Battery fault diagnosis method based on knowledge graph

Abstract

The invention discloses a battery fault diagnosis method based on a knowledge graph, and relates to the technical field of battery fault diagnosis, and the method comprises the following steps: obtaining real-time data and historical data of a to-be-diagnosed battery, and carrying out the preprocessing of the obtained data; according to the preprocessed data, identifying whether the battery has a fault symptom; inputting the recognized fault symptom as input into a preset knowledge graph for query to obtain a query result, and obtaining a reasoning result according to the query result and based on Bayesian network reasoning of the preset knowledge graph; and outputting a battery fault diagnosis report according to the query result and the reasoning result. According to the method, multiple fault symptoms and related factors can be comprehensively considered for diagnosis, the knowledge graph can be combined with other symptoms appearing at the same time and historical information of the battery to more accurately judge the fault type, the potential relation is mined through the Bayesian network reasoning function, the diagnosis precision is improved, and the reliability and efficiency of fault diagnosis are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery fault diagnosis, and in particular to a battery fault diagnosis method based on a knowledge graph. Background Art

[0002] With the increasing attention paid to environmental issues, new energy industries represented by electric vehicles have developed rapidly in line with the trend. As the core of electric vehicles, battery safety issues are often the primary concern of consumers. There are many types of battery failures that induce battery safety issues, the causes of failures are complex, the troubleshooting time is low, the fault detection is difficult, and the maintenance time is long. Therefore, there are many difficulties in battery fault diagnosis, and it also puts forward higher requirements for practitioners engaged in battery maintenance.

[0003] There are three main challenges in fault diagnosis of lithium-ion batteries: first, it is difficult to locate and handle battery faults, which is mainly due to the complex internal chemical reactions of lithium-ion batteries and the lack of obvious fault characterization in the early stages of a fault; second, the efficiency of fault diagnosis is low, which is because the diagnosis of battery faults currently relies heavily on manual diagnosis and the experience of maintenance personnel, there are relatively few testing methods, and there is a lack of battery full life cycle monitoring methods; finally, the technical threshold is high, lithium-ion battery fault diagnosis involves multidisciplinary professional knowledge, the learning cycle is long, and there is a lack of experienced senior talents.

[0004] With the rise of artificial intelligence in recent years, the technological progress of intelligent algorithms has been promoted, as well as the generation and application of knowledge graphs, providing ideas for solving the current difficulties in lithium-ion battery fault diagnosis.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In response to the problems in the related technology, the present invention proposes a battery fault diagnosis method based on knowledge graph to overcome the above-mentioned technical problems existing in the existing related technology.

[0007] The technical solution of the present invention is achieved in this way:

[0008] A battery fault diagnosis method based on knowledge graph includes the following steps:

[0009] Acquire real-time data and historical data of the battery to be diagnosed, and pre-process the acquired data;

[0010] Identify whether the battery has a fault symptom based on the preprocessed data, which at least includes: comparing the preprocessed data with a preset safety threshold range, and if the processed data does not meet the safety threshold range, it indicates that the current processed data has a fault symptom;

[0011] The identified fault symptom is used as input and input into a preset knowledge graph for query to obtain a query result, wherein the query result at least includes: a fault type, a fault cause and associated battery component information related to the fault symptom;

[0012] Obtaining inference results according to the query results and based on the Bayesian network inference of the preset knowledge graph, wherein the inference results at least include: calculating the posterior probabilities of different fault types by using known fault symptoms and prior probabilities;

[0013] Based on the query results and inference results, a battery fault diagnosis report is output.

[0014] Furthermore, the real-time data includes the current operating temperature, voltage, and current; the historical data includes past charging records and discharging records.

[0015] Furthermore, the preprocessing of the acquired data includes: performing data cleaning to remove noise data and outliers.

[0016] Furthermore, the diagnostic report includes: fault type, fault cause, fault severity and corresponding maintenance suggestions.

[0017] Furthermore, the preset knowledge graph includes the following steps:

[0018] Acquire battery characteristic data and battery failure data in advance;

[0019] Extract entity types, relationship types and attributes based on battery feature data and battery failure data;

[0020] The extracted entity types, relationship types and attributes are fused, and a graph database is used to store the knowledge graph.

