Vehicle remote intelligent diagnosis method, system, storage medium and device

Abstract

The invention discloses a remote intelligent diagnosis method for a vehicle. The method comprises the following steps: obtaining whole vehicle state data and vehicle end diagnosis data of the vehicle; screening and extracting description signals from the whole vehicle state data to form a description signal matrix, and analyzing the description signal matrix to obtain a preliminary result; forming a plurality of phenomenon description entries associated with the description information according to the preliminary result, determining the fault category of the diagnosis, and determining the range of a diagnostic analysis case analysis model corresponding to the fault category in a case library; performing matching analysis on the feature data of each case analysis model and the vehicle end diagnosis data of the diagnosis, and decomposing the diagnosis into at least one independent fault event; and performing classification analysis according to each fault reason associated with the case analysis model corresponding to each independent fault event, and determining the fault reason of the diagnosis. The invention further provides a corresponding system, a storage medium and a device. According to the invention, the vehicle maintenance difficulty of after-sales personnel can be reduced, and the accuracy of complex fault analysis can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle fault diagnosis, and in particular to a vehicle remote intelligent diagnosis method, system, storage medium and device. Background Art

[0002] In the past, many automotive functions were realized by a single control unit. When a fault occurred in an automotive function, the control unit corresponding to the faulty function could be diagnosed through a diagnostic instrument, and then the fault code could be obtained to know the general fault. However, with the development of current automotive intelligent networking technology, the integration of vehicle and cloud, and the integration of control units are increasingly becoming new development trends. More and more automotive functions tend to be realized by multiple nodes and multiple systems. Therefore, the past method of diagnosing single-node functional faults is becoming increasingly unsuitable for the new development trend.

[0003] Meanwhile, with the development of remote diagnosis technology, 4S stores and maintenance personnel can remotely diagnose the vehicles of market customers through network communication instead of on-site troubleshooting and analysis. However, a large number of fault codes and vehicle condition data are transmitted to the cloud. Without seeing the actual vehicle, it is not easy for maintenance personnel to remotely rely on cloud data to determine what faults have occurred in the vehicle. Therefore, many maintenance personnel of vehicle manufacturers can only rely on the single-node fault codes read by remote diagnosis to analyze one by one and guess the possible faults of the vehicle, which is inefficient and does not meet the development trend of functional diagnosis.

[0004] In existing technical solutions, the following solutions have also emerged: A fault tree of the vehicle's entire vehicle electronics and electrical appliances is established in the cloud, a fault knowledge graph is built, and then relevant fault recognition rule models and fault recognition algorithm models are established based on the fault knowledge graph. The vehicle condition fault data obtained in the cloud is imported into the models for analysis to obtain possible causes.

[0005] However, existing vehicle diagnosis solutions all have some deficiencies:

[0006] First of all, for complex functions that span multiple domains and multiple systems in a vehicle, sometimes the function trigger or execution conditions involve multi-domain and multi-system trigger signals, so there are quite a lot of logical links involved. It is very difficult for maintenance personnel to reverse-analyze which systems in which links are faulty from the functional fault phenomenon, which poses a rather high requirement for the ability of maintenance personnel.

[0007] Secondly, the existing cloud-based diagnostic solutions are only applicable to the failure analysis of simple and forward-designed systems. For fault cases where the fault mechanism is not clear but the fault causes are already clear, technicians are unable to establish a complete knowledge graph, and the system needs to allow technicians to establish an analysis model based on diagnostic experience and case accumulation. In fact, for most market fault cases, the reverse analysis of the fault failure mechanism is the main approach; only for fault cases with strong pertinence and failure analysis carried out from the forward perspective, and fault cases that conform to the DFEMA principle, can a knowledge graph be established or even a fault analysis case be constructed, but the number of such cases is very limited.

[0008] At the same time, the existing cloud-based diagnostic solutions do not have the ability to distinguish independent events. Because a vehicle fault may be caused by a single or multiple fault causes, but the fault manifestation data is diverse, which is because the relationship between fault causes and fault manifestations is one-to-many. However, the existing diagnostic solutions cannot recognize the one-to-many relationship between fault causes and fault manifestations, so the diagnostic accuracy is poor. Summary of the Invention

[0009] The technical problem to be solved by the present invention is to provide a vehicle remote intelligent diagnosis method, system, storage medium and device, which can reduce the maintenance difficulty of vehicles for after-sales personnel and improve the accuracy of complex fault analysis.

[0010] To solve the above technical problems, as one aspect of the present invention, a vehicle remote intelligent diagnosis method is provided, which at least includes the following steps:

[0011] Receive a vehicle fault diagnosis request or obtain the vehicle's overall vehicle status data and vehicle-end diagnosis data according to periodic triggering;

[0012] Screen and extract description signals from the overall vehicle status data to form a description signal matrix; and analyze the description signal matrix according to preset analysis rules and standard thresholds to obtain a preliminary result;

[0013] Form multiple phenomenon description entries respectively associated with each description information according to the preliminary result, respond to the selection of the phenomenon description entries by the staff, determine the fault category of this diagnosis, and determine the scope of the diagnostic analysis case analysis model corresponding to the fault category in the case library;

[0014] Match and analyze the characteristic data of each case analysis model of the fault category with the vehicle-end diagnosis data of this diagnosis, and decompose this diagnosis into at least one independent fault event;

[0015] Conduct classification analysis according to the various fault causes associated with the case analysis model corresponding to each independent fault event to determine the fault cause of this diagnosis.

[0016] Among them, receiving a vehicle fault diagnosis request or being triggered periodically to obtain the vehicle's overall vehicle status data and vehicle - end diagnosis data further includes:

[0017] Receiving a repair work order uploaded by after - sales personnel, and according to the vehicle VIN code in the repair work order, obtaining the corresponding overall vehicle status data and vehicle - end diagnosis data of the vehicle from the vehicle or from the cloud; or

[0018] Periodically obtaining the corresponding overall vehicle status data and vehicle - end diagnosis data of the vehicle from the vehicle or from the cloud according to a pre - set vehicle VIN code.

