Battery fault detection method and device based on energy storage power station

Abstract

The invention provides a fault detection method and device for a battery of an energy storage power station, and relates to the technical field of fault detection of the battery, and the method comprises the steps: setting a time window, sampling and preprocessing battery parameters of a battery cluster and a corresponding single cell in the battery cluster in the time window, and constructing a battery output state coefficient; based on the historical battery parameters, training a LightGBM model, and inputting the battery output state coefficient into the trained LightGBM model to obtain a fault probability; and based on the fault probability, adopting a dynamic threshold adaptive algorithm to realize threshold exceeding alarm. The method comprises the following steps of: sampling a battery cluster and a single battery cell in a time window, preprocessing sampling point data, constructing an output state coefficient, and respectively inputting an output state coefficient of the battery cluster and an output state coefficient of the single battery cell into a Light GBM model to obtain a battery cluster fault probability and a single battery cell fault probability; and meanwhile, dynamic threshold alarm is realized based on the fault probability of the battery clusters and the single battery cells.

Description

Technical Field

[0001] The present invention relates to the technical field of battery fault detection methods and devices, and particularly to a battery fault detection method and device based on an energy storage power station. Background Art

[0002] Traditional energy storage power stations mainly include battery clusters and single cells. Among them, each battery cluster contains multiple single cells. Due to the high power consumption and long service life of energy storage power stations, faults are likely to occur. Therefore, it is necessary to evaluate the states of battery clusters and single cells separately. The traditional battery fault detection methods and devices mainly rely on threshold detection and traditional machine learning, and are gradually developing towards multi-parameter fusion, unsupervised learning, and dynamic adaptive models, aiming to solve the problems of early detection, complex fault identification, and real-time performance.

[0003] In the prior art, the publication number CN117949828A discloses a method for determining the fault condition of a battery based on an unsupervised isolation forest algorithm. By obtaining the battery SOC from a preset time to the current time and performing dimensional division, the current SOC of each dimension is obtained, and the current SOC is input into the isolation forest algorithm model to further determine whether the battery has a fault. However, this method has low segmentation efficiency for multi-dimensional time series data. Battery data usually includes multi-dimensional parameters such as voltage, current, temperature, and internal resistance. It is difficult for the isolation forest to effectively capture the complex relationships between multi-variables during random feature segmentation, resulting in a decrease in the accuracy of anomaly detection. Faults caused by joint anomalies of multi-parameters may be missed, and the false alarm rate is relatively high. Therefore, it is necessary to sample the battery data in time windows, obtain the sampled data of battery parameters, and organize the multi-variable data into vectors that can be input into the model to effectively avoid the difficulty in distinguishing complex relationships between multi-variables. Moreover, the model detection should be used separately for battery clusters and single cells to improve the detection accuracy and reduce the false alarm rate.

[0004] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and therefore it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide a battery fault detection method and device based on an energy storage power station to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] A battery fault detection method based on an energy storage power station, the specific steps include:

[0008] S1: Set a time window, sample the battery parameters of the battery cluster and the corresponding individual battery cells inside it within the time window, and perform preprocessing to construct battery output state coefficients, where the battery output state coefficients include the output state coefficient of the battery cluster and the output state coefficients of the individual battery cells.

[0009] S2: Based on historical battery parameters, train a LightGBM model, and input the battery output state coefficients into the trained LightGBM model to obtain the failure probability of the battery cluster and the failure probabilities of each individual battery cell inside.

[0010] S3: Based on the failure probability of the battery cluster and the failure probabilities of the individual battery cells, count the number of failed individual battery cells, set an alarm threshold, and implement dynamic alarm for the battery cluster.

[0011] Further, the time window is [T0 - T, T0], where T0 represents the time stamp of the current moment and T represents the length of the time window.

[0012] When collecting battery parameters, collect the battery parameters of the battery cluster at a time interval of 10k, and collect the battery parameters of the individual battery cells at a time interval of k.

[0013] Further, the battery parameters of the individual battery cells include individual voltage, individual current, individual temperature, individual battery internal resistance, and individual self-discharge rate, and the battery parameters of the battery cluster include total voltage, total current, and temperature difference, where the temperature difference is the difference between the highest individual temperature and the lowest individual temperature of the individual battery cells inside the battery cluster.

[0014] Further, the steps of the preprocessing are as follows: For the battery parameters of the battery cluster, set the window of median filtering to a + 1, perform median filtering on the voltage, total current, and temperature difference data of the battery cluster, and perform normalization processing on the voltage, total current, and temperature difference data of the battery cluster after median filtering respectively according to the data type;

[0015] For the battery parameters of the individual battery cells, set the window of median filtering to 2a + 1, perform median filtering on the individual voltage, individual current, individual temperature, individual battery internal resistance, and individual self-discharge rate, and perform normalization processing on the individual voltage, individual current, individual temperature, individual battery internal resistance, and individual self-discharge rate after median filtering respectively according to the data type.

