The inter-class distance range is used as the inconsistency evaluation index to analyze the trend of temperature inconsistency. However, this technique does not preprocess the cell temperature and does not consider that equipment failure will lead to the failure of thermal management strategy and the occurrence of abnormal cell temperature. In this case, the calculated cell temperature dispersion is large, and the final temperature consistency evaluation result has a large deviation and cannot accurately describe the temperature consistency.

[0005] In conclusion, developing a method for evaluating the temperature consistency of battery cells in energy storage power stations remains a critical issue that urgently needs to be addressed in the field of energy storage power station technology. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies that fail to further analyze the causes of extreme temperatures. Cell temperature is not only affected by charging and discharging conditions, but also by cooling devices such as fans and air conditioners. When cooling equipment malfunctions, the temperature control strategy fails, leading to abnormally high cell temperatures. In this case, the root mean square error of the extreme values ​​is large, resulting in significant deviations in temperature consistency assessments and failing to accurately reflect the power plant's performance. Furthermore, the invention does not preprocess cell temperatures and does not consider that equipment failures can cause thermal management strategies to fail, leading to abnormal cell temperatures. Consequently, the calculated cell temperatures exhibit large dispersion, resulting in significant deviations in the final temperature consistency assessment and an inability to accurately describe the temperature consistency problem.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] This invention provides a method for evaluating the consistency of cell temperature in an energy storage power station, comprising the following steps: S1, filtering out abnormal cell temperatures from real-time collected cell temperature data according to a preset threshold, and recording the location of equipment faults with abnormal cell temperatures using a positioning device;

[0009] S2. By associating the operating status data of the refrigeration equipment, alarms are generated for the location and cause of the equipment fault, and alarm results are output.

[0010] S3. A temperature and alarm linkage strategy is adopted to perform temperature and alarm linkage preprocessing, and the preprocessed cell temperature data is then filtered for operating conditions.

[0011] S4. Based on the working condition screening classification, it is further divided into 6 indicator factors, which are used as input feature quantities, i.e., the criterion layer, and a comparison matrix is ​​constructed by comparison.

[0012] S5. Combining the range of each of the aforementioned index factors with their importance weights relative to the battery temperature consistency parameter, a fuzzy threshold is used to construct a standard matrix.

[0013] Further, in step S1, the method for filtering abnormal cell temperatures from the real-time collected cell temperature data according to a preset threshold and recording the equipment fault location of the abnormal cell temperature using a positioning device is as follows:

[0014] A distributed thermal sensing network deployed on and inside the battery cell surface collects real-time battery cell temperature data streams. A weighted sliding window filtering algorithm is used for noise smoothing, with thresholds including the highest temperature and maximum temperature difference. Multiple feature vectors are constructed for each battery cell temperature data point. Abnormal battery cell temperatures are detected using a multi-dimensional anomaly detection method based on Mahalanobis distance, expressed as:

[0015]

[0016] In the formula, x i (t) represents the eigenvector of cell i at time t. This represents the actual measured temperature of cell i at time t. This indicates the rate of change of cell temperature over time. D represents the rate of change of cell temperature over time. M,i (t) represents the Mahalanobis distance of cell i at time t, μ x Σ represents the mean vector of all cell feature vectors in the statistical sample. x Reflecting the correlation and variance among various features, Standardization and weighting used to calculate Mahalanobis distance, reflecting features across various dimensions. This means that the column vector is transposed into a row vector. When the i-th cell exceeds the threshold and is judged as abnormal, the positioning device module is activated to mark the device location information. The location information adopts six-degree-of-freedom spatial positioning.

