Vehicle status monitoring method, system, terminal and medium based on Internet of Vehicles data

Abstract

The embodiment of the present invention provides a vehicle state monitoring method, system, terminal and medium based on Internet of Vehicles data, which belongs to the field of Internet of Vehicles technology. The method includes: obtaining initial Internet of Vehicles data of a target vehicle; performing curve fitting on the initial Internet of Vehicles data according to dynamic time bending combined with preset data of the target vehicle to obtain a target fitting curve; performing data feature recognition on the target fitting curve to obtain initial abnormal data; performing group detection on the initial abnormal data according to the maximum information coefficient to obtain first abnormal data and a first abnormal probability of the first abnormal data; performing abnormal recognition on the initial abnormal data according to a convolutional neural network to obtain second abnormal data and a second abnormal probability of the second abnormal data; fusing the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain target abnormal data; performing vehicle state prediction according to the target abnormal data to obtain a vehicle state prediction result.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Vehicles, and in particular to a vehicle status monitoring method, system, terminal and medium based on Internet of Vehicles data. Background Art

[0002] With the development of artificial intelligence, artificial intelligence is also applied to vehicle status monitoring in the field of vehicle technology. Current vehicle status monitoring mainly relies on complex algorithm models, such as machine learning and deep learning, which can analyze a large amount of data to achieve accurate prediction and monitoring of vehicle status. However, the effective operation of these advanced algorithm models requires a large training data set. Only through sufficient learning and training can the model achieve a high prediction accuracy. However, training and optimizing these models is a time-consuming and computationally resource-intensive process. It may take a lot of time to adjust parameters and optimize algorithms, which not only increases the cost of technical implementation, but also becomes a bottleneck restricting the rapid iteration and application deployment of the model. In actual applications, due to factors such as sensor failure and environmental interference, the collected data may not always be accurate and reliable, which further exacerbates the uncertainty of model prediction. Therefore, the reliability of the monitoring results of existing vehicle status monitoring technology in the face of complex and changeable actual working conditions still needs to be improved. Summary of the invention

[0003] The main purpose of the embodiments of the present invention is to provide a vehicle status monitoring method, system, terminal and medium based on Internet of Vehicles data, aiming to solve the problem of low reliability of vehicle status monitoring results in related technologies.

[0004] In a first aspect, an embodiment of the present invention provides a vehicle state monitoring method based on Internet of Vehicles data, comprising:

[0005] Obtain the initial Internet of Vehicles data corresponding to the target vehicle under the target sensor;

[0006] Performing curve fitting on the initial Internet of Vehicles data according to dynamic time warping combined with preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor;

[0007] Performing data feature recognition on the target fitting curve to obtain initial abnormal data corresponding to the initial Internet of Vehicles data;

[0008] Performing group detection on the initial abnormal data according to the maximum information coefficient to obtain first abnormal data and a first abnormal probability corresponding to the first abnormal data;

[0009] Performing abnormality identification on the initial abnormal data according to a convolutional neural network to obtain second abnormal data and a second abnormal probability corresponding to the second abnormal data;

[0010] According to the first abnormal probability and the second abnormal probability, the first abnormal data and the second abnormal data are merged to obtain target abnormal data corresponding to the initial Internet of Vehicles data;

[0011] A vehicle state prediction is performed on the target vehicle according to the target abnormal data to obtain a vehicle state prediction result corresponding to the target vehicle.

[0012] In a second aspect, an embodiment of the present invention provides a vehicle state monitoring system based on Internet of Vehicles data, including:

[0013] A data acquisition module is used to obtain the initial Internet of Vehicles data corresponding to the target vehicle under the target sensor;

[0014] A curve fitting module, used for performing curve fitting on the initial Internet of Vehicles data according to dynamic time bending combined with preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor;

[0015] A feature recognition module, used for performing data feature recognition on the target fitting curve to obtain initial abnormal data corresponding to the initial Internet of Vehicles data;

[0016] A first detection module, configured to perform group detection on the initial abnormal data according to a maximum information coefficient to obtain first abnormal data and a first abnormal probability corresponding to the first abnormal data;

[0017] A second detection module, configured to perform anomaly recognition on the initial abnormal data according to a convolutional neural network to obtain second abnormal data and a second abnormal probability corresponding to the second abnormal data;

[0018] an abnormal fusion module, configured to fuse the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain target abnormal data corresponding to the initial Internet of Vehicles data;

[0019] The state prediction module is used to predict the vehicle state of the target vehicle according to the target abnormal data to obtain the vehicle state prediction result corresponding to the target vehicle.

[0020] In a third aspect, an embodiment of the present invention further provides a terminal, comprising a processor, a memory, a computer program stored in the memory and executable by the processor, and a data bus for realizing connection and communication between the processor and the memory, wherein when the computer program is executed by the processor, the steps of any one of the vehicle status monitoring methods based on Internet of Vehicles data provided in the specification of the present invention are implemented.

[0021] In a fourth aspect, an embodiment of the present invention further provides a storage medium for computer-readable storage, characterized in that the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement any step of a vehicle status monitoring method based on Internet of Vehicles data as provided in the specification of the present invention.

[0022] The embodiment of the present invention provides a vehicle state monitoring method, system, terminal and medium based on Internet of Vehicles data, the method comprising: obtaining initial Internet of Vehicles data corresponding to a target vehicle under a target sensor; performing curve fitting on the initial Internet of Vehicles data according to dynamic time bending combined with preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor, by performing dynamic time bending on the initial Internet of Vehicles data combined with preset data to perform curve fitting, the accuracy and applicability of the data are ensured, unnecessary noise is eliminated, and subsequent data processing is more effective; then data feature recognition is performed on the target fitting curve to obtain initial abnormal data corresponding to the initial Internet of Vehicles data; grouping and detecting the initial abnormal data according to the maximum information coefficient to obtain first abnormal data and a first abnormal probability corresponding to the first abnormal data; according to The convolutional neural network performs abnormal identification on the initial abnormal data to obtain the second abnormal data and the second abnormal probability corresponding to the second abnormal data; thereby, the initial abnormal data is grouped and detected for abnormal identification using the maximum information coefficient and the convolutional neural network is used for abnormal identification, thereby analyzing the data from different angles, improving the accuracy and reliability of abnormal detection, thereby fusing the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain the target abnormal data corresponding to the initial Internet of Vehicles data; by fusing the first abnormal probability with the second abnormal probability, the results of the two algorithms based on the maximum information coefficient and the convolutional neural network are comprehensively considered, providing a more comprehensive data abnormality assessment, improving the credibility of the target abnormal data, and providing a strong data basis for subsequent decision support. Finally, the vehicle state is predicted for the target vehicle according to the target abnormal data, and the vehicle state prediction result corresponding to the target vehicle is obtained. Therefore, based on the accurate abnormal detection results, the vehicle state can be predicted more accurately, thereby effectively improving the processing capacity of Internet of Vehicles data, enhancing the prediction accuracy of the vehicle state, and thereby improving the operating efficiency, safety and user experience of the vehicle. This method also solves the problem of low reliability of vehicle state monitoring results in related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0024] Figure 1 A flow chart of a vehicle status monitoring method based on Internet of Vehicles data provided by an embodiment of the present invention;

[0025] Figure 2 A schematic diagram of a scenario of a vehicle status monitoring method based on Internet of Vehicles data provided by an embodiment of the present invention;

[0026] Figure 3 A schematic diagram of the module structure of a vehicle status monitoring system based on Internet of Vehicles data provided by an embodiment of the present invention;

[0027] Figure 4 A schematic block diagram of the structure of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0030] It should be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0031] The embodiment of the present invention provides a vehicle state monitoring method, system, terminal and medium based on Internet of Vehicles data. The vehicle state monitoring method based on Internet of Vehicles data can be applied to a terminal, which can be an electronic device such as a tablet computer, a laptop computer, a desktop computer, a personal digital assistant and a wearable device. The terminal can be a server or a server cluster.