[0019] Among them, screening and extracting some description signals from the overall vehicle status data to form a description signal matrix; and analyzing the description signal matrix according to pre - set analysis rules and standard thresholds to obtain a preliminary result further includes:

[0020] According to pre - set extraction rules, screening and extracting some key description signals that can express each electrical system and functional module of the vehicle from the overall vehicle status data, and filling them into a general description signal matrix to form the description signal matrix for this diagnosis.

[0021] According to pre - set analysis rules and standard thresholds, comparing and verifying each signal in the description signal matrix, and obtaining a preliminary result based on each description signal that exceeds the standard value.

[0022] Among them, forming multiple phenomenon description entries according to the preliminary result, responding to the staff's selection of the phenomenon description entries, determining the fault category of this diagnosis, and determining the range of diagnostic analysis case analysis models corresponding to the fault category in the case library further includes:

[0023] Obtaining multiple pre - associated phenomenon description entries according to the preliminary result and displaying them;

[0024] Responding to the staff's selection of the phenomenon description entries, and determining the fault category of this diagnosis according to the selected phenomenon description entries;

[0025] Searching in the case library according to the fault category to obtain multiple diagnostic analysis case analysis models corresponding to the fault category.

[0026] Among them, matching and analyzing the characteristic data of each case analysis model of the fault category with the vehicle - end diagnosis data of this diagnosis, and decomposing this diagnosis into at least one independent fault event further includes:

[0027] By using traversal matching combined with bubble analysis method, the characteristic data of the multiple diagnostic analysis case models are matched and analyzed with the fault data of this diagnosis, and this diagnosis is decomposed into at least one independent fault event;

[0028] If there is fault data that cannot be matched, the function analysis model corresponding to each unmatched fault data is obtained.

[0029] Among them, according to the various fault causes associated with the case analysis model corresponding to each independent fault event, classification analysis is carried out to determine the fault cause of this diagnosis, which further includes:

[0030] Obtain the various fault causes in the case analysis model corresponding to each independent fault event, and obtain the various fault causes in the function analysis model corresponding to each unmatched fault data;

[0031] Classify the various fault causes, and at least determine the fault causes with intersections as the fault causes of this diagnosis.

[0032] Among them, it further includes:

[0033] If the preliminary result, fault classification or fault cause cannot be obtained, a label is added to this diagnosis and handed over to manual for analysis and judgment.

[0034] Among them, it further includes:

[0035] When periodically monitoring the vehicle for faults, through the analysis of the vehicle's description information matrix, when it is considered that the vehicle has a fault trend, according to the preset function analysis model, it is determined that the change trend of the precondition is most relevant to the change trend of the description signal, and potential fault causes are obtained.

[0036] Among them, according to the preset function analysis model, it is determined that the change trend of the precondition is most relevant to the change trend of the description signal, and potential fault causes are obtained, which specifically includes:

[0037] Analyze the data of the description matrix, confirm the vehicle signals with fault trends, and according to the function analysis model corresponding to the vehicle signals, calculate the covariance result between the vehicle signals and the corresponding preconditions. If the covariance result is the first predetermined positive number (such as 1) or the first predetermined negative number (such as -1), it is determined that the vehicle signal is correlated with the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted;

[0038] Among them, the covariance Cov(X,Y) is calculated by using the following formula:

[0039] Cov(X,Y) = E[XY] - E[X]E[Y];

[0040] Wherein, X and Y are two random signal variables to be detected, namely the vehicle signal to be detected and the precondition; E[X] is the expected value of the random signal variable; E[Y] is the expected value of the random signal variable; E[XY] represents the product of the expected values of the random signal variables X and Y, that is, the average of the products of all possible results.

[0041] Wherein, according to the preset functional analysis model, it is determined that the change trend of the precondition is most relevant to the change trend of the description signal, and potential fault causes are obtained, further including:

[0042] Calculating the correlation coefficient of the covariance between the vehicle signal and the corresponding precondition; if the correlation coefficient is greater than the second predetermined positive number, it is determined that there is a correlation between the vehicle signal and the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted.

[0043] Wherein, the correlation coefficient Corr(X,Y) of the covariance obtained by using the following formula:

[0044]

[0045] Wherein, X and Y are two random signal variables to be detected, namely the vehicle signal to be detected and the precondition; D[X] is the variance of a random signal variable; E[Y] is the variance of another random signal variable; Cov(X,Y) is the covariance of the two random signal variables.

[0046] Correspondingly, on the other hand, the present invention further provides a vehicle remote intelligent diagnosis system, which at least includes the following steps:

[0047] A vehicle data acquisition unit, configured to receive a vehicle fault diagnosis request or obtain the vehicle's overall vehicle state data and vehicle-end diagnosis data according to periodic triggering.

[0048] A preliminary result analysis unit, configured to screen and extract description signals from the overall vehicle state data to form a description signal matrix; and analyze the description signal matrix according to preset analysis rules and standard thresholds to obtain a preliminary result.

[0049] A fault category determination unit, configured to form a plurality of phenomenon description entries respectively associated with each description information according to the preliminary result, respond to the selection of the phenomenon description entries by the staff, determine the fault category of this diagnosis, and determine the range of the diagnostic analysis case analysis model corresponding to the fault category in the case library.

[0050] A fault event decomposition unit, which is used to perform matching analysis on the feature data of each case analysis model of the fault category and the vehicle-end diagnosis data of this diagnosis, and decompose this diagnosis into at least one independent fault event;

[0051] A fault cause induction unit, which is used to perform classification analysis according to the various fault causes associated with the case analysis model corresponding to each independent fault event, and determine the fault cause of this diagnosis.

