Engine fault diagnosis method, device and equipment based on large model and medium

Abstract

The invention provides an engine fault diagnosis method and device based on a large model, equipment and a medium, and relates to the technical field of engines, and the method comprises the steps: obtaining multi-source data of a to-be-diagnosed engine, extracting fault description information, and inputting the fault description information into a judgment model; when the collection instruction is output, engine operation data under the idling working condition are collected; calculating an average value of the collected engine operation data, and converting the average value into a natural language description of the operation data; combining the natural language description of the multi-source data with the natural language description of the operation data to obtain a final natural language description, inputting the final natural language description into a pre-training model, and outputting feature vector data; and inputting the feature vector data into a pre-trained engine fault diagnosis model, and outputting fault causes and maintenance suggestions. According to the method and the device, diagnosis can be carried out by using more source data, so that the fault cause part can be positioned more quickly and more accurately.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of engines, in particular to an engine fault diagnosis method and device based on a large model, a diagnosis equipment and a storage medium. BACKGROUND

[0002] In the fields of industrial production and transportation, the engine as the core power device directly determines the overall working efficiency and safety performance of the equipment. Once the engine fails, not only will it cause the equipment to shut down, resulting in huge economic losses, but also may cause safety accidents, threatening the safety of personnel and property. Therefore, quickly and accurately determining the high-incidence fault cause of the engine, and then carrying out targeted fault troubleshooting and maintenance work, has always been the core demand in the field of engine operation and maintenance. With the continuous penetration of intelligent technology in the industrial field, the fault diagnosis method based on data and model gradually replaces the traditional diagnosis method relying on manual experience, and becomes the mainstream technical direction of determining the fault cause of the engine. In the prior art, a method for determining the high-incidence fault cause of an engine has been disclosed. The method obtains the work order, basic information, maintenance report and sales archives of the current engine, extracts the current engine fault information from the work order, and inputs the above fault information, basic information, maintenance report and sales archives into a pre-trained engine high-incidence fault cause determination model. The model outputs the high-incidence fault cause of the current engine and the corresponding fault probability, and finally determines the target fault cause according to the fault probability and generates the component fault troubleshooting information, aiming to realize efficient and accurate diagnosis of the high-incidence fault cause of the engine and provide guidance for subsequent maintenance work. However, in actual application, the above prior art still has obvious defects, which greatly limits the upper limit of the fault diagnosis effect and performance, and it is difficult to meet the high-precision and high-reliability requirements of engine fault diagnosis under complex working conditions, which is specifically reflected in the following two aspects: On the one hand, the lack of data diversity leads to poor fault diagnosis effect. The prior art only uses the current engine fault information, the current engine basic information, the current engine maintenance report and the current engine sales archives as the data input source of the model. The information dimension contained in such data is relatively single, and it cannot fully and effectively represent the complex characteristics of engine faults. For example, the occurrence of engine failure is often closely related to the working condition change, environmental influence and aging trend of parts during long-term operation. However, the existing input data lack information reflecting these dynamic change processes, making it difficult for the model to capture the potential associated factors of fault occurrence. When facing some atypical faults or faults induced by multiple factors, the model is prone to fault cause judgment deviation due to the limitation of input information, resulting in a significant reduction in fault diagnosis effect. On the other hand, model selection limits the data input range, which in turn restricts the upper limit of model performance. In the prior art, the selection of engine high-failure cause determination models focuses on traditional classification models and conventional machine learning models, such as support vector machine (SVM) models, gradient boosting (XGBoost) models, deep neural networks (DNN), convolutional neural networks (CNN), BP neural networks (BPNN), probabilistic neural networks (PNN), and random forest models. Such models have significant limitations in data input, only supporting numerical feature input. For a large number of non-numerical features existing in the engine operation process, such as fault description (such as "engine water leakage" and "abnormal noise"), production plant identification (such as "Plant 1" and "Plant 2"), and equipment purpose (such as "tractor" and "loader"), etc. Type features, tedious feature encoding preprocessing is required to convert them into 0, 1, 2, etc. Numerical form before inputting into the model, which not only increases the complexity of data processing, but also may cause loss of part of the feature information. More importantly, such models do not support time series data as input at all, while structured time series data such as real-time running data (such as speed, oil pressure, and temperature sequence data changing over time) and historical working condition fluctuation data generated during engine operation contain key information of engine running state changes and are important basis for determining fault cause. Since the model cannot use such data, it cannot analyze the development process and causes of the fault from a dynamic perspective, resulting in insufficient learning of fault features by the model, severely limiting the upper limit of performance, and making it difficult to adapt to complex and variable engine fault diagnosis scenarios. In summary, the existing engine high-failure cause determination method cannot fully mine fault-related information and effectively utilize multi-type data due to insufficient data diversity and model selection limitations, resulting in fault diagnosis effect and performance that cannot meet actual needs, and there is an urgent need for a technical solution that can overcome the above defects to improve the accuracy and reliability of engine high-failure cause determination. SUMMARY