[0021] Furthermore, the battery characteristic data includes: basic parameters, production data and operation data of the battery.

[0022] Furthermore, the battery fault data includes: fault type and fault symptoms.

[0023] Beneficial effects of the present invention:

[0024] The present invention obtains real-time data and historical data of a battery to be diagnosed, pre-processes the obtained data, identifies whether the battery has fault symptoms based on the pre-processed data, compares the pre-processed data with a preset safety threshold range, and if the processed data does not meet the safety threshold range, it indicates that the current processed data has fault symptoms, and uses the identified fault symptoms as input, inputs them into a preset knowledge graph for query, obtains query results, fault types, fault causes, and associated battery component information related to the fault symptoms, obtains reasoning results based on the query results and based on Bayesian network reasoning of the preset knowledge graph, and outputs a battery fault diagnosis report based on the query results and reasoning results, so as to clearly understand the relationship between the various components of the battery and how these components interact with external factors to cause faults, which helps diagnostic personnel to grasp the occurrence mechanism of battery failure as a whole. Once the fault symptoms are identified, all information related to them can be quickly queried, including possible fault causes, fault locations, solutions, etc. At the same time, it can display the reasoning process and basis of fault diagnosis. During the diagnosis process, each diagnostic conclusion is drawn based on the entities, relationships and reasoning rules in the knowledge graph, avoiding the tedious process of manually consulting a large amount of information to find fault clues in traditional diagnostic methods, greatly saving time and energy, thereby improving the efficiency of fault diagnosis.

[0025] In addition, not only can multiple fault symptoms and related factors be considered comprehensively for diagnosis, but in battery fault diagnosis, a single parameter change may be caused by multiple reasons. The knowledge graph can combine other co-occurring symptoms and the historical information of the battery to more accurately determine the fault type. At the same time, the Bayesian network reasoning function is used to mine potential relationships to further improve the diagnostic accuracy. The cause of the fault hidden behind the known symptoms can be discovered, thereby improving the reliability and efficiency of fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0027] Figure 1 is a flow chart of a battery fault diagnosis method based on a knowledge graph according to an embodiment of the present invention;

[0028] Figure 2 is a schematic diagram of a process scenario of a battery fault diagnosis method based on a knowledge graph according to an embodiment of the present invention;

[0029] Figure 3It is a knowledge graph scenario diagram of a battery fault diagnosis method based on a knowledge graph according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0031] According to an embodiment of the present invention, a battery fault diagnosis method based on a knowledge graph is provided.

[0032] like Figure 1-Figure 3 As shown, the battery fault diagnosis method based on the knowledge graph according to an embodiment of the present invention includes the following steps:

[0033] Step S1, obtaining real-time data and historical data of the battery to be diagnosed, and preprocessing the obtained data;

[0034] Among them, real-time data includes current operating temperature, voltage, and current, and historical data includes past charging records and discharging records.

[0035] This technical solution pre-processes the acquired data, including: performing data cleaning to remove noise data and outliers.

[0036] Specifically, in the application, if the current sensor has a short-term fault, resulting in a spike in the collected current data, these abnormal data are removed by filtering. At the same time, data standardization is used to convert data of different units and magnitudes into the same standard range to facilitate subsequent analysis.

[0037] Step S2, identifying whether the battery has a fault symptom based on the preprocessed data, which at least includes: comparing the preprocessed data with a preset safety threshold range, and if the processed data does not meet the safety threshold range, it indicates that the current processed data has a fault symptom;

[0038] Specifically, in application, if the battery temperature exceeds the safety threshold range, which is represented by the normal operating temperature of -20°C to 60°C, it is considered that abnormal temperature symptoms have occurred; if the battery voltage drops too quickly and exceeds the safety threshold range, which is represented by the drop speed corresponding to the normal discharge curve, it is judged to be abnormal.

[0039] Step S3, taking the identified fault symptom as input, inputting it into a preset knowledge graph for query, and obtaining the query result, wherein the query result at least includes: the fault type, fault cause and associated battery component information related to the fault symptom.

[0040] This technical solution is specific, for example: when a symptom of battery heating is found, a query in the knowledge graph can reveal that it may be caused by short circuit, overcharging or abnormal chemical reaction inside the battery, and these causes may be related to components such as the battery's electrodes and electrolyte.