[0016] Further, the logic for constructing the output state coefficient of the battery cluster is as follows:

[0017] Based on the preprocessed battery parameters of the battery cluster, construct a feature matrix of the battery cluster, and the specific formula is:

[0018]

[0019] Among them, X represents the characteristic matrix of the battery cluster, V″ c (j), I″ c (j), and I″ c (j) respectively represent the total voltage, total current, and temperature difference data after preprocessing at the j-th sampling point. j and N respectively represent the number and total number of sampling points when collecting the battery parameters of the battery cluster;

[0020] Based on the total voltage, total current, and temperature difference data after preprocessing at each sampling point, the battery cluster state index of each sampling point is generated, and the output state coefficient of each battery cluster is formed. The formula is as follows:

[0021]

[0022] Among them, Xcl represents the output state coefficient of the battery cluster, and x j represents the battery cluster state index of the j-th sampling point;

[0023]

[0024] Among them, T0 represents the set temperature difference threshold.

[0025] Furthermore, the logic for constructing the output state coefficient of a single cell is as follows:

[0026] Based on the battery parameters of the single cell after preprocessing, the characteristic matrix of the single cell is constructed. The specific formula is

[0027]

[0028] Among them, Xce represents the characteristic matrix of the single cell; V″ n (i) and I″ n (i) respectively represent the single cell voltage value and single cell current value after preprocessing at the i-th sampling point of the single cell; T″ n (i), R″ n (i), and S″ n (i) respectively represent the single cell temperature, single cell internal resistance, and single cell self-discharge rate after preprocessing at the i-th sampling point of the single cell; i and M respectively represent the number and total number of sampling points when collecting the battery parameters of the single cell;

[0029] Based on the single cell voltage, single cell current, single cell temperature, single cell battery internal resistance, and single cell self-discharge rate data after preprocessing at each sampling point, the cell state index of each sampling point is generated, and the output state coefficient of each single cell is formed. The formula is as follows:

[0030]

[0031] Among them, Y cell represents the output state coefficient of the single cell, and y i represents the cell state index of the i-th sampling point, and:

[0032]

[0033] Among them, C represents the internal resistance adjustment constant of the single cell; T op represents the optimal operating temperature of the cell.

[0034] Furthermore, the LightGBM model includes a battery cluster model and a single cell model. The historical battery parameters include normal battery parameter samples and faulty battery parameter samples. The normal battery parameter samples include the output state coefficient samples of normal battery clusters and the output state coefficient samples of normal single cells. The faulty battery parameter samples include the output state coefficient samples of faulty battery clusters and the output state coefficient samples of faulty single cells;

[0035] When training the LightGBM model, the output state coefficient samples of the battery cluster are used as the feature inputs of the battery cluster model, and whether the battery cluster is faulty is used as the feature output to train the battery cluster model;

[0036] The output state coefficient samples of the single cell are used as the feature inputs of the single cell model, and whether the single cell is faulty is used as the feature output of the single cell model to train the single cell model;

[0037] The output state coefficients of the battery cluster are input into the trained battery cluster model to obtain the fault probability of each battery cluster. The output state coefficients of each corresponding single cell of the battery cluster are input into the trained single cell model to obtain the fault probability of each single cell.

[0038] Furthermore, the logic for realizing the dynamic alarm of the battery cluster is as follows:

[0039] According to the battery cluster fault probability and the fault probability of each single cell inside the battery cluster, the overall fault probability of the battery cluster is obtained. The formula is as follows:

[0040]

[0041] Among them, P total represents the overall fault probability of the battery, represents the fault probability of the q-th single cell inside the battery cluster, n represents the total number of single cells inside the battery cluster; q represents the single cell index number; p c represents the battery cluster fault probability;

[0042] Count the number of single battery cells with a failure probability greater than 0.5 and label it as A, dynamically set the alarm threshold, and compare the overall battery failure probability with the alarm threshold. If a battery cluster failure alarm signal is sent, where T represents the set threshold.

[0043] The present invention also provides a failure detection device for the batteries of an energy storage power station. The detection device is used to execute the above detection method and includes:

[0044] A state evaluation module for setting a time window, sampling the battery parameters of the battery cluster and the corresponding single battery cells inside it within the time window, performing preprocessing, and constructing a battery output state coefficient, where the battery output state coefficient includes the output state coefficient of the battery cluster and the output state coefficient of the single battery cells;

[0045] A failure prediction module that trains a LightGBM model based on historical battery parameters and inputs the battery output state coefficient into the trained LightGBM model to obtain the failure probability of the battery cluster and the failure probabilities of each single battery cell inside;

[0046] A failure warning module for counting the number of single battery cell failures based on the failure probability of the battery cluster and the failure probabilities of single battery cells, setting an alarm threshold, and realizing dynamic alarm for the battery cluster.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] By introducing a time window, sampling the battery cluster and single battery cells within the time window, performing median filtering and normalization processing on the sampled point data, effectively eliminating noise interference, retaining key failure features, constructing an output state coefficient, inputting the output state coefficient of the battery cluster and the output state coefficient of the single battery cells into the LightGBM model respectively to obtain the battery cluster failure probability and the single battery cell failure probability, avoiding cross-level interference, effectively improving the failure location accuracy, and at the same time, based on the failure probabilities of the battery cluster and single battery cells, adopting a dynamic threshold method to realize dynamic threshold alarm. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a schematic diagram of the overall method flow of the present invention;

[0050] Figure 2 is a schematic diagram of the overall device structure of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in combination with specific embodiments.