[0032] Some embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0033] Please refer to Figure 1 , Figure 1 A flowchart of a vehicle status monitoring method based on Internet of Vehicles data provided by an embodiment of the present invention.

[0034] like Figure 1 As shown, the vehicle status monitoring method based on Internet of Vehicles data includes steps S101 to S107.

[0035] Step S101: Obtain initial Internet of Vehicles data corresponding to a target vehicle under a target sensor.

[0036] Exemplarily, the target sensor type required for the target vehicle is determined, such as a speed sensor, an acceleration sensor, a GPS, a temperature sensor, etc., so as to establish a communication connection with the target vehicle through an on-board communication module (such as a CAN bus, an OBD-II interface, a 5G / 4G / Wi-Fi communication module, etc.), thereby ensuring that the data transmission channel between the target vehicle and the Internet of Vehicles platform is unobstructed, supporting real-time or batch data transmission, and setting the frequency of data collection according to demand (for example, collection per second, per minute, or on-demand), and then collecting the data of the target sensor in real time through the on-board system, and storing it in the in-vehicle cache, so as to upload the data in the on-board cache to the cloud server through the Internet of Vehicles communication module, so as to obtain the initial Internet of Vehicles data corresponding to the target vehicle under the target sensor through the cloud server.

[0037] Step S102: curve fitting is performed on the initial Internet of Vehicles data according to dynamic time warping combined with preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor.

[0038] Exemplarily, preset data corresponding to the target vehicle when the vehicle state is normal is obtained, so that the timestamps corresponding to the preset data and the initial Internet of Vehicles data are converted to the same dimension, so as to calculate the distance of each point between the preset data and the initial Internet of Vehicles data through dynamic time bending to form a distance matrix, and then find the optimal alignment path through dynamic programming, and then filter the initial Internet of Vehicles data according to the alignment path to obtain the target Internet of Vehicles data, and then perform curve fitting on the target Internet of Vehicles data according to curve fitting algorithms such as polynomial fitting, spline interpolation, etc. to obtain the target fitting curve corresponding to the target vehicle under the target sensor.

[0039] In some embodiments, the curve fitting of the initial Internet of Vehicles data according to dynamic time warping in combination with preset data of the target vehicle to obtain the target fitting curve corresponding to the target vehicle under the target sensor includes: setting an initial window width, and calculating the minimum mapping relationship between the preset data and the initial Internet of Vehicles data using the dynamic time warping according to the initial window width; determining the first time delay feature corresponding to the initial Internet of Vehicles data according to the minimum mapping relationship, and judging whether the first time delay feature and the initial window width satisfy the preset relationship; when the first time delay feature and the initial window width satisfy the preset relationship, curve fitting of the initial Internet of Vehicles data according to the minimum mapping relationship to obtain the target fitting curve corresponding to the target vehicle under the target sensor; when the first time delay feature and the initial window width do not satisfy the preset relationship, adjusting the initial window width until the latest mapping relationship and the second time delay feature corresponding to the adjusted initial window width satisfy the preset relationship; curve fitting of the initial Internet of Vehicles data according to the latest mapping relationship to obtain the target fitting curve corresponding to the target vehicle under the target sensor.

[0040] Exemplarily, the initial window width is set according to expert experience or historical experience, so that the preset data and the initial Internet of Vehicles data are used as inputs of the dynamic time warping algorithm, and the initial window width is used to calculate the minimum distance between the preset data and the initial Internet of Vehicles data through the dynamic time warping algorithm, that is, the cumulative sum of the distances passed by the optimal path is minimized, thereby obtaining the minimum mapping relationship between the preset data and the initial Internet of Vehicles data under the optimal path. In the calculation process, the initial window width is applied to constrain the path search to improve the calculation efficiency and reduce unnecessary matching.

[0041] Exemplarily, curve fitting is performed according to the minimum mapping relationship to obtain an initial fitting curve, and then the horizontal and vertical movements corresponding to the initial Internet of Vehicles data are calculated according to the initial fitting curve to quantify the degree of time distortion, so as to aggregate these delay information into a first delay feature, such as average delay, delay standard deviation, etc.

[0042] Exemplarily, the absolute value of the first delay characteristic is divided by the initial window width. When the division result is between 0 and 1, it means that the first delay characteristic and the initial window width satisfy a preset relationship, that is, the preset relationship is that the division result between the absolute value of the first delay characteristic and the initial window width is between 0 and 1.

[0043] Exemplarily, if the first time delay characteristic and the initial window width satisfy a preset relationship, the minimum mapping relationship is directly used to perform curve fitting on the initial Internet of Vehicles data according to curve fitting techniques (such as polynomial regression, spline interpolation, etc.) to directly generate a target fitting curve.

[0044] For example, if the first delay feature and the initial window width do not satisfy the preset relationship, that is, the division result between the absolute value of the first delay feature and the initial window width is not between 0 and 1, the window width needs to be adjusted, which may be increased or decreased, to find a more suitable alignment path, and then repeat the dynamic time bending calculation, delay feature extraction and condition judgment until a window width is found so that the corresponding latest mapping relationship and the second delay feature satisfy the preset relationship. Thus, the initial Internet of Vehicles data is aligned and fitted using the latest mapping relationship that meets the preset conditions to generate the final target fitting curve.

[0045] In some embodiments, determining the first delay characteristic corresponding to the initial Internet of Vehicles data based on the minimum mapping relationship includes: determining the first time information corresponding to the preset data and the second time information corresponding to the initial Internet of Vehicles data based on the minimum mapping relationship; calculating the time difference between the first time information and the second time information, and counting the difference distribution corresponding to the time difference; determining the first delay characteristic corresponding to the initial Internet of Vehicles data based on the time difference and the difference distribution.

[0046] Exemplarily, the minimum mapping relationship is the best alignment path between the preset data and the initial Internet of Vehicles data obtained by the dynamic time warping algorithm, and then the first time information and the second time information corresponding to the preset data and the initial Internet of Vehicles data at the matching point are extracted from the minimum mapping relationship. Before obtaining the first time information and the second time information, the timestamps corresponding to the preset data and the initial Internet of Vehicles data need to be normalized, that is, the timestamps corresponding to the preset data are converted to start from 0, and the timestamps corresponding to the initial Internet of Vehicles data are also converted to start from 0. Thus, the first time information corresponding to the preset data and the second time information corresponding to the initial Internet of Vehicles data are obtained according to the minimum mapping relationship.

[0047] Exemplarily, for each matching point, the time difference between the first time information of the preset data and the second time information of the initial Internet of Vehicles data is calculated, and the time differences of all matching points are recorded to form a time difference list.