[0017] Furthermore, in step S2, the method for alarming the location and cause of the equipment fault based on the operating status data of the refrigeration equipment and outputting the alarm result is as follows:

[0018] The evolution of the refrigeration equipment's operating state data over time is modeled using a nonlinear dynamic system approach, and a fault location-cause matching function is constructed, with the expression:

[0019]

[0020] In the formula, ψ(P) i S(t) represents the characteristic parameter P of cell i. i The degree of correlation between P and the current device state S(t) is used to quantify the contribution of the cell's current state to the alarm. i Let p represent the feature parameter vector of cell i, and p represent the reference feature parameter vector. ‖P i -p‖ 2 σ represents the Euclidean distance between the cell's characteristic and the reference characteristic. 2 The variance parameter controls the width of the Gaussian kernel function, and exp(·) is the exponential function used to convert Euclidean distance into similarity weights. This is an indicator function that takes a value of 1 when the condition within the parentheses is met, and 0 otherwise. It is used to filter out abnormal temperature conditions. T represents the actual measured temperature of cell i at time t. ref This represents the set temperature reference value, and ∈ represents the allowable temperature deviation range.

[0021] Furthermore, in step S2, the method for alarming the location and cause of the equipment fault based on the operating status data of the refrigeration equipment and outputting the alarm result is as follows:

[0022] Based on Bayesian networks and graph reasoning, a causal graph model is constructed, and alarm results are output through a multi-scale joint alarm function. The expression is:

[0023]

[0024] In the formula, AET(t) represents the alarm result output at time t. This indicates that the summation is applied to all battery cells, N. cl It is the total number of battery cells. This indicates that all fault types are summed, M. ft It is the total number of fault categories, β ij ψ(P) represents the association weight of cell i with fault type j. i S(t) represents the characteristic parameter P of cell i. i The degree of correlation between the current device state S(t) and the current state of the battery cell is used to quantify the contribution of the current state of the cell to the alarm, P(F j |S(t)) represents the fault type F that occurs under the current device state S(t). j The conditional probability reflects the likelihood of a failure.

[0025] Furthermore, in step S3, the temperature and alarm linkage strategy is used for temperature and alarm linkage preprocessing, and the preprocessed cell temperature data is then filtered according to operating conditions as follows:

[0026] The temperature and alarm linkage strategy assumes that the battery cells are deployed in a hierarchical battery pack system, and the device hierarchy is abstracted as a directed acyclic graph. The preprocessing is as follows: if there is a cooling device alarm result at the device level where the abnormal battery cell temperature is located, then this abnormal battery cell temperature is filtered out; otherwise, the abnormal battery cell temperature is retained. Each alarm result is defined as a triplet, and the alarm filtering function is defined as follows:

[0027]

[0028] In the formula, Φ(t) is a Boolean coefficient representing the dynamic weighting coefficient of cell i at time t. This indicates the existence of the k-th alarm event. devk This indicates the device Anc associated with the k-th alarm event. i t represents the set of upstream devices of the device containing cell i. k ∈[t-Δt,t] indicates that the alarm event occurred within the time period Δt prior to the current time t. Φ(t)=0 indicates that an alarm has been generated by a device at the level to which the cell belongs, therefore the temperature of the cell is unreliable. Φ(t)=1 indicates that no alarm has been generated at the level to which the cell belongs within this time window, so temperature data can be retained. Δt represents the length of the time window before the current time. The preprocessed cell temperature data after linkage is constructed as follows:

[0029]

[0030] In the formula, Let Φ(t) represent the estimated temperature of cell i at time t, and let Φ(t) be a Boolean coefficient representing the dynamic weighting coefficient of cell i at time t. This represents the actual measured temperature of cell i at time t.

[0031] Furthermore, in step S3, the temperature and alarm linkage strategy is used for temperature and alarm linkage preprocessing, and the preprocessed cell temperature data is then filtered according to operating conditions as follows:

[0032] Based on the cell temperature data, the operating conditions are screened, categorized into charging, discharging, and resting states. An operating condition state function is constructed, and a three-state hidden Markov model is introduced to smoothly determine the operating condition sequence. The Viterbi algorithm is then used to infer the optimal operating condition sequence. The highest temperature and maximum temperature difference for different operating conditions are statistically analyzed. The expression is:

[0033]

[0034] In the formula, This represents the highest temperature under operating condition c. This means finding the cell with the highest temperature among all cell indices i that satisfy operating condition c. This represents the set of all cell numbers corresponding to operating condition c. t represents the smoothed temperature value of the i-th cell. i This is the current time point of the battery cell i. This represents the maximum temperature difference under operating condition c. This represents the set of temperature data for all battery cells collected under operating condition c. This represents the maximum temperature value in the set of cell temperatures under operating condition c. This represents the minimum temperature value in the set of cell temperatures under operating condition c.