[0052] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the ordinary meanings understood by those of ordinary skill in the field to which the present invention pertains. The "first", "second" and similar terms used in the present invention do not denote any order, quantity or importance, but are only used to distinguish different components. Words such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. Words such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Upper", "lower", "left", "right", etc. are only used to represent relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0053] Embodiment:

[0054] Please refer to Figure 1 , the present invention provides a technical solution:

[0055] A coal mine supervision method based on video images, the specific steps include:

[0056] S1: Set a time window, sample the battery parameters of the battery cluster and the corresponding single battery cells inside it within the time window, and perform preprocessing to construct a battery output state coefficient, where the battery output state coefficient includes the output state coefficient of the battery cluster and the output state coefficient of the single battery cell.

[0057] The time window is [T0 - T, T0], where T0 represents the time stamp of the current moment, T represents the length of the time window. When collecting the battery parameters, the battery parameters of the battery cluster are collected at a time interval of 10k, and the battery parameters of the single battery cell are collected at a time interval of k; where k ∈ N + .

[0058] The battery parameters of the single battery cell include single battery voltage, single battery current, single battery temperature, single battery internal resistance and single battery self-discharge rate. The battery parameters of the battery cluster include total voltage, total current and temperature difference. The temperature difference is the difference between the highest single battery temperature and the lowest single battery temperature inside the battery cluster.

[0059] Collecting the above battery parameters is because the battery parameters can comprehensively represent the operating state of the battery and are highly sensitive to fault characteristics. For a single battery cell, the single battery voltage, single battery current, single battery temperature, single battery internal resistance and single battery self-discharge rate can accurately locate the faulty battery cell; for the battery cluster, the total voltage, total current and temperature difference can judge the system-level fault risk; and, the parameters of the single battery cell change relatively fast and require high-frequency sampling, while the parameters of the battery cluster change relatively slow and require low-frequency sampling;

[0060] The steps of the preprocessing are as follows: for the battery parameters of the battery cluster, set the window of median filtering to a + 1, perform median filtering on the voltage, total current, and temperature difference data of the battery cluster, and perform normalization processing on the voltage, total current, and temperature difference data of the battery cluster after median filtering respectively according to the data type;

[0061] Since the sampling of the total voltage, total current, and temperature difference of the battery cluster is low-frequency sampling, a relatively small window of median filtering is selected;

[0062] Calibrate the number of sampling points of the battery cluster parameters within the time window as N, N = T / 10k - 1, set the filtering window width to a + 1, and perform median filtering on the battery cluster parameters of each sampling point j:

[0063] V′ ic (j) = median({V ic (j - a), V ic (j - a + 1), …, V ic (j)});

[0064] I′ ic (j) = median({I ic (j - a), I ic (j - a + 1), …, I ic (j)});

[0065] ΔT′ ic (j) = median({ΔT ic (j - a), ΔT ic (j - a + 1), …, ΔT ic (j)});

[0066] where j ∈ [1, N]; V′ ic (j) represents the total voltage value after median filtering at the jth sampling point of the ith battery cluster; I′ ic (j) represents the total current value after median filtering at the jth sampling point of the ith battery cluster; ΔT′ ic (j) represents the temperature difference value after median filtering at the jth sampling point of the ith battery cluster; V ic (j) represents the original total voltage at the jth sampling point of the ith battery cluster; I ic (j) represents the original total current at the jth sampling point of the ith battery cluster; ΔT ic (j) represents the original temperature difference at the jth sampling point of the ith battery cluster;

[0067] Median filtering is used to process the total voltage, total current, and temperature difference of the battery cluster, which can filter out the jumps in cluster-level parameters caused by external interference, such as sudden drops in total voltage and abnormal temperature differences. Moreover, median filtering only eliminates isolated noise points and retains the true changes in continuous data. Based on the filtered parameter sequence, linear normalization is performed according to the maximum and minimum values of each parameter within the time window:

[0068]

[0069] m represents the sampling point index variable; V″ ic (j) represents the normalized total voltage value of the j-th sampling point of the i-th battery cluster; I″ ic (j) represents the normalized total current value of the j-th sampling point of the i-th battery cluster; ΔT″ c (j) represents the normalized temperature difference value of the j-th sampling point of the i-th battery cluster; V′ ic (m) represents the filtered total voltage value of the m-th sampling point of the i-th battery cluster within the time window; I′ ic (m) represents the filtered total current value of the m-th sampling point of the i-th battery cluster within the time window; ΔT′ ic (m) represents the filtered temperature difference value of the m-th sampling point of the i-th battery cluster within the time window;

