A lithium battery pack multi-fault diagnosis method

Abstract

The present application belongs to the field of lithium battery pack fault diagnosis, and is a kind of lithium battery pack multi-fault diagnosis method, which comprises: establishing a diagnosis topology structure, the diagnosis topology structure comprises N sub-modules with the same structure, and the N sub-modules are connected in series;The important features in the fault classification are found out by computer simulation of the diagnosis topology structure;Important characteristics are taken as input, different fault types corresponding to the change curve of important characteristics are taken as output, ET algorithm is trained, and the optimal classification positioning model is obtained by adjusting the parameters of ET algorithm in the training process;The measurement values collected by the sensors in the diagnosis process are polynomial transformed, and important features are obtained, and the important features are input into the optimal classification positioning model for fault classification and positioning.The problem that each cell needs to be monitored and a large number of monitoring sensors are required in the existing test method is solved, which helps to improve the safety and stability of the lithium battery pack.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of lithium battery pack fault diagnosis, and particularly relates to a lithium battery pack multi-fault diagnosis method. BACKGROUND

[0002] With the gradual increase of the proportion of renewable energy in new power systems, the use of lithium battery packs becomes more common. Since thermal runaway in lithium ion batteries can cause irreversible damage to the entire system, it is important to analyze and detect faults that can lead to thermal runaway. At the same time, timely determination of the location of the faulty battery also improves the overall service life of the electric vehicle.

[0003] Thermal runaway in lithium ion batteries mainly includes internal short circuit (ISC) faults and cell open circuit (COC) faults. In practical applications, when the battery system is running, the fault type is not fixed. The current detection method needs to detect for several hours, or it is necessary to monitor every unit in the system, which means that many sensors must be used and a large amount of data must be processed, which is very inconvenient for fault diagnosis. SUMMARY

[0004] The technical problem to be solved by the present application is to provide a lithium battery pack multi-fault diagnosis method, which solves the problem of the need to monitor each cell in the existing test method and the need for a large number of monitoring sensors.

[0005] The present application is implemented in the following way,

[0006] A lithium battery pack multi-fault diagnosis method, the method comprising:

[0007] A diagnostic topology is established, the diagnostic topology comprising N sub-modules with the same structure, the sub-module comprising 16 lithium iron phosphate batteries, a first current sensor, a second current sensor and a voltage sensor, two lithium iron phosphate batteries in series forming a series group, the connection end between the two lithium iron phosphate batteries forming a node, all series groups being connected in parallel, wherein four series groups form a first array group and another four series groups form a second array group, one measurement end of the first current sensor being connected to the four nodes of the first array group, the other measurement end of the first current sensor being connected to the four nodes of the second array group; one measurement end of the other second current sensor being connected to one node of the first array group, the other measurement end being connected to one node of the second array group; the voltage sensor being connected between any two nodes in the first array group or the second array group;

[0008] The N sub-modules are connected in series;

[0009] The diagnostic topology is simulated by computer, and the change curves of the measurement values of the current sensor and the voltage sensor and the polynomial transformation of the measurement values over time under different fault types are simulated, the change curves that are prominently changed corresponding to different faults and the current sensor, the voltage sensor or / and the polynomial transformation corresponding to the change curves are found, the change curves of the measurement values of the current sensor and the voltage sensor and the polynomial transformation of the measurement values over time under a plurality of faults are analyzed in importance, and important features in fault classification are found;

[0010] The ET algorithm is trained by taking the important features as input and different fault types corresponding to the change curves of the important features as output, in the training process, the parameters of the ET algorithm are adjusted by using the particle swarm algorithm, and an optimal classification positioning model is obtained.

[0011] The measurement values collected by the diagnostic process sensor are subjected to polynomial transformation, and important features are obtained, the important features are input into the optimal classification positioning model for fault classification and positioning.

[0012] Further, the polynomial transformation refers to that the measurement values of the current sensor and the voltage sensor are expanded into a plurality of features by difference and ratio.

[0013] Further, the important features refer to features that contribute to all faults when faults occur, and the contribution forms are different for different faults.

[0014] Further, the measurement values collected by the diagnostic process sensor are subjected to polynomial transformation before pre-judgment, and the pre-judgment includes judging which measurement values of the sensors in the sub-modules the changed measurement values belong to.

[0015] Further, the pre-judgment also includes that when it is judged that the measurement values in the sub-modules are not changed, it is regarded as no fault.

[0016] Further, the pre-judgment also includes judging the number of sudden changes of the measurement values of the sensors in the sub-modules, when the number is equal to 1, it is regarded as a sensor fault, and when the number is greater than or equal to 2, it is regarded as a battery fault.