[0003] The purpose of the present application is to provide a large model-based engine fault diagnosis method, device, diagnosis equipment and storage medium to improve the accuracy and reliability of engine high-failure cause determination.

[0004] In a first aspect, the embodiments of the present application provide a large model-based engine fault diagnosis method, comprising: Obtaining multi-source data of an engine to be diagnosed, and integrating the multi-source data into wide table data; the multi-source data includes engine basic data, fault code basic information, work order information, sales archives and engine maintenance reports; extracting fault description information from the wide table data, and inputting the fault description information into a pre-trained engine operation data collection and determination model to output a collection instruction or a non-collection instruction; when the collection instruction is output, the engine to be diagnosed is controlled to start and collect engine operation data under an idle speed working condition, the engine operation data being time series structured data; calculating an average value of the collected engine operation data according to a preset time window, converting the average value into a natural language description of the operation data, and storing the natural language description of the operation data; textualizing the wide table data to obtain a natural language description of multi-source data, merging the natural language description of the multi-source data and the natural language description of the operation data to obtain a final natural language description, and inputting the final natural language description into a pre-training model to output feature vector data; inputting the feature vector data into a pre-trained engine fault diagnosis model to output a fault cause, a corresponding fault probability, and a maintenance suggestion.

[0005] In a possible implementation, the engine basic data includes an engine number, an engine series, a production factory, and a production date. The fault code basic information includes a fault code, a fault reason, a fault component, and a fault solution method. The work order information includes a work order number, a fault description, and a fault date. The sales archives include a factory date and a device purpose. The engine maintenance report includes a work order number, an engine number, and a spare part name.

[0006] In a possible implementation, the engine operation data includes at least one of an engine speed, a torque, a water temperature, and an oil pressure.

[0007] In a possible implementation, after the natural language description of the operation data is stored, the method further includes: sending prompt information indicating that the natural language description conversion is completed.

[0008] In a possible implementation, the pre-training model adopts a Transformer model, a Bag-of-Words model, a Word2Vec model, or a Doc2Vec model.

[0009] In a second aspect, an engine fault diagnosis device based on a large model is provided, and the device includes: The acquisition module is configured to acquire multi-source data of an engine to be diagnosed, and integrate the multi-source data into wide table data; the multi-source data includes engine basic data, fault code basic information, work order information, sales archives, and engine maintenance reports; The determination module is configured to extract fault description information from the wide table data, and input the fault description information into a pre-trained engine operation data acquisition determination model to output an acquisition instruction or a non-acquisition instruction; when the acquisition instruction is output, the engine to be diagnosed is started and engine operation data under an idle speed working condition is acquired, and the engine operation data is time series structured data. The conversion module is configured to calculate an average value of the acquired engine operation data according to a preset time window, convert the average value into a natural language description of the operation data, and store the natural language description of the operation data. The feature module is configured to perform text processing on the wide table data to obtain a natural language description of the multi-source data, combine the natural language description of the multi-source data with the natural language description of the operation data to obtain a final natural language description, and input the final natural language description into a pre-training model to output feature vector data. The diagnosis module is configured to input the feature vector data into a pre-trained engine fault diagnosis model to output a fault cause, a corresponding fault probability, and a maintenance suggestion.

[0010] In a possible implementation, the engine basic data includes an engine number, an engine series, a production factory, and a production date. The fault code basic information includes a fault code, a fault reason, a fault component, and a fault solving method. The work order information includes a work order number, a fault description, and a fault date. The sales archives include a factory date and a device purpose. The engine maintenance report includes a work order number, an engine number, and a spare part name.