[0070] Normalizing the filtered total voltage, total current, and temperature difference is to adjust the dynamic range of the filtered data, making the input data closer to the real situation and reducing the false alarm rate of the battery cluster model;

[0071] For the battery parameters of single cells, set the window of median filtering to 2a + 1, perform median filtering on the single cell voltage, single cell current, single cell temperature, single cell battery internal resistance, and single cell self-discharge rate, and perform normalization processing on the single cell voltage, single cell current, single cell temperature, single cell battery internal resistance, and single cell self-discharge rate after median filtering respectively according to the data type;

[0072] Since the sampling of the single cell voltage, single cell current, single cell temperature, single cell battery internal resistance, and single cell self-discharge rate of single cells is high-frequency sampling, a larger window of median filtering is selected; calibrate the number of sampling points of single cell parameters within the time window as M, M = T / k - 1, set the width of the filtering window to 2a + 1, and perform median filtering on the single cell parameter of each sampling point l:

[0073] V′ n (i) = median({V(i - a), V(i - a + 1), …, V(i + a)});

[0074] I′ n(i) = median({I(i - a), I(i - a + 1), …, I(i + a)});

[0075] T′ n (i) = median({T(i - a), T(i - a + 1), …, T(i + a)});

[0076] R′ n (i) = median({R(i - a), R(i - a + 1), …, R(i + a)});

[0077] S′ n (i) = median({S(i - a), S(i - a + 1), …, S(i + a)});

[0078] where i ∈ [1, M]; V′ n (i) represents the median-filtered voltage value at the i-th sampling point of the n-th single cell; I′ n (i) represents the median-filtered current value at the i-th sampling point of the n-th single cell; T′ n (i) represents the median-filtered temperature value at the i-th sampling point of the n-th single cell; R′ n (i) represents the median-filtered internal resistance value at the i-th sampling point of the n-th single cell; S′ n (i) represents the median-filtered self-discharge rate value at the i-th sampling point of the n-th single cell; V(i + a) represents the voltage value at the (i + a)-th sampling point of the single cell; I(i + a) represents the current value at the (i + a)-th sampling point of the single cell; T(i + a) represents the temperature value at the (i + a)-th sampling point of the single cell; R(i + a) represents the internal resistance value at the (i + a)-th sampling point of the single cell; S(i + a) represents the self-discharge rate value at the (i + a)-th sampling point of the single cell;

[0079] Median filtering is used to process the single cell voltage, single cell current, single cell temperature, single cell internal resistance, and single cell self-discharge rate to suppress high-frequency sampling noise, retain the fault characteristics of the single cell, effectively eliminate the random spikes of the single cell voltage and single cell current, and avoid misjudging connection faults;

[0080] Based on the filtered parameter sequence, linear normalization is performed according to the maximum and minimum values of each parameter within the time window:

[0081]

[0082] V″ n (i) and I″ n (i) represent the preprocessed single cell voltage value and single cell current value at the i-th sampling point of the single cell respectively; T″n (i), R″ n (i) and S″ n (i) respectively represent the monomer temperature value, the monomer internal resistance value, and the monomer self-discharge rate value after pretreatment of the i-th sampling point of the monomer battery cell; i represents the number and the total number of sampling points when collecting the battery parameters of the monomer battery cell; V′ n (m) represents the filtered voltage value of the m-th sampling point of the monomer battery cell within the time window; I′ n (m) represents the filtered current value of the m-th sampling point of the monomer battery cell within the time window; T′ n (m) represents the filtered temperature value of the m-th sampling point of the monomer battery cell within the time window; R′ n (m) represents the filtered internal resistance value of the m-th sampling point of the monomer battery cell within the time window; S′ n (m) represents the filtered self-discharge rate value of the m-th sampling point of the monomer battery cell within the time window;

[0083] Normalizing the filtered monomer voltage, monomer current, monomer temperature, monomer battery internal resistance, and monomer self-discharge rate is to adjust the dynamic range of the filtered data, making the input data closer to the real situation and reducing the false alarm rate of the monomer battery cell model;

[0084] S2: Based on historical battery parameters, train the LightGBM model, and input the battery output state coefficient into the trained LightGBM model to obtain the failure probability of the battery cluster and the failure probabilities of each individual monomer battery cell inside;

[0085] The logic for constructing the battery cluster output state coefficient is:

[0086] Based on the battery parameters of the battery cluster after pretreatment, construct the feature matrix of the battery cluster. The specific formula is:

[0087]