[0017] Further, adjusting the parameters of the ET algorithm by using the particle swarm algorithm includes the number of decision trees, the maximum depth of the decision tree, the minimum number of samples required for node splitting and the minimum number of samples required for the leaf node.

[0018] Compared with the prior art, the present application has the beneficial effects that:

[0019] The topology structure of the present application can detect and locate by one voltage sensor and two current sensors in each sub-module, and can detect internal faults, open circuit faults and sensor faults in the battery pack composed of 16 lithium iron phosphate batteries in each sub-module. With a small number of sensors, the present application method reduces the amount of calculation while improving the accuracy and efficiency of fault diagnosis. At the same time, the present application adds the diagnosis of sensor faults to ensure the accuracy and reliability of the monitoring data. The present application method will help to improve the safety and stability of lithium battery pack and reduce the risk and loss caused by lithium battery failure. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 The structure diagram of the diagnostic topology structure in the lithium battery pack multi-fault diagnosis method provided by the embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical scheme and advantages of the present application more clear and explicit, the present application will be further described in detail below in combination with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0022] As shown in Figure 1 A lithium battery pack multi-fault diagnosis method is described by taking two sub-modules as an example. The faults of 32 lithium iron phosphate batteries can be detected and located at the same time, and the method comprises:

[0023] establishing a diagnostic topology structure, the diagnostic topology structure comprising a first sub-module and a second sub-module with the same structure, the first sub-module comprising 16 lithium iron phosphate batteries, a first current sensor A1, a second current sensor A2 and a first voltage sensor V1, two lithium iron phosphate batteries in series forming a series group, the connection end between the two lithium iron phosphate batteries forming a node, all series groups being connected in parallel, wherein four series groups form a first array group and another four series groups form a second array group, one measuring end of the first current sensor A1 being connected with four nodes of the first array group, the other measuring end of the first current sensor A1 being connected with four nodes of the second array group; one measuring end of the other second current sensor A2 being connected with one node of the first array group, the other measuring end being connected with one node of the second array group; the two ends of the first voltage sensor V1 being connected between any two nodes in the first array group or the second array group;

[0024] The second sub-module comprises a third current sensor A3, a fourth current sensor A4 and a second voltage sensor V2. The measuring end of the third current sensor A3 is connected with four nodes of the first array group, and the other measuring end of the third current sensor A3 is connected with four nodes of the second array group. One measuring end of the fourth current sensor A4 is connected with one node of the first array group, and the other measuring end of the fourth current sensor A4 is connected with one node of the second array group. The two ends of the second voltage sensor V2 are connected between any two nodes in the first array group or the second array group.

[0025] The first sub-module and the second sub-module are in a series connection. When there are multiple sub-modules, the sub-modules are in series connection. In the embodiment, two sub-modules are taken as an example, but the number of sub-modules is not limited to two.

[0026] The diagnostic topology is simulated by computer. In the embodiment, MATLAB is used for simulation. The measurement values of the current sensors and the voltage sensors in the diagnostic topology and the polynomial transformation of the measurement values are simulated under different fault types, and the change curves of the measurement values with time are simulated, for example, the change curve of the measurement value I1 of the first current sensor A1 with time. The change curve here can also be a straight line with constant values. The change curves corresponding to different faults are found, and the current sensors, the voltage sensors or / and the polynomial transformation corresponding to the change curves are found. It should be noted that one fault corresponds to multiple change curves. For all the current sensors and the voltage sensors in the diagnostic topology, some change curves have great influence, and some change curves have little influence. The current sensors, the voltage sensors or / and the polynomial transformation corresponding to the change curves with great influence are found. The polynomial transformation here refers to the difference and ratio between the measurement values of the first current sensor A1, the second current sensor A2, the first voltage sensor V1, the third current sensor A3, the fourth current sensor A4 and the second voltage sensor V2, which are expanded into multiple features.

[0027] Through the above process, multiple change curves corresponding to different faults and the features corresponding to the change curves can be simulated. The features here refer to which current sensor or voltage sensor or polynomial transformation the change curve belongs to.

[0028] The change curves of the measurement values of the current sensors and the voltage sensors and the polynomial transformation of the measurement values with time under multiple faults are analyzed for importance, and important features in fault classification are found.

[0029] The important feature here refers to a feature that contributes to all faults when a fault occurs, and has differences in contribution forms for different faults. For example: fault A and fault B, the change curve corresponding to the first current sensor A1 changes significantly, and the change curve of the current sensor A1 can be distinguished as belonging to fault A or fault B for fault A and fault B. Then the first current sensor A1 can be used as an important feature. This important feature can be a single feature or a combination of multiple features.