[0011] In a possible implementation, the engine operation data includes at least one of an engine speed, a torque, a water temperature, and an oil pressure.

[0012] In a possible implementation, the conversion module is further configured to, after storing the natural language description of the operation data, send prompt information indicating that the natural language description conversion is completed.

[0013] In a possible implementation, the pre-training model adopts a Transformer model, a Bag-of-Words model, a Word2Vec model, or a Doc2Vec model.

[0014] In a third aspect, an embodiment of the present application provides a diagnosis device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the method of the first aspect of the present application when executing the computer program.

[0015] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores computer readable instructions executable by a processor to implement the method of the first aspect of the present application.

[0016] The engine fault diagnosis method, device, diagnosis device and storage medium based on a large model provided by the present application are provided. The multi-source data of an engine to be diagnosed is acquired, and the multi-source data is integrated into wide table data. Fault description information is extracted from the wide table data, and the fault description information is input into a pre-trained engine operation data acquisition and determination model to output an acquisition instruction or a non-acquisition instruction. When the acquisition instruction is output, the engine to be diagnosed is started and controlled to acquire engine operation data under an idle speed operating condition. The average value of the acquired engine operation data is calculated according to a preset time window, and the average value is converted into a natural language description of the operation data, and the natural language description of the operation data is stored. The wide table data is textually processed to obtain a natural language description of the multi-source data, and the natural language description of the multi-source data is merged with the natural language description of the operation data to obtain a final natural language description. The final natural language description is input into a pre-training model to output feature vector data. The feature vector data is input into a pre-trained engine fault diagnosis model to output a fault cause, a corresponding fault probability and a repair suggestion. Compared with the prior art, the engine fault diagnosis model of the present application can utilize more source data for diagnosis, especially fault code data and engine operation data, to provide high-value information for fault diagnosis, and to realize faster and more accurate fault cause positioning. BRIEF DESCRIPTION OF DRAWINGS

[0017] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The accompanying drawings are included to provide a description of the preferred embodiments, and are not intended to limit the scope of the present application. Moreover, like reference numerals designate like parts throughout the several views in the drawings. In the drawings: Figure 1 A flowchart of an engine fault diagnosis method based on a large model provided by the present application is shown; Figure 2 A structural schematic diagram of an engine fault diagnosis device based on a large model provided by the present application is shown. DETAILED DESCRIPTION

[0018] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be embodied in various forms without being limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure will be thoroughly and completely understood, and will fully convey the scope of the present disclosure to those skilled in the art.

[0019] It should be noted that the technical terms or scientific terms used in the present application should be understood as their ordinary meanings understood by those skilled in the art, unless otherwise specified.

[0020] In addition, the terms "first" and "second" and the like are used to distinguish different objects, rather than to describe a particular order. Furthermore, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus including a series of steps or units is not limited to the listed steps or units, but can optionally include other steps or units not listed or can optionally include other steps or units inherent to such processes, methods, products, or apparatus.

[0021] The embodiments of the present application provide a large model-based engine fault diagnosis method and device, a diagnosis device, and a computer readable storage medium, which will be described below with reference to the accompanying drawings.

[0022] Please refer to Figure 1 which shows a flowchart of a large model-based engine fault diagnosis method provided by the present application. As Figure 1 indicated, the method includes the following steps: S101, obtaining multi-source data of an engine to be diagnosed, and integrating the multi-source data into wide table data; the multi-source data includes engine basic data, fault code basic information, work order information, sales archives, and engine repair reports; Specifically, the fault diagnosis device can be connected with an engine OBD interface to obtain the engine basic data, fault code basic information, work order information, sales archives, and repair reports. The OBD interface is a vehicle-mounted diagnosis interface conforming to relevant standards, through which bidirectional communication with the engine ECU is established, not only to obtain multi-source data of the engine, but also to send instructions to control the engine to enter a preset test working condition.

[0023] In actual application, the OBD interface can be integrated with a level conversion circuit and an anti-interference module. The level conversion circuit converts the 5V signal output by the ECU into 3.3V for processing by the data acquisition module, and the anti-interference module uses a TVS diode and a common-mode inductor to suppress electromagnetic interference.