[0048] Exemplarily, a statistical analysis is performed on the time difference list, and then the frequency of occurrence of each time difference is counted to obtain a difference distribution corresponding to the time difference.

[0049] Exemplarily, the first delay feature corresponding to the initial Internet of Vehicles data is determined by combining the time difference and the difference distribution according to the following formula:

[0050] ;

[0051] in, represents the first delay feature, sum represents the total number of time differences, represents the i-th time difference, Represents the difference distribution corresponding to the i-th time difference.

[0052] Exemplarily, the difference distribution represents the frequency of occurrence of time differences, and then the common time differences when calculating the first delay feature according to the above formula will have a greater impact on the final delay feature. The first delay feature integrates the information of all time differences, rather than relying solely on a specific time difference, so it can more comprehensively reflect the time difference characteristics of the initial Internet of Vehicles data. If some time differences are outliers, but their difference distribution is very low, then their impact on the final delay feature will be weakened. Therefore, this feature may have a certain robustness to outliers, so that the first delay feature provides a quantitative time difference measurement standard, which can be used to provide good support for subsequent data analysis.

[0053] It should be noted that the second delay feature can also be implemented according to the above steps, which will not be repeated in this application.

[0054] Step S103: performing data feature recognition on the target fitting curve to obtain initial abnormal data corresponding to the initial Internet of Vehicles data.

[0055] Exemplarily, anomaly detection methods such as median absolute deviation method, local outlier factor, Gaussian mixture model, etc. are used to perform data feature recognition on the target fitting curve to obtain initial abnormal data corresponding to the initial Internet of Vehicles data.

[0056] For example, through the local outlier factor, the data value of each time point is first extracted from the initial Internet of Vehicles data, and then for each data point, its deviation from the target fitting curve at the same time point is calculated. The deviation can be defined as the absolute difference or square difference between the data point value and the fitting curve value, so that all absolute differences or square differences are identified as abnormal data according to the local outlier factor, and then after obtaining the abnormal absolute difference or square difference, the initial Internet of Vehicles data corresponding to the absolute difference or square difference is determined as the initial abnormal data.

[0057] Step S104: performing group detection on the initial abnormal data according to the maximum information coefficient to obtain first abnormal data and a first abnormal probability corresponding to the first abnormal data.

[0058] Exemplarily, normal data corresponding to the target vehicle is obtained by manual labeling, and then the correlation value between the normal data and the initial abnormal data is calculated according to the maximum information coefficient, and then the initial abnormal data is grouped according to the correlation value to obtain multiple groups of abnormal data, and then each group of abnormal data is identified for abnormality using a Gaussian mixture model to obtain the final abnormal data in each group of abnormal data and the abnormal probability corresponding to each abnormal data, and then the abnormal data of each group and the abnormal probability corresponding to each abnormal data are merged to obtain the first abnormal data and the first abnormal probability corresponding to the first abnormal data.

[0059] In some embodiments, the grouping detection of the initial abnormal data according to the maximum information coefficient to obtain the first abnormal data and the first abnormal probability corresponding to the first abnormal data includes: obtaining the vehicle operation type corresponding to the initial abnormal data from the target fitting curve, and obtaining the corresponding ideal Internet of Vehicles data according to the vehicle operation type; calculating the information coefficient value between the initial abnormal data and the ideal Internet of Vehicles data according to the maximum information coefficient, and obtaining the maximum information value corresponding to the information coefficient value; constructing a target kernel function according to the maximum information value, and calculating the corresponding distance information between the initial abnormal data and the ideal Internet of Vehicles data according to the target kernel function; screening the first abnormal data from the initial abnormal data according to the distance information, and determining the first abnormal probability corresponding to the first abnormal data according to the distance information.

[0060] Exemplarily, the target fitting curve is a curve obtained by combining preset data with a dynamic time bending algorithm, which can reflect the operating behavior of the target vehicle under the target sensor, so as to judge the vehicle operation type of the target vehicle, such as acceleration, deceleration, turning, parking, etc., according to the shape and trend (such as stability, fluctuation, rapid change, etc.) of the target fitting curve, and then extract the time period corresponding to the initial abnormal data from the target fitting curve, judge the curve characteristics of the time period, and thus determine the vehicle operation type.

[0061] Exemplarily, ideal Internet of Vehicles data refers to data collected when the vehicle operates in an ideal state under the same vehicle operation type, and then ideal Internet of Vehicles data with the same type of current vehicle operation is filtered out from historical data.

[0062] Exemplarily, the initial abnormal data and the ideal Internet of Vehicles data are calculated according to the maximum information coefficient to obtain an information coefficient value, and the range of the information coefficient value is [0, 1]. The larger the information coefficient value, the stronger the correlation between the two. Then, the largest one is selected from all the information coefficient values ​​as the maximum information value, so as to design a target kernel function according to the maximum information value, such as a Gaussian kernel function or other nonlinear kernel function. For example, the parameters of the target kernel function, such as variance or bandwidth, are set according to the maximum information value.

[0063] Exemplarily, the target kernel function is used to calculate the distance information between the initial abnormal data and the ideal Internet of Vehicles data, so as to set the distance information threshold according to actual needs, for example, the data point whose distance information exceeds a certain value is regarded as the first abnormal data. The first abnormal probability can be defined as the relative size of the distance information and the threshold, or the first abnormal probability is that the distance information conforms to a certain statistical distribution (such as a normal distribution), and the probability density value of the first abnormal data is calculated, and then the probability density value corresponding to the first abnormal data is determined as the first abnormal probability corresponding to the first abnormal data.

[0064] In some embodiments, constructing a target kernel function based on the maximum information value includes: determining a Gaussian kernel function and a multi-style kernel function; determining a correlation type corresponding to the initial abnormal data based on the maximum information value, and determining a first weight corresponding to the Gaussian kernel function based on the correlation type and the maximum information value; determining a second weight corresponding to the multi-style kernel function based on the first weight; and determining the target kernel function by fusing the Gaussian kernel function and the multi-style kernel function based on the first weight and the second weight.

[0065] For example, the Gaussian kernel function can capture the smooth changes between data points and is suitable for modeling nonlinear relationships. The polynomial kernel function can capture the polynomial relationship between data points and is suitable for modeling linear and low-order nonlinear relationships.

[0066] Exemplarily, the correlation type is determined according to the maximum information value. If the maximum information value is high (such as > 0.7), it indicates that there is a strong nonlinear correlation between the data, and it is suitable to use the Gaussian kernel function. If the maximum information value is low (such as < 0.5), it indicates that there may be a weak linear correlation between the data, and it is suitable to use the polynomial kernel function. If the maximum information value is moderate (such as 0.5 ≤ maximum information value ≤ 0.7), it indicates that there may be a medium-strength correlation between the data, and it is suitable to combine the Gaussian kernel function and the polynomial kernel function.

[0067] Exemplarily, to ensure that the target kernel function is applicable to different correlation types, the corresponding weight when fusing the Gaussian kernel function and the polynomial kernel function is determined according to the maximum information value, and then the maximum information value is squared, and the square root result is determined as the first weight corresponding to the Gaussian kernel function, and then the result of 1 minus the first weight is determined as the second weight, and then the second weight is determined as the fusion weight of the polynomial kernel function in the target kernel function.