[0088] Among them, X represents the feature matrix of the battery cluster, V″ c (j), I″ c (j) and I″ c (j) respectively represent the total voltage, total current, and temperature difference data after pretreatment of the j-th sampling point. j and N respectively represent the number and the total number of sampling points when collecting the battery parameters of the battery cluster;

[0089] Constructing the feature matrix of the battery cluster in the form of sampling points effectively retains the continuity of the original data and provides a basis for the construction of the battery cluster state index;

[0090] Based on the processed total voltage, total current, and temperature difference data of each sampling point, the battery cluster state index of each sampling point is generated to form the output state coefficient of each battery cluster. The formula is as follows:

[0091]

[0092] Among them, Xcl represents the output state coefficient of the battery cluster, and x j represents the battery cluster state index of the j-th sampling point;

[0093]

[0094] The performance of the battery can be measured by power. Power is defined as the product of voltage and current. This product can directly reflect the working state of the battery at a specific moment. The changes in voltage and current can indicate the health status of the battery. Under normal circumstances, the voltage and current of the battery should be maintained within a certain range. During a fault, they may decrease or fluctuate. Therefore, combining them into a product can more comprehensively reflect the overall state of the battery. The product of the total voltage V″ c (j) and the total current I″ c (j) reflects the power output ability of the battery cluster.

[0095] The temperature difference reflects the temperature uniformity inside the battery cluster. A large temperature difference means that some battery cells have faults or uneven discharge. During the actual operation of the battery, the temperature exceeding a certain threshold will cause a significant decline in battery performance or increase the risk of failure. By setting a temperature difference threshold, the impact of temperature on the battery state can be clearly defined, which is convenient for understanding and use. The exponential function form is used to introduce the non-linear impact of the temperature difference on the state index. The impact of temperature changes on battery performance is non-linear. Using an exponential function can effectively capture this non-linear characteristic. T0 in this part is a temperature difference threshold used to distinguish the degree of influence of the temperature difference on the state index. When the temperature difference is greater than T0, the state index will rise rapidly, indicating that the battery may be in a higher risk state, which can better reflect the rapid and significant impact of too high or too low temperature on battery performance in actual situations.

[0096] The form of the denominator is which can effectively prevent the denominator from being zero. Since is always a non-negative number (i.e., greater than or equal to 0), the denominator is always greater than or equal to 1, thus avoiding uncertainties and potential mathematical errors in the calculation, ensuring that when the temperature difference is 0 or in a lower state, the denominator will not be too small, so as to keep the numerical range of the state index within a reasonable level. Even when the temperature difference is not large, the state index can remain at a stable base value.

[0097] When the temperature difference is relatively low compared to the temperature difference threshold, the value approaches 1, and the state index will be close to with little impact. When the temperature difference approaches the threshold, the change in the state index is gradual rather than abrupt, which reflects the response characteristics of the battery to temperature difference changes, making the state assessment smoother and avoiding unnecessary sharp fluctuations. When the temperature difference is relatively high compared to the temperature difference threshold, the value of the battery cluster state index decreases rapidly, resulting in a significant increase in the battery cluster state index, reflecting the greater impact of temperature difference on the battery cluster state and being able to reflect the significant impact of temperature difference on battery performance and health status in the real world.

[0098] The logic for constructing the output state coefficient of a single cell is as follows:

[0099] Based on the battery parameters of the single cell after preprocessing, construct the feature matrix of the single cell. The specific formula is

[0100]

[0101] where Xce represents the feature matrix of the single cell; M represents the total number of sampling points when collecting the battery parameters of the single cell; V″ n (i) and I″ n (i) respectively represent the preprocessed single cell voltage value and single cell current value at the i-th sampling point of the single cell; T″ n (i), R″ n (i) and S″ n (i) respectively represent the preprocessed single cell temperature, internal resistance of the single cell, and self-discharge rate of the single cell at the i-th sampling point; i and M respectively represent the sampling point number and the total number of sampling points when collecting the battery parameters of the single cell;

[0102] Constructing the feature matrix of the single cell in the form of sampling points effectively retains the continuity of the original data and provides a basis for the construction of the cell state index later;

[0103] Based on the preprocessed single cell voltage, single cell current, single cell temperature, internal resistance of the single cell, and self-discharge rate data at each sampling point for processing, generate the cell state index at each sampling point, constituting the output state coefficient of each single cell. The formula is as follows:

[0104]

[0105] where Y cell represents the output state coefficient of the single cell, y i represents the cell state index at the i-th sampling point, and:

[0106]

[0107] g[T″ n (i) represents the temperature influence function, C represents a constant, and T op represents the optimal operating temperature of the battery cell.

