Automobile data analysis method and system based on intelligent diagnostic instrument

Abstract

The invention relates to the field of vehicle data analysis, in particular to an automobile data analysis method and system based on an intelligent diagnostic instrument. The method comprises the following steps: acquiring vehicle multi-modal monitoring data based on an intelligent diagnostic instrument; operation power change calculation is conducted on the vehicle multi-mode monitoring data, time sequence state feature mining is conducted, and vehicle time sequence state features are obtained; carrying out multi-frequency noise reduction optimization on the vehicle multi-modal monitoring data, and carrying out normalized periodic evolution mining so as to generate normalized vibration periodic evolution characteristics; potential feature association deep mining is carried out on the vehicle time sequence state features and the normalized vibration period evolution features, then association graph modeling is carried out, and a vehicle multi-modal feature knowledge graph is constructed; and abnormal parameter deviation calculation and fault positioning traceability are carried out on the real-time operation parameters of the vehicle based on the vehicle multi-modal feature knowledge graph, so that a potential abnormal fault positioning point is obtained. According to the invention, efficient and accurate vehicle fault risk prediction analysis is realized.

Description

Technical Field

[0001] The present invention relates to the field of vehicle data analysis, and in particular to an automobile data analysis method and system based on an intelligent diagnostic instrument. Background Art

[0002] With the continuous development of the automobile industry and the improvement of its intelligence level, modern automobiles are no longer simple mechanical products, but complex systems that highly integrate multiple technologies such as electronics, information, and control. Vehicle fault prediction and health management have become key issues in improving automobile safety, reliability, and operational efficiency. Automobiles include multiple complex components such as engines, electronic control systems, sensors, braking systems, and power systems. These systems will be affected by various internal and external factors during long-term operation, such as temperature fluctuations, voltage changes, sensor failures, component wear, etc., which will lead to system performance degradation or even failure. Therefore, how to accurately and timely predict and diagnose these potential faults has become a major challenge to ensure vehicle safety and performance.

[0003] Traditional automobile data analysis usually relies on manual inspection and traditional troubleshooting methods, mainly detecting faults through inspection of mechanical parts or regular maintenance. Although these methods can ensure the normal operation of the vehicle to a certain extent, they often have problems such as long inspection cycles, inaccurate detection results, and low efficiency. In addition, traditional methods cannot provide real-time and accurate fault prediction, resulting in vehicle failures often being unable to be warned in advance before they occur, causing serious safety accidents or high maintenance costs. Therefore, a more intelligent and safer automobile data analysis method is needed. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes an automobile data analysis method and system based on an intelligent diagnostic instrument to solve at least one of the above technical problems.

[0005] To achieve the above object, the present invention provides an automobile data analysis method based on an intelligent diagnostic instrument, comprising the following steps: Step S1: obtaining vehicle multimodal monitoring data based on the intelligent diagnostic instrument; calculating the operating power change of the vehicle multimodal monitoring data, and mining the time series state characteristics to obtain the vehicle time series state characteristics; Step S2: performing multi-frequency noise reduction optimization on the vehicle multi-modal monitoring data and mining the normalized periodic evolution to generate normalized vibration periodic evolution features; Step S3: Perform deep mining of potential feature associations on the vehicle's time series state features and normalized vibration cycle evolution features, then perform association graph modeling to construct a vehicle multimodal feature knowledge graph; Step S4: obtaining the real-time operating parameters of the vehicle; performing abnormal parameter deviation calculation and fault location tracing on the real-time operating parameters of the vehicle based on the vehicle multimodal feature knowledge graph, thereby obtaining potential abnormal fault location points; Step S5: performing multi-period fault simulation on the potential abnormal fault location point and predicting the vehicle fault risk, thereby generating vehicle potential fault risk prediction data; Step S6: Make cloud-edge collaborative decisions on the vehicle's potential fault risk prediction data, perform dynamic fault diagnosis scheduling, and build an intelligent fault diagnosis scheduling model.

[0006] The present invention obtains vehicle multimodal monitoring data through an intelligent diagnostic instrument, and can obtain various aspects of vehicle operating status information. The operation power change calculation and time series state feature mining are helpful to understand the dynamic changes of vehicle operating status and provide basic data for subsequent analysis. Multi-frequency noise reduction optimization helps to clarify the signals in vehicle monitoring data and mine the normalized vibration cycle evolution characteristics. The generation of normalized vibration cycle evolution characteristics helps to understand the evolution law of vehicle vibration behavior and provide support for anomaly detection and fault prediction. The deep mining of potential feature associations and the modeling of association graphs help to discover the correlation between different features and construct a vehicle multimodal feature knowledge graph. The knowledge graph provides a comprehensive feature relationship. The joint perspective provides a basis for subsequent anomaly detection and fault location. Based on the vehicle's multimodal feature knowledge graph, it calculates deviations and locates faults in the vehicle's real-time operating parameters. Fault location tracing helps to find potential abnormal fault locations and discover potential fault risks in advance. Through multi-period fault simulation and fault risk prediction, it generates vehicle potential fault risk prediction data. Predicting fault risks helps to take preventive measures and reduce the impact of faults on vehicle operation. Through cloud-edge collaborative decision-making and dynamic fault diagnosis scheduling, it builds an intelligent fault diagnosis scheduling model. The model can help to achieve rapid response to fault events, improve the efficiency and accuracy of fault diagnosis, and ensure the safety and stability of vehicle operation.

[0007] Preferably, step S1 comprises the following steps: Step S11: acquiring vehicle multimodal monitoring data based on the intelligent diagnostic instrument, wherein the vehicle multimodal monitoring data includes vehicle engine all-round monitoring parameters, vehicle circuit monitoring data and vehicle running status sound signals; Step S12: Performing temperature fluctuation analysis on all-round monitoring parameters of the vehicle engine and extracting temperature fluctuation data; Step S13: performing discrete fitting of time series fluctuations on the temperature fluctuation data to construct an engine temperature fluctuation curve; Step S14: calculating the operating power variation of the vehicle engine's all-round monitoring parameters to generate engine operating power variation characteristics; Step S15: mining the time series state characteristics of the engine operating power variation characteristics, the engine temperature fluctuation curve and the vehicle circuit monitoring data to obtain the vehicle time series state characteristics.

[0008] The present invention provides comprehensive vehicle monitoring information by acquiring all-round monitoring parameters of the vehicle engine, vehicle circuit monitoring data and vehicle operating status sound signals. The comprehensive analysis of multimodal data can provide a more comprehensive understanding of the vehicle's operating status and provide basic data for subsequent analysis. Temperature fluctuation analysis helps to understand the changing law of engine temperature and extract key temperature fluctuation data. The extracted temperature fluctuation data can be used for subsequent feature construction and fault prediction. Through discrete fitting of time series fluctuations, an engine temperature fluctuation curve is constructed to present the specific trend of temperature fluctuations. The establishment of the temperature fluctuation curve helps to conduct a more in-depth analysis of engine temperature changes and provide a basis for abnormality detection. By calculating the operating power change of the engine's all-round monitoring parameters, the engine operating power change characteristics are generated. These characteristics can reflect the changes in engine performance and provide important information for fault risk analysis. In combination with the engine operating power change characteristics, the engine temperature fluctuation curve and the vehicle circuit monitoring data, the time series state feature mining is performed. The mining of the time series state feature can comprehensively consider information from multiple aspects, more comprehensively describe the vehicle's operating status, and help identify potential fault risk points.

[0009] Preferably, step S15 specifically comprises the following steps: Identify multiple circuit element nodes of the vehicle; Based on multiple circuit element nodes of the vehicle, the interaction analysis between the element nodes is performed to obtain the node interaction circuit; Match the circuit parameters of the node interaction circuit based on the vehicle circuit monitoring data, so as to obtain the monitoring parameters of each vehicle circuit; extracting voltage timing parameters based on monitoring parameters of the vehicle circuit; Performing voltage timing waveform fitting on the voltage timing parameters to obtain a voltage timing waveform curve; Performing Fourier transformation on the voltage time series waveform curve to generate voltage time series waveform spectrum data; Calculate the voltage variation amplitude on the voltage time series waveform spectrum data to obtain the voltage variation amplitude value; Evolving the voltage change amplitude value, thereby generating voltage change trend evolution characteristics; Carry out vehicle voltage load situation analysis based on the voltage change trend evolution characteristics to obtain vehicle voltage load situation characteristics; Based on the vehicle voltage load situation characteristics, the voltage load-operating power correlation change analysis is performed on the engine operating power change characteristics to generate a voltage load-operating power correlation characteristic; The time-series state quantity evolution of the voltage load-operating power correlation characteristics and the engine temperature fluctuation curve is performed to obtain the vehicle time-series state characteristics.

[0010] The present invention can lay a foundation for subsequent circuit fault analysis and monitoring by identifying each component node of the vehicle circuit. Each node represents a functional point in the vehicle circuit system, such as an engine sensor, a headlight, a control module, etc. Identifying these nodes helps to clarify the working status and interaction relationship of each component in the circuit. The interaction analysis between component nodes can help understand the signal transmission and current flow mode in the circuit. By analyzing the mutual influence between the nodes, the operation mode of the circuit system under different working conditions can be understood, and a reference can be provided for circuit fault location. Circuit parameter matching can accurately compare the monitored circuit data with the characteristics of each component node, so as to obtain the specific monitoring parameters of each circuit, which helps to further clarify the status of each circuit component, such as voltage, current, etc., for subsequent analysis. The voltage timing parameters reflect the voltage change of the circuit at different times. By extracting these timing parameters, the dynamic changes of the circuit can be analyzed, which provides a basis for subsequent voltage waveform analysis and helps to identify whether the circuit has the risk of fluctuation, poor stability or voltage anomaly. The voltage timing waveform fitting can convert the voltage change trend into a visual waveform curve, which helps us to more clearly identify the law and anomaly of voltage fluctuation. The fitted waveform can accurately describe Describe the changing trend of voltage and identify the pattern of abnormal fluctuation. Fourier transform converts the voltage waveform from time domain to frequency domain, which can reveal the frequency components in the voltage fluctuation, thereby discovering some periodic problems hidden in the time domain data. Spectral data can help detect some hidden fault modes, such as high-frequency noise or low-frequency oscillation of the circuit. The calculation of the voltage change amplitude can quantify the degree of voltage fluctuation, which is an important indicator for judging the stability and load status of the circuit. By calculating the voltage change amplitude value, it can be identified whether the circuit has abnormal fluctuations, overload and other problems. Analyzing the trend evolution of the voltage change amplitude helps to understand the long-term change pattern of voltage fluctuation, which can help predict circuit failures and avoid sudden failure. By correlating the voltage load with the engine power change, the impact of voltage change on engine performance can be revealed. This correlation analysis helps to understand how power supply affects the operating status of the engine, especially under high load or abnormal voltage conditions, the engine's working efficiency and performance are affected. By analyzing the evolution of the time series state quantity of voltage load, operating power and temperature fluctuations, the overall health status of the vehicle can be comprehensively evaluated, and the fault trend can be identified in time. Combining electrical and thermodynamic characteristics in multiple dimensions helps to build a more accurate fault prediction model.