[0068] Exemplarily, the Gaussian kernel function and the polynomial kernel function are linearly combined according to the first weight and the second weight to obtain the target kernel function, so that the target kernel function can simultaneously capture the characteristics of the Gaussian kernel function and the polynomial kernel function, and is suitable for processing data of different correlation types, thereby better capturing the relationship between the initial abnormal data and the ideal Internet of Vehicles data.

[0069] Step S105: performing abnormality identification on the initial abnormal data according to a convolutional neural network to obtain second abnormal data and a second abnormal probability corresponding to the second abnormal data.

[0070] Exemplarily, the convolutional neural network extracts high-order features layer by layer through convolutional layers, pooling layers and fully connected layers, and then converts historical abnormal data into a format suitable for input into the convolutional neural network, and then divides the historical abnormal data into training sets and test sets to perform abnormality classification training on the convolutional neural network to obtain the corresponding network parameters of the convolutional neural network.

[0071] Exemplarily, the preprocessed initial abnormal data is input into the convolutional neural network, and then the network parameters are used to obtain the classification results through the convolution layer, the pooling layer and the fully connected layer, so as to judge whether each initial abnormal data is abnormal according to the output probability corresponding to the classification result. For example, if the output probability is greater than a certain threshold (such as 0.5), the data point is considered to be abnormal, and then the data points determined as abnormal by the convolutional neural network are screened out from the initial abnormal data as the second abnormal data, and the second abnormal probability corresponding to the second abnormal data is determined according to the output probability.

[0072] Step S106: fusing the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain target abnormal data corresponding to the initial Internet of Vehicles data.

[0073] Exemplarily, the first abnormal data and the second abnormal data are merged to obtain the corresponding overlapping data, and then the first probability corresponding to the overlapping data is obtained from the first abnormal probability and the second probability corresponding to the overlapping data is obtained from the second abnormal probability, and then the first probability and the second probability are weighted averaged to obtain the target probability corresponding to the overlapping data, and then the target probability is compared with the first preset threshold to obtain the first target data from the overlapping data, and then the first remaining data after excluding the overlapping data from the first abnormal data is obtained, and then the third probability corresponding to the first remaining data is obtained from the first abnormal probability, and then the third probability is compared with the second preset threshold to obtain the corresponding second target data in the first remaining data, and then the second remaining data after excluding the overlapping data from the second abnormal data is obtained, and then the fourth probability corresponding to the second remaining data is obtained from the second abnormal probability, and then the fourth probability is compared with the second preset threshold to obtain the corresponding third target data in the second remaining data, and finally the first target data, the second target data and the third target data are merged to obtain the target abnormal data corresponding to the initial Internet of Vehicles data.

[0074] In some embodiments, the fusion of the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain the target abnormal data corresponding to the initial Internet of Vehicles data includes: obtaining a first abnormal time corresponding to the first abnormal data and a second abnormal time corresponding to the second abnormal data from the initial Internet of Vehicles data; merging the first abnormal data and the second abnormal data according to the first abnormal time and the second abnormal time to obtain third abnormal data; removing the third abnormal data from the first abnormal data to obtain fourth abnormal data and removing the third abnormal data from the second abnormal data to obtain fifth abnormal data; obtaining a third abnormal probability corresponding to the third abnormal data from the first abnormal probability and a third abnormal time corresponding to the third abnormal data from the second abnormal data; obtaining a third abnormal probability corresponding to the third abnormal data from the first abnormal probability and a third abnormal time corresponding to the third abnormal data from the second abnormal time. The fourth abnormal probability corresponding to the third abnormal data is obtained from the second abnormal probability; the fifth abnormal probability corresponding to the fourth abnormal data is obtained from the first abnormal probability and the sixth abnormal probability corresponding to the fifth abnormal data is obtained from the second abnormal probability; the third abnormal probability and the fourth abnormal probability are fused using evidence theory to obtain the seventh abnormal probability corresponding to the third abnormal data; the fourth abnormal data is screened according to the fifth abnormal probability to obtain the first data, and the fifth abnormal data is screened according to the sixth abnormal probability to obtain the second data; the third abnormal data is screened according to the seventh abnormal probability to obtain the third data; and the first data, the second data and the third data are fused to determine the target abnormal data corresponding to the initial Internet of Vehicles data.

[0075] Exemplarily, the first abnormal data is extracted from the initial Internet of Vehicles data, and the first abnormal time corresponding to each abnormal data point is extracted. And the second abnormal data is extracted from the initial Internet of Vehicles data, and the second abnormal time corresponding to each abnormal data point is extracted.

[0076] Exemplarily, based on the first abnormal time and the second abnormal time, abnormal data points with the same or similar timestamps are found, and then these abnormal data points with the same or similar timestamps are merged into third abnormal data, and then the third abnormal data is eliminated from the first abnormal data to obtain fourth abnormal data, and the third abnormal data is eliminated from the second abnormal data to obtain fifth abnormal data.

[0077] Exemplarily, a third abnormal probability corresponding to the third abnormal data is extracted from the first abnormal probability, and a fourth abnormal probability corresponding to the third abnormal data is extracted from the second abnormal probability. A fifth abnormal probability corresponding to the fourth abnormal data is extracted from the first abnormal probability, and a sixth abnormal probability corresponding to the fifth abnormal data is extracted from the second abnormal probability.

[0078] Exemplarily, the third abnormal probability and the fourth abnormal probability are fused using evidence theory to obtain the seventh abnormal probability corresponding to the fused third abnormal data, so that the fourth abnormal data is screened according to the fifth abnormal probability and the third preset threshold to obtain the first data. The fifth abnormal data is screened according to the sixth abnormal probability and the third preset threshold to obtain the second data. And the third abnormal data is screened according to the seventh abnormal probability and the fourth preset threshold to obtain the third data.

[0079] Exemplarily, the first data, the second data and the third data are combined to obtain target abnormal data corresponding to the initial Internet of Vehicles data.

[0080] Step S107: predicting the vehicle state of the target vehicle according to the target abnormal data to obtain a vehicle state prediction result corresponding to the target vehicle.

[0081] Exemplarily, prediction models such as decision trees, random forests, support vector machines (SVM), etc. are used to take historical abnormal data and the actual vehicle states corresponding to the historical abnormal data (such as normal, slightly abnormal, seriously abnormal, etc.) as training sets, and then use the training sets to train the selected prediction model to obtain the target prediction model.

[0082] Exemplarily, a vehicle state prediction is performed on the target abnormal data according to the target prediction model to obtain a vehicle state prediction result corresponding to the target vehicle, and the vehicle state prediction result is one of normal, slightly abnormal, and seriously abnormal.

[0083] In some embodiments, predicting the vehicle state of the target vehicle based on the target abnormal data to obtain a vehicle state prediction result corresponding to the target vehicle includes: using the self-encoding layer of the state classification model to encode and decode the target abnormal data to obtain a first encoding result and a first decoding result corresponding to the target abnormal data; using the feature preservation layer of the state classification model to adjust the first encoding result and the first decoding result according to the data consistency constraint within the same sensor and the data consistency constraint between different sensors to obtain a second encoding result; using the feature fusion layer of the state classification model to weightedly fuse the second encoding result to obtain a target encoding result; using the state classification layer of the state classification model to predict the vehicle state according to the target encoding result to obtain the vehicle state prediction result corresponding to the target vehicle.