[0024] The specific content of each type of data is as follows: The engine basic data includes engine number, engine series, manufacturer, and production date; The fault code basic information includes fault code, fault reason, fault component, and fault solution method; The work order information includes work order number, fault description, and fault date; The sales archives include factory date and equipment purpose; The engine maintenance report includes work order number, engine number, and spare part name.

[0025] In the field of data processing and model application, wide table data is an important data organization form, and its core feature is to integrate all information related to the same analysis object (such as an engine) originally scattered in multiple data tables or data sources into a data table. In short, wide table data is not a "new data", but a "data integration form", and its core role is to break data silos, enrich information dimensions, and provide more comprehensive and concentrated input data for engine fault diagnosis models, thereby improving the accuracy of model diagnosis.

[0026] For easy understanding, the sample data of the above wide table data is shown in Table 1 as follows.

[0027] Table 1

[0028] S102, extracting fault description information from the wide table data, and inputting the fault description information into a pre-trained engine running data collection and determination model to output a collection instruction or a non-collection instruction; when the collection instruction is output, the control starts the engine to be diagnosed and collects engine running data under idle speed working condition, and the engine running data is time series structured data; For easy understanding, the sample data of the fault description information is shown in Table 2 as follows.

[0029] Table 2

[0030] The fault description information is input into a pre-trained engine running data collection and determination model to determine whether to collect engine running data. If needed, a "yes" indicating lamp is lit on the fault diagnosis equipment terminal, and if not needed, a "no" indicating lamp is lit. If the "yes" indicating lamp is lit, the engine is started, and the running data of the engine under idle speed working condition is collected. The engine running data is structured data (time series), and the fields that need to be collected include but are not limited to one or more of engine speed, torque, water temperature, and oil pressure.

[0031] This embodiment uses an engine operation data acquisition and determination model to determine whether to collect engine operation data, and collects engine operation data as needed.

[0032] S103. Calculate the average value of the collected engine operating data according to a preset time window, then convert the average value into a natural language description of the operating data, and store the natural language description of the operating data. For example, the average value of the data collected over the last five minutes can be calculated, and the calculated average value can be converted into a natural language description of the running data using D2T (Data-to-Text) technology.

[0033] In some embodiments, after storing the natural language description of the runtime data, a prompt message indicating that the natural language description conversion is complete can be issued.

[0034] To facilitate understanding, the following example is provided: The data is as follows: { Engine number: 123J456; "Rotation speed": [748, 750, 752, 749, 753, 751, 747]; Water temperature: [99.0, 99.1, 99.1, 99.0, 99.0, 99.1, 99.1]; Oil pressure: [223, 225, 226, 220, 215, 224, 218] } The average value of the structured data in the above data was calculated, and the results are as follows: { Engine number: 123J456; "Rotation speed": 750; Water temperature: 99.1°C; Oil pressure: 222 } The above results were then input into the D2T model and converted into a natural language description: Engine 123J456 has a current average speed of 750 rpm, an average coolant temperature of 99.1 ℃, and an average oil pressure of 222 kPa.

[0035] After completing the above D2T operation, the natural language structure is stored in the fault diagnosis equipment terminal, and an indicator light is used to indicate that the D2T operation has been completed.

[0036] S104. The wide table data is processed into text to obtain a natural language description of the multi-source data. The natural language description of the multi-source data is merged with the natural language description of the running data to obtain a final natural language description. The final natural language description is input into the pre-trained model to output feature vector data. Specifically, the pre-trained model can be a Transformer model, a Bag-of-Words model, a Word2Vec model, or a Doc2Vec model, etc.

[0037] For ease of understanding, the following is a natural language description of the combined D2T processing results of the wide table data and engine operating data for engine number 123J456: Engine 123J456, part of the WPXNG series, was manufactured in Factory 1 on November 2, 2019, and shipped on January 1, 2020. It is intended for use in construction machinery. The fault is insufficient engine power and vibration, with a projected failure date of January 1, 2025. The fault code is PK256-00, corresponding to single / multi-cylinder misfire. The cause is an ignition system malfunction, affecting the ignition and valve train systems. Troubleshooting steps include checking the system voltage, verifying the main relay output voltage matches the system voltage, ensuring good contact between the main relay and all coil pins, checking the contact between all coil pins and the ECU, checking the spark plug output for proper spark, checking the spark plug gap, and inspecting the wiring harness for water (if present, clean it). Engine cylinder pressure was also checked. On-site engine operating data was collected: average engine speed 750 rpm, average coolant temperature 99.1 ℃, and average oil pressure 222 kPa.

