Fault detection method for string PCS

Abstract

The embodiment of the invention relates to a fault detection method for a string PCS. The fault detection method comprises the following steps: judging a fault shutdown probability index based on a first operation parameter of an AC bus and a first fault judgment model; when the fault shutdown probability index meets a first condition, performing energy storage line positioning on a fault based on a second operation parameter of the energy storage line and a second fault judgment model; and for the fault energy storage line, predicting the fault degree of the battery pack based on a third fault judgment model, and carrying out fault processing according to the fault degree. According to the technical scheme provided by the embodiment of the invention, a preliminary judgment basis is provided for fault detection based on the first fault judgment model, and the second fault judgment model and the third fault judgment model are adopted to carry out fault positioning and fault degree prediction respectively under the condition that the system can continue to operate, so that layered detection and positioning of the fault are realized, and the fault detection efficiency is improved. The fault detection efficiency can be improved under the condition that the system safety is ensured.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of energy storage system control, and in particular to a fault detection method for a string PCS. Background Art

[0002] In the prior art, fault location for string-type PCS energy storage systems is relatively complex. Because the energy storage system contains multiple interrelated parts, such as the AC bus, multiple energy storage lines, and battery packs, fault signals may affect and propagate each other in the system. Traditional fault detection methods may only detect a single component or a single level. When a fault occurs, it is difficult to quickly and accurately determine where the fault occurs in the system, or the complex positioning measures may cause the fault to escalate and cause the system to be unable to operate. Summary of the invention

[0003] Based on the above situation of the prior art, an object of an embodiment of the present invention is to provide a fault detection method for a string PCS, which can improve the efficiency and accuracy of fault detection while ensuring system safety.

[0004] To achieve the above object, according to one aspect of the present invention, a fault detection method for a string PCS is provided, wherein a plurality of energy storage lines are respectively connected to an AC bus, and each energy storage line includes a battery pack and a string PCS; the method comprises the following steps:

[0005] Determine a fault shutdown probability index based on a first operating parameter of the AC bus and a first fault judgment model;

[0006] When the fault shutdown probability index meets the first condition, the fault is located in the energy storage line based on the second operating parameter of the energy storage line and the second fault judgment model;

[0007] For the faulty energy storage circuit, predict the fault degree of the battery pack based on the third fault judgment model, and perform fault processing according to the fault degree;

[0008] Among them, the first operating parameters of the AC bus include voltage, current, active power, reactive power, voltage amplitude change rate, current amplitude change rate and power change rate; the second operating parameters of the energy storage line include output voltage, output current and output power.

[0009] Furthermore, the first fault judgment model is an LSTM network model, and the output features of the first fault judgment model are fault probability and emergency shutdown probability;

[0010] The loss function L of the failure probability f for:

[0011]

[0012] The loss function L of the emergency shutdown probability e for:

[0013]

[0014] Where M represents the number of training samples, y f-i represents the true fault label of the i-th training sample, P f-i represents the failure probability of the model output, y e-i represents the true emergency stop label of the i-th training sample, P e-i Represents the emergency shutdown probability output by the model.

[0015] Further, the failure shutdown probability indicator includes the failure probability and the emergency shutdown probability;

[0016] The first condition is expressed as the failure probability is greater than or equal to the failure threshold and the emergency shutdown probability is less than the emergency shutdown threshold;

[0017] The second condition is represented by the failure probability being greater than or equal to the failure threshold and the emergency shutdown probability being greater than or equal to the emergency shutdown threshold.

[0018] Furthermore, when the fault shutdown probability index meets the second condition, an emergency shutdown operation is performed.

[0019] Furthermore, the second fault judgment model is expressed as:

[0020]

[0021] Among them, F i represents the probability of failure of the i-th energy storage line, r i (t) represents the power proportion of the i-th energy storage line at time t, ΔV bi (t) represents the output voltage change rate of the battery pack of the i-th energy storage circuit at time t, ΔI bi (t) represents the output current change rate of the battery pack of the i-th energy storage line at time t, Δt represents the sampling time interval, r th Indicates the power change threshold, V bn Indicates the rated voltage of the battery pack, I bn represents the rated current of the battery pack, and α and β represent the threshold adjustment parameters respectively.