[0084] For example, historical abnormal data and the real vehicle status corresponding to the historical abnormal data (such as normal, mild abnormality, severe abnormality, etc.) are used as training sets, and then the autoencoder layer of the state classification model is designed to extract the key features of the data and reconstruct the data. The feature retention layer of the state classification model is designed to ensure the data consistency within the same sensor and between different sensors. The feature fusion layer of the state classification model is designed, and the weighted fusion mechanism is determined to fuse multiple features. The state classification layer of the state classification model is designed, such as a fully connected layer + softmax, to output the category of the vehicle state, so as to train the model on the training set, and use the validation set to monitor the model performance, so as to obtain an efficient state classification model for real-time prediction of vehicle status.

[0085] Exemplarily, the self-encoding layer of the state classification model can map the target abnormal data from the high-dimensional space to the low-dimensional space, and then map it back from the low-dimensional space to the high-dimensional space, and then input the target abnormal data into the encoder part of the self-encoding layer. The encoder compresses the target abnormal data into a low-dimensional representation to obtain a first encoding result, and then inputs the first encoding result into the decoder part of the self-encoding layer. The decoder reconstructs the low-dimensional representation into high-dimensional data to obtain a first decoding result.

[0086] Exemplarily, the feature preservation layer is used to ensure the rationality of the encoding result and the decoding result in terms of data consistency. The encoding result and the decoding result are adjusted by introducing data consistency constraints within the same sensor and data consistency constraints between different sensors. Ensure that the data of the same sensor remains consistent during the encoding and decoding process. For example, the data of the same sensor should be continuous in time. The data consistency constraints between different sensors ensure that the data of different sensors remain consistent during the encoding and decoding process. For example, the data of different sensors should be correlated in space. Thus, according to the above constraints, the first encoding result is adjusted to obtain the second encoding result.

[0087] Exemplarily, the feature fusion layer is used to weightedly fuse multiple features in the second encoding result to generate a target encoding result. Through weighted fusion, important features can be highlighted and irrelevant features can be suppressed. Finally, the state classification layer is used to map the target encoding result to a specific vehicle state category, so that the vehicle state category is determined as the vehicle state prediction result.

[0088] See also Figure 2 , Figure 2 A schematic diagram of a scenario for realizing a vehicle status monitoring method based on vehicle network data provided in an embodiment of the present application, before implementing the monitoring method, the vehicle-side vehicle computer 11 monitors in real time whether the one-key data upload operation button of the vehicle computer is pressed by the user, and if it is pressed, the vehicle-side vehicle computer 11 reads the real-time vehicle condition data (the buried point data of each ECU of the vehicle, such as: vehicle speed, battery voltage, current, motor torque and other data), thereby packaging the above data and sending it to the vehicle networking module 12, and then the vehicle networking module 12 receives the data request for packaging and uploading by the vehicle-side vehicle computer 11, and then transparently transmits the packaged data of the vehicle-side vehicle computer 11 to the vehicle networking big data platform 13, and the vehicle networking big data platform 13 unpacks, cleans, and parses the uploaded data, and sends the processed data to the terminal 300, that is, the initial vehicle networking data corresponding to the target sensor is sent to the terminal Terminal 300, and then terminal 300 performs curve fitting on the initial Internet of Vehicles data according to dynamic time bending combined with preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor; performs data feature recognition on the target fitting curve to obtain initial abnormal data corresponding to the initial Internet of Vehicles data; performs group detection on the initial abnormal data according to the maximum information coefficient to obtain first abnormal data and a first abnormal probability corresponding to the first abnormal data; performs abnormal recognition on the initial abnormal data according to the convolutional neural network to obtain second abnormal data and a second abnormal probability corresponding to the second abnormal data; fuses the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain target abnormal data corresponding to the initial Internet of Vehicles data; performs vehicle state prediction on the target vehicle according to the target abnormal data to obtain a vehicle state prediction result corresponding to the target vehicle. Therefore, the terminal 300 sends the vehicle status prediction result to the Internet of Vehicles big data platform 13, and the Internet of Vehicles big data platform 13 pushes the vehicle status prediction result to the Internet of Vehicles module 12, and the Internet of Vehicles module 12 sends the vehicle status prediction result to the vehicle-side computer 11, and after the vehicle-side computer 11 receives the vehicle status prediction result, the vehicle diagnosis result is presented on the large screen corresponding to the vehicle-side computer 11.

[0089] See also Figure 3 , Figure 3A vehicle state monitoring system 200 based on Internet of Vehicles data is provided in an embodiment of the present application. The vehicle state monitoring system 200 based on Internet of Vehicles data includes a data acquisition module 201, a curve fitting module 202, a feature recognition module 203, a first detection module 204, a second detection module 205, an abnormal fusion module 206, and a state prediction module 207, wherein the data acquisition module 201 is used to obtain the initial Internet of Vehicles data corresponding to the target vehicle under the target sensor; the curve fitting module 202 is used to perform curve fitting on the initial Internet of Vehicles data according to dynamic time bending combined with preset data of the target vehicle to obtain the target fitting curve corresponding to the target vehicle under the target sensor; the feature recognition module 203 is used to perform data feature recognition on the target fitting curve to obtain the target fitting curve. Initial abnormal data corresponding to the initial Internet of Vehicles data; a first detection module 204, used to perform group detection on the initial abnormal data according to the maximum information coefficient to obtain first abnormal data and a first abnormal probability corresponding to the first abnormal data; a second detection module 205, used to perform abnormal identification on the initial abnormal data according to a convolutional neural network to obtain second abnormal data and a second abnormal probability corresponding to the second abnormal data; an abnormal fusion module 206, used to fuse the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain target abnormal data corresponding to the initial Internet of Vehicles data; a state prediction module 207, used to perform vehicle state prediction on the target vehicle according to the target abnormal data to obtain a vehicle state prediction result corresponding to the target vehicle.

[0090] In some implementations, the curve fitting module 202 performs, during the process of performing curve fitting on the initial Internet of Vehicles data according to dynamic time warping combined with preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor:

[0091] Setting an initial window width, and calculating a minimum mapping relationship between the preset data and the initial Internet of Vehicles data using the dynamic time warping according to the initial window width;

[0092] Determining a first delay feature corresponding to the initial Internet of Vehicles data according to the minimum mapping relationship, and judging whether the first delay feature and the initial window width satisfy a preset relationship;

[0093] When the first time delay feature and the initial window width satisfy the preset relationship, curve fitting is performed on the initial Internet of Vehicles data according to the minimum mapping relationship to obtain the target fitting curve corresponding to the target vehicle under the target sensor;

[0094] When the first delay characteristic and the initial window width do not satisfy the preset relationship, the initial window width is adjusted until the latest mapping relationship corresponding to the adjusted initial window width and the second delay characteristic satisfy the preset relationship;

[0095] Curve fitting is performed on the initial Internet of Vehicles data according to the latest mapping relationship to obtain the target fitting curve corresponding to the target vehicle under the target sensor.