[0038] The combined natural language descriptions are input into a pre-trained Transformer model for vectorization to obtain feature vector data. Commonly used Transformer models include the BERT Embeddings series, BGE Embeddings series, and OpenAI Embeddings series.

[0039] S105. Input the feature vector data into the pre-trained engine fault diagnosis model and output the fault cause, the corresponding fault probability, and maintenance suggestions.

[0040] Specifically, the acquired data is input into a pre-trained engine fault diagnosis model, which outputs the causative components of engine faults, the corresponding fault probabilities, and maintenance recommendations.

[0041] For ease of understanding, the output of the example in step S104 is provided: { Engine number: 123J456; "Cause of failure": ["Spark plug", "Ignition coil", "Excessive valve clearance"]; "Probability of failure": [0.80, 0.12, 0.07]; Repair Recommendations: "Disassemble and inspect all spark plugs, checking electrode shape, carbon buildup, and erosion; use an ignition energy analyzer to compare waveforms for each cylinder to identify the weaker cylinder; check the tightness of the spark plug cap and coil connection; measure and adjust valve clearance." } Repair technicians can conduct troubleshooting and repair based on repair recommendations, achieving efficient engine fault diagnosis and repair execution.

[0042] In the engine fault diagnosis method based on a large model provided in this application embodiment, the engine fault diagnosis model can utilize more source data for diagnosis, especially fault code data and engine operation data, which provides high-value information for fault diagnosis and enables faster and more accurate location of the fault cause.

[0043] Fault codes typically follow standard coding conventions, combined with OEM-defined content, to provide partially standardized information. This helps quickly locate the system or component range and pinpoint the fault area. Fault codes also include a fault description, cause, and solution, supporting rapid repair decisions. OEM-defined content refers to the fact that, in addition to following industry-standard coding conventions, the fault code coding specifications incorporate proprietary rules or supplementary information defined by the engine brand. These brand-specific definitions are tailored to the structure and performance characteristics of their own engine products, allowing fault codes to more accurately match the fault scenarios of that brand's engines.

[0044] Engine operating data is highly diverse, capable of collecting dozens or even hundreds of signals, which enhances the interpretability of fault code information and helps pinpoint the root cause of faults. Using D2T technology to convert this structured engine operating data into natural language descriptions is a key technological aspect enabling the use of engine operating data.

[0045] Existing technologies only utilize basic engine data, work order information, sales records, and maintenance reports. This insufficient information leads to poor implementation effectiveness and fails to meet current needs. This application abandons type feature encoding in existing technologies and instead uses natural language vectorization technology, which possesses stronger semantic capture capabilities, reduces information loss, and contributes to the accuracy of fault diagnosis.

[0046] For ease of understanding, the following example is provided: Taking the domestic Plant 1, Domestic Plant 2, and Vietnam Plant 1 from the engine's basic information as examples, we perform type feature encoding and vectorization on each. Type feature encoding encodes the three plants as the numbers 0, 1, and 2 respectively, while vectorization (using the BGE Embedding model as an example) encodes each plant as a 1024-dimensional vector. The distance between these three 1024-dimensional vectors in the vector space represents their correlation; therefore, the vectors corresponding to domestic Plant 1 and Domestic Plant 2 are closer than the vector corresponding to Vietnam Plant 1. However, the type feature encoding of 0, 1, and 2 completely loses the potential information about the domestic and foreign plants, thus completely negating the impact of the differences between domestic and foreign plants on fault diagnosis.

[0047] Therefore, this application can combine engine operating data, fault code information and other relevant data to achieve accurate fault diagnosis and user-friendly maintenance guidance, thereby improving the maintenance efficiency of maintenance technicians.