[0030] The ET algorithm is trained with the important features as input and different fault types corresponding to the change curve of the important features as output. In the training process, the particle swarm algorithm is used to adjust the parameters of the ET algorithm to obtain the optimal classification positioning model. The particle swarm algorithm is used to adjust the parameters of the ET algorithm, including the number of decision trees, the maximum depth of the decision tree, the minimum number of samples required for node splitting, and the minimum number of samples required for leaf nodes.

[0031] The measurement values collected by the diagnostic process sensor are subjected to polynomial transformation, and important features are obtained. The important features are input into the optimal classification positioning model for fault classification and positioning.

[0032] In another embodiment, before polynomial transformation, all features need to be pre-judged to determine whether the changed measurement values belong to the measurement values of the sensors in the first sub-module or the measurement values of the sensors in the second sub-module.

[0033] This can further narrow down the fault range.

[0034] The pre-judgment also includes considering that no fault has occurred when the measurement values in the sub-modules are all unchanged.

[0035] The pre-judgment also includes determining the number of sudden changes in the measurement values of the sensors in the sub-modules. When the number is equal to 1, it is considered that the sensor is faulty, and when the number is greater than or equal to 2, it is considered that the battery is faulty.

[0036] The method can extract the measurement values collected by the sensor and the polynomial transformation of the measurement values as features, analyze the differences between normal state, internal fault, and open circuit fault. By comparing the change curves under different fault states, abnormal changes during faults can be found. Thus, the location of the fault can be accurately positioned, and effective fault diagnosis can be achieved.

[0037] The measurement values collected by the current sensor and the voltage sensor in this embodiment are denoted as A1, A2, A3, A4, V1 and V2 as features. In order to reduce the calculation complexity in fault detection, the collected features are subjected to polynomial transformation, and through simulation simulation, the importance analysis is performed on the current sensor and voltage sensor measurement values and the polynomial transformation of the measurement values over time under multiple faults, and the important features in fault classification are found. In this embodiment, “V1”, “A1-A3”, “A3-A4”, “V1-V2” and “A1 / A2” are obtained as important features for multiple fault detection. By selecting important features and performing feature processing, the effects of reducing unnecessary calculations and redundant information are achieved, thereby improving the efficiency and accuracy of fault diagnosis. Therefore, in the classification process of the ET algorithm, targeted processing and selection can be performed according to the importance of the features, thereby optimizing the performance of the fault detection system.

[0038] In this embodiment, when the first sub-module has an ISC (internal short circuit) or COC (cell open circuit) fault, A1, A2 and V1 will change at the same time. Therefore, when only one sensor reading shows a change, there may be a fault problem with this sensor. However, when all these sensor readings change, it indicates that a fault has occurred in the first sub-module. First, the extracted features are subjected to polynomial transformation processing to obtain four important features: “V1”, “A1-A3”, “A3-A4”, “V1-V2” and “A1 / A2”. Then, these extracted important features and other features are taken as inputs, the important characteristics are taken as inputs, and the change curves of the important characteristics correspond to different fault types as outputs. The ET algorithm is trained, and in the training process, the particle swarm algorithm is used to adjust the parameters of the ET algorithm to obtain the optimal classification positioning model. It should be noted that when the readings of the three sensors do not change significantly, it indicates that no fault has occurred during normal operation.

[0039] The process of adjusting the parameters of the ET algorithm using the particle swarm algorithm is as follows: first, assign initial random positions and initial random velocities to all particles in the space. Then, each particle advances according to its velocity, the known best global position in the problem space and the known best position of the particle. As the calculation proceeds, the particles explore and exploit the known favorable positions in the search space, gathering or condensing around one or more best points.

[0040] The formula used in the ET algorithm planning node is:

[0041]

[0042] where P mkis the percentage of k-class observations in node m, and H is the Gini coefficient used to split the nodes of ET algorithm;

[0043] The formula for adjusting the parameters of the ET algorithm using the particle swarm algorithm is:

[0044]

[0045] wherein and are the velocity and position of particle i in the kth generation, pbest is the single particle extreme value, gbest is the extreme value of all particles, r1 and r2 are random numbers between [0, 1]; d represents the dimension; c1 and c2 are acceleration coefficients.

[0046] The features need to be normalized when input, and the following normalization formula is used:

[0047]

[0048] wherein, x ′ mj is the normalized result of the feature, x mj is the ith data of the jth feature of the battery, is the average value of the jth feature data, S j is the standard deviation of the jth feature data.