[0052] Among them, the preliminary result analysis unit further includes:

[0053] A description signal matrix establishment unit, which is used to screen and extract some key description signals that can express each electrical system and functional module of the whole vehicle from the whole vehicle state data according to the preset extraction rules, and fill them into the general description signal matrix to form the description signal matrix of this diagnosis;

[0054] A comparison and analysis unit, which is used to perform comparison and verification on each signal of the description signal matrix according to the preset analysis rules and standard thresholds, and obtain the preliminary result according to each description signal that exceeds the standard value.

[0055] Among them, the fault category determination unit further includes:

[0056] A phenomenon description entry association unit, which is used to obtain a plurality of pre-associated phenomenon description entries according to the preliminary result and display them;

[0057] A category determination unit, which is used to respond to the selection of the staff for the phenomenon description entry, and determine the fault category of this diagnosis according to the selected phenomenon description entry;

[0058] A case analysis model selection unit, which is used to retrieve in the case library according to the fault category and obtain a plurality of diagnostic analysis case analysis models corresponding to the fault category.

[0059] Among them, the fault event decomposition unit further includes:

[0060] A matching analysis unit, which is used to perform matching analysis on the feature data of the plurality of diagnostic analysis case analysis models and the fault data of this diagnosis by using traversal matching combined with bubble analysis method, and decompose this diagnosis into at least one independent fault event;

[0061] A function analysis model determination unit, which is used to obtain the function analysis model corresponding to each unmatched fault data when there is unmatched fault data in the matching result.

[0062] Among them, the fault cause induction unit further includes:

[0063] A fault cause acquisition unit, configured to acquire each fault cause in the case analysis model corresponding to each independent fault event, and acquire each fault cause in the function analysis model corresponding to each unmatched fault data;

[0064] A summarization processing unit, configured to classify the fault causes, and determine at least the fault causes with an intersection as the fault causes of this diagnosis.

[0065] Wherein, it further includes:

[0066] A fault trend analysis unit, configured to, when periodically monitoring the vehicle for faults, through the analysis of the vehicle's description information matrix, when it is considered that the vehicle has a fault trend, according to a preset function analysis model, determine that the change trend of the precondition has the highest correlation with the change trend of the description signal, and obtain potential fault causes;

[0067] Wherein, according to a preset function analysis model, determining that the change trend of the precondition has the highest correlation with the change trend of the description signal, and obtaining potential fault causes specifically includes:

[0068] Analyze the data of the description matrix, confirm the vehicle signals with a fault trend, according to the function analysis model corresponding to the vehicle signals, calculate the covariance result between the vehicle signals and the corresponding preconditions. If the covariance result is a first predetermined positive number or a first predetermined negative number, it is determined that the vehicle signals are correlated with the preconditions, and the preconditions are determined as potential fault causes, and a corresponding inspection is prompted; or

[0069] Calculate the correlation coefficient of the covariance between the vehicle signals and the corresponding preconditions; if the correlation coefficient is greater than a second predetermined positive number, it is determined that the vehicle signals are correlated with the preconditions, and the preconditions are determined as potential fault causes, and a corresponding inspection is prompted; wherein, the covariance Cov(X,Y) obtained by using the following formula:

[0070] Cov(X,Y) = E[XY] - E[X]E[Y];

[0071] The correlation coefficient Corr(X,Y) of the covariance obtained by using the following formula:

[0072]

[0073] Among them, X and Y are two random signal variables to be detected, that is, the vehicle signal to be detected and the preconditions; E[X] is the expected value of the random signal variable; E[Y] is the expected value of the random signal variable; E[XY] represents the product of the expected values of the random signal variables X and Y, that is, the average value of the products of all possible results; D[X] is the variance of a random signal variable; E[Y] is the variance of another random signal variable; Cov(X,Y) is the covariance of the two random signal variables.

[0074] As another aspect of the present invention, there is also provided a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method as described above are implemented.

[0075] As another aspect of the present invention, there is also provided a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method as described above are implemented.

[0076] Implementing the embodiments of the present invention has the following beneficial effects:

[0077] The present invention provides a vehicle remote intelligent diagnosis method, system, storage medium and device. By establishing a description matrix based on the vehicle's overall vehicle state data and vehicle-end diagnosis data to identify and judge vehicle fault conditions, analyze possible fault independent events, and then distinguish between calling a case diagnosis model or a function analysis model according to the matching degree of the description matrix, so as to analyze possible causes and help after-sales personnel provide intelligent diagnosis services; it can reduce the maintenance difficulty of after-sales personnel for vehicles and improve the accuracy of analyzing complex faults.

[0078] In the embodiments of the present invention, the description matrix is not simply established based on a knowledge graph, but is established from both forward and reverse perspectives, and supports technicians to optimize and iterate the description matrix according to the actual combat cases, optimize and expand the elements of the description matrix, and optimize the description coverage of the description matrix for fault events. At the same time, the solution of the present invention does not simply establish a diagnosis model from a knowledge graph, but allows technicians to establish a case diagnosis model based on diagnostic experience and a function analysis model based on a knowledge graph and forward design respectively, and fuse these two analysis models to comprehensively analyze and judge problems, thereby improving the accuracy of diagnosis.

[0079] At the same time, in the embodiments of the present invention, for vehicles without clear fault codes and fault feature data, the intelligent diagnosis analysis system uses the method of combining the description matrix with the function analysis model to help after-sales personnel analyze and discover potential fault correlation signals, so as to find possible fault causes.