[0048] In the above embodiments, a large-model-based engine fault diagnosis method is provided. Correspondingly, this application also provides a large-model-based engine fault diagnosis device, which can be implemented through software, hardware, or a combination of both. For example, the large-model-based engine fault diagnosis device may include integrated or separate functional modules or units to perform the corresponding steps in the above methods. Please refer to... Figure 2 This illustration shows a schematic diagram of a large-model-based engine fault diagnosis device provided by some embodiments of this application. Since the device embodiments are basically similar to the method embodiments, the description is relatively simple; relevant details can be found in the description of the method embodiments. The device embodiments described below are merely illustrative.

[0049] like Figure 2 As shown, the engine fault diagnosis device 10 based on a large model provided in this application may include: The acquisition module 101 is used to acquire multi-source data of the engine to be diagnosed and integrate the multi-source data into a wide table data; the multi-source data includes engine basic data, fault code basic information, work order information, sales records and engine maintenance reports; The judgment module 102 is used to extract fault description information from the wide table data, input the fault description information into a pre-trained engine operation data acquisition and judgment model, and output an acquisition command or a non-acquisition command; when an acquisition command is output, the engine to be diagnosed is started and engine operation data under idling conditions is acquired, wherein the engine operation data is time series structured data. The conversion module 103 is used to calculate the average value of the collected engine operating data according to a preset time window, then convert the average value into a natural language description of the operating data, and store the natural language description of the operating data. The feature module 104 is used to perform textual processing on the wide table data to obtain a natural language description of the multi-source data, merge the natural language description of the multi-source data with the natural language description of the running data to obtain a final natural language description, input the final natural language description into the pre-trained model, and output feature vector data. The diagnostic module 105 is used to input the feature vector data into a pre-trained engine fault diagnosis model and output the fault cause, the corresponding fault probability, and maintenance suggestions.

[0050] In one possible implementation, the engine basic data includes the engine number, engine series, manufacturer, and production date; The basic information of the fault code includes the fault code, the cause of the fault, the faulty component, and the solution to the fault. The work order information includes the work order number, fault description, and fault date; The sales records include the manufacturing date and the intended use of the equipment; The engine repair report includes the work order number, engine serial number, and part name.

[0051] In one possible implementation, the engine operating data includes at least one of engine speed, torque, coolant temperature, and oil pressure.

[0052] In one possible implementation, the conversion module 103 is further configured to: after storing the natural language description of the running data, issue a prompt message indicating that the natural language description conversion is complete.

[0053] In one possible implementation, the pre-trained model employs a Transformer model, a Bag-of-Words model, a Word2Vec model, or a Doc2Vec model.

[0054] The engine fault diagnosis device based on a large model provided in this application acquires multi-source data of the engine to be diagnosed and integrates the multi-source data into a wide table data; extracts fault description information from the wide table data and inputs the fault description information into a pre-trained engine operation data acquisition and judgment model, outputting an acquisition command or a non-acquisition command; when an acquisition command is output, the device controls the engine to be diagnosed to start and acquires engine operation data under idling conditions; calculates the average value of the acquired engine operation data according to a preset time window, then converts the average value into a natural language description of the operation data and stores the natural language description of the operation data; performs textual processing on the wide table data to obtain the natural language description of the multi-source data, merges it with the natural language description of the operation data to obtain the final natural language description, inputs the final natural language description into a pre-trained model, and outputs feature vector data; inputs the feature vector data into a pre-trained engine fault diagnosis model, and outputs the fault cause, the corresponding fault probability, and maintenance suggestions. Compared with existing technologies, the engine fault diagnosis model of this application can utilize more source data for diagnosis, especially fault code data and engine operation data, which provides high-value information for fault diagnosis and enables faster and more accurate location of fault-causing components.

[0055] This application also provides a diagnostic device corresponding to the engine fault diagnosis method based on a large model provided in the foregoing embodiments. The diagnostic device may be a mobile phone, laptop computer, tablet computer, desktop computer, etc., to execute the engine fault diagnosis method based on a large model described above.

[0056] The diagnostic equipment provided in this application embodiment and the engine fault diagnosis method based on a large model provided in this application embodiment are based on the same inventive concept and have the same beneficial effects as the methods they adopt, operate or implement.

[0057] This application also provides a computer-readable storage medium corresponding to the engine fault diagnosis method based on a large model provided in the foregoing embodiments, wherein a computer program (i.e., a program product) is stored thereon. When the computer program is run by a processor, it executes the engine fault diagnosis method based on a large model provided in any of the foregoing embodiments.

