Search

CN-122024355-A - Real-time vehicle running state monitoring method and system based on Internet of things

CN122024355ACN 122024355 ACN122024355 ACN 122024355ACN-122024355-A

Abstract

The invention relates to the field of vehicle detection, and provides a real-time vehicle running state monitoring method and system based on the Internet of things, wherein the method comprises the steps of carrying out multistage preprocessing on an original signal acquired by a vehicle multisource sensor to obtain a standardized multi-mode data stream; the method comprises the steps of inputting an edge computing unit, carrying out feature extraction on a standardized multi-mode data stream through a multi-domain feature extraction algorithm to obtain a comprehensive feature vector, carrying out anomaly detection through a manifold alignment algorithm to obtain a feature data packet with a priority mark, carrying out channel allocation on a multi-channel communication interface through a game theory model to obtain a target transmission scheme, uploading the feature data packet to a cloud according to the target transmission scheme to obtain a time sequence feature database, inputting the time sequence feature database into a cloud digital twin model, carrying out fault diagnosis on the time sequence feature database through a residual error fusion algorithm to obtain a structured fault diagnosis report. The invention improves the driving safety of the vehicle.

Inventors

  • GUO JINXING
  • ZHAO XIN

Assignees

  • 江苏文睿信息科技有限公司

Dates

Publication Date
20260512
Application Date
20260414

Claims (10)