[0080] In summary, implementing the embodiments of the present invention provides an efficient and convenient data processing solution, which well realizes the problem of extracting independent fault events from a large amount of vehicle fault data, efficiently invoking a diagnostic model to diagnose and analyze vehicle fault events, greatly reducing the diagnostic pressure on after-sales personnel and improving the work efficiency of diagnostic analysis; at the same time, it can also realize fault trend analysis, improve brand competitiveness, increase user viscosity, and enhance product quality. Implementing the embodiments of the present invention can greatly reduce the maintenance cost of users and improve vehicle use safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, obtaining other drawings based on these drawings without creative efforts still belongs to the scope of the present invention;

[0082] Figure 1 FIG. is a schematic main flow chart of an embodiment of a vehicle remote intelligent diagnosis method provided by the present invention;

[0083] Figure 2 is a schematic flow chart of a specific example of the method provided by the present invention;

[0084] Figure 3 is Figure 2 a more detailed schematic flow chart of vehicle remote intelligent diagnosis after receiving a repair work order in ;

[0085] Figure 4 FIG. is a schematic structural diagram of an embodiment of a vehicle remote intelligent diagnosis system provided by the present invention;

[0086] Figure 5 is Figure 4 a schematic structural diagram of the preliminary result analysis unit in ;

[0087] Figure 6 is Figure 4 a schematic structural diagram of the fault category determination unit in ;

[0088] Figure 7 is Figure 4 a schematic structural diagram of the fault event decomposition unit in ;

[0089] Figure 8 is Figure 5 a schematic structural diagram of the fault cause induction unit in. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0090] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings.

[0091] As Figure 1 shown, a schematic diagram of the main process of an embodiment of a vehicle remote intelligent diagnosis method provided by the present invention is shown, and in combination with Figure 2 and Figure 3 the more detailed process shown. In this embodiment, the method is implemented by an intelligent diagnosis system built on a cloud platform, and the method at least includes the following steps:

[0092] Step S10, receive a vehicle fault diagnosis request or obtain the vehicle's overall vehicle status data and vehicle terminal diagnosis data according to periodic triggering;

[0093] Step S11, screen and extract description signals from the overall vehicle status data to form a description signal matrix; and analyze the description signal matrix according to preset analysis rules and standard thresholds to obtain a preliminary result;

[0094] Step S12, form multiple phenomenon description entries respectively associated with each description information according to the preliminary result, respond to the selection of the phenomenon description entries by the staff, determine the fault category of this diagnosis, and determine the range of the diagnostic analysis case analysis model corresponding to the fault category in the case library;

[0095] Step S13, perform a matching analysis on the characteristic data of each case analysis model of the fault category and the vehicle terminal diagnosis data of this diagnosis, and decompose this diagnosis into at least one independent fault event;

[0096] Step S14, perform a classification analysis according to the various fault causes associated with the case analysis model corresponding to each independent fault event to determine the fault cause of this diagnosis.

[0097] The following will, in combination with Figure 2 and Figure 3 and specific examples, elaborate on each step involved in the present invention in detail.

[0098] More specifically, in the step S10, receiving a vehicle fault diagnosis request or obtaining the vehicle's overall vehicle status data and vehicle terminal diagnosis data according to periodic triggering further includes:

[0099] Step S100, receive a repair work order uploaded by after-sales personnel, and obtain the corresponding overall vehicle status data and vehicle terminal diagnosis data of the vehicle from the vehicle or from the cloud according to the vehicle VIN code in the repair work order;

[0100] When after-sales personnel receive a message push from the intelligent diagnosis system or a customer reports a problem, a repair work order will be created in the cloud. The repair work order includes at least a description of the fault phenomenon and keywords. Specifically, the repair work order contains the vehicle VIN code, the time of the case occurrence (including the date), the time of work order creation, and the fault phenomenon and keyword information that allows after-sales technicians to describe and record.

[0101] The intelligent diagnosis system set in the cloud will actively collect vehicle condition data from the cloud according to the VIN code information of the repair work order, and judge the vehicle condition status; in other examples, it can also send a collection command to the vehicle to collect corresponding data in real time.

[0102] In a more specific application scenario, at this time, if the intelligent diagnosis system in the cloud analyzes from the vehicle condition signal that the vehicle condition is in a pre-defined unsuitable diagnosis state (such as the vehicle is driving, the vehicle is in a power deficit state, the vehicle is in a dangerous scenario and environment, etc.), it will prompt the after-sales personnel of the actual state of the vehicle, and the after-sales personnel will decide whether to continue diagnosing the vehicle.

[0103] If the intelligent diagnosis system in the cloud analyzes and judges from the vehicle condition signal that the vehicle condition is in a suitable diagnosis state (such as the vehicle is in a stationary non-operating state), the intelligent diagnosis system will actively issue a diagnosis service instruction, read the vehicle's bus data and diagnostic data, and package them into vehicle status data and transmit them to the cloud.

[0104] The cloud can read the vehicle data and vehicle-end diagnostic data of the vehicle end. Among them, the vehicle-end bus data includes bus data containing at least vehicle condition function information and instruction information (which may not include audio and video media stream data in in-vehicle Ethernet) within the front and rear time intervals T (such as 10s, 15s, 30s, etc.) when the vehicle receives the send instruction.

[0105] The vehicle-end diagnostic data refers to the diagnostic data of each electrical system collected by the vehicle-end diagnostic engine according to the diagnostic script or set diagnostic instructions, including at least diagnostic trouble codes, electrical system version information, signals and values of internal sensors and key modules of the electrical system, status signals and values of function modules, etc.

[0106] In another case, the step S100 may include step S101, and periodically obtain the corresponding vehicle status data and vehicle-end diagnostic data of the vehicle from the vehicle or from the cloud according to the pre-set vehicle VIN code.