[0011] Preferably, the specific steps of step S2 are: Step S21: performing multi-frequency noise reduction optimization on the vehicle running state sound signal, thereby generating a noise reduction reconstructed sound signal; Step S22: performing a signal vibration characteristic time series analysis on the noise reduction and reconstructed sound signal to generate signal time series vibration characteristic data; Step S23: performing time series fitting on the signal time series vibration characteristic data to construct a signal time series vibration diagram; Step S24: performing vibration change trend evolution on the signal time series vibration diagram to extract vibration change trend features; Step S25: performing vibration peak-valley value distribution analysis on the signal time series vibration diagram to extract vibration peak-valley value distribution data; Step S26: Perform normalized periodic evolution mining on the vibration peak-valley value distribution data and the vibration change trend characteristics, thereby generating normalized vibration periodic evolution characteristics.

[0012] The present invention generates a noise-reduced and reconstructed sound signal by performing multi-frequency noise reduction optimization on the sound signal of the vehicle running state. The sound signal after noise reduction is clearer, can reduce interference, and more accurately reflect the vehicle running state. The vibration characteristics of the noise-reduced and reconstructed sound signal are analyzed to generate signal time series vibration characteristic data. Time series analysis can reveal the change of the vibration characteristics of the sound signal over time, which is helpful to discover abnormal vibration modes. Through time series fitting, a signal time series vibration diagram is constructed to show the time series characteristics of the vibration signal. The time series vibration diagram helps to intuitively understand the change law of the vibration signal, provides visualization support for subsequent analysis, and performs vibration change analysis on the signal time series vibration diagram. The vibration trend evolution analysis is performed to extract the vibration change trend characteristics. The vibration change trend characteristics can reflect the evolution law of the vibration signal and help to judge the stability and abnormality of the vibration signal. The peak-to-valley value distribution of the vibration graph is analyzed and the vibration peak-to-valley value distribution data is extracted. The vibration peak-to-valley value distribution data can reveal the amplitude change of the vibration signal and provide a basis for abnormality detection. Combined with the vibration peak-to-valley value distribution data and the vibration change trend characteristics, normalized periodic evolution mining is performed to generate normalized vibration periodic evolution characteristics. Normalized vibration periodic evolution characteristics can help identify the periodic changes of vibration signals and provide more in-depth information support for fault prediction and maintenance.

[0013] Preferably, the specific steps of step S21 are: Perform multi-time-frequency decomposition on the vehicle running state sound signal to obtain multiple time-frequency spectrograms; Calculate the noise component of each frequency on multiple time-spectrum graphs, thereby extracting the noise component of each frequency; Perform noise change identification on the noise component of each frequency to obtain the noise component change characteristics of each frequency; Adaptive filtering and denoising are performed according to the variation characteristics of the noise components of each frequency, thereby obtaining adaptive filtering and denoising frequency domain data; Define sliding time windows; Based on the sliding time window, the adaptive filtering denoising frequency domain data is segmented into sliding time windows to obtain denoising frequency domain data of multiple time windows; According to the vehicle running state sound signal, the noise reduction frequency domain data is subjected to signal distortion identification, and the time window frequency domain data of the signal distortion is extracted; Calculating the distortion amplitude of the signal distorted time window frequency domain data to generate a distortion amplitude value; Based on the distortion amplitude value, nonlinear distortion amplitude correction is performed on the adaptive filtering noise reduction frequency domain data, thereby obtaining distortion correction frequency domain data; The distortion-corrected frequency domain data is inversely reconstructed in time-frequency order to generate a noise-reduced reconstructed sound signal.

[0014] The present invention can decompose the sound signal in the time domain and frequency domain through multi-time-frequency decomposition, provide more detailed signal characteristics, and obtain multiple time-frequency spectrum diagrams, which is helpful for in-depth understanding of the frequency change and timing characteristics of the sound signal. The calculation of the noise component of each frequency can determine the noise level at each frequency, and provide accurate noise information for subsequent processing. The extraction of the noise component of each frequency is helpful for analyzing the noise situation of different frequency bands in the sound signal. The noise change recognition can capture the noise change characteristics of each frequency and help analyze the dynamic changes of noise. The noise component change characteristics are helpful for understanding the fluctuation of noise levels at different frequencies. Adaptive filtering noise reduction can be based on the change characteristics of the noise component. Effectively denoising the sound signal to obtain adaptive filtering denoising frequency domain data helps to reduce the impact of noise on the signal and improve the accuracy of the data. Defining a sliding time window helps to segment the signal, making signal processing more refined and effective. Signal distortion identification can accurately capture the distortion in the signal and provide a basis for subsequent processing. Extracting the time window frequency domain data of the distorted signal helps to locate and correct the distortion problem in the signal. Inverse time-frequency reconstruction can restore the processed frequency domain data to time domain data and generate the final denoised reconstructed sound signal. Generating a denoised reconstructed sound signal helps to restore the original signal and provide clearer and more accurate sound information.

[0015] Preferably, step S3 specifically comprises the following steps: Step S31: performing feature space mapping on the vehicle time series state characteristics and normalized vibration period evolution characteristics, thereby obtaining time series state feature nodes and normalized vibration feature nodes; Step S32: performing deep mining of potential feature associations on the time series state feature nodes and the normalized vibration feature nodes, thereby extracting potential feature association data; Step S33: performing semantic feature evolution on the potential feature association data, thereby generating feature association semantic logic; Step S34: Based on the feature association semantic logic, the vehicle time series state characteristics and the normalized vibration cycle evolution characteristics are modeled as a correlation graph to construct a vehicle multimodal feature knowledge graph.

[0016] The present invention helps to convert complex time-series state features and vibration cycle evolution features into feature nodes that are easy to understand and process through feature space mapping. Obtaining time-series state feature nodes and normalized vibration feature nodes can help extract key features and provide a basis for subsequent analysis. Deep mining of potential feature association data helps to discover potential associations hidden in the data and provide deeper data insights. Extracting potential feature association data can reveal the complex relationship between time-series state features and vibration features, providing more information for subsequent analysis. Semantic feature evolution converts data into semantic features that are easy to understand and explain, improving the interpretability and applicability of the data. Generating feature association semantic logic helps to understand and analyze the data at a deeper level and provide more information for subsequent modeling. Association graph modeling can visualize and model the relationship between time-series state features and vibration cycle features, helping to understand the complex associations between data. Building a vehicle multimodal feature knowledge graph helps to integrate different types of feature information, provide a comprehensive data perspective, and provide more comprehensive information support for fault prediction.

[0017] Preferably, the specific steps of step S4 are: Step S41: obtaining real-time operating parameters of the vehicle; Step S42: Calculate the abnormal parameter deviation of the real-time operating parameters of the vehicle based on the vehicle multimodal feature knowledge graph, and extract the abnormal deviation data; Step S43: performing potential abnormal fault analysis on the abnormal deviation data to obtain potential abnormal fault data; Step S44: tracing the potential abnormal fault data to locate the fault, thereby obtaining the potential abnormal fault location point.

[0018] The present invention can provide the latest vehicle status information by acquiring the vehicle's operating parameters in real time, provide real-time data support for fault prediction, obtain real-time parameters, help to monitor vehicle status changes in a timely manner, and provide an accurate data basis for subsequent analysis and processing. Abnormal deviation calculation for real-time parameters can detect the deviation of parameter values ​​from the normal range and discover abnormal data. Extracting abnormal deviation data helps to identify potential abnormal situations and provide important clues for further fault prediction. Potential abnormal fault analysis can deeply explore abnormal patterns existing in abnormal deviation data and find potential fault signals. Obtaining potential abnormal fault data helps to identify existing fault patterns and provide a strong basis for fault prediction and location. Fault location tracing can track the source of potential abnormal fault data and help accurately locate existing fault points. Obtaining potential abnormal fault location points helps to quickly locate and solve vehicle fault problems and improve fault handling efficiency and accuracy.

[0019] Preferably, the specific steps of step S5 are: Step S51: performing multi-period fault simulation on the potential abnormal fault location point to generate fault simulation data of multiple periods; Step S52: performing fault point operating condition analysis on the fault simulation data of multiple time periods to generate fault point operating condition data of multiple time periods; Step S53: performing fault diffusion path mining based on the fault point working condition data of multiple time periods to obtain the fault diffusion path of each time period; Step S54: performing fault diffusion path distribution analysis on the fault diffusion path in each time period, and constructing fault path distribution graphs for multiple time periods; Step S55: Perform vehicle failure risk prediction on the failure path distribution diagrams of multiple time periods, thereby generating vehicle potential failure risk prediction data.

[0020] The present invention helps to simulate the fault conditions that occur in different time periods through multi-period fault simulation, so as to more comprehensively evaluate the vehicle fault risk. Generating fault simulation data for multiple time periods can help identify the time series characteristics and change trends of faults, and provide more information for subsequent analysis. Fault point working condition analysis can help understand the working state and characteristics of the fault point in the fault simulation data for different time periods, and provide a basis for the analysis of the fault diffusion path. Generating fault point working condition data for multiple time periods helps to reveal the performance of the fault under different conditions, and provides more details for subsequent analysis. Fault diffusion path mining can identify the propagation path of the fault in the vehicle system, help understand the propagation law and impact range of the fault, and obtain each time. The fault diffusion path of each segment helps to gain a deeper understanding of the propagation of faults in different time periods and provide a basis for fault risk assessment. The fault diffusion path distribution analysis can count the distribution of fault paths in different time periods and help identify the main propagation paths and frequencies of faults. Constructing fault path distribution maps for multiple time periods helps to visualize the characteristics of fault diffusion paths and provide intuitive information for fault risk assessment. Vehicle fault risk prediction is based on multi-time period fault path distribution maps, which can comprehensively consider the propagation characteristics and risk levels of faults in different time periods, improve the accuracy of risk prediction, and generate vehicle potential fault risk prediction data to help identify the risks of faults that occur and provide an important reference for formulating prevention and maintenance strategies.