  1. 1. The real-time vehicle running state monitoring method based on the Internet of things is characterized by comprising the following steps of: s1, carrying out multistage preprocessing on an original signal acquired by a vehicle multisource sensor to obtain a standardized multi-mode data stream; S2, inputting the standardized multi-mode data stream into an edge computing unit, extracting features of the standardized multi-mode data stream through a multi-domain feature extraction algorithm to obtain a comprehensive feature vector, and detecting abnormality of the comprehensive feature vector through a manifold alignment algorithm to obtain a feature data packet with a priority mark; s3, carrying out channel allocation on a multi-channel communication interface through a game theory model according to the priority identification in the characteristic data packet to obtain a target transmission scheme, and uploading the characteristic data packet to a cloud according to the target transmission scheme to obtain a time sequence characteristic database; s4, inputting the time sequence feature database into a cloud digital twin model, and performing fault diagnosis on the time sequence feature database through a residual error fusion algorithm to obtain a structured fault diagnosis report.
  2. 2. The method for monitoring the running state of a vehicle in real time based on the internet of things according to claim 1, wherein the step S1 further comprises: S11, performing hardware filtering processing on the original acquisition signals of the plurality of sensors through a low-pass filter circuit, and filtering high-frequency electromagnetic interference components to obtain primary filtering signals; s12, denoising the vibration signal in the primary filtering signal through a wavelet threshold denoising algorithm; smoothing the slowly-varying signals in the primary filtering signals through sliding average filtering to obtain a denoising signal set; And S13, according to the denoising signal set, uniformly aligning all channel signals to a preset sampling granularity through a time interpolation alignment algorithm, and packaging the channel signals into a standardized multi-mode data stream.
  3. 3. The method for monitoring the running state of a vehicle in real time based on the internet of things according to claim 1, wherein the step S2 further comprises: S21, respectively extracting time domain statistical features, frequency domain power spectrum density features and time domain semantic features from sliding window data of the standardized multi-mode data stream to obtain three feature subsets; S22, splicing and fusing the three types of feature subsets to obtain the comprehensive feature vector; S23, performing anomaly detection on the comprehensive feature vector through a manifold alignment algorithm to obtain a feature data packet.
  4. 4. The method for monitoring the running state of a vehicle in real time based on the internet of things according to claim 3, wherein the step S21 specifically includes: Calculating the mean value, variance, peak value factor, waveform factor and kurtosis from the time sequence data in the sliding window to obtain time domain statistical characteristics; Performing spectrum analysis on the time sequence data through fast Fourier transformation, dividing a plurality of frequency bands by Mel scales, and calculating the energy duty ratio of each frequency band to obtain the frequency domain power spectral density characteristic; And generating a time-frequency diagram through short-time Fourier transform, inputting the time-frequency diagram into a lightweight convolutional neural network, and extracting and obtaining a time-frequency domain semantic feature vector.
  5. 5. The method for monitoring the running state of a vehicle in real time based on the internet of things according to claim 3, wherein the step S23 further comprises: S231, constructing a k neighbor graph for the history window data of the comprehensive feature vector through an incremental local linear embedding algorithm, and performing manifold modeling on the topological structure of the data point in a low-dimensional space to obtain a current manifold structure model; s232, aiming at the newly arrived comprehensive feature vector data points, calculating the embedding distance and the local curvature variation relative to the current manifold structure model to obtain manifold deviation metric values; S233, carrying out weighted fusion on the manifold deviation metric value and the abnormal score output by the isolated forest model, and dividing the health state according to a preset threshold value to obtain a local initial judgment label; S234, the local initial label and the manifold deviation metric package generate a characteristic data packet with a priority mark.
  6. 6. The method for monitoring the running state of a vehicle in real time based on the internet of things according to claim 1, wherein the step S3 further comprises: S31, taking the delay reciprocal, the channel reliability score and the transmission cost as components, dynamically adjusting the weight of each component according to the priority identification in the characteristic data packet, and constructing a multi-objective utility function; S32, regarding the C-V2X, the cellular network, the Wi-Fi and the low-power-consumption wide area network as game participants, continuously optimizing a plurality of channel strategies through an iterative optimal response algorithm, and solving Nash equilibrium points to obtain a preliminary transmission scheme; S33, negotiating a transmission degradation strategy in the dimensionality comprising timeliness and data integrity through the bargaining game model when all the channel utility values are lower than a preset critical value so as to correct the preliminary transmission scheme and obtain a target transmission scheme; S34, uploading according to the target transmission scheme in a grading mode according to the priority, and obtaining a time sequence characteristic database.
  7. 7. The method for monitoring the running state of a vehicle in real time based on the internet of things according to claim 1, wherein the step S4 further comprises: S41, driving physical mechanism sub-model simulation operation by using actual measurement parameters in the time sequence characteristic database based on a vehicle dynamics equation and a thermodynamic model to obtain theoretical expected state values of a plurality of vehicle components; S42, inputting the historical comprehensive feature vector sequence in the time sequence feature database into a time sequence Transformer network, and learning time sequence dependency relations of a plurality of feature dimensions in a normal driving mode to obtain an abnormal probability vector; s43, calculating a residual vector between the actually measured comprehensive feature vector and the theoretical expected state value in the time sequence feature database, and carrying out residual fusion on the residual vector and the abnormal probability vector through a multi-layer perceptron to obtain a fault type probability distribution vector covering multiple types of fault modes; S44, generating the structural fault diagnosis report according to the fault type probability distribution vector and manifold evolution track analysis.
  8. 8. Real-time vehicle running state monitoring system based on thing networking, its characterized in that includes: the preprocessing module is used for carrying out multistage preprocessing on the original signals acquired by the vehicle multisource sensor to obtain a standardized multi-mode data stream; the detection module is configured based on an edge computing unit and is used for receiving the standardized multi-mode data stream, extracting the characteristics of the standardized multi-mode data stream through a multi-domain characteristic extraction algorithm to obtain a comprehensive characteristic vector, and detecting the abnormality of the comprehensive characteristic vector through a manifold alignment algorithm to obtain a characteristic data packet with a priority mark; The transmission module is used for carrying out channel allocation on the multi-channel communication interface through a game theory model according to the priority identification in the characteristic data packet to obtain a target transmission scheme, and uploading the characteristic data packet to a cloud according to the target transmission scheme to obtain a time sequence characteristic database; The diagnosis module is configured based on a cloud digital twin model and is used for receiving the time sequence feature database, performing fault diagnosis on the time sequence feature database through a residual fusion algorithm and obtaining a structured fault diagnosis report.
  9. 9. Vehicle running state real-time monitoring equipment based on thing networking, its characterized in that includes: a memory and at least one processor, the memory having instructions stored therein; At least one of the processors invokes the instructions in the memory to cause an internet of things based vehicle operation state real-time monitoring device to perform an internet of things based vehicle operation state real-time monitoring method as claimed in any one of claims 1 to 7.
  10. 10. A computer readable storage medium, wherein instructions are stored on the computer readable storage medium, and when executed by a processor, the instructions implement a real-time monitoring method for vehicle operation status based on internet of things according to any one of claims 1 to 7.