[0096] In some implementations, during the process of determining the first delay feature corresponding to the initial Internet of Vehicles data according to the minimum mapping relationship, the curve fitting module 202 executes:

[0097] Determine, according to the minimum mapping relationship, the first time information corresponding to the preset data and the second time information corresponding to the initial Internet of Vehicles data;

[0098] Calculating a time difference between the first time information and the second time information, and collecting statistics on a difference distribution corresponding to the time difference;

[0099] The first time delay characteristic corresponding to the initial Internet of Vehicles data is determined according to the time difference and the difference distribution.

[0100] In some implementations, the first detection module 204, in the process of performing group detection on the initial abnormal data according to the maximum information coefficient to obtain the first abnormal data and the first abnormal probability corresponding to the first abnormal data, performs:

[0101] Obtaining the vehicle operation type corresponding to the initial abnormal data from the target fitting curve, and obtaining corresponding ideal Internet of Vehicles data according to the vehicle operation type;

[0102] Calculating an information coefficient value between the initial abnormal data and the ideal Internet of Vehicles data according to the maximum information coefficient, and obtaining a maximum information value corresponding to the information coefficient value;

[0103] Constructing a target kernel function according to the maximum information value, and calculating the corresponding distance information between the initial abnormal data and the ideal Internet of Vehicles data according to the target kernel function;

[0104] The first abnormal data is obtained by screening the initial abnormal data according to the distance information, and the first abnormal probability corresponding to the first abnormal data is determined according to the distance information.

[0105] In some implementations, the first detection module 204, during the process of constructing the target kernel function according to the maximum information value, performs:

[0106] Determine Gaussian kernel function and polyhedral kernel function;

[0107] Determining a correlation type corresponding to the initial abnormal data according to the maximum information value, and determining a first weight corresponding to the Gaussian kernel function according to the correlation type and the maximum information value;

[0108] Determine a second weight corresponding to the diverse kernel function according to the first weight;

[0109] The target kernel function is determined by fusing the Gaussian kernel function and the diverse kernel function according to the first weight and the second weight.

[0110] In some implementations, the abnormal fusion module 206 performs, during the process of fusing the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain the target abnormal data corresponding to the initial Internet of Vehicles data:

[0111] Obtaining, from the initial Internet of Vehicles data, a first abnormal time corresponding to the first abnormal data and a second abnormal time corresponding to the second abnormal data;

[0112] Merging the first abnormal data and the second abnormal data according to the first abnormal time and the second abnormal time to obtain third abnormal data;

[0113] Eliminating the third abnormal data from the first abnormal data to obtain fourth abnormal data and eliminating the third abnormal data from the second abnormal data to obtain fifth abnormal data;

[0114] Obtain a third abnormal probability corresponding to the third abnormal data from the first abnormal probability and obtain a fourth abnormal probability corresponding to the third abnormal data from the second abnormal probability;

[0115] Obtain a fifth abnormal probability corresponding to the fourth abnormal data from the first abnormal probability and obtain a sixth abnormal probability corresponding to the fifth abnormal data from the second abnormal probability;

[0116] Using evidence theory to fuse the third abnormal probability and the fourth abnormal probability to obtain a seventh abnormal probability corresponding to the third abnormal data;

[0117] Screening the fourth abnormal data according to the fifth abnormal probability to obtain first data, and screening the fifth abnormal data according to the sixth abnormal probability to obtain second data;

[0118] Screening the third abnormal data according to the seventh abnormal probability to obtain third data;

[0119] The first data, the second data and the third data are integrated to determine the target abnormal data corresponding to the initial Internet of Vehicles data.

[0120] In some implementations, the state prediction module 207 performs, during the process of predicting the vehicle state of the target vehicle according to the target abnormal data and obtaining the vehicle state prediction result corresponding to the target vehicle:

[0121] Using the self-encoding layer of the state classification model to encode and decode the target abnormal data to obtain a first encoding result and a first decoding result corresponding to the target abnormal data;

[0122] Using the feature preservation layer of the state classification model, adjusting the first encoding result and the first decoding result according to the data consistency constraint within the same sensor and the data consistency constraint between different sensors to obtain a second encoding result;

[0123] Using the feature fusion layer of the state classification model to perform weighted fusion on the second encoding result to obtain a target encoding result;

[0124] The state classification layer of the state classification model is used to predict the vehicle state according to the target encoding result to obtain the vehicle state prediction result corresponding to the target vehicle.

[0125] In some embodiments, the vehicle status monitoring system 200 based on Internet of Vehicles data can be applied to a terminal.

[0126] It should be noted that technical personnel in the relevant field can clearly understand that, for the convenience and conciseness of description, the specific working process of the vehicle status monitoring system 200 based on Internet of Vehicles data described above can refer to the corresponding process in the aforementioned vehicle status monitoring method based on Internet of Vehicles data embodiment, and will not be repeated here.

[0127] See also Figure 4 , Figure 4 A schematic block diagram of the structure of a terminal provided by an embodiment of the present invention.

[0128] like Figure 4 As shown, the terminal 300 includes a processor 301 and a memory 302 , and the processor 301 and the memory 302 are connected via a bus 303 , such as an I2C (Inter-integrated Circuit) bus.

[0129] Specifically, the processor 301 is used to provide computing and control capabilities to support the operation of the entire terminal. The processor 301 can be a central processing unit (CPU), and the processor 301 can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0130] Specifically, the memory 302 may be a Flash chip, a read-only memory (ROM) disk, an optical disk, a USB flash drive, or a mobile hard disk.

[0131] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a partial structure related to the embodiment of the present invention, and does not constitute a limitation on the terminal to which the embodiment of the present invention is applied. The specific server may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0132] The processor is used to run a computer program stored in a memory, and implement any one of the vehicle status monitoring methods based on Internet of Vehicles data provided by an embodiment of the present invention when executing the computer program.

[0133] In one embodiment, the processor is used to run a computer program stored in the memory, and implements the following steps when executing the computer program:

[0134] Obtain the initial Internet of Vehicles data corresponding to the target vehicle under the target sensor;

[0135] Performing curve fitting on the initial Internet of Vehicles data according to dynamic time warping combined with preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor;

[0136] Performing data feature recognition on the target fitting curve to obtain initial abnormal data corresponding to the initial Internet of Vehicles data;

[0137] Performing group detection on the initial abnormal data according to the maximum information coefficient to obtain first abnormal data and a first abnormal probability corresponding to the first abnormal data;

[0138] Performing abnormality identification on the initial abnormal data according to a convolutional neural network to obtain second abnormal data and a second abnormal probability corresponding to the second abnormal data;

[0139] According to the first abnormal probability and the second abnormal probability, the first abnormal data and the second abnormal data are merged to obtain target abnormal data corresponding to the initial Internet of Vehicles data;

[0140] A vehicle state prediction is performed on the target vehicle according to the target abnormal data to obtain a vehicle state prediction result corresponding to the target vehicle.