[0035] Furthermore, in step S4, the working condition screening classification is further divided into 6 index factors, which serve as input features, i.e., the criterion layer, and the comparison matrix is ​​constructed by comparing them as follows:

[0036] Based on the aforementioned operating condition screening, three operating condition categories have been obtained. For each operating condition, two key thermal control feature values ​​are extracted, namely the highest temperature and the maximum temperature difference. The six index factors are the highest charging temperature, the maximum charging temperature difference, the highest discharging temperature, the maximum discharging temperature difference, the highest standing temperature, and the maximum standing temperature difference, which are used as input feature quantities, i.e., the criterion layer. A comparison matrix A is constructed based on engineering experience, historical risk probability, and expert knowledge.

[0037] Furthermore, in step S4, the working condition screening classification is further divided into 6 index factors, which serve as input features, i.e., the criterion layer, and the comparison matrix is ​​constructed by comparing them as follows:

[0038] Calculate the maximum eigenvalue and the normalized eigenvector of the comparison matrix. The maximum eigenvalue is defined as w = [w1, w2, ..., w6], which is the principal eigenvector of the comparison matrix. Here, w represents the weight vector, w1 represents the weight of the highest charging temperature, w2 represents the weight of the maximum charging temperature difference, w3 represents the weight of the highest discharging temperature, w4 represents the weight of the maximum discharging temperature difference, w5 represents the weight of the highest resting temperature, and w6 represents the weight of the maximum resting temperature difference. The corresponding maximum eigenvalue λ... max The power method is used to approximate the solution and estimate the largest eigenvalue. The expression is:

[0039]

[0040] In the formula, λ max Let represent the largest eigenvalue, w represent the eigenvector, and A represent the pairwise comparison matrix of the relative importance of each indicator. The transpose of the eigenvector represents the eigenvector normalized by the comparison matrix. Normalizing the eigenvector yields the weight vector, expressed as:

[0041]

[0042] In the formula, w A Let w represent the normalized weight vector and w represent the feature vector. This indicates that the weights of the six indicators are summed for normalization, and a consistency check is performed on the comparison matrix. The expression is:

[0043]

[0044] In the formula, CR represents the consistency ratio, CI represents the consistency index, and RI represents the random consistency index. When CR < 0.1, the consistency of the comparison matrix is ​​considered acceptable.

[0045] Further, in step S5, the method for constructing a standard matrix by combining the range of each index factor with its importance weight relative to the battery temperature consistency parameter using a fuzzy threshold is as follows:

[0046] Define the range of each of the aforementioned indicator factors, construct fuzzy membership functions, and form a set of fuzzy indicator functions, expressed as:

[0047]

[0048] In the formula, η y This represents the fuzzy evaluation values ​​of six temperature indicators. y represents the transposed list of fuzzy memberships. i The i-th temperature index value, Indicates the relationship between y i The constructed fuzzy membership function, the constructed standard matrix, and the establishment of a standard matrix G for each of the index factors using fuzzy triangular numbers. AC .

[0049] Further, in step S5, the method is as follows:

[0050] The highest temperature and maximum temperature difference of each of the aforementioned index factors are combined with the standard matrix to form the index matrix G′. AC Calculate the eigenvector and the index matrix, and use the analytic hierarchy process (AHP) to calculate the temperature consistency evaluation coefficient, expressed as:

[0051]

[0052] In the formula, C tp Indicates the temperature consistency evaluation coefficient. Let R represent the transpose of the weight vector, and let R represent the membership vector. It is the accumulation symbol that adds from the 1st term to the 6th term, w i y represents the weight of the i-th indicator. i The i-th temperature index value, Represents the i-th temperature index y i The fuzzy membership value is output as a C. tp The consistency coefficient C for the scoring results ∈[0,1] tp The closer to 0, the more severe the inconsistency.