[0021] Preferably, the specific steps of step S6 are: Step S61: Sharing the potential failure risk prediction data of the vehicle with the cloud to obtain failure risk prediction shared data; Step S62: performing fault risk type diagnosis on the fault risk prediction shared data to generate fault risk diagnosis data; Step S63: Perform cloud-edge collaborative decision-making on the fault risk diagnosis data, thereby generating a fault risk collaborative diagnosis strategy; Step S64: Perform dynamic fault diagnosis scheduling based on the fault risk collaborative diagnosis strategy and build an intelligent fault diagnosis scheduling model.

[0022] Through cloud data sharing, the present invention can share fault risk prediction data with other relevant teams or organizations, improve information sharing efficiency and collaboration, make fault risk prediction shared data accessible to multiple parties, facilitate wider application and analysis, and promote cross-departmental cooperation and decision-making. Fault risk type diagnosis helps to gain an in-depth understanding of the characteristics and potential impacts of different types of faults in fault risk data, providing a basis for further decision-making. Generating fault risk diagnosis data can help distinguish different types of fault risks, and provide targeted suggestions for subsequent decision-making and processing. Cloud-edge collaborative decision-making can integrate fault risk diagnosis data from different sources, promote multi-faceted collaboration and joint decision-making, and generate fault risk collaborative diagnosis strategies. It helps to integrate opinions and data from all parties and improve the comprehensiveness and accuracy of decision-making. Dynamic fault diagnosis scheduling can realize intelligent fault diagnosis scheduling management according to the fault risk collaborative diagnosis strategy, improve fault processing efficiency and accuracy, and construct an intelligent fault diagnosis scheduling model to help flexibly arrange fault diagnosis tasks according to actual conditions, optimize resource utilization, and improve vehicle maintenance efficiency.

[0023] In this specification, a vehicle data analysis system based on an intelligent diagnostic instrument is provided, which is used to execute the vehicle data analysis method based on the intelligent diagnostic instrument as described above, comprising: The time series state feature module is used to obtain vehicle multi-modal monitoring data based on the intelligent diagnostic instrument; calculate the operating power change of the vehicle multi-modal monitoring data, and mine the time series state features to obtain the vehicle time series state features; The vibration period feature module is used to optimize multi-frequency noise reduction of vehicle multi-modal monitoring data and mine normalized period evolution to generate normalized vibration period evolution features; The knowledge graph module is used to conduct deep mining of potential feature associations of vehicle time series state characteristics and normalized vibration cycle evolution characteristics, and then conduct association graph modeling to construct a vehicle multimodal feature knowledge graph; The fault location and tracing module is used to obtain the real-time operating parameters of the vehicle; based on the vehicle multimodal feature knowledge graph, the abnormal parameter deviation of the real-time operating parameters of the vehicle is calculated and the fault location and tracing are performed to obtain the potential abnormal fault location point; The risk prediction module is used to perform multi-period fault simulation on potential abnormal fault location points and predict vehicle fault risks, thereby generating vehicle potential fault risk prediction data; The collaborative decision-making module is used to make collaborative decisions on the cloud edge for vehicle potential fault risk prediction data, perform dynamic fault diagnosis scheduling, and build an intelligent fault diagnosis scheduling model.

[0024] By acquiring vehicle multimodal monitoring data and mining time series state features, the present invention can fully understand the changing trends of vehicle operating states, provide strong support for subsequent fault prediction, and the obtained vehicle time series state features are helpful to identify potential problems and take preventive maintenance measures, thereby improving vehicle reliability and safety. Through multi-frequency noise reduction optimization and normalized periodic evolution mining, the evolution law of vehicle vibration characteristics can be identified to help predict the occurrence of potential vibration problems. Generating normalized vibration periodic evolution features can help to timely discover abnormal vibration conditions and improve the accuracy of diagnosis of vehicle vibration problems. By deeply mining potential feature associations and constructing a vehicle multimodal feature knowledge graph, a large amount of data can be integrated and analyzed to help identify the associations and laws between features. The constructed knowledge graph can help to better understand the relationship between vehicle data features, providing more accurate information for subsequent fault location and prediction. Based on the real-time operating parameters of the vehicle, abnormal parameter deviation calculation and fault location tracing can be performed to quickly and accurately locate potential abnormal fault points, reducing the time and cost of troubleshooting. This module can detect potential fault points in advance, reduce the impact of faults on vehicle operation, and improve the reliability and safety of the vehicle. Multi-period fault simulation and risk prediction of potential abnormal fault location points can help prevent faults in advance and reduce the impact of sudden faults on vehicle operation. The generated vehicle potential fault risk prediction data is helpful to formulate corresponding maintenance plans and measures, improve the availability and operation efficiency of the vehicle, and realize intelligent fault handling management through cloud-edge collaborative decision-making and dynamic fault diagnosis and scheduling, improve fault handling efficiency and accuracy, and build an intelligent fault diagnosis and scheduling model to help optimize resource utilization, improve vehicle maintenance efficiency, and ensure safe vehicle operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A schematic diagram of the steps of a method for analyzing automobile data based on an intelligent diagnostic instrument according to the present invention; Figure 2 Detailed implementation flow chart of step S1; Figure 3 Detailed implementation flow chart of step S2; Figure 4 Detailed implementation flow chart of step S3. DETAILED DESCRIPTION

[0026] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0027] The example of this application provides a method and system for analyzing automobile data based on an intelligent diagnostic instrument. The execution subject of the method and system for analyzing automobile data based on an intelligent diagnostic instrument includes but is not limited to: mechanical equipment, data processing platform, cloud server node, network upload device, etc. equipped with the system can be regarded as the general computing node of this application, and the data processing platform includes but is not limited to: at least one of an audio image management system, an information management system, and a cloud data management system.

[0028] See also Figures 1 to 4 The present invention provides a method for analyzing automobile data based on an intelligent diagnostic instrument, and the method for analyzing automobile data based on an intelligent diagnostic instrument comprises the following steps: Step S1: obtaining vehicle multimodal monitoring data based on the intelligent diagnostic instrument; calculating the operating power change of the vehicle multimodal monitoring data, and mining the time series state characteristics to obtain the vehicle time series state characteristics; Step S2: performing multi-frequency noise reduction optimization on the vehicle multi-modal monitoring data and mining the normalized periodic evolution to generate normalized vibration periodic evolution features; Step S3: Perform deep mining of potential feature associations on the vehicle's time series state features and normalized vibration cycle evolution features, then perform association graph modeling to construct a vehicle multimodal feature knowledge graph; Step S4: obtaining the real-time operating parameters of the vehicle; performing abnormal parameter deviation calculation and fault location tracing on the real-time operating parameters of the vehicle based on the vehicle multimodal feature knowledge graph, thereby obtaining potential abnormal fault location points; Step S5: performing multi-period fault simulation on the potential abnormal fault location point and predicting the vehicle fault risk, thereby generating vehicle potential fault risk prediction data; Step S6: Make cloud-edge collaborative decisions on the vehicle's potential fault risk prediction data, perform dynamic fault diagnosis scheduling, and build an intelligent fault diagnosis scheduling model.

[0029] The present invention obtains vehicle multimodal monitoring data through an intelligent diagnostic instrument, and can obtain various aspects of vehicle operating status information. The operation power change calculation and time series state feature mining are helpful to understand the dynamic changes of vehicle operating status and provide basic data for subsequent analysis. Multi-frequency noise reduction optimization helps to clarify the signals in vehicle monitoring data and mine the normalized vibration cycle evolution characteristics. The generation of normalized vibration cycle evolution characteristics helps to understand the evolution law of vehicle vibration behavior and provide support for anomaly detection and fault prediction. The deep mining of potential feature associations and the modeling of association graphs help to discover the correlation between different features and construct a vehicle multimodal feature knowledge graph. The knowledge graph provides a comprehensive feature relationship. The joint perspective provides a basis for subsequent anomaly detection and fault location. Based on the vehicle's multimodal feature knowledge graph, it calculates deviations and locates faults in the vehicle's real-time operating parameters. Fault location tracing helps to find potential abnormal fault locations and discover potential fault risks in advance. Through multi-period fault simulation and fault risk prediction, it generates vehicle potential fault risk prediction data. Predicting fault risks helps to take preventive measures and reduce the impact of faults on vehicle operation. Through cloud-edge collaborative decision-making and dynamic fault diagnosis scheduling, it builds an intelligent fault diagnosis scheduling model. The model can help to achieve rapid response to fault events, improve the efficiency and accuracy of fault diagnosis, and ensure the safety and stability of vehicle operation.

