CN-116872732-B - Method and device for determining fault type of single battery and engineering vehicle

CN116872732BCN 116872732 BCN116872732 BCN 116872732BCN-116872732-B

Abstract

The embodiment of the application provides a method and device for determining a single battery fault type, an engineering vehicle, a processor and a storage medium. The method comprises the steps of collecting the running state of the engineering vehicle in a preset time period and the state data of each single battery in the battery pack, determining the target single battery with faults in the battery pack according to the running state and the state data, and sequentially inputting the state data and the running state of each target single battery into a trained long-short-period memory artificial neural network so as to determine the fault type of each target single battery through the trained long-short-period memory artificial neural network. According to the method, the running state of the new energy engineering vehicle is considered in the fault diagnosis process, and the situation of misdiagnosis caused by state transformation can be avoided.

Inventors

CHEN LIANG
TONG XING
Liao Deng
ZHOU ZHIZHONG
Zou chenyang

Assignees

中科云谷科技有限公司

Dates

Publication Date: 20260508
Application Date: 20230615

Claims (12)

1. A method for determining a type of cell failure, characterized by being applied to an engineering vehicle including at least one battery pack, the method comprising: collecting the running state of the engineering vehicle in a preset time period and the state data of each single battery in the battery pack, wherein the battery pack comprises a plurality of single batteries, the running state at least comprises a starting state, a distribution state, a normal transportation state and an emergency stop state, and the state data at least comprises the voltage, the current, the temperature change rate, the battery charge state and the internal resistance of the single batteries; determining a failed target single battery in the battery pack according to the running state and the state data; The state data and the running state of each target single battery are sequentially input into a trained long-short-period memory artificial neural network, so that the fault type of each target single battery is determined through the trained long-short-period memory artificial neural network; wherein the determining, according to the operation state and the state data, the failed target single battery in the battery pack includes: any two single batteries in the battery pack are formed into a battery set; According to the state data of each single battery in the battery set, the Manhattan distance, the Euclidean distance and the Chebyshev distance between two single batteries in the battery set are respectively determined; Determining a weighting coefficient corresponding to each type of feature distance; Carrying out weighted summation on each type of characteristic distance according to the weighting coefficient, and finally determining the obtained value as the characteristic distance between two single batteries in the battery set; Determining a target battery set with the characteristic distance larger than a preset distance threshold; and determining the single batteries included in at least two target battery sets as the target single batteries.
2. The method for determining a cell failure type according to claim 1, wherein the determining a characteristic distance between two cells in the battery set from the state data of each cell in the battery set includes calculating the characteristic distance according to formula (1) : (1) Wherein p is the number of battery characteristics for measuring distance, A feature value expressed as a p-th feature of the first unit cell in the battery set, The p-th characteristic value of the second single battery in the battery set is represented, h represents a coefficient of the characteristic distance, the value changes along with the type of the characteristic distance, and the values of h are respectively 1, 2 and infinity and respectively correspond to the Manhattan distance, the Euclidean distance and the Chebyshev distance.
3. The method for determining a cell failure type according to claim 1, wherein the operating conditions include at least a start-up condition, a distribution condition, a normal transport condition, and an emergency shutdown condition; The step of sequentially inputting the state data and the running state of each target single battery to a trained long-short-term memory artificial neural network, so as to determine the fault type of each target single battery through the trained long-short-term memory artificial neural network comprises the following steps: extracting state data of each target single battery and original characteristics of the running state of each target single battery; performing feature interaction on any two original features to obtain new features; combining the original features with the new features, and carrying out normalization processing after feature selection; And inputting the processed characteristics into a trained long-short-period memory artificial neural network so as to determine the fault type of each target single battery through the trained long-short-period memory artificial neural network.
4. The method for determining the fault type of the single battery according to claim 1, wherein the long-short-term memory artificial neural network is an LSTM neural network, wherein the hidden layers of the LSTM neural network include a first hidden layer and a second hidden layer, wherein the number of LSTM neural units in the first hidden layer is the number of delay variables, the second hidden layer includes 8 neurons, a ReLU activation function is adopted, the fault types are classified by Softmax, and the number of output nodes is the same as the number of the fault types and corresponds to each fault type.
5. The method for determining a cell failure type according to claim 4, further comprising a training step of a long-term memory artificial neural network, the training step comprising: acquiring historical operating states of battery packs of a plurality of engineering vehicles and historical state data of single batteries of each battery pack; extracting historical features of the historical operating state and the historical state data; determining the historical fault type of a single battery with faults, and adding a fault type mark for each historical feature according to the historical fault type; Classifying the historical features, and dividing the historical features into delay variables and non-delay variables; performing characteristic interaction on the delay variable and the non-delay variable to obtain a new delay variable and a new non-delay variable; inputting a delay variable to the first hidden layer to train the long-term and short-term memory artificial neural network; And inputting non-delay variables and the output of the first hiding layer to the second hiding layer so as to train the long-term and short-term memory artificial neural network.
6. The method for determining a cell failure type of claim 1, wherein the work vehicle includes a display device, the method further comprising: and sending the fault type, the running state and/or the state data to the display equipment for display.
7. The method for determining a cell failure type of claim 1, wherein the engineering vehicle is a pump truck.
8. A processor configured to perform the method for determining a cell failure type according to any one of claims 1 to 7.
9. An apparatus for determining a type of cell failure, comprising: the plurality of sensors are arranged on the battery pack and are used for collecting state data of each single battery of the battery pack; A communication module mounted on the battery pack for transmitting the status data and the operation status of the engineering vehicle, and The processor of claim 8.
10. Engineering vehicle, characterized in that it comprises a device for determining the type of single battery fault according to claim 9.
11. The work vehicle of claim 10, wherein the work vehicle is a pump truck.
12. A machine-readable storage medium having instructions stored thereon, which when executed by a processor, cause the processor to be configured to perform the method for determining a cell failure type according to any of claims 1 to 7.