[0053] Beneficial effects

[0054] Compared with known public technologies, the technical solution provided by this invention has the following beneficial effects:

[0055] When in use, this invention integrates temperature indicators from multiple dimensions and operating conditions, fully considering the impact of equipment status on temperature anomalies, avoiding misjudgments due to external cooling failures. By combining the analytic hierarchy process (AHP) and fuzzy thresholding, it achieves accurate quantitative evaluation of battery temperature consistency in energy storage power stations, enhancing the accuracy and reliability of anomaly detection. It alarms for equipment failures through abnormal temperatures, and then filters abnormal temperatures using a temperature-fault alarm linkage strategy. The pre-processed temperature can truly reflect the temperature consistency performance of the power station. Attached Figure Description

[0056] Figure 1 This is a flowchart of a method for evaluating the temperature consistency of battery cells in an energy storage power station according to the present invention. Detailed Implementation

[0057] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0058] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but includes other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0059] The present invention will now be described in further detail with reference to the accompanying drawings:

[0060] Example:

[0061] like Figure 1 As shown, the present invention provides a method for evaluating the temperature consistency of battery cells in an energy storage power station, comprising the following steps:

[0062] S1. Based on a preset threshold, abnormal cell temperatures are filtered out from the real-time collected cell temperature data, and the location of the equipment fault with the abnormal cell temperature is recorded by a positioning device.

[0063] S2. By associating the operating status data of the refrigeration equipment, alarms are generated for the location and cause of the equipment fault, and alarm results are output.

[0064] S3. A temperature and alarm linkage strategy is adopted to perform temperature and alarm linkage preprocessing, and the preprocessed cell temperature data is then filtered for operating conditions.

[0065] S4. Based on the working condition screening classification, it is further divided into 6 indicator factors, which are used as input feature quantities, i.e., the criterion layer, and a comparison matrix is ​​constructed by comparison.

[0066] S5. Combining the range of each of the aforementioned index factors with their importance weights relative to obtaining the battery temperature consistency parameter, a fuzzy threshold is used to construct a standard matrix;

[0067] Taking the data of an energy storage power station containing 16 battery clusters in each container as an example, the container number and battery cluster number of the abnormal cell temperature are first found based on the threshold of the highest temperature of 40℃ and the maximum temperature difference of 7℃. Combined with the operating status data of air conditioner and fan, it is checked whether there is a malfunction of the refrigeration equipment. If there is a malfunction, the cause of the malfunction is recorded and the battery cluster is marked. Then the cell temperature of the marked battery cluster is removed from the cell temperature data.

[0068] The preprocessed temperature data was filtered according to operating conditions, divided into charging, discharging, and resting states, and the highest charging temperature Tmax was calculated. char Maximum temperature difference during charging (Tdiff) char Maximum discharge temperature Tmax dischar Maximum discharge temperature difference Tdiff dischar Maximum static temperature Tmax stand and the maximum temperature difference Tdiff when stationary stand By conducting in-depth analysis of rich historical data, the interrelationships and degree of influence among various indicators are identified, and the importance of these six indicator factors is compared to construct a comparison matrix A.

[0069]

[0070] a ij This is the result of comparing the importance of indicator i and indicator j. Calculate the largest eigenvalue of the comparison matrix and its normalized eigenvector w. A .

[0071] Based on the existing classification of single-cell battery consistency status into four types: healthy, sub-healthy, severe, and poor, and considering the statistical characteristics of the temperature consistency index threshold of single-cell batteries in energy storage power stations, the parameters of the battery temperature consistency evaluation index for energy storage power stations are shown in Table 1.