[0030] In the embodiment of the present invention, refer to Figure 1 , is a schematic flow chart of the steps of a method for analyzing automobile data based on an intelligent diagnostic instrument according to the present invention. In this example, the steps of the method for analyzing automobile data based on an intelligent diagnostic instrument include: Step S1: obtaining vehicle multimodal monitoring data based on the intelligent diagnostic instrument; calculating the operating power change of the vehicle multimodal monitoring data, and mining the time series state characteristics to obtain the vehicle time series state characteristics; In this embodiment, an intelligent diagnostic instrument (such as an OBD-II adapter, a CAN bus monitor or a dedicated data acquisition device) is selected to connect to the electronic control unit (ECU) of the vehicle to ensure that the device is properly installed and can read the vehicle's operating data in real time. Determine the type of multimodal monitoring data that needs to be collected, such as engine speed, throttle position, vehicle speed, oil pressure, temperature, etc. Set the frequency of data collection (such as once per second) to ensure the timeliness and accuracy of the data, start the intelligent diagnostic instrument, and start real-time collection of multimodal monitoring data. Use the corresponding software tools to read and record the data to ensure the integrity of the data. In Python, you can use the python-OBD library or python-can library for data collection, and store the collected multimodal monitoring data in a database (such as SQLite, MySQL, or MongoDB) for subsequent analysis. According to the relevant parameters of the vehicle, determine the calculation formula for the operating power, P=(T×ω) / η, where P is power, T is torque, ω is angular velocity (which can be calculated from the speed), and η is efficiency (usually a constant or adjusted according to the working conditions). Extract the required parameters, such as torque and speed, from the stored multimodal monitoring data, perform necessary data cleaning, and process missing values. and noise to ensure the accuracy of the data. Use a programming language (such as Python) to calculate the power. Substitute the extracted parameters into the power calculation formula to obtain the operating power at each time point. Use the numpy library for efficient array calculations. Store the calculated operating power data together with the timestamp in the database for time series analysis. Sort the operating power data and align the timestamps to ensure the time series of the data. Use the pandas library to convert the data into a time series format for subsequent analysis. Select a suitable time series feature extraction method, such as the sliding window method, Fourier transform, wavelet transform, etc., to extract time series state features. Perform feature extraction on the time series data of the operating power to generate time series state features. For example, calculate statistical features such as the mean, variance, maximum and minimum values ​​through a sliding window. Use scipy and numpy libraries for signal processing and feature extraction. Store the extracted time series state features in the database to form a feature data set for subsequent analysis and use.

[0031] Step S2: performing multi-frequency noise reduction optimization on the vehicle multi-modal monitoring data and mining the normalized periodic evolution to generate normalized vibration periodic evolution features; In this embodiment, vibration-related parameters, such as acceleration, velocity or displacement data, are extracted from the stored multimodal monitoring data to ensure the timing and integrity of the data. A suitable noise reduction technology is selected, such as: wavelet transform: suitable for noise reduction of multi-frequency signals, which can retain the instantaneous characteristics of the signal. Fourier transform: used for frequency domain analysis, which can effectively suppress high-frequency noise. Kalman filter: a recursive filter in a dynamic system, which can update the state estimate in real time. A suitable wavelet basis function (such as Daubechies, Symlets) is selected to perform wavelet decomposition, extract different frequency components of the signal, set a threshold, remove high-frequency noise components (usually using a soft threshold method), perform an inverse wavelet transform on the denoised signal, and reconstruct the denoised signal. In Python, the PyWavelets library can be used to implement wavelet transform, determine the key indicators of normalized periodic characteristics, such as vibration amplitude, frequency, period, etc., and refer to the normal operating status of the vehicle to reconstruct the denoised signal. The signal is analyzed in time domain and frequency domain: Time domain analysis: Calculate the mean, variance, maximum value, minimum value, etc. of the vibration signal; Frequency domain analysis: Use fast Fourier transform (FFT) to extract the frequency components of the signal, determine the main frequency and harmonics, and use the sliding window method to analyze the noise reduction signal, extract the vibration period characteristics of each time period, and calculate the period, amplitude, frequency and other characteristics within each window. You can use numpy and scipy libraries for window sliding and feature calculation, and normalize the extracted period features to eliminate the influence of different sensors or different conditions: Use Z-score standardization or Min-Max standardization to convert the eigenvalues ​​to a unified scale. In Python, use StandardScaler or MinMaxScaler of the scikit-learn library for standardization, and store the generated normalized vibration period evolution features in the database to form a feature data set for subsequent analysis and use.

[0032] Step S3: Perform deep mining of potential feature associations on the vehicle's time series state features and normalized vibration cycle evolution features, then perform association graph modeling to construct a vehicle multimodal feature knowledge graph; In this embodiment, ensure that the vehicle time series state characteristics (such as speed, acceleration, speed, throttle position, etc.) and normalized vibration period evolution characteristics (such as vibration frequency, amplitude, etc.) have been collected. Data integration: Integrate data from different sources into a unified data frame (such as PandasDataFrame) to ensure that each row represents the characteristic data of a time point. Clean data: Check missing values ​​and outliers in the data, fill or delete them to ensure data quality. Standardization: Standardize all features (for example, use Z-score standardization) to eliminate dimensional effects in subsequent analysis and select appropriate association rules. Then use learning algorithms, such as Apriori algorithm or FP-Growth algorithm, which can effectively mine the correlation between features and construct feature matrix: integrate the time series state features and normalized vibration features into a feature matrix, ensure that each column represents a feature and each row represents a sample, which is convenient for subsequent analysis, binarization processing: when necessary, binarize the features (such as converting continuous features into discrete features according to thresholds) to adapt to the association rule algorithm, use the selected association mining algorithm to analyze the feature matrix, identify potential feature correlations, record indicators such as support, confidence, and lift, and ensure extraction The association relationship between the mined potential features is of practical significance. The mined potential feature association relationship is organized into a structured format (such as DataFrame), and information such as feature pairs, support, and confidence are recorded to facilitate subsequent modeling. Select a suitable knowledge graph construction tool (such as Neo4j, GraphDB, or RDF), prepare the environment required to build the graph, and define the time series state features and normalized vibration features as nodes. For example, nodes can include "speed", "vibration frequency", etc. According to the mined feature association relationship, define the edges between the nodes to represent the association relationship between them. For example, an edge can represent "speed and vibration frequency". In the selected knowledge graph tool, create nodes and edges to form the preliminary structure of the vehicle multimodal feature knowledge graph, ensure that each node and edge is accompanied by relevant attribute information (such as support and confidence), import the potential feature association data into the knowledge graph database, and form a complete multimodal feature knowledge graph. You can use the API or batch import tool provided by the graph database to execute queries to verify whether the constructed knowledge graph can correctly return the expected results and confirm the correctness of the nodes and edges. According to the test results, optimize the structure and content of the knowledge graph, and adjust the attributes of the nodes and edges to improve the usability and information value of the graph.

[0033] Step S4: obtaining the real-time operating parameters of the vehicle; performing abnormal parameter deviation calculation and fault location tracing on the real-time operating parameters of the vehicle based on the vehicle multimodal feature knowledge graph, thereby obtaining potential abnormal fault location points; In this embodiment, necessary sensors are installed on the vehicle, including speed sensors, accelerometers, throttle position sensors, engine temperature sensors, etc., to ensure that various operating parameters can be collected in real time. A data acquisition system (such as an OBD-II interface or a CAN bus device) is used to read the vehicle's operating parameters in real time, to ensure that the device can process high-frequency data and transmit the data to a computer or the cloud, to convert the collected real-time operating parameters into a unified format (such as JSON or CSV) to ensure that the data structure is standardized for subsequent processing, to clean the real-time operating parameters, to check missing values ​​and outliers, to make necessary filling or corrections, to ensure data accuracy, to store the cleaned real-time operating parameters in a database (such as MySQL, MongoDB or a data warehouse) for subsequent analysis, to use a graph database API (such as Neo4j's Cypher query language) to connect to the previously constructed vehicle multimodal feature knowledge graph, to extract normal state parameters corresponding to the real-time operating parameters from the knowledge graph, which are based on historical The data and feature association are derived by comparing the real-time operating parameters with the normal state parameters, calculating the deviation, setting the threshold (such as ±10%), judging whether the deviation exceeds the normal range, thereby identifying the abnormal parameters, and storing the identified abnormal parameters and their deviation values ​​in the database for subsequent analysis and fault location. The feature association logic extracted from the knowledge graph is used to analyze other features related to the abnormal parameters (such as vibration features, temperature features, etc.), and the relevant feature nodes are obtained through the query function of the graph database. The known fault mode data (which can be a database based on historical fault cases) are matched to identify the fault mode. For example, machine learning models (such as decision trees and random forests) can be used for fault mode recognition. According to the identified fault mode, the location point of the potential abnormal fault is determined, such as the failure of a sensor, the wear of a component, etc. The analysis results are organized into a fault diagnosis report, including information such as abnormal parameters, deviation values, potential fault modes and location points, for subsequent repair and maintenance.

[0034] Step S5: performing multi-period fault simulation on the potential abnormal fault location point and predicting the vehicle fault risk, thereby generating vehicle potential fault risk prediction data; In this embodiment, fault simulation software (such as MATLAB Simulink, CarSim or other vehicle dynamics simulation software) is used to build a vehicle model and perform fault simulation. A detailed vehicle model is built according to the actual parameters of the vehicle (such as the power system, suspension system, etc.) to ensure that the model can accurately reflect the operating characteristics of the vehicle. The relevant feature data of the potential abnormal fault location point extracted from the knowledge graph is imported into the simulation tool, including the fault type, the affected sensor and the expected fault behavior. The time period of the fault simulation is determined (for example, a short-term simulation is a few minutes, and a long-term simulation is a few hours or a few days). The simulation period is determined according to the vehicle usage scenario. Different fault conditions (such as sensor failure, component wear, system failure, etc.) are set according to the potential abnormal fault location point, and how these faults are manifested in the simulation is defined. The simulation is started, the performance of the vehicle under different fault conditions is observed, and the changes in key operating parameters (such as speed, acceleration, throttle position, vibration, etc.) are recorded. The data generated during the simulation process is recorded, including the state when the fault occurs, the dynamic response of the vehicle and the feedback of the system. Feedback information, use machine learning algorithms (such as decision trees, random forests, support vector machines) or statistical models (such as regression analysis) to establish a fault risk prediction model, extract key features that affect the fault risk from the simulation data, select highly correlated features for subsequent analysis, use historical fault data and data generated by simulation to train the model, optimize model parameters to improve prediction accuracy, use cross-validation to evaluate the model, ensure that the model performs stably on different data sets, use the trained model to predict the fault risk of new simulation data, calculate the risk level of each fault location point (such as low, medium, and high), and organize the prediction results into a report, including the risk level of each potential fault location point, the predicted fault type and its impact, organize the results of the fault risk prediction, and write a detailed report, including simulation settings, risk assessment, prediction results, and recommended measures. Apply the prediction results to vehicle maintenance and management decisions, take timely measures to reduce the risk of failure, establish a feedback mechanism, collect actual fault data during vehicle operation, compare it with simulation predictions, and continuously improve the model and prediction accuracy. Step S6: Make cloud-edge collaborative decisions on the vehicle's potential fault risk prediction data, perform dynamic fault diagnosis scheduling, and build an intelligent fault diagnosis scheduling model.