[0107] In a specific example, the intelligent diagnostic system supports presetting some monitoring scripts and storing them in the system's script library. When a specific case occurs, after-sales personnel can retrieve the relevant monitoring scripts and send them to the vehicle to monitor some key parameters on the vehicle. Among them, the monitoring script is a set of automatic detection processes formed by a series of monitoring instructions and diagnostic service instructions based on preset vehicle conditions and execution conditions, and exists in the form of a script file. It is understandable that the monitoring signal range of the script file is some key vehicle data preset by technicians, which is used to verify the vehicle's fault trend in more dimensions.

[0108] The specific monitoring process will eventually generate different fault codes based on different conditions and record them in the script results. This allows the development of vehicle fault trends to be analyzed based on the fault codes in the script results.

[0109] More specifically, in step S11, the process of screening and extracting some descriptive signals from the vehicle status data to form a descriptive signal matrix, and analyzing the descriptive signal matrix according to preset analysis rules and standard thresholds to obtain preliminary results further includes:

[0110] Step S110: According to the preset extraction rules, the key descriptive signals that can express the electrical systems and functional modules of the vehicle are screened and extracted from the vehicle status data, and the key descriptive signals are filled into the general descriptive signal matrix to form the descriptive signal matrix for this diagnosis.

[0111] The description matrix is composed of a bunch of characteristic signals that can describe the vehicle electrical system and functional status. The signal may be in the form of a bus signal, a diagnostic signal, or a pre-buried signal in a log.

[0112] These signals are vehicle-side data collected by a vehicle-side electrical system (such as a gateway, on-board data cache module, etc.) according to certain conditions (such as a certain sampling period) and cached locally on the vehicle. When the network signal strength is good and meets the preset conditions (such as time), the data will be uploaded to the cloud platform in a sub-package and analyzed by the cloud platform's intelligent diagnostic system.

[0113] In this step, the intelligent diagnostic system will filter and extract key descriptive signals that can express the vehicle's various electrical systems and functional modules from the vehicle status data according to preset extraction rules, forming a descriptive signal matrix. Specifically, these can be vehicle power characteristic signals, air conditioning system characteristic signals, entertainment system characteristic signals, and intelligent driving system characteristic signals, etc. For example, the health status data of the air conditioning system can be described by the high-voltage sensor signal and the low-voltage sensor signal, and the functional health status of the tire pressure system can be described by the four tire pressure values and the four tire temperature values.

[0114] Step S111: Compare and verify each signal in the description signal matrix according to the preset analysis rules and standard thresholds, and obtain a preliminary result based on each description signal that exceeds the standard value.

[0115] In this step, the intelligent diagnosis system in the cloud will compare and verify the signals in the description matrix according to the preset analysis rules and standard thresholds. After the intelligent diagnosis system extracts the vehicle-end data for the vehicle to be monitored, it will automatically form a description matrix according to the preset rules, analyze the characteristic signals in the description matrix, and add labels according to the analysis results. The analysis results are divided into 5 levels:

[0116] Level 0: The vehicle has no faults;

[0117] Level 1: The vehicle has serious faults;

[0118] Level 2: The vehicle has perceptible faults;

[0119] Level 3: The vehicle has faults, but it does not affect the user's use;

[0120] Level 4: The vehicle has no faults temporarily, but there is a trend of faults occurring.

[0121] If the intelligent diagnosis system analyzes that the vehicle has no abnormalities based on the signals in the description matrix, which contradicts the result of the after-sales personnel creating a maintenance work order believing that this vehicle has faults, the system will pop up a selection box, prompting the after-sales personnel "The system cannot perform intelligent analysis on this case. Do you suggest switching to manual analysis?" After the after-sales personnel select the "OK" button, the system will add a time label to the data of this case and store it, and wait for in-depth analysis by subsequent technical experts before incorporating it into the diagnosis system.

[0122] For example, a certain car owner feedbacks to the maintenance personnel that the window lifting function of his vehicle is abnormal, and the after-sales personnel create a maintenance work order, but the intelligent diagnosis system analyzes that the vehicle is healthy, which indicates that the analysis result of the description matrix does not match the actual situation, and it is necessary to perform manual diagnosis and correct the description matrix or analysis rules of the intelligent diagnosis system.

[0123] More specifically, in the step S12, the forming of multiple phenomenon description entries respectively associated with each description information according to the preliminary result, and in response to the selection of the phenomenon description entries by the staff, determining the fault category of this diagnosis, and determining the range of the diagnostic analysis case analysis model corresponding to the fault category in the case library, further includes:

[0124] According to the preliminary result, obtain multiple pre-associated phenomenon description entries and display them. Among them, various types of description information are analyzed in advance, and at least one phenomenon description entry is associated with each type of description information. This phenomenon description entry can be, for example, some questions used to describe possible fault phenomena;

[0125] In response to the selection of the phenomenon description entry by the staff, determine the fault category of this diagnosis according to the selected phenomenon description entry;

[0126] According to the fault category, retrieve in the case library to obtain multiple diagnostic analysis case analysis models corresponding to the fault category.

[0127] In a specific example, if the intelligent diagnosis system analyzes that the vehicle is abnormal according to the signals of the description matrix, then the system will predict some possible faults of the vehicle according to the analysis rules and the system corresponding to the signals, and form some questions (fuzzified phenomenon descriptions) for the after-sales personnel to select and confirm. These questions are the aforementioned phenomenon description entries, and the after-sales personnel click on the corresponding phenomenon description entries to let the system judge whether the signals of the description matrix can cover the fault events of this case. In this process, the after-sales personnel judge the coincidence degree of the phenomenon description, and select multiple options to pick out some prediction phenomenon description keywords that are relatively most in line with the fault phenomenon. Behind each phenomenon description keyword, there are some fault characteristics. The intelligent diagnosis system can obtain the fault direction according to the keywords not selected by the after-sales personnel, eliminate these fault characteristics, and thus can form a description formula of the fault characteristic negative item inside, complete the second-round selection and more accurate description of the fault category of this time, obtain clearer fault characteristics, and help to screen the diagnostic analysis model in the next step;

[0128] For example, a car owner feedbacks to the maintenance personnel that his vehicle is not cooling. The intelligent diagnosis system checks the signals of the description matrix and finds that the signal of the low-pressure sensor of the air conditioner is too low, which indicates the possibility of refrigerant leakage. The intelligent diagnosis system will pop up a dialog box for the after-sales personnel, showing "Is there any abnormality in the air conditioning refrigeration of this vehicle?" If the after-sales personnel select "Yes", the intelligent diagnosis system can confirm that the description matrix can cover this fault and will continue to analyze.