[0072] Table 1. Evaluation Indicators for Battery Temperature Consistency in Energy Storage Power Stations

[0073]

[0074] By combining the ranges of each indicator with their importance weights relative to obtaining battery temperature consistency parameters, a fuzzy threshold is used to construct a standard matrix G. AC The highest temperature and maximum temperature difference obtained under different operating conditions are combined with the standard matrix to form the index matrix G′. AC Calculate the eigenvector and the index matrix, and use the analytic hierarchy process (AHP) to calculate the temperature consistency evaluation coefficient, expressed as:

[0075]

[0076] In the formula, C tp Indicates the temperature consistency evaluation coefficient. Let R represent the transpose of the weight vector, and let R represent the membership vector. It is the accumulation symbol that adds from the 1st term to the 6th term, w i y represents the weight of the i-th indicator. i The i-th temperature index value, Represents the i-th temperature index y i The fuzzy membership value is output as a C. tp The consistency coefficient C for the scoring results ∈[0,1] tp The closer to 0, the more serious the inconsistency, and at this point, it is necessary to pay close attention to the operation of the power plant.

[0077] In this embodiment, based on the fusion of multi-dimensional and multi-condition temperature indicators, the influence of equipment status on temperature anomalies is fully considered to avoid misjudgment due to external cooling failures. By combining the analytic hierarchy process (AHP) and fuzzy thresholding, accurate quantitative evaluation of battery temperature consistency in energy storage power stations is achieved, enhancing the accuracy and reliability of anomaly detection. Abnormal temperatures trigger equipment fault alarms, and the temperature-fault alarm linkage strategy filters abnormal temperatures. The pre-processed temperature can truly reflect the temperature consistency performance of the power station. By quantitatively reflecting the temperature consistency of individual batteries, theoretical support is provided for the operation and maintenance analysis of energy storage power stations, thereby improving the value of energy storage power station operation analysis.

[0078] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating temperature consistency of an energy storage power station cell, characterized in that, The method comprises the following steps: S1, according to the preset threshold value, the abnormal battery temperature is screened out from the real-time collected battery temperature data, and the equipment failure position of the abnormal battery temperature is recorded through the positioning device; S2, the device failure position and the cause are alarmed through the running state data of the refrigeration equipment, and the alarm result is output; S3, a temperature and alarm linkage strategy is adopted for temperature and alarm linkage preprocessing, and the battery temperature data after preprocessing is subjected to working condition screening; S4, according to the working condition screening classification, six index factors are further divided into six index factors as input characteristic quantities, that is, the criterion layer, and a comparison matrix is constructed by comparison; S5, a fuzzy threshold value is adopted in combination with the range of each index factor and the importance weight relative to the obtained battery temperature consistency parameter to construct a standard matrix.

2. The energy storage plant cell temperature uniformity evaluation method of claim 1, wherein, In step S1, according to the preset threshold value, the abnormal battery temperature is screened out from the real-time collected battery temperature data, and the equipment failure position of the abnormal battery temperature is recorded through the positioning device: By deploying a distributed thermal sensor network on the surface and inside of the battery, real-time battery temperature data flow is collected, and a weighted sliding window filtering algorithm is used for noise smoothing. The threshold value includes the maximum temperature and the maximum temperature difference. A multi-feature vector is constructed for each battery temperature data. The abnormal battery temperature is detected based on the Mahalanobis distance discrimination method, and the expression is: where x i (t) denotes the feature vector of the cell i at time t, denotes the actual measured temperature of the cell i at time t, denotes the rate of change of the cell temperature over time, denotes the rate of change of the cell temperature rate over time, D M,i (t) denotes the Mahalanobis distance of the cell i at time t, μ x denotes the mean vector of all cell feature vectors in the statistical sample, Σ x reflects the correlation and variance between each feature, for calculating the Mahalanobis distance embodies the standardization and weighting of each dimensional feature, denotes the transposition of the column vector to the row vector, when the i-th cell exceeds the threshold value and is judged to be abnormal, the positioning device module is activated to mark the equipment position information, and the position information adopts six-degree-of-freedom space positioning.