[0108] The product of the single - cell voltage and the single - cell current, V″ n (i)*I″ n (i) represents the output power of the single battery cell. The power output of a single battery cell is a direct indicator for evaluating its performance. Power can reflect the output ability of the battery at a certain moment. Under normal circumstances, a single battery cell should be able to output a certain single - cell voltage and single - cell current. If these two values decrease significantly, it may mean that there is a fault in the battery. Therefore, the product of the single - cell voltage and the single - cell current can effectively indicate the health status of the battery cell. The internal resistance of the single cell is an important indicator of battery health. The larger the internal resistance, the greater the power loss and the more heat generated, which affects the efficiency and safety of the battery. Putting the internal resistance of the single cell in the denominator means that an increase in the internal resistance of the single cell will cause the state index to decrease, reflecting the trend of the decline in battery performance. To avoid division - by - zero errors, a small constant C is added to ensure that the formula can still be calculated even when the internal resistance of the single cell is zero.

[0109] The self - discharge rate of the single cell represents the speed at which the battery loses power without a load. An excessively high self - discharge rate of the single cell means poor battery performance. Taking it as part of the denominator means that an increase in the self - discharge rate of the single cell will cause the state index to decrease, reflecting the decline in the effective utilization rate of the battery, 1 + S″ n (i) By adding a constant 1 in the form, it is ensured that regardless of the value of the self - discharge rate, the denominator is always greater than or equal to 1, which can prevent the formula from crashing in extreme cases and provides mathematical stability. Using 1 + S″ n (i) in the form ensures that the influence of the self - discharge rate of the single cell is non - negative. Even when the self - discharge rate of the single cell is 0, the state index can still reflect other performances of the battery. When the self - discharge rate of the single cell increases, S″ n (i) will increase, resulting in an increase in the entire denominator, thereby reducing the state index, reflecting the inhibitory effect of the self - discharge rate of the single cell on the performance of the battery cell, which is in line with the actual situation.

[0110] The state of a single battery cell is not determined by a single factor. The internal resistance of the single cell and the self - discharge rate of the single cell jointly affect the effective energy output of the battery cell. The multiplicative form effectively reflects the linkage relationship between the two. Even if the internal resistance of the single cell is very small or the self - discharge rate of the single cell is very low, the influence of the other factor will correspondingly amplify or reduce the state index. The internal resistance of the single cell and the self - discharge rate of the single cell can be regarded as two aspects of the effective energy loss of the battery. The internal resistance of the single cell reflects the loss during use, while the self - discharge rate of the single cell reflects the loss in the static state. By combining these two, a composite energy loss index is formed to provide a comprehensive assessment of the battery health status.

[0111] The performance of the battery is significantly affected by the cell temperature. Excessive or too low temperature will accelerate battery aging and reduce efficiency. The cell temperature influence function is selected in the form of an exponential function, T op represents the optimal operating temperature of the battery cell. The battery cell performs best at this temperature. The temperature influence function has central symmetry, which means that when the cell temperature is within the optimal range, the function value is close to 1. When the cell temperature deviates slightly, the function value drops rapidly, reflecting the non-linear influence of the cell temperature on the battery performance.

[0112] The cell temperature influence function is essentially a Gaussian function, which has good smoothness and continuity. The influence of temperature on battery performance is non-linear and varies with temperature. The Gaussian function can naturally capture this change. The Gaussian function is symmetric about its center, that is, it performs best near the optimal operating temperature, and the function value drops rapidly when deviating from this temperature, which conforms to the physical characteristics that the battery cell performs best near the optimal operating temperature.

[0113] Through [T″ n (i)-T op 2 It emphasizes that the influence of temperature deviation from the optimal value is quadratic. That is, when the cell temperature deviates slightly, the influence is small; while when the cell temperature deviates greatly, the influence will increase rapidly, making the influence of the cell temperature on the battery performance gradually increase. The square term makes the influence when the cell temperature deviates show non-linear characteristics, that is, the greater the deviation, the more significant the influence, which can effectively reflect the sensitivity of the battery performance to the change of the cell temperature in reality. For example, at extreme temperatures, the battery performance may drop sharply, and the square term can well reflect this phenomenon. When the cell temperature is too low, the chemical reaction rate of the battery decreases, resulting in a reduction in power output; when the cell temperature is too high, it may cause the battery to overheat, accelerate aging or damage. Therefore, the influence of the cell temperature on the battery performance presents a similar Gaussian-like characteristic.

[0114] The LightGBM model includes a battery cluster model and a cell model. The historical battery parameters include normal battery parameter samples and faulty battery parameter samples. The normal battery parameter samples include output state coefficient samples of normal battery clusters and output state coefficient samples of normal cells. The faulty battery parameter samples include output state coefficient samples of faulty battery clusters and output state coefficient samples of faulty cells;

[0115] When training the LightGBM model, the output state coefficient samples of the battery cluster are used as the feature input of the battery cluster model, and whether the battery cluster is faulty is used as the feature output to train the battery cluster model;

[0116] ​Taking the output state coefficient sample of a single cell as the feature input of the single cell model, and whether the single cell is faulty as the feature output of the single cell model, training the single cell model;

[0117] Input the output state coefficient of the battery cluster into the trained battery cluster model to obtain the failure probability of each battery cluster. Input the output state coefficient of each single cell corresponding to the battery cluster into the trained single cell model to obtain the failure probability of each single cell.