[0041] Step S4, obtaining a reasoning result according to the query result and based on the Bayesian network reasoning of the preset knowledge graph, wherein the reasoning result at least includes: calculating the posterior probability of different fault types by using known fault symptoms and prior probabilities;

[0042] This technical solution, through Bayesian network reasoning, can more accurately locate the cause and location of the fault.

[0043] Step S5: outputting a battery fault diagnosis report according to the query result and the inference result.

[0044] The technical solution, specifically, the diagnostic report includes: fault type, fault cause, fault severity and corresponding maintenance suggestions, among which the fault type such as determining whether it is a short circuit fault or a capacity attenuation fault, the fault cause such as defects in the battery production process, long-term incorrect usage, etc., the fault severity such as minor, moderate, severe, which can be divided according to the impact of the fault on battery performance, and corresponding maintenance suggestions such as replacing battery components and adjusting charging strategies.

[0045] In addition, if Figure 3 As shown, for the above knowledge graph, the following steps are included:

[0046] Step S301, pre-acquiring battery characteristic data and battery fault data;

[0047] Among them, battery characteristic data include: basic battery parameters, such as battery type, capacity, and nominal voltage; production data, such as production date, production batch, and production process; operating data, such as number of charges, depth of discharge, operating temperature, operating current, and operating voltage.

[0048] Among them, battery failure data includes failure types, such as short circuit, overcharging, over-discharging, increased internal resistance, and various parameter values ​​when the failure occurs; failure symptoms, such as battery heating, sudden reduction in cruising range, abnormal charging speed, and corresponding maintenance records.

[0049] Specifically, in application, the data sources are divided into three methods: laboratory testing equipment, laboratory testing equipment host computer and manual import report. Based on these three methods, various requirements for battery characteristic data and battery failure data can be basically covered. The specific methods are described as follows:

[0050] Among them, laboratory testing equipment: equipped with a data acquisition smart box (referred to as data acquisition box), based on some standard data acquisition protocols, such as Web API, Modbus / PLC, serial port protocol (RS-485, etc.), etc., the data acquisition box can actively collect or accept data push from the testing equipment, collect the test data generated during the experiment into the data acquisition box, and then forward the data to the server.

[0051] Among them, the host computer of laboratory testing equipment: If the testing equipment is equipped with a host computer, data collection can be performed through plug-ins. Install the plug-in on the host computer, and after the host computer obtains the test data, it transfers the data to the plug-in, and the plug-in then forwards the data to the server.

[0052] Among them, manual import report: supports manual import of experimental test data into the server, and supports xlsx / xls / CSV format data.

[0053] Step S302, extracting entity types, relationship types and attributes according to battery feature data and battery fault data.

[0054] Among them, entity types include: batteries, battery components (such as electrodes, electrolytes, and diaphragms), fault types, and fault symptoms.

[0055] Among them, the relationship types include: the composition relationship between batteries and battery components, for example, batteries contain electrodes; the association relationship between parameters and batteries, for example, the capacity of a battery is a certain value; the association relationship between fault types and fault symptoms, specifically, short circuits can cause battery heating.

[0056] Among them, attributes include: mapping characteristics of entity types and relationship types, such as: battery capacity, severity of fault symptoms, etc.

[0057] Step S303, fusing the extracted entity types, relationship types and attributes, and using a graph database to store the knowledge graph;

[0058] In this technical solution, if different data sources have different descriptions of the same fault symptom, they need to be unified. The ontology method is used to define the hierarchy and semantic relationship of knowledge to ensure the consistency of knowledge. At the same time, the graph database Neo4j is used, which can efficiently store and query the relationship between entities.

[0059] In addition, for the above-mentioned knowledge graph construction, when applied, it can also provide:

[0060] Query and retrieval services, including: Query interface: Provides a query interface to allow users to query information in the knowledge graph, such as test results, device status, etc. Retrieval engine: Uses a retrieval engine (Elasticsearch) for efficient data retrieval.

[0061] Visualization services, visualization tools: Use visualization tools (Gephi, D3.js) to display entities and relationships in the knowledge graph. Interactive interface: Develop an interactive interface to allow users to browse and explore the knowledge graph more intuitively.