[0129] Another example, a car owner feedbacks to the maintenance personnel that the fault light has ever come on during the vehicle driving. After the after-sales personnel create a work order, the intelligent diagnosis system extracts the data of the vehicle end including this period of time to the cloud for analysis according to the case occurrence time described in the work order. The intelligent diagnosis system reads the bus message with the time label, checks the bus data with the vehicle speed not being zero, and finds that a certain status signal of the ESP is error. Then the system will pop up a dialog box and ask the after-sales personnel: "Has there ever been a fault light or an abnormality related to braking during the driving of this vehicle?" If the after-sales personnel click "Yes", it means that the system can continue to analyze.

[0130] If the after-sales personnel click "No", it means that the description matrix does not apply to this fault case, indicating that the signals in the description matrix cannot cover this case. The intelligent diagnosis system will, as described in point 2 above, add a timestamp to this part of the data, store it, and transfer it for manual diagnosis.

[0131] More specifically, in the step S13, the feature data of each case analysis model of the fault category is matched and analyzed with the vehicle-end diagnosis data of this diagnosis, and this diagnosis is decomposed into at least one independent fault event, which further includes:

[0132] Step S130, using a traversal matching combined with a bubble analysis method, the feature data of the multiple diagnostic analysis case analysis models is matched and analyzed with the fault data of this diagnosis, and this diagnosis is decomposed into at least one independent fault event;

[0133] In a specific example, the intelligent diagnosis system needs to distinguish how many independent events there are, because each 1 independent event corresponds to the invocation of 1 kind of diagnostic analysis case analysis model.

[0134] According to the characteristic performance of the description matrix, in the case library in the system, find a case analysis model with consistent characteristics for diagnostic analysis. Use the method of matching traversal combined with bubble sorting to select the most consistent case analysis model. If the characteristics of a case analysis model A are 50% consistent with the characteristic performance of the description matrix, then the case analysis model A can be recorded first, and the characteristic consistency of the case analysis model A is considered to be 50%. Then continue to traverse and compare the characteristic consistency of the next case analysis model with the description matrix until the characteristic consistency of the next case analysis model B is greater than 50%, and then we record the case analysis model B and delete A. And so on, until all case analysis models are traversed at the end, and the case analysis model x with the highest characteristic consistency is found, and the fault of this vehicle is first considered to have 1 independent fault event, which is the most matched with x.

[0135] Then, in the same way, the remaining feature data can be matched with the case analysis models in the case library to sequentially find the case analysis models with the largest remaining characteristic consistency.

[0136] Step S131, if there is fault data that cannot be matched, obtain the function analysis model corresponding to each piece of fault data that cannot be matched.

[0137] If the finally remaining feature data cannot find the most matched case analysis model in the way of combined features, it is considered that each 1 remaining feature data corresponds to 1 fault independent event, and the corresponding function analysis model is invoked.

[0138] It can be understood that in step S131 of the present invention, the case analysis model with the highest matching degree is preferentially searched for analysis, and the remaining abnormal signals are regarded as independent events and analyzed according to the function analysis model. It can be known that in the intelligent diagnosis system, the case analysis model corresponding to a single fault code or a single fault feature data has the lowest priority.

[0139] For example, there are 30 signals in the description matrix showing abnormalities. First, search for a similar case analysis model in the case library. It is found that there is one Model A, and 8 characteristic signals are the same as the abnormal signals in the description matrix. There is another Model B with 27 characteristic signals consistent with the signals in the description matrix. Then, Model B is preferentially retrieved, and the 27 matching characteristic signals are regarded as an independent fault event, and intelligent diagnosis is performed for the vehicle according to the analysis rules of Model B. For the remaining 3 characteristic signals, these 3 characteristic signals can be regarded as 3 independent fault events respectively, and the function analysis model corresponding to the characteristic signal is retrieved.

[0140] In this step S13, the case analysis model refers to an automatic diagnosis process formed by technicians based on empirical cases, knowledge graphs, and functional logical relationships, including characteristic signal descriptions, diagnostic processes, and diagnostic results. Among them, the characteristic signal description is used to identify the characteristic performance of the case analysis model. For example, for the case analysis model of engine startup failure, the characteristic signals may be engine speed = 0, startup status = not started, vehicle electrical status = non-crank status, startup switch = has been pressed, and EMS_errorST = 1.

[0141] The diagnostic process consists of multiple steps, and each step is a judgment and analysis of a certain vehicle-end signal according to a standard threshold and a preset processing and diagnostic rule.

[0142] The diagnostic result is a unique diagnostic result code finally formed according to the diagnostic process and judgment conditions.

[0143] According to the logical mechanism, the case diagnosis model can be divided into two categories: the case analysis model and the function analysis model. Among them, the case analysis model is formed by technicians through sorting and screening based on actual cases, and each step of the diagnostic process may not have a correlation. The function analysis model is compiled by technicians according to the function design document and the function implementation logic. The previous step of the diagnostic process is the precondition or sufficient condition or trigger condition for the next step, and there is a logical correlation between each step.

[0144] Each case analysis model corresponds to an independent fault event.

[0145] As for the case library, it is the database area in the intelligent diagnosis system that stores and manages a large number of case analysis models and functional analysis models.