[0022] Further, predicting the fault degree of the battery pack based on the third fault judgment model includes the following steps:

[0023] Construct state transition probability matrix and observation matrix;

[0024] Based on the real-time state parameters of the faulty energy storage line, the state transition probability matrix and the observation matrix, a forward prediction vector and a backward prediction vector are calculated;

[0025] Calculating a fault degree probability value of a battery pack of a faulty energy storage circuit based on the forward prediction vector and the backward prediction vector;

[0026] The failure degree of the battery pack is predicted according to the failure degree probability value.

[0027] Furthermore, the state transition probability matrix is ​​expressed as:

[0028]

[0029] Among them, P ij Represents the probability of the battery pack transferring from state i to state j, i = 0, 1, 2, 3; j = 0, 1, 2, 3; the states include 0: normal state, 1: slightly abnormal state, 2: moderately abnormal state, 3: severely abnormal state.

[0030] Furthermore, the observation matrix is ​​an N×M matrix, M=M1×M2×M3, M1 represents the number of intervals divided for the first observation value, M2 represents the number of intervals divided for the second observation value, M3 represents the number of intervals divided for the third observation value, and N represents the number of battery pack states; the element B of the observation matrix ij It is expressed as:

[0031]

[0032] Among them, let the observation interval combination corresponding to j be n ij Indicates that in the historical data, when the battery pack is in state i, the first observation is observed in the interval The second observation is in the interval The third observation is in the interval The number of times.

[0033] Furthermore, the forward prediction vector is obtained recursively according to the following formula:

[0034]

[0035] Among them, α t (j) is the forward prediction vector, which indicates that at time t, based on the observation sequence O1, O2, ..., O t , the probability that the battery pack is in state j;

[0036] The backward prediction vector is obtained recursively according to the following formula:

[0037]

[0038] Among them, β t (i) is the backward prediction vector, which means at time t, based on the observation sequence O t+1 , O t+2 , ..., O T , the probability that the battery pack is in state i.

[0039] Furthermore, the failure degree probability value is calculated according to the following formula:

[0040]

[0041] Among them, γ t (i) is the probability value of the failure degree, which indicates the probability of the battery pack being in each state.

[0042] In summary, the embodiment of the present invention provides a fault detection method for a string PCS, the method comprising the following steps: judging a fault shutdown probability index based on a first operating parameter of an AC bus and a first fault judgment model; when the fault shutdown probability index meets the first condition, locating the energy storage line based on the second operating parameter of the energy storage line and the second fault judgment model; predicting the fault degree of the battery pack based on the third fault judgment model for the fault storage line, and performing fault processing according to the fault degree. The technical solution provided by the embodiment of the present invention determines whether a fault may occur and whether the fault is a fault requiring an emergency shutdown operation based on the first fault judgment model, providing a preliminary judgment basis for fault detection, and using the second fault judgment model and the third fault judgment model to respectively locate the fault and predict the fault degree when the system can continue to operate, thereby realizing hierarchical fault detection and positioning, and improving the efficiency of fault detection while ensuring system safety. In addition, corresponding fault judgment models are provided for fault detection at different levels, and targeted detection can be performed according to the characteristics of the operating parameters and fault modes at each level, thereby improving the accuracy of the fault detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of a charging and discharging control method of a string PCS provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0045] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.

[0046] The technical solution of the present invention is described in detail below with reference to the accompanying drawings. An embodiment of the present invention provides a fault detection method for a string PCS. The string PCS (Power Conversion System) is a key core component for realizing the bidirectional flow of electric energy between the energy storage system and the power grid. In an embodiment of the present invention, the energy storage system includes a plurality of energy storage lines, and the plurality of energy storage lines are respectively connected to the AC bus. Each energy line includes a battery pack and a string PCS. Preferably, the battery pack can be connected to the string PCS through a DC / DC conversion circuit. The AC bus can be connected to the power grid, for example, through a transformer or other equipment. Figure 1 A flow chart of a fault detection method for a string PCS provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the fault detection method includes the following steps:

[0047] S202, judging the fault shutdown probability index based on the first operating parameter of the AC bus and the first fault judgment model. The fault shutdown probability index includes the fault probability and the emergency shutdown probability. The first operating parameter of the AC bus includes the voltage of the AC bus, the current of the AC bus, the active power of the AC bus, the reactive power of the AC bus, the voltage amplitude change rate of the AC bus, the current amplitude change rate of the AC bus, and the power change rate of the AC bus. The first operating parameter can be monitored by various sensors set on the AC bus, and calculated according to the monitoring data of the sensor. The first fault judgment model adopts a long short-term memory (Long Short-Term Memory, hereinafter referred to as "LSTM") network model, and the output features are the fault probability and the emergency shutdown probability, that is, the first fault judgment model judges the current probability of a fault and the probability of an emergency shutdown according to the current first operating parameter. The core of the LSTM network model is the memory unit, which can effectively remember information in a longer time series data. When the embodiment of the present invention is used to process the time series data of the AC bus (such as voltage, current, change rate and other data that change over time), LSTM can remember the operating status of the AC bus over a long period of time in the past, so as to better capture the long-term trend and periodic changes in the data. Since the occurrence of a fault is a gradual process and is closely related to the past operating status, the LSTM network model can be used to determine whether a fault may occur based on the current operating parameters, and whether the fault is a fault that requires an emergency shutdown operation, providing a preliminary judgment basis for fault detection.

[0048] In an embodiment of the present invention, the LSTM network model is a multi-layer LSTM network, including an input layer, multiple LSTM hidden layers and an output layer. The number of input layer nodes is equal to the number of input features (i.e., the first operating parameter), the number of hidden layer nodes can be determined based on experiments and data complexity, and the output layer has two nodes, which are used to output the failure probability and the emergency shutdown probability, respectively. A standard activation function is used in the LSTM network model, and a sigmoid function can be used in the output layer to map the output to an interval to represent the probability. The loss function of the LSTM network model is set to the loss function of the failure probability and the loss function of the emergency shutdown probability. The loss function of the failure probability L f It is expressed as:

[0049]

[0050] The loss function L of the emergency shutdown probability e It is expressed as:

[0051]

[0052] Where M represents the number of training samples, y f-irepresents the true fault label of the i-th training sample, P f-i represents the failure probability of the model output, y e-i represents the true emergency stop label of the i-th training sample, P e-i Represents the emergency shutdown probability output by the model.

[0053] The Adam optimization algorithm can be used to train the network model, adjust the weight and bias of the network model, and minimize the loss function. The time series data of the first operating parameter collected and preprocessed in real time is input into the trained LSTM network model, that is, the first fault judgment model, to obtain the fault probability and emergency shutdown probability.

[0054] S204. When the fault shutdown probability index meets the first condition, the energy storage line is located based on the second operating parameters and the second fault judgment model of the energy storage line. The first condition refers to the situation where the fault probability is greater than or equal to the fault threshold and the emergency shutdown probability is less than the emergency shutdown threshold, that is, under the first condition, it is judged that the energy storage system has a fault, but the fault does not belong to a fault that has a significant impact on the system and does not need to be shut down immediately. At this time, the fault can be located in the line through the second fault judgment model and further processed. If the output of the first fault judgment model shows that the fault probability is greater than or equal to the fault threshold and the emergency shutdown probability is greater than or equal to the emergency shutdown threshold, that is, when the fault shutdown probability index meets the second condition, the emergency shutdown operation is performed.

[0055] The second operating parameters of the energy storage line include output voltage, output current and output power, and collect the DC side voltage, DC side current, AC side voltage and AC side current of the string PCS in the energy storage line, as well as the output voltage and output current of the battery pack. The power contribution of the line to the AC bus is calculated using the AC side voltage and AC side current of the string PCS, and the power contribution is divided by the total power of the AC bus to obtain the power share of the energy storage line; the output voltage change rate and output current change rate of the battery pack are calculated based on the output voltage and output current of the battery pack at each sampling moment.

[0056] The second fault judgment model is used to judge which energy storage line the fault occurs in. The second fault judgment model is expressed as:

[0057]

[0058] Among them, F i represents the probability of failure of the i-th energy storage line, r i (t) represents the power proportion of the i-th energy storage line at time t, ΔV bi (t) represents the output voltage change rate of the battery pack of the i-th energy storage circuit at time t, ΔI bi(t) represents the output current change rate of the battery pack of the i-th energy storage line at time t, Δt represents the sampling time interval, r th Indicates the power change threshold, V bn Indicates the rated voltage of the battery pack, I bn represents the rated current of the battery pack, and α and β represent the threshold adjustment parameters respectively. When the power proportion of the energy storage line drops beyond the set threshold, and the output voltage change rate and output current change rate of the battery pack both exceed the set threshold, it is considered that the energy storage line has a fault. According to some optional embodiments, the AC side voltage and AC side current change rate of the string PCS can also be used as a basis for judgment.