[0035] In this embodiment, all data obtained from fault risk prediction (such as risk level, predicted fault type and its potential impact) are summarized to form a structured data set (such as JSON or CSV format), and cloud computing resources (such as AWS, Azure or self-built servers) are configured for large-scale data processing and model training. At the same time, edge devices (such as on-board computers or edge gateways) are configured for real-time data processing and decision-making. The real-time operation data and fault risk prediction data of the vehicle are transmitted to the cloud and edge devices using protocols such as MQTT or HTTP to ensure the security and real-time nature of data transmission. The received prediction data is cleaned and preprocessed in the cloud to ensure data integrity and consistency to prepare for subsequent analysis. A pre-established fault risk prediction model is used to perform risk assessment on the integrated prediction data, and a priority (such as high, medium, and low) is assigned to each potential fault point. According to the risk assessment results, a corresponding fault response strategy is formulated. For example, for high-risk faults, inspection and repair are arranged first; for medium and low-risk faults, a monitoring plan can be formulated to process the vehicle's operation data in real time on the edge device, and perform fault diagnosis and response according to the decision instructions transmitted from the cloud. Scheduling: Use dynamic scheduling algorithms (such as priority queue scheduling, genetic algorithms, etc.) to schedule fault diagnosis tasks. According to real-time data and risk assessment results, dynamically adjust the priority and resource allocation of fault diagnosis. According to the scheduling results, the edge device performs the corresponding fault diagnosis task, and can use embedded diagnostic algorithms (such as fault tree analysis, fuzzy logic control, etc.) for real-time fault detection. Based on the collaborative mechanism of cloud decision-making and edge execution, design an intelligent fault diagnosis scheduling model. The model should consider multiple factors, such as fault type, risk level, resource availability and time constraints. Use historical fault data and real-time operation data to train the scheduling model, optimize model parameters to improve scheduling efficiency, and apply optimization algorithms (such as particle swarm optimization, simulated annealing, etc.) to optimize the scheduling strategy to ensure the timeliness and accuracy of fault diagnosis. Verify the effectiveness of the scheduling model through simulation tests and actual scenarios to ensure the adaptability and response speed of the model in different fault scenarios. Collect data and feedback information during the fault diagnosis process, analyze the performance of the scheduling model, identify potential improvement points, regularly update and optimize the scheduling model, and combine new fault data and operation data to ensure the accuracy and real-time performance of the model.

[0036] In this embodiment, refer to Figure 2 , is a flowchart of detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include: Step S11: acquiring vehicle multimodal monitoring data based on the intelligent diagnostic instrument, wherein the vehicle multimodal monitoring data includes vehicle engine all-round monitoring parameters, vehicle circuit monitoring data and vehicle running status sound signals; Step S12: Performing temperature fluctuation analysis on all-round monitoring parameters of the vehicle engine and extracting temperature fluctuation data; Step S13: performing discrete fitting of time series fluctuations on the temperature fluctuation data to construct an engine temperature fluctuation curve; Step S14: calculating the operating power variation of the vehicle engine's all-round monitoring parameters to generate engine operating power variation characteristics; Step S15: mining the time series state characteristics of the engine operating power variation characteristics, the engine temperature fluctuation curve and the vehicle circuit monitoring data to obtain the vehicle time series state characteristics.

[0037] In this embodiment, the interface of the intelligent diagnostic instrument is determined, and the formats and types of multimodal monitoring data supported by it are understood, including engine monitoring parameters, circuit monitoring data, and sound signals. The API or data interface (such as OBD-II or CAN bus) of the diagnostic instrument is used to obtain the multimodal monitoring data of the vehicle in real time, and the frequency of data acquisition is ensured to be high enough to capture changes in key parameters (such as obtaining data once per second). The obtained multimodal monitoring data is stored in a database (such as MongoDB, PostgreSQL) for subsequent analysis, and the timestamp, data type, and source of the data are recorded to facilitate data tracking and cleaning. The temperature monitoring data of the vehicle engine is extracted from the stored multimodal monitoring data. Measure parameters such as coolant temperature and oil temperature, clean the extracted temperature data, remove outliers and missing values, ensure the accuracy and completeness of the data, use statistical methods (such as Z-score) to detect and remove outliers, calculate temperature fluctuations, use standard deviation and mean methods to evaluate the amplitude of temperature changes, record the time period of temperature fluctuations, prepare for subsequent analysis, organize the cleaned temperature fluctuation data into time series, ensure that the data is continuous and orderly, select appropriate time series analysis models (such as ARIMA, exponential smoothing, etc.) to fit the temperature fluctuation data, use the selected model to fit the temperature fluctuation data, generate the engine temperature fluctuation curve, and use Python's stats models or R's forecast package for fitting and visualization, verify the accuracy of the fitting results, check the distribution of residuals and the predictive ability of the model, ensure that the fitting curve can reflect the actual fluctuations, extract parameters related to engine operating power, such as speed, torque, etc. from multimodal monitoring data, calculate the engine operating power change according to the engine power calculation formula (such as power = torque × speed / 5252), record the calculated power change data and store it in the database, analyze the calculated engine operating power characteristics, identify the trend and abnormal points of power change, generate a time series chart of power change for easy observation and analysis, and extract the engine temperature fluctuation curve and operating power change characteristics. The vehicle circuit monitoring data are integrated into a unified data set, and the time series feature extraction technology (such as sliding window and Fourier transform) is applied to analyze the integrated data to extract key features. The tsfresh library of Python or the TSA package of R is used for feature extraction. The extracted features are analyzed to identify the time series state characteristics of the vehicle, including the distinction between normal state and abnormal state. A time series state feature report is generated to record the identified state characteristics and their corresponding time periods. The mined time series state features are stored in the database for subsequent query and analysis. The time series state features are displayed using visualization tools (such as Matplotlib or Tableau) to help understand the vehicle performance and health status.

[0038] In this embodiment, the specific steps of step S15 are: Identify multiple circuit element nodes of the vehicle; Based on multiple circuit element nodes of the vehicle, the interaction analysis between the element nodes is performed to obtain the node interaction circuit; Match the circuit parameters of the node interaction circuit based on the vehicle circuit monitoring data, so as to obtain the monitoring parameters of each vehicle circuit; extracting voltage timing parameters based on monitoring parameters of the vehicle circuit; Performing voltage timing waveform fitting on the voltage timing parameters to obtain a voltage timing waveform curve; Performing Fourier transformation on the voltage time series waveform curve to generate voltage time series waveform spectrum data; Calculate the voltage variation amplitude on the voltage time series waveform spectrum data to obtain the voltage variation amplitude value; Evolving the voltage change amplitude value, thereby generating voltage change trend evolution characteristics; Carry out vehicle voltage load situation analysis based on the voltage change trend evolution characteristics to obtain vehicle voltage load situation characteristics; Based on the vehicle voltage load situation characteristics, the voltage load-operating power correlation change analysis is performed on the engine operating power change characteristics to generate a voltage load-operating power correlation characteristic; The time-series state quantity evolution of the voltage load-operating power correlation characteristics and the engine temperature fluctuation curve is performed to obtain the vehicle time-series state characteristics.

[0039] In this embodiment, the documents of the vehicle circuit diagram and related circuit components are collected to understand the function and layout of each circuit component, use circuit analysis tools (such as LTspice or AltiumDesigner) to identify the circuit component nodes, record the identifier of each component (such as resistor, sensor, control module, etc.), determine the connection relationship between each component node, form a circuit topology diagram, store the identified component nodes and their connection relationship in a database, form a node information table, and facilitate subsequent analysis. The electrical connection between the nodes is analyzed, the signal flow direction and interaction relationship are determined, and the data flow path is recorded. Based on the analysis results, a node interaction circuit model is constructed to clarify the input and output signals of each node, and the circuit simulation software is used to simulate the interactive behavior of the circuit. , verify the validity of the interactive logic, store the interactive circuit model and parameters in the database for subsequent circuit parameter matching, extract circuit parameter data from the vehicle circuit monitoring system, including voltage, current, power and other monitoring data, compare the node interactive circuit, match the circuit monitoring parameters of each component node to ensure data consistency, use Python and other programming languages ​​to write scripts to automate the parameter matching process, extract voltage-related timing data from the circuit monitoring parameters to ensure data integrity and continuity, organize the voltage timing data into a time series format, ensure that the timestamp corresponds to the voltage value, select a suitable waveform fitting model (such as polynomial fitting, sine wave fitting) to adapt to the characteristics of the voltage timing data, use Python The NumPy or SciPy library is used to fit the voltage time series data, generate a voltage time series waveform curve, evaluate the accuracy of the fitting, check the fitting error, use Matplotlib and other visualization tools to draw the voltage time series waveform curve to facilitate the observation of waveform characteristics, perform Fourier transform on the fitted voltage time series waveform curve, extract frequency domain information, use NumPy's fft module to implement fast Fourier transform (FFT), record the spectrum data obtained after Fourier transform, including amplitude and frequency information, calculate the voltage change amplitude value based on the spectrum data, identify the amplitude of the main frequency components, store the calculated voltage change amplitude value in the database for subsequent analysis and comparison, perform statistical analysis on the voltage change amplitude value, and determine Change trends and their evolutionary characteristics, use time series analysis techniques (such as moving average, exponential smoothing) to extract trend characteristics, generate voltage change trend evolution characteristic reports, and store them in the database. Based on the voltage change trend evolution characteristics, establish a voltage load situation analysis framework, combine voltage monitoring parameters and load data, analyze the state characteristics of vehicle voltage load, identify normal and abnormal load states, collect vehicle voltage load situation characteristics and engine operating power change characteristics, form a comprehensive analysis data set, use correlation analysis methods (such as Pearson correlation coefficient) to study the relationship between voltage load and operating power, determine the degree of influence of voltage load state on engine operating power, generate voltage load-operating power correlation characteristic reports, and store them in the database.Based on the voltage load-operating power correlation characteristics and the engine temperature fluctuation curve, a time series state quantity evolution model is established to dynamically analyze the time series state quantity, identify the overall state characteristics and evolution trend of the vehicle, record the analysis results, and use visualization tools to display the vehicle's time series state characteristics for easy understanding and decision-making.