[0146] It can be understood that the true cause of a single fault in a vehicle may produce one or more fault phenomena, corresponding to one or more fault characteristic signal values. There is an independent relationship between different true fault causes. Therefore, a vehicle may have a very large number of fault phenomena and fault signals, but according to causal logic analysis, they are all jointly generated by several independent fault events behind, which are essentially caused by the true cause of the fault. Therefore, we can regard the fault events corresponding to each true cause as independent events.

[0147] More specifically, in the step S14, according to the various fault causes associated with the case analysis model corresponding to each independent fault event, a classification analysis is carried out to determine the fault cause of this diagnosis, which further includes:

[0148] Obtain the various fault causes in the case analysis model corresponding to each independent fault event, and obtain the various fault causes in the functional analysis model corresponding to each unmatched fault data;

[0149] Classify the various fault causes, and at least determine the fault causes with intersections as the fault causes of this diagnosis.

[0150] For example, in an actual example, the intelligent diagnosis system analyzes a certain vehicle and retrieves three case diagnosis models A, B, and C. Model A believes that the fault may be caused by wire harness, fuse, and switch failures; Model B believes that the fault may be caused by fuse, display screen, and control unit ECU failures; Model C believes that the fault may be caused by motor and fuse failures. Then, the intersection of the causes of the three case analysis models A, B, and C all has fuses. Therefore, the intelligent diagnosis system believes that the fuse is faulty and is a possible cause of the vehicle's fault, and outputs the result as a prompt to the after-sales personnel. So that the after-sales personnel can repair the vehicle according to the analysis result.

[0151] It can be understood that in each step of the present invention, if no preliminary result, fault classification, or fault cause can be obtained, a label is added to this diagnosis and handed over to manual for analysis and judgment.

[0152] It can be understood that in the scenario of active monitoring in the foregoing step S101. When periodically monitoring the vehicle for faults, through the analysis of the vehicle's description information matrix, when it is considered that the vehicle has a fault trend, according to the preset functional analysis model, it is determined that the change trend of the precondition has the highest correlation with the change trend of the description signal, and potential fault causes are obtained.

[0153] Specifically, if the vehicle owner has not reported any faults, the after-sales staff can also add the VIN codes of some vehicles to the intelligent diagnosis system, and the intelligent diagnosis system will monitor the health of these vehicles. The intelligent diagnosis system will actively and periodically collect vehicle data from the cloud according to the VIN code, extract the description matrix, and analyze the description matrix according to the preset analysis rules and standard thresholds.

[0154] If the intelligent diagnosis system discovers that the vehicle may have serious faults through the description matrix, it will actively push the information to the after-sales staff. The after-sales staff will then actively contact the vehicle owner to confirm whether there are faults and take emergency rescue measures.

[0155] If the intelligent diagnosis system discovers potential faults through the description matrix but they do not affect the current use of the vehicle owner, the intelligent diagnosis system will actively push the information to the after-sales staff. The after-sales staff will then actively contact the vehicle owner and create a maintenance work order for the vehicle owner, repeating the subsequent work processes based on the maintenance work order.

[0156] If the intelligent diagnosis system discovers that the vehicle has no faults through the description matrix, it will not give any alarm prompts.

[0157] If the intelligent diagnosis system discovers that the vehicle has no faults through the description matrix, but there are fault trends in some systems or functions, the intelligent diagnosis system will actively push the result information to the after-sales staff. The after-sales staff will issue a monitoring script to the vehicle terminal in the cloud to monitor and collect key vehicle data, so as to verify whether it conforms to the trend. If it conforms, it means that the vehicle indeed has a fault trend, and then the after-sales staff will actively contact the vehicle owner and create a maintenance work order for the vehicle owner, repeating the work processes of the above steps. If it does not conform, the system will add labels and hand it over to the manual for analysis and optimization of trend judgment.

[0158] If the intelligent diagnosis system analyzes through the description matrix and believes that the vehicle has a fault trend, then the system will, according to the function analysis model, see which precondition's change trend has the highest correlation with the change trend of the description signal (calculate using the covariance formula to analyze the most influential cause).

[0159] In a specific example, according to the preset function analysis model, it is determined that the change trend of the precondition has the highest correlation with the change trend of the description signal, and the potential fault cause is obtained. The following method is used to achieve it:

[0160] The intelligent diagnosis system has the function of vehicle health trend analysis, providing a technical means for fault trend analysis for users of vehicles without obvious faults, enabling maintenance personnel to find the reasons not only when there are fault codes or the vehicle has faults, but also to eliminate faults as much as possible before they occur.

[0161] Analyze the data describing the matrix, confirm the vehicle signals with a fault trend, calculate the covariance result between the vehicle signals and the corresponding preconditions according to the function analysis model corresponding to the vehicle signals. If the covariance result is the first predetermined positive value (such as 1) or the first predetermined negative value (such as -1), it is determined that there is a correlation between the vehicle signal and the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted;

[0162] Among them, the covariance Cov(X,Y) obtained by using the following formula:

[0163] Cov(X,Y) = E[(X - E[X])(Y - E[Y])]

[0164] = E[XY] - 2E[Y]E[X] combined with E[X]E[Y]

[0165] = E[XY] - E[X]E[Y];

[0166] Among them, X and Y are two random signal variables to be detected, that is, the vehicle signal to be detected and the precondition; E[X] is the expected value of the random signal variable; E[Y] is the expected value of the random signal variable; E[XY] represents the product of the expected values of the random signal variables X and Y, that is, the average of the products of all possible results.

[0167] Furthermore, in some examples, the correlation coefficient (i.e., the correlation strength) of the covariance between the vehicle signal and the corresponding precondition can also be calculated at the same time; if the correlation coefficient is greater than the second predetermined positive number, it is determined that there is a correlation between the vehicle signal and the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted; among them, if the correlation coefficient is close to 1 or -1, it means that the signal change tends to be a linear relationship. In the present invention, the second predetermined positive number can be set to 0.5, and all correlated signals with a correlation coefficient greater than 0.5 need to be inspected.