Description

Method and device for determining fault type of single battery and engineering vehicle Technical Field The application relates to the technical field of battery fault diagnosis, in particular to a method and device for determining a single battery fault type, an engineering vehicle, a storage medium and a processor. Background New energy automobiles are one of the most important ways to achieve carbon emission reduction. In recent years, new energy automobiles in China enter a high-speed development stage. The power battery is used as the most important component of the new energy pump truck, and the running state of the power battery directly influences the overall performance of the vehicle. The power battery fault is light and affects the construction efficiency and operation experience of the new energy pump truck, and the power battery fault is heavy and causes short circuit or explosion, so that fire disaster is caused, and the safety of the new energy pump truck and constructors is greatly threatened. The fault diagnosis is carried out on the power battery of the new energy pump truck, so that the fault diagnosis can be carried out, major accidents caused by the fault of the power battery can be prevented, the service life of the power battery of the new energy pump truck can be prolonged, the damage to other parts of the new energy pump truck caused by the fault of the battery can be reduced, and the running cost of the vehicle related to the power battery can be reduced. However, in the prior art, fault diagnosis is generally performed on the battery pack of the whole vehicle, so that the fault diagnosis accuracy is low, and the maintenance difficulty is increased. Disclosure of Invention The embodiment of the application aims to provide a method and device for determining the fault type of a single battery, an engineering vehicle, a storage medium and a processor. To achieve the above object, a first aspect of the present application provides a method for determining a type of a single battery fault, applied to an engineering vehicle, the engineering vehicle including at least one battery pack, the method comprising: Collecting the running state of the engineering vehicle in a preset time period and the state data of each single battery in a battery pack, wherein the battery pack comprises a plurality of single batteries; determining a failed target single battery in the battery pack according to the running state and the state data; the state data and the running state of each target single battery are sequentially input into the trained long-period memory artificial neural network, so that the fault type of each target single battery is determined through the trained long-period memory artificial neural network. In the embodiment of the application, the running state at least comprises a starting state, a distribution state, a normal transportation state and an emergency stop state, the state data at least comprises voltages, currents, temperatures, temperature change rates, battery charge states and internal resistances of the single batteries, the target single batteries which are in failure in the battery pack are determined according to the running state and the state data, any two single batteries in the battery pack are formed into a battery set, the characteristic distance between the two single batteries in the battery set is determined according to the state data of each single battery in the battery set, the target battery set with the characteristic distance larger than a preset distance threshold is determined, and the single batteries at least included in the two target battery sets are determined to be target single batteries. In the embodiment of the application, the types of the characteristic distances at least comprise Manhattan distance, euclidean distance and Chebyshev distance, the characteristic distances between two single batteries in the battery set are determined according to the state data of each single battery in the battery set, the Manhattan distance, euclidean distance and Chebyshev distance between the two single batteries in the battery set are respectively determined according to the state data of each single battery in the battery set, the weighting coefficient corresponding to each type of characteristic distance is determined, the weighting summation is carried out on each type of characteristic distance according to the weighting coefficient, and the obtained value is finally determined as the characteristic distance between the two single batteries in the battery set. In an embodiment of the present application, determining the feature distance between two cells in the battery set from the state data of each cell in the battery set includes calculating the feature distance d (i, j) according to formula (1): Wherein p is the number of battery characteristics for measuring the distance, x ip is the characteristic value of the p-th characteristic of the first