[0141] In some implementations, the processor 301 performs, during the process of performing curve fitting on the initial Internet of Vehicles data according to dynamic time warping combined with preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor:

[0142] Setting an initial window width, and calculating a minimum mapping relationship between the preset data and the initial Internet of Vehicles data using the dynamic time warping according to the initial window width;

[0143] Determining a first delay feature corresponding to the initial Internet of Vehicles data according to the minimum mapping relationship, and judging whether the first delay feature and the initial window width satisfy a preset relationship;

[0144] When the first time delay feature and the initial window width satisfy the preset relationship, curve fitting is performed on the initial Internet of Vehicles data according to the minimum mapping relationship to obtain the target fitting curve corresponding to the target vehicle under the target sensor;

[0145] When the first delay characteristic and the initial window width do not satisfy the preset relationship, the initial window width is adjusted until the latest mapping relationship corresponding to the adjusted initial window width and the second delay characteristic satisfy the preset relationship;

[0146] Curve fitting is performed on the initial Internet of Vehicles data according to the latest mapping relationship to obtain the target fitting curve corresponding to the target vehicle under the target sensor.

[0147] In some implementations, during the process of determining the first delay feature corresponding to the initial Internet of Vehicles data according to the minimum mapping relationship, the processor 301 executes:

[0148] Determine, according to the minimum mapping relationship, the first time information corresponding to the preset data and the second time information corresponding to the initial Internet of Vehicles data;

[0149] Calculating a time difference between the first time information and the second time information, and collecting statistics on a difference distribution corresponding to the time difference;

[0150] The first time delay characteristic corresponding to the initial Internet of Vehicles data is determined according to the time difference and the difference distribution.

[0151] In some implementations, the processor 301, in the process of performing group detection on the initial abnormal data according to the maximum information coefficient to obtain the first abnormal data and the first abnormal probability corresponding to the first abnormal data, executes:

[0152] Obtaining the vehicle operation type corresponding to the initial abnormal data from the target fitting curve, and obtaining corresponding ideal Internet of Vehicles data according to the vehicle operation type;

[0153] Calculating an information coefficient value between the initial abnormal data and the ideal Internet of Vehicles data according to the maximum information coefficient, and obtaining a maximum information value corresponding to the information coefficient value;

[0154] Constructing a target kernel function according to the maximum information value, and calculating the corresponding distance information between the initial abnormal data and the ideal Internet of Vehicles data according to the target kernel function;

[0155] The first abnormal data is obtained by screening the initial abnormal data according to the distance information, and the first abnormal probability corresponding to the first abnormal data is determined according to the distance information.

[0156] In some implementations, the processor 301, during the process of constructing the target kernel function according to the maximum information value, executes:

[0157] Determine Gaussian kernel function and polyhedral kernel function;

[0158] Determining a correlation type corresponding to the initial abnormal data according to the maximum information value, and determining a first weight corresponding to the Gaussian kernel function according to the correlation type and the maximum information value;

[0159] Determine a second weight corresponding to the diverse kernel function according to the first weight;

[0160] The target kernel function is determined by fusing the Gaussian kernel function and the diverse kernel function according to the first weight and the second weight.

[0161] In some implementations, during the process of fusing the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain the target abnormal data corresponding to the initial Internet of Vehicles data, the processor 301 executes:

[0162] Obtaining, from the initial Internet of Vehicles data, a first abnormal time corresponding to the first abnormal data and a second abnormal time corresponding to the second abnormal data;

[0163] Merging the first abnormal data and the second abnormal data according to the first abnormal time and the second abnormal time to obtain third abnormal data;

[0164] Eliminating the third abnormal data from the first abnormal data to obtain fourth abnormal data and eliminating the third abnormal data from the second abnormal data to obtain fifth abnormal data;

[0165] Obtain a third abnormal probability corresponding to the third abnormal data from the first abnormal probability and obtain a fourth abnormal probability corresponding to the third abnormal data from the second abnormal probability;

[0166] Obtain a fifth abnormal probability corresponding to the fourth abnormal data from the first abnormal probability and obtain a sixth abnormal probability corresponding to the fifth abnormal data from the second abnormal probability;

[0167] Using evidence theory to fuse the third abnormal probability and the fourth abnormal probability to obtain a seventh abnormal probability corresponding to the third abnormal data;

[0168] Screening the fourth abnormal data according to the fifth abnormal probability to obtain first data, and screening the fifth abnormal data according to the sixth abnormal probability to obtain second data;

[0169] Screening the third abnormal data according to the seventh abnormal probability to obtain third data;

[0170] The first data, the second data and the third data are integrated to determine the target abnormal data corresponding to the initial Internet of Vehicles data.

[0171] In some implementations, the processor 301 performs, during the process of predicting the vehicle state of the target vehicle according to the target abnormal data and obtaining the vehicle state prediction result corresponding to the target vehicle:

[0172] Using the self-encoding layer of the state classification model to encode and decode the target abnormal data to obtain a first encoding result and a first decoding result corresponding to the target abnormal data;

[0173] Using the feature preservation layer of the state classification model, adjusting the first encoding result and the first decoding result according to the data consistency constraint within the same sensor and the data consistency constraint between different sensors to obtain a second encoding result;

[0174] Using the feature fusion layer of the state classification model to perform weighted fusion on the second encoding result to obtain a target encoding result;

[0175] The state classification layer of the state classification model is used to predict the vehicle state according to the target encoding result to obtain the vehicle state prediction result corresponding to the target vehicle.

[0176] It should be noted that technicians in the relevant field can clearly understand that for the convenience and simplicity of description, the specific working process of the terminal described above can refer to the corresponding process in the aforementioned vehicle status monitoring method based on Internet of Vehicles data, and will not be repeated here.

[0177] An embodiment of the present invention also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement any step of a vehicle status monitoring method based on Internet of Vehicles data as provided in the description of the embodiment of the present invention.

[0178] The storage medium may be an internal storage unit of the terminal described in the foregoing embodiment, such as a hard disk or memory of the terminal. The storage medium may also be an external storage device of the terminal, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the terminal.

[0179] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware embodiment, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0180] It should be understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations. It should be noted that, in this article, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "including a..." does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0181] The serial numbers of the embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. The above description is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A vehicle status monitoring method based on Internet of Vehicles data, characterized in that: The method comprises: Obtain the initial Internet of Vehicles data corresponding to the target vehicle under the target sensor; Obtaining preset data corresponding to the target vehicle when the vehicle state is normal, performing curve fitting on the initial Internet of Vehicles data according to dynamic time warping combined with the preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor; Performing data feature recognition on the target fitting curve to obtain initial abnormal data corresponding to the initial Internet of Vehicles data; Performing group detection on the initial abnormal data according to the maximum information coefficient to obtain first abnormal data and a first abnormal probability corresponding to the first abnormal data; Performing abnormality identification on the initial abnormal data according to a convolutional neural network to obtain second abnormal data and a second abnormal probability corresponding to the second abnormal data; According to the first abnormal probability and the second abnormal probability, the first abnormal data and the second abnormal data are merged to obtain target abnormal data corresponding to the initial Internet of Vehicles data; Performing vehicle state prediction on the target vehicle according to the target abnormal data to obtain a vehicle state prediction result corresponding to the target vehicle; The step of performing group detection on the initial abnormal data according to the maximum information coefficient to obtain the first abnormal data and the first abnormal probability corresponding to the first abnormal data includes: Obtaining the vehicle operation type corresponding to the initial abnormal data from the target fitting curve, and obtaining corresponding ideal Internet of Vehicles data according to the vehicle operation type; Calculating an information coefficient value between the initial abnormal data and the ideal Internet of Vehicles data according to the maximum information coefficient, and obtaining a maximum information value corresponding to the information coefficient value; Constructing a target kernel function according to the maximum information value, and calculating the distance information corresponding to the initial abnormal data and the ideal Internet of Vehicles data according to the target kernel function; The first abnormal data is obtained by screening the initial abnormal data according to the distance information, and the first abnormal probability corresponding to the first abnormal data is determined according to the distance information.