[0118] S3: Based on the failure probability of the battery cluster and the failure probability of the single cells, count the number of faulty single cells, set an alarm threshold, and implement dynamic alarm for the battery cluster.

[0119] The logic for implementing the dynamic alarm of the battery cluster is as follows:

[0120] According to the failure probability of the battery cluster and the failure probability of each single cell inside the battery cluster, obtain the overall failure probability of the battery cluster. The formula is as follows:

[0121]

[0122] Among them, P total represents the overall failure probability of the battery, represents the failure probability of the q-th single cell in the battery cluster, n represents the total number of single cells in the battery cluster; q represents the index number of the single cell; p c represents the failure probability of the battery cluster;

[0123] Count the number of single cells with a failure probability greater than 0.5 and label it as A. Dynamically set the alarm threshold, and compare the overall failure probability of the battery with the alarm threshold. If then send a battery cluster failure alarm signal, T represents the set threshold;

[0124] P total Combines several key factors such as the failure probability of the battery cluster and the failure probability of each single cell to generate the overall failure probability of the battery, which can comprehensively reflect the overall failure probability of a single battery cluster. By considering that the failure probabilities of each single cell are relatively independent, the probability that a single cell does not fail is calculated in the form of a product, and then 1 minus the probability that the battery does not fail is used to calculate the overall failure probability. To achieve dynamic threshold alarm, count the number of single cells with a failure probability greater than 0.5. Based on the known number of single cells, consider introducing a trigonometric function, and the value of this trigonometric function changes between 0 and 1. Set the threshold, which is set according to expert experience or historical failure data, and express the two in the form of a product to achieve dynamic threshold alarm;

[0125] Please refer to Figure 2, the present invention further provides a fault detection device for the battery of an energy storage power station. The detection device is used to execute the above detection method and includes:

[0126] A state evaluation module, which is used to set a time window, sample the battery parameters of the battery cluster and the corresponding individual battery cells inside it within the time window, and perform preprocessing to construct a battery output state coefficient. The battery output state coefficient includes the output state coefficient of the battery cluster and the output state coefficient of the individual battery cells;

[0127] A fault prediction module, which trains a LightGBM model based on historical battery parameters, and inputs the battery output state coefficient into the trained LightGBM model to obtain the fault probability of the battery cluster and the fault probability of each individual battery cell inside;

[0128] A fault warning module, which is used to count the number of faulty individual battery cells based on the fault probability of the battery cluster and the fault probability of the individual battery cells, set an alarm threshold, and realize the dynamic alarm of the battery cluster.

[0129] All the above formulas are dimensionless and take their numerical values for calculation. The formula is obtained by collecting a large amount of data and performing software simulation to get a formula closest to the actual situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0130] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or by the combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0131] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units. They can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0132] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present application, and all should be covered within the protection scope of the present application.

Claims

1. A method for fault detection of a battery in an energy storage power station, characterized in that, Including: S1: Set a time window, sample the battery parameters of the battery cluster and the corresponding single battery cells inside it within the time window, and perform preprocessing to construct battery output state coefficients, where the battery output state coefficients include the output state coefficient of the battery cluster and the output state coefficients of single battery cells; S2: Based on historical battery parameters, train a LightGBM model, and input the battery output state coefficients into the trained LightGBM model to obtain the failure probability of the battery cluster and the failure probabilities of each single battery cell inside; S3: Based on the failure probability of the battery cluster and the failure probabilities of single battery cells, count the number of failed single battery cells, set an alarm threshold, and implement dynamic alarm for the battery cluster.

2. The fault detection method for the battery of an energy storage power station according to claim 1, wherein The time window is [T0 - T, T0], where T0 represents the time stamp of the current moment, and T represents the length of the time window; When collecting battery parameters, collect the battery parameters of the battery cluster at a time interval of 10k, and collect the battery parameters of single battery cells at a time interval of k.

3. The fault detection method for the battery of an energy storage power station according to claim 2, wherein The battery parameters of single battery cells include single cell voltage, single cell current, single cell temperature, single cell battery internal resistance, and single cell self-discharge rate. The battery parameters of the battery cluster include total voltage, total current, and temperature difference, where the temperature difference is the difference between the highest single cell temperature and the lowest single cell temperature inside the battery cluster.

4. The fault detection method for a battery of an energy storage power station according to claim 3, characterized in that, The steps of the preprocessing are as follows: For the battery parameters of the battery cluster, set the window of median filtering to a + 1, perform median filtering on the total voltage, total current, and temperature difference data of the battery cluster, and for the total voltage, total current, and temperature difference data of the battery cluster after median filtering, perform normalization processing according to the data type respectively; For the battery parameters of single battery cells, set the window of median filtering to 2a + 1, perform median filtering on the single cell voltage, single cell current, single cell temperature, single cell battery internal resistance, and single cell self-discharge rate, and for the single cell voltage, single cell current, single cell temperature, single cell battery internal resistance, and single cell self-discharge rate after median filtering, perform normalization processing according to the data type respectively.