[0040] In this embodiment, refer to Figure 3 , is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include: Step S21: performing multi-frequency noise reduction optimization on the vehicle running state sound signal, thereby generating a noise reduction reconstructed sound signal; Step S22: performing a signal vibration characteristic time series analysis on the noise reduction and reconstructed sound signal to generate signal time series vibration characteristic data; Step S23: performing time series fitting on the signal time series vibration characteristic data to construct a signal time series vibration diagram; Step S24: performing vibration change trend evolution on the signal time series vibration diagram to extract vibration change trend features; Step S25: performing vibration peak-valley value distribution analysis on the signal time series vibration diagram to extract vibration peak-valley value distribution data; Step S26: Perform normalized periodic evolution mining on the vibration peak-valley value distribution data and the vibration change trend characteristics, thereby generating normalized vibration periodic evolution characteristics.

[0041] In this embodiment, a running state sound signal is obtained from the sound monitoring system of the vehicle, and it is ensured that the collected signal has sufficient duration and quality. A spectrum analysis (such as fast Fourier transform, FFT) is used to identify the noise frequency component in the signal, and the frequency range of the noise and the effective signal is divided. An appropriate noise reduction technology is selected, such as wavelet transform, spectral subtraction or adaptive filtering, and optimization is performed for noises of different frequencies. The sound signal is processed by the selected noise reduction algorithm to generate a noise-reduced reconstructed sound signal. The signal is processed using Python's SciPy library or MATLAB. The effectiveness of the noise reduction effect is verified by comparing the signal waveform and spectrum before and after noise reduction. Feature extraction is performed on the noise-reduced reconstructed sound signal to identify vibration features, such as amplitude, frequency and phase information. A time series analysis method (such as a sliding window method) is used to calculate the vibration features of the signal to generate signal time series vibration feature data, and the vibration feature value at each time point is recorded to form a time series data set. The signal time series vibration feature data is organized into a time series format to ensure the integrity and continuity of the data. A suitable time series fitting model (such as polynomial fitting, sine wave fitting) is selected. ) to fit the vibration feature data, use Python's NumPy or SciPy library to fit the vibration feature data, construct a signal time series vibration diagram, evaluate the precision and accuracy of the fitting, use visualization tools such as Matplotlib to draw the signal time series vibration diagram, and show the changing trend of the vibration characteristics. Based on the signal time series vibration diagram, establish a vibration change trend analysis framework, use time series analysis techniques (such as moving average, exponential smoothing) to extract the trend characteristics of vibration changes, generate a vibration change trend characteristic report, and store the results in a database, analyze the signal time series vibration diagram, identify the peaks and valleys in the vibration signal, count the distribution of peaks and valleys, record the amplitude, frequency and occurrence time of peaks and valleys, establish a normalized periodic evolution analysis framework based on the vibration peak and valley distribution data and vibration change trend characteristics, apply periodic analysis methods (such as Fourier transform, periodogram analysis) to analyze the vibration data, extract the normalized vibration periodic evolution characteristics, record the normalized vibration periodic evolution characteristics in the database, generate an analysis report, and use visualization tools to display the evolution characteristics for easy understanding and decision-making.

[0042] In this embodiment, the specific steps of step S21 are: Perform multi-time-frequency decomposition on the vehicle running state sound signal to obtain multiple time-frequency spectrograms; Calculate the noise component of each frequency on multiple time-spectrum graphs, thereby extracting the noise component of each frequency; Perform noise change identification on the noise component of each frequency to obtain the noise component change characteristics of each frequency; Adaptive filtering and denoising are performed according to the variation characteristics of the noise components of each frequency, thereby obtaining adaptive filtering and denoising frequency domain data; Define sliding time windows; Based on the sliding time window, the adaptive filtering denoising frequency domain data is segmented into sliding time windows to obtain denoising frequency domain data of multiple time windows; According to the vehicle running state sound signal, the noise reduction frequency domain data is subjected to signal distortion identification, and the time window frequency domain data of the signal distortion is extracted; Calculating the distortion amplitude of the signal distorted time window frequency domain data to generate a distortion amplitude value; Based on the distortion amplitude value, nonlinear distortion amplitude correction is performed on the adaptive filtering noise reduction frequency domain data, thereby obtaining distortion correction frequency domain data; The distortion-corrected frequency domain data is inversely reconstructed in time-frequency order to generate a noise-reduced reconstructed sound signal.

[0043] In this embodiment, a high-quality microphone or sensor is used to collect the sound signal of the vehicle running status in real time to ensure the clarity and integrity of the signal, and a suitable sampling frequency (such as 44.1kHz or 48kHz) is set to ensure that sufficient frequency components can be captured. Short-time Fourier transform (STFT) or wavelet transform (WaveletTransform) is used for multi-time-frequency decomposition to decompose the sound signal into multiple time-frequency spectrograms, and generate a spectrum corresponding to each time window. The scipy.signal library in Python or the spectrogram function in MATLAB is used for implementation. Each time-frequency spectrogram shows the signal at different times. And the energy distribution on the frequency, analyze each time-frequency spectrum, identify the noise components corresponding to different frequencies, and use frequency domain filtering (such as bandpass filter) to extract the noise components of each frequency for each time-frequency spectrum. Scipy.signal.butter in Python can be used to design filters, extract and record the noise components of each frequency, and form a noise feature data set for subsequent analysis. Statistical methods (such as mean and standard deviation calculation) or machine learning methods (such as decision trees and support vector machines) are used to identify changes in noise components. The noise components of each frequency are analyzed, the moments of significant changes are identified, and the change characteristics are recorded to generate the noise components of each frequency. The changing characteristic data is used as the basis for subsequent noise reduction processing. Adaptive filtering algorithms, such as Least Mean Squares (LMS) or Recursive Least Squares (RLS), are used to dynamically adjust the filter parameters. The noise component change characteristics are input into the adaptive filter for real-time noise reduction. Adaptive filtering noise reduction frequency domain data is generated. The size of the sliding window (such as 0.5 seconds or 1 second) is determined to adapt to the changing frequency of the signal. The sliding step size (such as 0.1 seconds each time) is set to ensure that the window covers different parts of the signal. The programming language (such as Python) is used to perform sliding window segmentation on the noise reduction frequency domain data to generate noise reduction for multiple time windows. The frequency domain data block uses correlation analysis, signal reconstruction error and other methods to identify signal distortion, analyzes the noise reduction frequency domain data of each time window, identifies the distorted part, calculates the amplitude of the distorted time window frequency domain data, identifies the distortion amplitude value, stores the calculated distortion amplitude value in the database, uses nonlinear correction methods such as polynomial regression or piecewise linear regression to correct the distortion, corrects the adaptive filtering noise reduction frequency domain data according to the calculated distortion amplitude value, generates distortion corrected frequency domain data, uses inverse short-time Fourier transform (ISTFT) or inverse wavelet transform to reconstruct the signal back to the time domain, and inversely transforms the corrected frequency domain data to generate a noise reduction reconstructed sound signal.

[0044] In this embodiment, refer to Figure 4, is a flowchart of detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include: Step S31: performing feature space mapping on the vehicle time series state characteristics and normalized vibration period evolution characteristics, thereby obtaining time series state feature nodes and normalized vibration feature nodes; Step S32: performing deep mining of potential feature associations on the time series state feature nodes and the normalized vibration feature nodes, thereby extracting potential feature association data; Step S33: performing semantic feature evolution on the potential feature association data, thereby generating feature association semantic logic; Step S34: Based on the feature association semantic logic, the vehicle time series state characteristics and the normalized vibration cycle evolution characteristics are modeled as a correlation graph to construct a vehicle multimodal feature knowledge graph.

[0045] In this embodiment, the time series state feature data and normalized vibration period evolution feature data of the vehicle are collected to ensure the integrity and consistency of the data. A mapping method is selected, such as principal component analysis (PCA), t-SNE or UMAP. These methods can map high-dimensional data to low-dimensional space for visualization and analysis. The time series state features and normalized vibration features are processed using the selected mapping method. In Python, PCA or t-SNE in the sklearn library can be used to achieve mapping. During the mapping process, the mapping result of each feature is recorded to form a time series state feature node and a normalized vibration feature node. The mapping results are stored in a database to form a feature node set in preparation for subsequent analysis. A suitable feature association mining algorithm is selected, such as the Apriori algorithm, FP-Growth or graph neural network (GNN). These methods can identify the potential relationship between features. The generated time series state feature nodes and normalized vibration feature nodes are subjected to association analysis using the selected algorithm. Network analysis is performed to extract potential feature association data, record the association strength and pattern between features, and construct a semantic network based on the potential feature association data. Framework, determine the semantic labels and attributes of features (such as time, frequency, amplitude, etc.), perform semantic analysis on potential features, identify the changes and evolution laws of features, which can be achieved using natural language processing technology or logical reasoning methods, record the semantic logic of each feature, including its evolution history and current status, store the generated feature-associated semantic logic documents in the database for subsequent query and analysis, select suitable knowledge graph construction tools, such as Neo4j, GraphDB or RDF, which can effectively manage and query graph structure data, based on the semantic logic of feature association, associate vehicle time series state features and normalized vibration cycle evolution features into association graph modeling, define nodes (features) and edges (relationships) in the graph, ensure accurate mapping of each feature and its associated logic, verify the validity of the constructed knowledge graph, display the graph structure through visualization tools (such as Gephi or Cytoscape), check the connection relationship between nodes and edges, optimize and adjust the graph based on feedback to ensure that it reflects accurate feature associations, store the constructed knowledge graph in the database, and provide a query interface for subsequent analysis and decision support.

[0046] In this embodiment, step S4 includes the following steps: Step S41: obtaining real-time operating parameters of the vehicle; Step S42: Calculate the abnormal parameter deviation of the real-time operating parameters of the vehicle based on the vehicle multimodal feature knowledge graph, and extract the abnormal deviation data; Step S43: performing potential abnormal fault analysis on the abnormal deviation data to obtain potential abnormal fault data; Step S44: tracing the potential abnormal fault data to locate the fault, thereby obtaining the potential abnormal fault location point.