[0059] S206. For the faulty energy storage line, predict the fault degree of the battery pack based on the third fault judgment model, and handle the fault according to the fault degree. In the embodiment of the present invention, a third fault judgment model based on the state transition probability matrix and the observation matrix is ​​used to predict the possible fault degree based on the current state of the battery pack, so as to handle it in a targeted manner in advance. The prediction of the fault degree of the battery pack based on the third fault judgment model can be based on the following steps:

[0060] S2061, construct a state transition probability matrix and an observation matrix. The state transition probability matrix can be expressed as:

[0061]

[0062] Among them, P ij represents the probability of the battery pack transitioning from state i to state j, i = 0, 1, 2, 3; j = 0, 1, 2, 3; the calculation method is the number of transitions from state i to state j divided by the total number of transitions from state j, for example, P 01 =The number of transitions from state 0 to state 1 / the total number of transitions from state 0. In the state transition probability matrix, the sum of the elements in each row is 1. When the state value is 0, it indicates a normal state; when it is 1, it indicates a slightly abnormal state; when it is 2, it indicates a moderately abnormal state; when it is 3, it indicates a severely abnormal state. The parameter range of the battery pack under each state can be preset. For example, in a normal state, the battery pack SOC is between 40% and 60%, the battery internal resistance is stable and within the normal resistance range, and the temperature is maintained at 20°C-35°C; in a slightly abnormal state, the battery pack SOC deviates from the normal range by 20%-40% or 60%-80%, the battery internal resistance increases within a certain threshold, and the temperature is between 35°C-45°C; the moderately abnormal state and the severely abnormal state are similar and can be set according to actual needs. The above state transition probability matrix can be generated based on the historical operation data of the battery pack, including the state records at different time points and the transition between states.

[0063] Select multiple observations that can reflect the state of the battery pack, such as SOC, temperature, internal resistance, the difference between the maximum and minimum values ​​of the single cell voltage, etc. Divide each observation into intervals. For example, for SOC, it can be divided into 5 intervals, namely [0-20%), (20%-40%), (40%-60%), (60%-80%), (80%-100%). In the embodiment of the present invention, three observations are selected to construct an observation matrix. The observation matrix is ​​an M×M matrix, M=M1×M2×M3, M1 represents the number of intervals divided for the first observation, M2 represents the number of intervals divided for the second observation, M3 represents the number of intervals divided for the third observation, and N represents the number of battery pack states (corresponding to the states in the above-mentioned state transition probability matrix, N=4). The element B of the observation matrix ij It is expressed as:

[0064]

[0065] Among them, let the observation interval combination corresponding to j be n ij Indicates that in the historical data, when the battery pack is in state i, the first observation is observed in the interval The second observation is in the interval The third observation is in the interval The number of times. ij represents the probability of observing a combination belonging to the jth observation interval in state i. Calculate B ij , count the number of times the observation combination falls into the jth interval combination when in state i in the historical data, divided by the total number of observations in state i.

[0066] S2062. Calculate the forward prediction vector and the backward prediction vector based on the real-time state parameters, state transition probability matrix and observation matrix of the faulty energy storage line. Determine the probability of the energy storage line being in each state according to the real-time state parameters of the faulty energy storage line as the initial state probability vector. For example, if the current state is normal, the initial state probability vector π(0)=[1,0,0,0]. Construct the observation sequence O=(O1,O2,……,O T ), T represents the length of the observation sequence (i.e., the number of time steps), O t Represents the observation value at time t. Collect the real-time measurement values ​​of each observation at the initial time, and determine the observation interval combination to which it belongs as the initial observation value O0.

[0067] The forward prediction vector α t (i) represents the time t based on the observation sequence O1, O2, ..., O t , the probability that the battery pack is in state i, the initial value is For t = 2, 3, ..., T, T represents the length of the observation sequence, and we can get it by recursion:

[0068]

[0069] Backward prediction vector β t (i) represents the time t based on the observation sequence O t+1 , O t+2 , ..., O T , the probability that the battery pack is in state i, the initial value is β(i) = 1 (for all i), for t = T-1, T-2, ..., 1, by recursion:

[0070]

[0071] S2063. Calculate the fault degree probability value of the battery pack of the faulty energy storage circuit based on the forward prediction vector and the backward prediction vector.