Description

Real-time vehicle running state monitoring method and system based on Internet of things Technical Field The invention relates to the technical field of vehicle detection, in particular to a real-time vehicle running state monitoring method and system based on the Internet of things. Background With the continuous deep degree of electronization and intellectualization of automobiles, real-time monitoring of the running state of the automobiles has become an important means for guaranteeing driving safety and reducing maintenance cost. Modern vehicles are provided with a large number of heterogeneous sensors, can continuously collect multidimensional operation data such as engine rotating speed, tire pressure, chassis posture, braking temperature and the like, and have potential value for reflecting the health state of the vehicles after the data are converged through a vehicle-mounted bus. Meanwhile, the rapid development of edge computing, internet of vehicles communication and cloud deep learning technologies provides a technical foundation for building an end-side-cloud collaborative vehicle health monitoring architecture. The existing vehicle health monitoring method generally extracts statistical characteristics of sensor data at a vehicle-mounted end or a cloud end, sets a fixed threshold to trigger fault early warning, or adopts a spectrum analysis method to judge abnormality of signals such as vibration, temperature and the like. However, all the above methods judge single or few characteristic dimensions at the numerical level, and it is difficult to perceive the change of the overall distribution structure of the multi-dimensional sensor data in the high-dimensional space. Secondly, in normal operation of the vehicle, the multidimensional sensor data presents a stable low-dimensional manifold structure in a high-dimensional space, and early faults often appear as weak distortions or local topological deviations of the manifold structure, and the structural changes are hardly perceived at the level of single numerical characteristics or simple statistics, so that the early perception capability of the traditional method in the fault germination stage is insufficient. In addition, in the existing internet of vehicles transmission scheme, a transmission channel is generally statically designated according to data priority, when a plurality of high-priority data packets compete for limited channel resources at the same time in a complex network environment with multiple channels, a static routing strategy cannot dynamically sense the real-time load state and transmission income of each channel, transmission delay and even loss of key early warning data are easy to cause, and the timeliness of sending key health data is difficult to guarantee in network congestion or signal instability scenes. Disclosure of Invention The invention provides a real-time vehicle running state monitoring method and system based on the Internet of things, which are used for solving the defects of the prior art. The invention provides a vehicle running state real-time monitoring method based on the Internet of things, which comprises the following steps: s1, carrying out multistage preprocessing on an original signal acquired by a vehicle multisource sensor to obtain a standardized multi-mode data stream; S2, inputting the standardized multi-mode data stream into an edge computing unit, extracting features of the standardized multi-mode data stream through a multi-domain feature extraction algorithm to obtain a comprehensive feature vector, and detecting abnormality of the comprehensive feature vector through a manifold alignment algorithm to obtain a feature data packet with a priority mark; s3, carrying out channel allocation on a multi-channel communication interface through a game theory model according to the priority identification in the characteristic data packet to obtain a target transmission scheme, and uploading the characteristic data packet to a cloud according to the target transmission scheme to obtain a time sequence characteristic database; s4, inputting the time sequence feature database into a cloud digital twin model, and performing fault diagnosis on the time sequence feature database through a residual error fusion algorithm to obtain a structured fault diagnosis report. According to the vehicle running state real-time monitoring method based on the Internet of things, the step S1 further comprises the following steps: S11, performing hardware filtering processing on the original acquisition signals of the plurality of sensors through a low-pass filter circuit, and filtering high-frequency electromagnetic interference components to obtain primary filtering signals; s12, denoising the vibration signal in the primary filtering signal through a wavelet threshold denoising algorithm; smoothing the slowly-varying signals in the primary filtering signals through sliding average filtering to obtain a denoisin