[0049] The method can well diagnose the ISC fault, COC fault and sensor fault of the lithium battery pack, and the corresponding position. The fault diagnosis mechanism can effectively ensure the safety and stability of the lithium battery pack. It can help the operator to find and handle the fault in time, reduce the influence on the whole system, and improve the reliability and life of the battery pack.

[0050] In order to prove that the method has very good performance effect in multi-fault detection, tests are carried out, and the model trained by the data set under NEDC condition (European standard test process for evaluating automobile fuel consumption and emission. NEDC test is an experimental method based on simulated road driving cycle, which covers two different driving modes of low-speed urban driving and highway driving) of the present application performs very well in the verification data set under NEDC condition and the data set of constant current discharge, and the result accuracy is 98% and 100% respectively.

[0051] In order to verify that the method has good generalization performance, three data sets are established respectively under WLTP3 (a new WLTP working condition has been started in Europe in September 2017, the WLTP working condition is jointly formulated by experts in Japan, the United States, Europe and other countries, and the characteristic is to collect real driving working condition data in the world), ARL (urban working condition in Common Artemis Driving Cycles (CADC) working condition, wherein CADC refers to statistical analysis of a large database of European real world driving patterns in the European Artemis project) and AU (rural working condition in CADC working condition), and then the trained data set is tested. The test results show that the accuracy of fault diagnosis under the data sets of WLTP3, ARL and AU is 94%, 95% and 96% respectively. It is proved that the method has good fault detection effect under different working condition data sets.

[0052] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for diagnosing multiple faults in a lithium battery pack, characterized in that, The method includes: A diagnostic topology is established, comprising N identical sub-modules. Each sub-module includes 16 lithium iron phosphate batteries, a first current sensor, a second current sensor, and a voltage sensor. Two lithium iron phosphate batteries are connected in series to form a series group, and the connection between two lithium iron phosphate batteries forms a node. All series groups are connected in parallel. Four series groups form a first array group, and the other four series groups form a second array group. One measuring terminal of the first current sensor is connected to the four nodes of the first array group, and the other measuring terminal of the first current sensor is connected to the four nodes of the second array group. One measuring terminal of the second current sensor is connected to one node of the first array group, and the other measuring terminal is connected to one node of the second array group. The two ends of the voltage sensor are connected between any two nodes within the first or second array group. N sub-modules are connected in series; The diagnostic topology is simulated by computer, and the measured values ​​of current and voltage sensors and the polynomial transformations of these values ​​are simulated over time under different fault types. The prominent change curves corresponding to different faults and the corresponding current sensor, voltage sensor, and / or polynomial transformation are identified. An importance analysis is performed on the measured values ​​of current and voltage sensors and the polynomial transformations of these values ​​over time under multiple faults to identify important features in fault classification. These important features refer to those that contribute to all faults when a fault occurs, and whose contribution varies depending on the specific fault. The ET algorithm is trained by taking key characteristics as input and the different fault types corresponding to the change curves of key characteristics as output. During the training process, the parameters of the ET algorithm are adjusted by particle swarm optimization to obtain the optimal classification and localization model. The measured values ​​collected by the sensors during the diagnostic process are subjected to polynomial transformation to obtain important features. These important features are then input into the optimal classification and localization model for fault classification and localization.

2. The method for diagnosing multiple faults in a lithium battery pack according to claim 1, characterized in that, The polynomial transformation refers to expanding the measured values ​​of the current sensor and voltage sensor into multiple features by subtracting and comparing each pair of values.

3. The method for diagnosing multiple faults in a lithium battery pack according to claim 1, characterized in that, Before performing a polynomial transformation on the measured values ​​collected by the sensors during the diagnostic process, a pre-judgment is performed. The pre-judgment includes determining which sub-module's sensor's measured value the changed measured value belongs to.

4. The method for diagnosing multiple faults in a lithium battery pack according to claim 3, characterized in that, The pre-judgment also includes considering that no fault has occurred when the measured values ​​in the judgment submodule have not changed.

5. The method for diagnosing multiple faults in a lithium battery pack according to claim 3, characterized in that, The pre-judgment also includes determining the number of sudden changes in the measured values ​​of the sensors in the submodule. When the number is equal to 1, it is considered a sensor fault; when the number is greater than or equal to 2, it is considered a battery fault.

6. The method for diagnosing multiple faults in a lithium battery pack according to claim 3, characterized in that, The parameters of the ET algorithm are adjusted using the particle swarm optimization algorithm, including the number of decision trees, the maximum depth of the decision trees, the minimum number of samples required for node splitting, and the minimum number of samples required for leaf nodes.

Images