[0072] According to the forward prediction vector and the backward prediction vector, the probability γ of the battery pack being in each state under a given observation sequence is calculated t (i)

[0073]

[0074] S2064: Predict the degree of failure of the battery pack according to the probability value of the degree of failure. t (i) to predict the possible failure degree of the battery pack. For example, if γ t (0) (i.e., the probability that the battery pack is in a normal state at time t) is greater than 0.8, then the battery pack is considered to be basically normal in the short term; if γ t (1) Maximum and greater than 0.5, or γ t (0)+γ t (1)>0.8 and γ t (1) is relatively large, it is judged that a minor fault may occur in the short term; if γ t (2) Maximum and greater than 0.5, or γ t (0)+γ t (1)+γ t (2)>0.8 and γ t (2) is relatively large, it is judged that a moderate fault may occur in the short term; if γ t (3) Maximum and greater than 0.5, or γ t (0)+γ t (1)+γ t (2)+γ t (3)>0.8 and γ t (3) If it is relatively large, it is judged that a serious fault may occur in the short term.

[0075] Fault processing is performed according to the fault severity. When the fault severity is basically normal or a minor fault may occur in the short term, it indicates that the battery pack failure will not affect the operation of the entire system in the short term. The fault condition can be recorded and each battery pack can be repaired according to the maintenance cycle. When the fault severity is a moderate fault or a severe fault may occur in the short term, the energy storage line can be disconnected from the AC bus according to actual needs to avoid a major impact of the fault of the energy storage line on the operation of the entire system. After cutting off the energy storage line, the power can be redistributed based on the output power of the remaining lines and the required total power.

[0076] In the embodiment of the present invention, a third fault judgment model is constructed based on the state transition probability matrix and the observation matrix to predict the degree of failure of the battery pack. The state transition probability matrix describes the probability of the battery pack switching from one state to another. The state transition probability matrix is ​​based on historical operating data and internal laws, and can reflect the dynamic process of the development of battery pack failures. The observation matrix links the real-time observed battery pack parameters with different states, and can quantify the probability of observing a specific parameter combination in each state. The observation matrix can infer the most likely state of the current battery pack based on the actual observed values ​​of these parameters. The state transition probability matrix is ​​combined with the observation matrix, while taking into account the individual differences and common laws of different battery packs, as well as the influence of historical deduction laws and real-time status conditions, thereby providing more accurate fault prediction.

[0077] In summary, the embodiment of the present invention relates to a fault detection method for a string PCS, the method comprising the following steps: judging a fault shutdown probability index based on a first operating parameter of an AC bus and a first fault judgment model; when the fault shutdown probability index meets the first condition, locating the energy storage line based on a second operating parameter of the energy storage line and a second fault judgment model; predicting the degree of fault of the battery pack based on a third fault judgment model for the faulty energy storage line, and performing fault processing according to the degree of fault. The technical solution provided by the embodiment of the present invention determines whether a fault may occur and whether the fault is a fault requiring an emergency shutdown operation based on the first fault judgment model, providing a preliminary judgment basis for fault detection, and using the second fault judgment model and the third fault judgment model to respectively locate the fault and predict the degree of fault when the system can continue to operate, thereby realizing hierarchical fault detection and positioning, and improving the efficiency of fault detection while ensuring system safety. In addition, corresponding fault judgment models are provided for fault detection at different levels, and targeted detection can be performed according to the characteristics of the operating parameters and fault modes at each level, thereby improving the accuracy of the fault detection results.

[0078] It should be understood that the discussion of any of the above embodiments is only exemplary and is not intended to imply that the scope of the present invention (including claims) is limited to these examples; under the idea of ​​the present invention, the technical features in the above embodiments or different embodiments may also be combined, the steps may be implemented in any order, and there are many other changes in different aspects of one or more embodiments of the present invention as described above, which are not provided in detail for the sake of simplicity. The above specific embodiments of the present invention are only used to illustrate or explain the principles of the present invention, and do not constitute a limitation of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc. made without departing from the spirit and scope of the present invention should be included in the scope of protection of the present invention. In addition, the claims attached to the present invention are intended to cover all changes and modifications that fall within the scope and boundaries of the attached claims, or the equivalent forms of such scope and boundaries.