[0047] In this embodiment, ensure that the vehicle is equipped with a real-time monitoring system (such as OBD-II or CAN bus) to obtain key operating parameters such as engine speed, vehicle speed, oil temperature, voltage, etc., use an appropriate programming interface (such as CAN interface or API) to obtain operating parameters from the vehicle's electronic control unit (ECU) in real time, use programming languages ​​such as Python or MATLAB, and use corresponding libraries (such as python-can) to achieve real-time data collection, store the acquired real-time operating parameters in a database, ensure the integrity and traceability of the data for subsequent analysis, define the parameter range and characteristics of normal operating status based on the vehicle multimodal feature knowledge graph, use statistical analysis methods to calculate the deviation between the real-time operating parameters and the normal parameter range in the knowledge graph, for example, Z-score, mean ± standard deviation and other methods can be used to calculate the abnormal deviation of the real-time operating parameters, extract deviation data, record which parameters are beyond the normal range, use Python's numpy and pandas libraries for data processing and deviation calculation, select appropriate abnormal Detection algorithms, such as IsolationForest, One-ClassSVM, or deep learning-based autoencoders, are specifically used to analyze abnormal deviation data. If machine learning methods are used, the model must first be trained with normal operating data and verified to ensure that it can effectively identify anomalies. The abnormal deviation data is input into the trained anomaly detection model for analysis, potential abnormal fault data is extracted, the detected anomaly type and its characteristics are recorded, the cause of the fault is analyzed, and appropriate fault location and tracing methods are selected, such as fault tree analysis (FTA), root cause analysis (RCA), or model-based methods. Based on potential abnormal fault data, the cause and influencing factors of the fault are analyzed, the location point of the fault is determined, and the associated information in the knowledge graph is used to identify the system modules related to the abnormal fault and their interactive relationships. The determined potential abnormal fault location points are recorded in the database to form a fault location report. If necessary, the analysis method or parameters are adjusted to improve the accuracy of fault location. According to the location results, corresponding maintenance or repair plans are formulated to ensure the normal operation of the vehicle.

[0048] In this embodiment, the specific steps of step S5 are: Step S51: performing multi-period fault simulation on the potential abnormal fault location point to generate fault simulation data of multiple periods; Step S52: performing fault point operating condition analysis on the fault simulation data of multiple time periods to generate fault point operating condition data of multiple time periods; Step S53: performing fault diffusion path mining based on the fault point working condition data of multiple time periods to obtain the fault diffusion path of each time period; Step S54: performing fault diffusion path distribution analysis on the fault diffusion path in each time period, and constructing fault path distribution graphs for multiple time periods; Step S55: Perform vehicle failure risk prediction on the failure path distribution diagrams of multiple time periods, thereby generating vehicle potential failure risk prediction data.

[0049] In this embodiment, the environment and parameters of the fault simulation, including the operating conditions and running status of the vehicle, are determined to ensure the validity of the simulation results. A suitable fault model, such as a fault tree model, a Markov model or a physics-based simulation model, is selected to simulate different types of faults. Multi-period fault simulation is performed on the potential abnormal fault location point to generate fault simulation data for multiple periods. Professional simulation software (such as MATLAB / Simulink, ANSYS or other tools) is used to perform fault simulation. The generated fault simulation data for multiple periods are stored in a database for subsequent analysis. Key indicators and parameters required for fault point working condition analysis, such as temperature, pressure, vibration, etc., are determined to ensure the comprehensiveness of the data. Suitable data analysis techniques, such as statistical analysis, time series analysis or machine learning methods, are selected to analyze the fault simulation data. The generated fault simulation data for multiple periods are analyzed for the working condition of the fault point, and the working condition feature data of the fault point is extracted. Python libraries such as pandas, numpy and scikit-learn are used for analysis. Suitable path mining algorithms, such as graph theory methods, network analysis or graph neural networks based on deep learning, are selected to perform fault diffusion paths based on the working condition data of the fault point for multiple periods. Mining, identifying the propagation path of faults in each time period, using Python's networkx library or deep learning framework (such as TensorFlow / PyTorch) to implement path mining, recording the fault diffusion path of each time period in the database for subsequent analysis and visualization, determining the indicators of fault diffusion path distribution analysis, such as the frequency and impact of the path, statistically analyzing the fault diffusion path of each time period, generating fault diffusion path distribution data, using data visualization tools (such as Matplotlib or Seaborn) to draw path distribution diagrams to display the distribution of fault paths in each time period, selecting a suitable fault risk prediction model, such as regression analysis, decision tree, random forest or deep learning model, using historical fault data and path distribution data to train the prediction model, and verifying it to ensure its accuracy, predicting vehicle failure risks based on the fault path distribution diagrams of multiple time periods, generating vehicle potential failure risk prediction data, using Python's scikit-learn or TensorFlow for model training and prediction, storing the generated vehicle potential failure risk prediction data in the database, and using visualization tools to display the prediction results for decision support.

[0050] In this embodiment, the specific steps of step S6 are: Step S61: Sharing the potential failure risk prediction data of the vehicle with the cloud to obtain failure risk prediction shared data; Step S62: performing fault risk type diagnosis on the fault risk prediction shared data to generate fault risk diagnosis data; Step S63: Perform cloud-edge collaborative decision-making on the fault risk diagnosis data, thereby generating a fault risk collaborative diagnosis strategy; Step S64: Perform dynamic fault diagnosis scheduling based on the fault risk collaborative diagnosis strategy and build an intelligent fault diagnosis scheduling model.

[0051] In this embodiment, the potential failure risk prediction data of the vehicle is collected to ensure the integrity and consistency of the data. A suitable cloud service platform, such as AWS, Azure, or Google Cloud, is selected for data storage and sharing. The failure risk prediction data is uploaded to the cloud using an API or SDK to ensure data security and privacy protection. In Python, boto3 (AWSSDK) or google-cloud-storage library can be used to upload data. A shared link is created or sharing permissions are set on the cloud so that relevant personnel or systems can access the shared failure risk prediction data. An analysis framework for failure risk type diagnosis is constructed, risk types and related indicators are defined, and a suitable failure risk diagnosis algorithm, such as a decision tree, support vector machine (SVM), or random forest, is selected to analyze the shared failure risk prediction data to generate failure risk diagnosis data. In Python, the scikit-learn library can be used to build a model and perform diagnosis. The generated failure risk diagnosis data is stored in a database for subsequent use. The coordination of cloud and edge computing nodes is determined. Decision-making framework, define the roles and responsibilities of each node, select appropriate collaborative decision-making algorithms, such as game theory models, swarm intelligence algorithms, or multi-agent systems, make collaborative decisions between cloud and edge nodes based on fault risk diagnosis data, generate fault risk collaborative diagnosis strategies, use distributed computing frameworks (such as Apache Kafka or Apache Flink) for data processing and decision-making, store the generated fault risk collaborative diagnosis strategies in the database for subsequent implementation, design intelligent fault diagnosis scheduling models, define scheduling objectives and constraints, such as time, resources, and priority, select appropriate scheduling algorithms, such as genetic algorithms, ant colony algorithms, or reinforcement learning algorithms, to optimize scheduling plans, perform dynamic fault diagnosis scheduling based on fault risk collaborative diagnosis strategies, optimize resource configuration and task allocation, use Python's DEAP library (for genetic algorithms) or stable-baselines3 (for reinforcement learning) to implement the model, monitor the scheduling execution effect, collect feedback data, and optimize and adjust the scheduling model to improve the efficiency and accuracy of fault diagnosis.

[0052] In this embodiment, a vehicle data analysis system based on an intelligent diagnostic instrument is provided, which is used to execute the vehicle data analysis method based on an intelligent diagnostic instrument as described above, including: The time series state feature module is used to obtain vehicle multi-modal monitoring data based on the intelligent diagnostic instrument; calculate the operating power change of the vehicle multi-modal monitoring data, and mine the time series state features to obtain the vehicle time series state features; The vibration period feature module is used to optimize multi-frequency noise reduction of vehicle multi-modal monitoring data and mine normalized period evolution to generate normalized vibration period evolution features; The knowledge graph module is used to conduct deep mining of potential feature associations of vehicle time series state characteristics and normalized vibration cycle evolution characteristics, and then conduct association graph modeling to construct a vehicle multimodal feature knowledge graph; The fault location and tracing module is used to obtain the real-time operating parameters of the vehicle; based on the vehicle multimodal feature knowledge graph, the abnormal parameter deviation of the real-time operating parameters of the vehicle is calculated and the fault location and tracing are performed to obtain the potential abnormal fault location point; The risk prediction module is used to perform multi-period fault simulation on potential abnormal fault location points and predict vehicle fault risks, thereby generating vehicle potential fault risk prediction data; The collaborative decision-making module is used to make collaborative decisions on the cloud edge for vehicle potential fault risk prediction data, perform dynamic fault diagnosis scheduling, and build an intelligent fault diagnosis scheduling model.

[0053] By acquiring vehicle multimodal monitoring data and mining time series state features, the present invention can fully understand the changing trends of vehicle operating states, provide strong support for subsequent fault prediction, and the obtained vehicle time series state features are helpful to identify potential problems and take preventive maintenance measures, thereby improving vehicle reliability and safety. Through multi-frequency noise reduction optimization and normalized periodic evolution mining, the evolution law of vehicle vibration characteristics can be identified to help predict the occurrence of potential vibration problems. Generating normalized vibration periodic evolution features can help to timely discover abnormal vibration conditions and improve the accuracy of diagnosis of vehicle vibration problems. By deeply mining potential feature associations and constructing a vehicle multimodal feature knowledge graph, a large amount of data can be integrated and analyzed to help identify the associations and laws between features. The constructed knowledge graph can help to better understand the relationship between vehicle data features, providing more accurate information for subsequent fault location and prediction. Based on the real-time operating parameters of the vehicle, abnormal parameter deviation calculation and fault location tracing can be performed to quickly and accurately locate potential abnormal fault points, reducing the time and cost of troubleshooting. This module can detect potential fault points in advance, reduce the impact of faults on vehicle operation, and improve the reliability and safety of the vehicle. Multi-period fault simulation and risk prediction of potential abnormal fault location points can help prevent faults in advance and reduce the impact of sudden faults on vehicle operation. The generated vehicle potential fault risk prediction data is helpful to formulate corresponding maintenance plans and measures, improve the availability and operation efficiency of the vehicle, and realize intelligent fault handling management through cloud-edge collaborative decision-making and dynamic fault diagnosis and scheduling, improve fault handling efficiency and accuracy, and build an intelligent fault diagnosis and scheduling model to help optimize resource utilization, improve vehicle maintenance efficiency, and ensure safe vehicle operation.