5. The fault detection method of the battery of an energy storage power station according to claim 3, characterized in that, The logic for constructing the output state coefficient of the battery cluster is: Based on the preprocessed battery parameters of the battery cluster, construct the feature matrix of the battery cluster. The specific formula is: Among them, X represents the characteristic matrix of the battery cluster, V″ c (j), I″ c (j), and I″ c (j) respectively represent the total voltage, total current, and temperature difference data after preprocessing at the j-th sampling point. j and N respectively represent the number and total number of sampling points when collecting the battery parameters of the battery cluster; Based on the processed total voltage, total current, and temperature difference data of each sampling point after preprocessing, generate the state index of the battery cluster for each sampling point, and form the output state coefficient of each battery cluster. The formula is: Among them, Xcl represents the output state coefficient of the battery cluster, and x j represents the battery cluster state index of the j-th sampling point; where T0 represents the set temperature difference threshold.

6. The fault detection method for a battery of an energy storage power station according to claim 3, characterized in that, The logic for constructing the output state coefficient of single battery cells is: Based on the preprocessed battery parameters of single battery cells, construct the feature matrix of single battery cells. The specific formula is Among them, Xce represents the characteristic matrix of the single cell; V″ n (i) and I″ n (i) respectively represent the single cell voltage value and the single cell current value after preprocessing at the i-th sampling point of the single cell; T″ n (i), R″ n (i) and S″ n (i) respectively represent the single cell temperature, the internal resistance of the single cell, and the self-discharge rate of the single cell after preprocessing at the i-th sampling point of the single cell; i and M respectively represent the number and the total number of sampling points when collecting the battery parameters of the single cell; Based on the processed single cell voltage, single cell current, single cell temperature, single cell battery internal resistance, and single cell self-discharge rate data of each sampling point after preprocessing, generate the cell state index for each sampling point, and form the output state coefficient of each single battery cell. The formula is: Among them, Y cell represents the output state coefficient of the single cell, and y i represents the cell state index of the i-th sampling point, and: Among them, C represents the monomer internal resistance regulation constant; T op represents the optimal operating temperature of the battery cell.

7. The fault detection method for the battery of an energy storage power station according to claim 1, characterized in that, The LightGBM model includes a battery cluster model and a single-cell model. The historical battery parameters include normal battery parameter samples and faulty battery parameter samples. The normal battery parameter samples include the output state coefficient samples of normal battery clusters and the output state coefficient samples of normal single cells. The faulty battery parameter samples include the output state coefficient samples of faulty battery clusters and the output state coefficient samples of faulty single cells. When training the LightGBM model, the output state coefficient samples of the battery cluster are used as the feature input of the battery cluster model, and whether the battery cluster is faulty is used as the feature output to train the battery cluster model. The output state coefficient samples of the single cell are used as the feature input of the single cell model, and whether the single cell is faulty is used as the feature output of the single cell model to train the single cell model. The output state coefficients of the battery cluster are input into the trained battery cluster model to obtain the failure probability of each battery cluster. The output state coefficients of each corresponding single cell of the battery cluster are input into the trained single cell model to obtain the failure probability of each single cell.

8. The fault detection method for the battery of an energy storage power station according to claim 1, characterized in that, The logic for realizing the dynamic alarm of the battery cluster is as follows: Based on the battery cluster failure probability and the failure probability of each single cell inside the battery cluster, the overall failure probability of the battery cluster is obtained. The formula is as follows: Among them, P total represents the overall battery failure probability, represents the failure probability of the q-th single cell in the battery cluster, n represents the total number of single cells in the battery cluster; q represents the single cell index number; p c represents the battery cluster failure probability; Count the number of single cell failures with a probability greater than 0.5 and label it as A. Dynamically set the alarm threshold, and compare the overall battery failure probability with the alarm threshold. If then issue a battery cluster failure alarm signal. T represents the set threshold.

9. A fault detection device for a battery of an energy storage power station, characterized in that: The detection device is used to execute the detection method according to any one of claims 1-8, including: A state evaluation module, which is used to set a time window, sample the battery parameters of the battery cluster and the corresponding single cells inside it within the time window, and perform preprocessing to construct the battery output state coefficients. The battery output state coefficients include the output state coefficients of the battery cluster and the output state coefficients of the single cells. A fault prediction module, which trains the LightGBM model based on the historical battery parameters, and inputs the battery output state coefficients into the trained LightGBM model to obtain the failure probability of the battery cluster and the failure probabilities of each single cell inside. A fault warning module, which is used to count the number of faulty single cells based on the failure probability of the battery cluster and the failure probability of the single cells, set an alarm threshold, and realize the dynamic alarm of the battery cluster.

Images