Claims

1. A fault detection method for a string PCS, characterized in that: A plurality of energy storage lines are respectively connected to the AC bus, each energy storage line comprising a battery pack and a string-type PCS; the method comprises the following steps: Determine a fault shutdown probability index based on a first operating parameter of the AC bus and a first fault judgment model; When the fault shutdown probability index meets the first condition, the fault is located in the energy storage line based on the second operating parameter of the energy storage line and the second fault judgment model; For the faulty energy storage circuit, predict the fault degree of the battery pack based on the third fault judgment model, and perform fault processing according to the fault degree; Among them, the first operating parameters of the AC bus include voltage, current, active power, reactive power, voltage amplitude change rate, current amplitude change rate and power change rate; the second operating parameters of the energy storage line include output voltage, output current and output power.

2. The method according to claim 1, characterized in that The first fault judgment model is an LSTM network model, and the output features of the first fault judgment model are fault probability and emergency shutdown probability; The loss function L of the failure probability f for: The loss function L of the emergency shutdown probability e for: Where M represents the number of training samples, y f-i represents the true fault label of the i-th training sample, P f-i represents the failure probability of the model output, y e-i represents the true emergency stop label of the i-th training sample, P e-i Represents the emergency shutdown probability output by the model.

3. The method according to claim 2, characterized in that The failure shutdown probability index includes the failure probability and the emergency shutdown probability; The first condition is expressed as the failure probability is greater than or equal to the failure threshold and the emergency shutdown probability is less than the emergency shutdown threshold; The second condition is represented by the failure probability being greater than or equal to the failure threshold and the emergency shutdown probability being greater than or equal to the emergency shutdown threshold.

4. The method according to claim 3, characterized in that When the fault shutdown probability index meets the second condition, an emergency shutdown operation is performed.

5. The method according to claim 1, characterized in that The second fault judgment model is expressed as: Among them, F i represents the probability of failure of the i-th energy storage line, r i (t) represents the power proportion of the i-th energy storage line at time t, ΔV bi (t) represents the output voltage change rate of the battery pack of the i-th energy storage circuit at time t, ΔI bi (t) represents the output current change rate of the battery pack of the i-th energy storage line at time t, Δt represents the sampling time interval, r th Indicates the power change threshold, V bn Indicates the rated voltage of the battery pack, I bn represents the rated current of the battery pack, and α and β represent the threshold adjustment parameters respectively.

6. The method according to claim 1, characterized in that Predicting the fault degree of the battery pack based on the third fault judgment model includes the following steps: Construct state transition probability matrix and observation matrix; Based on the real-time state parameters of the faulty energy storage line, the state transition probability matrix and the observation matrix, a forward prediction vector and a backward prediction vector are calculated; Calculating a fault degree probability value of a battery pack of a faulty energy storage circuit based on the forward prediction vector and the backward prediction vector; The failure degree of the battery pack is predicted according to the failure degree probability value.

7. The method according to claim 6, characterized in that The state transition probability matrix is ​​expressed as: Among them, P ij Represents the probability of the battery pack transferring from state i to state j, i = 0, 1, 2, 3; j = 0, 1, 2, 3; the states include 0: normal state, 1: slightly abnormal state, 2: moderately abnormal state, 3: severely abnormal state.

8. The method according to claim 7, characterized in that The observation matrix is ​​an N×M matrix, where M=M1×M2×M3, M1 represents the number of intervals divided for the first observation value, M2 represents the number of intervals divided for the second observation value, M3 represents the number of intervals divided for the third observation value, and N represents the number of battery pack states; the element B of the observation matrix ij It is expressed as: Among them, let the observation interval combination corresponding to j be n ij Indicates that in the historical data, when the battery pack is in state i, the first observation is observed in the interval The second observation is in the interval The third observation is in the interval The number of times.

9. The method according to claim 8, characterized in that The forward prediction vector is obtained recursively according to the following formula: Among them, α t (j) is the forward prediction vector, which indicates that at time t, based on the observation sequence O1, O2, ..., O t , the probability that the battery pack is in state j; The backward prediction vector is obtained recursively according to the following formula: Among them, β t (i) is the backward prediction vector, which means at time t, based on the observation sequence O t+1 , O t+2 , ..., O T , the probability that the battery pack is in state i.

10. The method according to claim 9, characterized in that The failure degree probability value is calculated according to the following formula: Among them, γ t (i) is the probability value of the failure degree, which indicates the probability of the battery pack being in each state.

Images