[0054] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0055] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for analyzing automobile data based on an intelligent diagnostic instrument, characterized in that: The following steps are involved: Step S1: Acquire vehicle multimodal monitoring data based on the intelligent diagnostic instrument; Calculate the operating power change of the vehicle multi-modal monitoring data and mine the time series state characteristics to obtain the vehicle time series state characteristics; Step S2: performing multi-frequency noise reduction optimization on the vehicle multi-modal monitoring data and mining the normalized periodic evolution to generate normalized vibration periodic evolution features; Step S3: Perform deep mining of potential feature associations on the vehicle's time series state features and normalized vibration cycle evolution features, then perform association graph modeling to construct a vehicle multimodal feature knowledge graph; Step S4: obtaining the real-time operating parameters of the vehicle; performing abnormal parameter deviation calculation and fault location tracing on the real-time operating parameters of the vehicle based on the vehicle multimodal feature knowledge graph, thereby obtaining potential abnormal fault location points; Step S5: performing multi-period fault simulation on the potential abnormal fault location point and predicting the vehicle fault risk, thereby generating vehicle potential fault risk prediction data; Step S6: Make cloud-edge collaborative decisions on the vehicle's potential fault risk prediction data, perform dynamic fault diagnosis scheduling, and build an intelligent fault diagnosis scheduling model.

2. The automobile data analysis method based on the intelligent diagnostic instrument according to claim 1 is characterized in that: The specific steps of step S1 are: Step S11: acquiring vehicle multimodal monitoring data based on the intelligent diagnostic instrument, wherein the vehicle multimodal monitoring data includes vehicle engine all-round monitoring parameters, vehicle circuit monitoring data and vehicle running status sound signals; Step S12: Performing temperature fluctuation analysis on all-round monitoring parameters of the vehicle engine and extracting temperature fluctuation data; Step S13: performing discrete fitting of time series fluctuations on the temperature fluctuation data to construct an engine temperature fluctuation curve; Step S14: calculating the operating power variation of the vehicle engine's all-round monitoring parameters to generate engine operating power variation characteristics; Step S15: mining the time series state characteristics of the engine operating power variation characteristics, the engine temperature fluctuation curve and the vehicle circuit monitoring data to obtain the vehicle time series state characteristics.

3. The automobile data analysis method based on the intelligent diagnostic instrument according to claim 2 is characterized in that: The specific steps of step S15 are: Identify multiple circuit element nodes of the vehicle; Based on multiple circuit element nodes of the vehicle, the interaction analysis between the element nodes is performed to obtain the node interaction circuit; Matching circuit parameters of node interaction circuits based on vehicle circuit monitoring data to obtain monitoring parameters of each vehicle circuit; extracting voltage timing parameters based on monitoring parameters of the vehicle circuit; Performing voltage timing waveform fitting on the voltage timing parameters to obtain a voltage timing waveform curve; Performing Fourier transformation on the voltage time series waveform curve to generate voltage time series waveform spectrum data; Calculate the voltage variation amplitude on the voltage time series waveform spectrum data to obtain the voltage variation amplitude value; Evolving the voltage change amplitude value, thereby generating voltage change trend evolution characteristics; Carry out vehicle voltage load situation analysis based on the voltage change trend evolution characteristics to obtain vehicle voltage load situation characteristics; Based on the vehicle voltage load situation characteristics, the voltage load-operating power correlation change analysis is performed on the engine operating power change characteristics to generate a voltage load-operating power correlation characteristic; The time-series state characteristics of the vehicle are obtained by performing time-series state quantity evolution on the voltage load-operating power correlation characteristics and the engine temperature fluctuation curve.

4. The automobile data analysis method based on the intelligent diagnostic instrument according to claim 1 is characterized in that: The specific steps of step S2 are: Step S21: performing multi-frequency noise reduction optimization on the vehicle running state sound signal, thereby generating a noise reduction reconstructed sound signal; Step S22: performing a signal vibration characteristic time series analysis on the noise reduction and reconstructed sound signal to generate signal time series vibration characteristic data; Step S23: performing time series fitting on the signal time series vibration characteristic data to construct a signal time series vibration diagram; Step S24: performing vibration change trend evolution on the signal time series vibration diagram to extract vibration change trend features; Step S25: performing vibration peak-valley value distribution analysis on the signal time series vibration diagram to extract vibration peak-valley value distribution data; Step S26: Perform normalized periodic evolution mining on the vibration peak-valley value distribution data and the vibration change trend characteristics, thereby generating normalized vibration periodic evolution characteristics.

5. The automobile data analysis method based on the intelligent diagnostic instrument according to claim 4 is characterized in that: The specific steps of step S21 are: Perform multi-time-frequency decomposition on the vehicle running state sound signal to obtain multiple time-frequency spectrograms; Calculate the noise component of each frequency on multiple time-spectrum graphs, thereby extracting the noise component of each frequency; Perform noise change identification on the noise component of each frequency to obtain the noise component change characteristics of each frequency; Adaptive filtering and denoising are performed according to the variation characteristics of the noise components of each frequency, thereby obtaining adaptive filtering and denoising frequency domain data; Define sliding time windows; Based on the sliding time window, the adaptive filtering denoising frequency domain data is segmented into sliding time windows to obtain denoising frequency domain data of multiple time windows; According to the vehicle running state sound signal, the noise reduction frequency domain data is subjected to signal distortion identification, and the time window frequency domain data of the signal distortion is extracted; Calculating the distortion amplitude of the signal distorted time window frequency domain data to generate a distortion amplitude value; Based on the distortion amplitude value, nonlinear distortion amplitude correction is performed on the adaptive filtering noise reduction frequency domain data, thereby obtaining distortion correction frequency domain data; The distortion-corrected frequency domain data is inversely reconstructed in time-frequency order to generate a noise-reduced reconstructed sound signal.

6. The automobile data analysis method based on the intelligent diagnostic instrument according to claim 1 is characterized in that: The specific steps of step S3 are: Step S31: performing feature space mapping on the vehicle time series state characteristics and normalized vibration period evolution characteristics, thereby obtaining time series state feature nodes and normalized vibration feature nodes; Step S32: performing deep mining of potential feature associations on the time series state feature nodes and the normalized vibration feature nodes, thereby extracting potential feature association data; Step S33: performing semantic feature evolution on the potential feature association data, thereby generating feature association semantic logic; Step S34: Based on the feature association semantic logic, the vehicle time series state characteristics and the normalized vibration cycle evolution characteristics are modeled as a correlation graph to construct a vehicle multimodal feature knowledge graph.

7. The automobile data analysis method based on the intelligent diagnostic instrument according to claim 1 is characterized in that: The specific steps of step S4 are: Step S41: obtaining real-time operating parameters of the vehicle; Step S42: calculating the abnormal parameter deviation of the real-time operating parameters of the vehicle based on the vehicle multimodal feature knowledge graph, and extracting the abnormal deviation data; Step S43: performing potential abnormal fault analysis on the abnormal deviation data to obtain potential abnormal fault data; Step S44: tracing the potential abnormal fault data to locate the fault, thereby obtaining the potential abnormal fault location point.

8. The automobile data analysis method based on the intelligent diagnostic instrument according to claim 1 is characterized in that: The specific steps of step S5 are: Step S51: performing multi-period fault simulation on the potential abnormal fault location point to generate fault simulation data of multiple periods; Step S52: performing fault point operating condition analysis on the fault simulation data of multiple time periods to generate fault point operating condition data of multiple time periods; Step S53: performing fault diffusion path mining based on the fault point working condition data of multiple time periods to obtain the fault diffusion path of each time period; Step S54: performing fault diffusion path distribution analysis on the fault diffusion path in each time period, and constructing fault path distribution graphs for multiple time periods; Step S55: Perform vehicle failure risk prediction on the failure path distribution diagrams of multiple time periods, thereby generating vehicle potential failure risk prediction data.

9. The automobile data analysis method based on the intelligent diagnostic instrument according to claim 1, characterized in that: The specific steps of step S6 are: Step S61: Sharing the potential failure risk prediction data of the vehicle with the cloud to obtain failure risk prediction shared data; Step S62: performing fault risk type diagnosis on the fault risk prediction shared data to generate fault risk diagnosis data; Step S63: Perform cloud-edge collaborative decision-making on the fault risk diagnosis data, thereby generating a fault risk collaborative diagnosis strategy; Step S64: Perform dynamic fault diagnosis scheduling based on the fault risk collaborative diagnosis strategy and build an intelligent fault diagnosis scheduling model.

10. An automobile data analysis system based on an intelligent diagnostic instrument, characterized in that: The method for analyzing automobile data based on the intelligent diagnostic instrument according to claim 1 comprises: The time series state feature module is used to obtain vehicle multi-modal monitoring data based on the intelligent diagnostic instrument; calculate the operating power change of the vehicle multi-modal monitoring data, and mine the time series state features to obtain the vehicle time series state features; The vibration period feature module is used to optimize multi-frequency noise reduction of vehicle multi-modal monitoring data and mine normalized period evolution to generate normalized vibration period evolution features; The knowledge graph module is used to conduct deep mining of potential feature associations of vehicle time series state characteristics and normalized vibration cycle evolution characteristics, and then conduct association graph modeling to construct a vehicle multimodal feature knowledge graph; The fault location and tracing module is used to obtain the real-time operating parameters of the vehicle; based on the vehicle multimodal feature knowledge graph, the abnormal parameter deviation of the real-time operating parameters of the vehicle is calculated and the fault location and tracing are performed to obtain the potential abnormal fault location point; The risk prediction module is used to perform multi-period fault simulation on potential abnormal fault location points and predict vehicle fault risks, thereby generating vehicle potential fault risk prediction data; The collaborative decision-making module is used to make collaborative decisions on the cloud edge for vehicle potential fault risk prediction data, perform dynamic fault diagnosis scheduling, and build an intelligent fault diagnosis scheduling model.

Images