3. The energy storage plant cell temperature uniformity evaluation method of claim 2, wherein, In step S2, the device failure position and the cause are alarmed through the running state data of the refrigeration equipment, and the alarm result is output: The evolution of the running state data of the refrigeration equipment with time is modeled by introducing a nonlinear dynamic system, and a device failure position-cause matching function is constructed, and the expression is: where ψ(P i , S(t)) represents the degree of association between the characteristic parameter P i of the battery cell i and the current device state S(t) and is used to quantify the contribution of the current state of the battery cell to the alarm, P i represents the characteristic parameter vector of the battery cell i, p represents the reference characteristic parameter vector, and ‖P i -p‖ 2 represents the Euclidean distance between the characteristics of the battery cell and the reference characteristics, σ 2 is a variance parameter representing the width of the Gaussian kernel function, and exp(·) is an exponential function used to convert the Euclidean distance into a similarity weight, is an indicator function that takes a value of 1 when the condition in the parentheses is satisfied and 0 otherwise and is used to filter abnormal temperature cases, represents the actual measured temperature of the battery cell i at time t, T ref represents a set temperature reference value, and ∈ represents an allowed temperature deviation range.

4. The energy storage plant cell temperature uniformity evaluation method of claim 3, wherein, In step S2, the device failure position and the cause are alarmed through the running state data of the refrigeration equipment, and the alarm result is output: Based on Bayesian network and graph reasoning, a causal graph model is constructed, and the alarm result is output through a multi-scale joint alarm function, and the expression is: where AET(t) represents the alarm result output at time t, represents the accumulation of all cells, N cl is the total number of cells, represents the accumulation of all fault types, M ft is the total number of fault categories, β ij represents the association weight of cell i to fault type j, ψ(P i , S(t)) represents the association degree between the characteristic parameter P i of cell i and the current device state S(t) for quantifying the contribution of the current state of the cell to the alarm, P(F j | S(t)) represents the conditional probability of the occurrence of fault type F j under the current device state S(t) reflecting the possibility of the fault.

5. The energy storage plant cell temperature uniformity evaluation method of claim 4, wherein, In step S3, a temperature and alarm linkage strategy is adopted for temperature and alarm linkage preprocessing, and the battery temperature data after preprocessing is subjected to working condition screening: The temperature and alarm linkage strategy is set to deploy the battery in a layered battery pack system. The device hierarchical structure is abstracted as a directed acyclic graph. The preprocessing is as follows: if there is a refrigeration equipment alarm result in the device hierarchy where the abnormal battery temperature is located, the abnormal battery temperature is filtered out, otherwise, the abnormal battery temperature is retained. Each alarm result is a triple, and the alarm filtering function is defined as: Φ(t) is a Boolean coefficient representing the dynamic weight coefficient of the battery cell i at time t, represents the existence of the kth alarm event dev k represents the device Anc associated with the kth alarm event i represents the upper device set of the device where the battery cell i is located, t k ∈[t-Δt,t] represents that the alarm event occurs within the time window of Δt before the current time t, Φ(t)=0 represents that the battery cell belongs to a device that has generated an alarm, so the temperature of the battery cell is unreliable, Φ(t)=1 represents that the battery cell is in a device level without any alarm within the time window, and Δt represents the length of the time window before the current time. The battery cell temperature data after the linkage preprocessing is constructed as follows: wherein, represents the estimated temperature of the battery cell i at time t, and Φ(t) is a Boolean coefficient representing the dynamic weight coefficient of the battery cell i at time t, represents the actual measured temperature of the battery cell i at time t.

6. The energy storage plant cell temperature uniformity evaluation method of claim 5, wherein, In step S3, a temperature and alarm linkage strategy is adopted for temperature and alarm linkage preprocessing, and the battery temperature data after preprocessing is subjected to working condition screening: According to the battery temperature data, the working condition screening is divided into charging, discharging and standing states, a working condition state function is constructed, a three-state hidden Markov model is introduced, a working condition sequence is smoothed, and the working condition sequence is optimally inferred through the Viterbi algorithm. The highest temperature and the maximum temperature difference of different working conditions are counted, and the expression is: In the formula, represents the highest temperature under the working condition c, represents finding the battery cell with the maximum temperature among all battery cell indexes i that meet the working condition c, represents the set of all battery cell numbers corresponding to the working condition c, represents the value after smoothing the temperature of the i-th battery cell, t i is the current time point of the battery cell i, represents the maximum temperature difference under the working condition c, represents the set of temperature data of all battery cells collected under the working condition c, represents the maximum temperature value in the set of battery cell temperatures under the working condition c, represents the minimum temperature value in the set of battery cell temperatures under the working condition c.