[0062] In summary, with the help of the above technical solution of the present invention, the following effects can be achieved:

[0063] The present invention obtains real-time data and historical data of a battery to be diagnosed, pre-processes the obtained data, identifies whether the battery has fault symptoms based on the pre-processed data, compares the pre-processed data with a preset safety threshold range, and if the processed data does not meet the safety threshold range, it indicates that the current processed data has fault symptoms, and uses the identified fault symptoms as input, inputs them into a preset knowledge graph for query, obtains query results, fault types, fault causes, and associated battery component information related to the fault symptoms, obtains reasoning results based on the query results and based on Bayesian network reasoning of the preset knowledge graph, and outputs a battery fault diagnosis report based on the query results and reasoning results, so as to clearly understand the relationship between the various components of the battery and how these components interact with external factors to cause faults, which helps diagnostic personnel to grasp the occurrence mechanism of battery failure as a whole. Once the fault symptoms are identified, all information related to them can be quickly queried, including possible fault causes, fault locations, solutions, etc. At the same time, it can display the reasoning process and basis of fault diagnosis. During the diagnosis process, each diagnostic conclusion is drawn based on the entities, relationships and reasoning rules in the knowledge graph, avoiding the tedious process of manually consulting a large amount of information to find fault clues in traditional diagnostic methods, greatly saving time and energy, thereby improving the efficiency of fault diagnosis.

[0064] In addition, not only can multiple fault symptoms and related factors be considered comprehensively for diagnosis, but in battery fault diagnosis, a single parameter change may be caused by multiple reasons. The knowledge graph can combine other co-occurring symptoms and the historical information of the battery to more accurately determine the fault type. At the same time, the Bayesian network reasoning function is used to mine potential relationships to further improve the diagnostic accuracy. The cause of the fault hidden behind the known symptoms can be discovered, thereby improving the reliability and efficiency of fault diagnosis.

[0065] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. After considering the disclosure of the specification and the examples, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and the examples are only regarded as exemplary, and the true scope and spirit of the present disclosure are indicated by the claims.

[0066] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A battery fault diagnosis method based on knowledge graph, characterized in that: The following steps are involved: Acquire real-time data and historical data of the battery to be diagnosed, and pre-process the acquired data; Identify whether the battery has a fault symptom based on the preprocessed data, which at least includes: comparing the preprocessed data with a preset safety threshold range, and if the processed data does not meet the safety threshold range, it indicates that the current processed data has a fault symptom; The identified fault symptom is used as input and input into a preset knowledge graph for query to obtain a query result, wherein the query result at least includes: a fault type, a fault cause and associated battery component information related to the fault symptom; Obtaining inference results according to the query results and based on the Bayesian network inference of the preset knowledge graph, wherein the inference results at least include: calculating the posterior probabilities of different fault types by using known fault symptoms and prior probabilities; Based on the query results and inference results, a battery fault diagnosis report is output.

2. The battery fault diagnosis method based on knowledge graph according to claim 1 is characterized in that: The real-time data includes the current operating temperature, voltage, and current; the historical data includes past charging records and discharging records.

3. The battery fault diagnosis method based on knowledge graph according to claim 1 is characterized in that: The preprocessing of the acquired data includes: performing data cleaning to remove noise data and abnormal values.

4. The battery fault diagnosis method based on knowledge graph according to claim 1 is characterized in that: The diagnostic report includes: fault type, fault cause, fault severity and corresponding maintenance suggestions.

5. The battery fault diagnosis method based on knowledge graph according to claim 1 is characterized in that: The preset knowledge graph includes the following steps: Acquire battery characteristic data and battery failure data in advance; Extract entity types, relationship types and attributes based on battery feature data and battery failure data; The extracted entity types, relationship types and attributes are fused, and a graph database is used to store the knowledge graph.

6. The battery fault diagnosis method based on knowledge graph according to claim 5 is characterized in that: The battery characteristic data includes: basic parameters, production data and operation data of the battery.

7. The battery fault diagnosis method based on knowledge graph according to claim 5 is characterized in that: The battery failure data includes: failure type and failure symptoms.

Images