2. The method according to claim 1, characterized in that: The performing curve fitting on the initial Internet of Vehicles data according to dynamic time warping combined with the preset data of the target vehicle to obtain a target fitting curve corresponding to the target vehicle under the target sensor includes: Setting an initial window width, and calculating a minimum mapping relationship between the preset data and the initial Internet of Vehicles data using the dynamic time warping according to the initial window width; Determining a first delay feature corresponding to the initial Internet of Vehicles data according to the minimum mapping relationship, and judging whether the first delay feature and the initial window width satisfy a preset relationship; When the first time delay feature and the initial window width satisfy the preset relationship, curve fitting is performed on the initial Internet of Vehicles data according to the minimum mapping relationship to obtain the target fitting curve corresponding to the target vehicle under the target sensor; When the first delay characteristic and the initial window width do not satisfy the preset relationship, the initial window width is adjusted until the latest mapping relationship corresponding to the adjusted initial window width and the second delay characteristic satisfy the preset relationship; Curve fitting is performed on the initial Internet of Vehicles data according to the latest mapping relationship to obtain the target fitting curve corresponding to the target vehicle under the target sensor.

3. The method according to claim 2, characterized in that The determining, according to the minimum mapping relationship, a first delay characteristic corresponding to the initial Internet of Vehicles data includes: Determine, according to the minimum mapping relationship, the first time information corresponding to the preset data and the second time information corresponding to the initial Internet of Vehicles data; Calculating a time difference between the first time information and the second time information, and collecting statistics on a difference distribution corresponding to the time difference; The first time delay characteristic corresponding to the initial Internet of Vehicles data is determined according to the time difference and the difference distribution.

4. The method according to claim 1, characterized in that: The constructing a target kernel function according to the maximum information value comprises: Determine Gaussian kernel function and polyhedral kernel function; Determining a correlation type corresponding to the initial abnormal data according to the maximum information value, and determining a first weight corresponding to the Gaussian kernel function according to the correlation type and the maximum information value; Determine a second weight corresponding to the diverse kernel function according to the first weight; The target kernel function is determined by fusing the Gaussian kernel function and the diverse kernel function according to the first weight and the second weight.

5. The method according to claim 1, characterized in that The step of fusing the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain target abnormal data corresponding to the initial Internet of Vehicles data includes: Obtaining, from the initial Internet of Vehicles data, a first abnormal time corresponding to the first abnormal data and a second abnormal time corresponding to the second abnormal data; Merging the first abnormal data and the second abnormal data according to the first abnormal time and the second abnormal time to obtain third abnormal data; Eliminating the third abnormal data from the first abnormal data to obtain fourth abnormal data and eliminating the third abnormal data from the second abnormal data to obtain fifth abnormal data; Obtain a third abnormal probability corresponding to the third abnormal data from the first abnormal probability and obtain a fourth abnormal probability corresponding to the third abnormal data from the second abnormal probability; Obtain a fifth abnormal probability corresponding to the fourth abnormal data from the first abnormal probability and obtain a sixth abnormal probability corresponding to the fifth abnormal data from the second abnormal probability; Using evidence theory to fuse the third abnormal probability and the fourth abnormal probability to obtain a seventh abnormal probability corresponding to the third abnormal data; Screening the fourth abnormal data according to the fifth abnormal probability to obtain first data, and screening the fifth abnormal data according to the sixth abnormal probability to obtain second data; Screening the third abnormal data according to the seventh abnormal probability to obtain third data; The first data, the second data and the third data are integrated to determine the target abnormal data corresponding to the initial Internet of Vehicles data.

6. The method according to claim 1, characterized in that The performing vehicle state prediction on the target vehicle according to the target abnormal data to obtain a vehicle state prediction result corresponding to the target vehicle includes: Using the self-encoding layer of the state classification model to encode and decode the target abnormal data to obtain a first encoding result and a first decoding result corresponding to the target abnormal data; Using the feature preservation layer of the state classification model, adjusting the first encoding result and the first decoding result according to the data consistency constraint within the same sensor and the data consistency constraint between different sensors to obtain a second encoding result; Using the feature fusion layer of the state classification model to perform weighted fusion on the second encoding result to obtain a target encoding result; The state classification layer of the state classification model is used to predict the vehicle state according to the target encoding result to obtain the vehicle state prediction result corresponding to the target vehicle.

7. A vehicle status monitoring system based on Internet of Vehicles data, characterized in that: include: A data acquisition module is used to obtain the initial Internet of Vehicles data corresponding to the target vehicle under the target sensor; A curve fitting module, used to obtain preset data corresponding to the target vehicle when the vehicle state is normal, and to perform curve fitting on the initial Internet of Vehicles data in combination with the preset data of the target vehicle according to dynamic time bending to obtain a target fitting curve corresponding to the target vehicle under the target sensor; A feature recognition module, used for performing data feature recognition on the target fitting curve to obtain initial abnormal data corresponding to the initial Internet of Vehicles data; A first detection module, configured to perform group detection on the initial abnormal data according to the maximum information coefficient to obtain the first abnormal data and the first abnormal probability corresponding to the first abnormal data; wherein, the group detection on the initial abnormal data according to the maximum information coefficient to obtain the first abnormal data and the first abnormal probability corresponding to the first abnormal data comprises: obtaining the vehicle operation type corresponding to the initial abnormal data from the target fitting curve, and obtaining the corresponding ideal Internet of Vehicles data according to the vehicle operation type; calculating the information coefficient value between the initial abnormal data and the ideal Internet of Vehicles data according to the maximum information coefficient, and obtaining the maximum information value corresponding to the information coefficient value; constructing a target kernel function according to the maximum information value, and calculating the corresponding distance information between the initial abnormal data and the ideal Internet of Vehicles data according to the target kernel function; screening the first abnormal data from the initial abnormal data according to the distance information, and determining the first abnormal probability corresponding to the first abnormal data according to the distance information; A second detection module, configured to perform anomaly recognition on the initial abnormal data according to a convolutional neural network to obtain second abnormal data and a second abnormal probability corresponding to the second abnormal data; an abnormal fusion module, configured to fuse the first abnormal data and the second abnormal data according to the first abnormal probability and the second abnormal probability to obtain target abnormal data corresponding to the initial Internet of Vehicles data; The state prediction module is used to predict the vehicle state of the target vehicle according to the target abnormal data to obtain a vehicle state prediction result corresponding to the target vehicle.

8. A terminal, characterized in that: The terminal includes a processor and a memory; The memory is used to store computer programs; The processor is used to execute the computer program and implement the vehicle status monitoring method based on Internet of Vehicles data as described in any one of claims 1 to 6 when executing the computer program.

9. A computer storage medium for computer storage, characterized in that: The computer storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the vehicle status monitoring method based on vehicle network data as described in any one of claims 1 to 6.

Images