[0058] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.

[0059] The computer-readable storage medium provided in the above embodiments of this application and the engine fault diagnosis method based on a large model provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application stored therein.

[0060] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and they should all be covered within the scope of the claims and specification of this application.

Claims

1. An engine fault diagnosis method based on a large model, characterized in that, include: Acquire multi-source data of the engine to be diagnosed and integrate the multi-source data into a wide table; the multi-source data includes engine basic data, fault code basic information, work order information, sales records and engine repair reports; Fault description information is extracted from the wide table data and input into a pre-trained engine operation data acquisition and judgment model. The model outputs an acquisition command or a non-acquisition command. When an acquisition command is output, the engine to be diagnosed is started and engine operation data under idling conditions is acquired. The engine operation data is time-series structured data. The average value of the collected engine operating data is calculated according to a preset time window, and then the average value is converted into a natural language description of the operating data and stored. The wide table data is processed into text to obtain a natural language description of the multi-source data. The natural language description of the multi-source data is merged with the natural language description of the running data to obtain the final natural language description. The final natural language description is input into the pre-trained model to output feature vector data. The feature vector data is input into a pre-trained engine fault diagnosis model, which outputs the fault-causing component, the corresponding fault probability, and maintenance suggestions.

2. The engine fault diagnosis method based on a large model according to claim 1, characterized in that, The engine's basic data includes the engine number, engine series, manufacturer, and production date; The basic information of the fault code includes the fault code, the cause of the fault, the faulty component, and the solution to the fault. The work order information includes the work order number, fault description, and fault date; The sales records include the manufacturing date and the intended use of the equipment; The engine repair report includes the work order number, engine serial number, and part name.

3. The engine fault diagnosis method based on a large model according to claim 1, characterized in that, The engine operating data includes at least one of the following: engine speed, torque, water temperature, and oil pressure.

4. The engine fault diagnosis method based on a large model according to claim 1, characterized in that, Following the natural language description storing the operational data, the system further includes: A notification message will be displayed indicating that the natural language description conversion is complete.

5. The engine fault diagnosis method based on a large model according to claim 1, characterized in that, The pre-trained model uses the Transformer model, Bag-of-Words model, Word2Vec model, or Doc2Vec model.

6. An engine fault diagnosis device based on a large model, characterized in that, include: The acquisition module is used to acquire multi-source data of the engine to be diagnosed and integrate the multi-source data into a wide table data; the multi-source data includes engine basic data, fault code basic information, work order information, sales records and engine repair reports; The judgment module is used to extract fault description information from the wide table data, input the fault description information into the pre-trained engine operation data acquisition and judgment model, and output acquisition command or non-acquisition command; when the acquisition command is output, the engine to be diagnosed is started and engine operation data under idling conditions is acquired, and the engine operation data is time series structured data. The conversion module is used to calculate the average value of the collected engine operating data according to a preset time window, then convert the average value into a natural language description of the operating data, and store the natural language description of the operating data. The feature module is used to perform textual processing on the wide table data to obtain a natural language description of the multi-source data, merge the natural language description of the multi-source data with the natural language description of the running data to obtain a final natural language description, input the final natural language description into the pre-trained model, and output feature vector data. The diagnostic module is used to input the feature vector data into a pre-trained engine fault diagnosis model and output the fault cause, the corresponding fault probability, and maintenance suggestions.

7. The engine fault diagnosis device based on a large model according to claim 6, characterized in that, The engine's basic data includes the engine number, engine series, manufacturer, and production date; The basic information of the fault code includes the fault code, the cause of the fault, the faulty component, and the solution to the fault. The work order information includes the work order number, fault description, and fault date; The sales records include the manufacturing date and the intended use of the equipment; The engine repair report includes the work order number, engine serial number, and part name.

8. The engine fault diagnosis device based on a large model according to claim 6, characterized in that, The engine operating data includes at least one of the following: engine speed, torque, water temperature, and oil pressure.

9. A diagnostic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the method as described in any one of claims 1 to 5.

10. A computer-readable storage medium, characterized in that, It stores computer-readable instructions that can be executed by a processor to implement the method as described in any one of claims 1 to 5.

Images