7. The energy storage plant cell temperature uniformity evaluation method of claim 6, wherein, In step S4, the working condition screening classification is further divided into 6 index factors, which serve as input features, i.e., the criterion layer. The method for constructing a comparison matrix by comparison is as follows: Based on the aforementioned operating condition screening, three operating condition categories have been obtained. For each operating condition, two key thermal control feature values ​​are extracted, namely the highest temperature and the maximum temperature difference. The six index factors are the highest charging temperature, the maximum charging temperature difference, the highest discharging temperature, the maximum discharging temperature difference, the highest standing temperature, and the maximum standing temperature difference, which are used as input feature quantities, i.e., the criterion layer. A comparison matrix A is constructed based on engineering experience, historical risk probability, and expert knowledge.

8. The energy storage plant cell temperature uniformity evaluation method of claim 7, wherein, In step S4, the working condition screening classification is further divided into 6 index factors, which serve as input features, i.e., the criterion layer. The method for constructing a comparison matrix by comparison is as follows: computing the largest eigenvalue of the comparison matrix and the normalized eigenvector of the comparison matrix, the largest eigenvalue, setting w = [w1, w2, …, w6] as the principal eigenvector of the comparison matrix, wherein w represents a weight vector, w1 represents the weight of the highest temperature of charging, w2 represents the weight of the maximum temperature difference of charging, w3 represents the weight of the highest temperature of discharging, w4 represents the weight of the maximum temperature difference of discharging, w5 represents the weight of the highest temperature of standing, and w6 represents the weight of the maximum temperature difference of standing, corresponding to the largest eigenvalue λ max , using the power method to approximate the solution, estimating the largest eigenvalue, expression: In the formula, λ max denotes the maximum eigenvalue, w denotes the eigenvector, A denotes the pair-wise comparison matrix of the relative importance between each index, denotes the transpose of the eigenvector, the eigenvector of the comparison matrix is normalized, the eigenvector is normalized to obtain the weight vector, and the expression is: In the formula, w A represents a normalized weight vector, w represents a feature vector, represents that the weight values of the 6 indexes are added for normalization processing, and the consistency of the comparison matrix is checked, expression: In the formula, CR represents the consistency ratio, CI represents the consistency index, and RI represents the random consistency index. When CR < 0.1, the consistency of the comparison matrix is ​​considered acceptable.

9. The energy storage plant cell temperature uniformity evaluation method of claim 8, wherein, In step S5, the method for constructing a standard matrix by combining the range of each index factor with its importance weight relative to the battery temperature consistency parameter using a fuzzy threshold is as follows: Define the range of each of the aforementioned indicator factors, construct fuzzy membership functions, and form a set of fuzzy indicator functions, expressed as: wherein η y represents the fuzzy evaluation value of 6 temperature indexes, y represents the transposed fuzzy membership list, y i the ith temperature index value, y represents the transposed fuzzy membership list, y i the constructed fuzzy membership function, the constructed standard matrix, the standard matrix G is established using fuzzy triangular numbers for each of the index factors AC .

10. The energy storage plant cell temperature uniformity evaluation method of claim 8, wherein, In step S5, the method is as follows: The highest temperature and the maximum temperature difference of each index factor and the standard matrix form an index matrix G AC The characteristic vector and the index matrix are calculated, and the temperature consistency evaluation coefficient is calculated by using the analytic hierarchy process, and the expression is: where C tp represents the temperature consistency evaluation coefficient, represents the transpose of the weight vector, R represents the index membership degree vector, is the cumulative sign from the first item to the sixth item, w i represents the weight of the i-th index, y i the i-th temperature index value, represents the fuzzy membership value of the i-th temperature index y i , outputs a C tp ∈[0,1] score result, the consistency coefficient C tp closer to 0, the more serious the inconsistency.

Images