[0168] Among them, the correlation coefficient Corr(X,Y) of the covariance obtained by using the following formula:

[0169]

[0170] Among them, X and Y are two random signal variables to be detected, that is, the vehicle signal to be detected and the precondition; D[X] is the variance of a random signal variable; E[Y] is the variance of another random signal variable; Cov(X,Y) is the covariance of the two random signal variables.

[0171] For example, when the ambient light of a vehicle shows a phenomenon of sudden brightening and dimming, the intelligent diagnosis system discovers that there are relatively abnormal fluctuations in the power supply voltage signal U0 of the ambient light. According to the functional logic, the ambient light module is affected by the BCM power supply U1 and the vehicle head unit drive signal S0. The intelligent diagnosis system calculates the vehicle-end data for a period of time and finds that although the amplitudes of U0 and U1 are very different, their covariance is 1, indicating a correlation. According to the functional logic, the BCM power supply is directly powered by the battery, and the signal U2 of the EBS battery sensor can be used to detect it. The intelligent diagnosis system analyzes U1 and U2 and finds that although the fluctuation amount of U2 is very small, the resulting covariance is 1, indicating a correlation. Therefore, the intelligent diagnosis system prompts the after-sales personnel that there may be an abnormality in the vehicle's battery, and the voltage fluctuation causes the brightness of the ambient light to suddenly brighten and dim.

[0172] It can be understood that the more detailed processes shown in Figure 2 and Figure 3 can be referred to together. In the method provided by the present invention, the intelligent diagnosis system first conducts a preliminary inspection of the vehicle as a whole based on the signals of the description matrix, and analyzes whether there are any faults.

[0173] Then, according to the comparison of the consistency between the description matrix and the characteristic signals of the case diagnosis model, it is determined which case diagnosis models to select for diagnostic analysis of the vehicle, which is equivalent to an in-depth inspection by an expert.

[0174] Based on the results of the two inspections and diagnoses, a final diagnostic conclusion is formed to assist the after-sales personnel in vehicle maintenance.

[0175] In the embodiment of the present invention, by having the following functions, the technical problem of the matching between a large amount of fault data and the analysis model in the industry can be well solved.

[0176] It has the function of preliminary screening of fault events. This function uses the data transmitted from the vehicle to the cloud to first establish a description signal matrix, make a preliminary judgment on the vehicle-end fault events, distinguish how many independent fault events there are, and use the description matrix to determine which diagnostic models to use

[0177] It has the function of fault analysis, screening, and selection. To improve the accuracy of the description of vehicle fault events by the description signal matrix, predictive phenomenon description keywords of possible fault phenomena are formed. The after-sales personnel help the intelligent diagnosis and analysis system exclude the most inconsistent fault directions by selecting keywords, so as to enable the intelligent diagnosis and analysis system to exclude some diagnostic models that are most inconsistent with the actual vehicle fault phenomena, and improve the analysis accuracy.

[0178] It has an independent event analysis function. Through the combined analysis of features and the comparison and judgment of feature consistency, the intelligent diagnosis and analysis system classifies a large amount of fault feature data in combination, and automatically analyzes how many independent fault events actually exist in the vehicle during the process of sequentially finding the case analysis model that best matches the vehicle fault features.

[0179] It has a vehicle health trend analysis function. For vehicles without clear fault codes and fault feature data, the intelligent diagnosis and analysis system can help after-sales personnel analyze and discover potential fault correlation signals by using the description matrix in combination with the function analysis model, so as to find possible fault causes.

[0180] As Figure 4 shown, a structural schematic diagram of an embodiment of a vehicle remote intelligent diagnosis system provided by the present invention is shown; in combination with Figures 5 to 8 shown, in this embodiment, the system 1 at least includes the following steps:

[0181] The vehicle data acquisition unit 10 is used to receive vehicle fault diagnosis requests or obtain the vehicle's overall vehicle state data and vehicle-end diagnosis data according to periodic triggering.

[0182] The preliminary result analysis unit 11 is used to screen and extract description signals from the overall vehicle state data to form a description signal matrix; and analyze the description signal matrix according to preset analysis rules and standard thresholds to obtain preliminary results.

[0183] The fault category determination unit 12 is used to form multiple phenomenon description entries respectively associated with each description information according to the preliminary results, respond to the selection of the phenomenon description entries by the staff, determine the fault category of this diagnosis, and determine the range of the diagnostic analysis case analysis model corresponding to the fault category in the case library.

[0184] The fault event decomposition unit 13 is used to perform matching analysis on the feature data of each case analysis model of the fault category and the vehicle-end diagnosis data of this diagnosis, and decompose this diagnosis into at least one independent fault event.

[0185] The fault cause induction unit 14 is used to perform classification analysis according to the various fault causes associated with the case analysis model corresponding to each independent fault event to determine the fault cause of this diagnosis.

[0186] As Figure 5 shown, in a specific example, the preliminary result analysis unit 11 further includes:

[0187] A description signal matrix establishment unit 110 is configured to screen and extract some key description signals that can represent each electrical system and functional module of the vehicle from the vehicle state data according to a preset extraction rule, and fill them into a general description signal matrix to form a description signal matrix for this diagnosis;

[0188] A comparison and analysis unit 111 is configured to perform comparison and verification on each signal in the description signal matrix according to a preset analysis rule and a standard threshold, and obtain a preliminary result based on each description signal that exceeds the standard value.

[0189] As Figure 6 shown, in a specific example, the fault category determination unit 12 further includes:

[0190] A phenomenon description entry association unit 120 is configured to obtain a plurality of pre-associated phenomenon description entries according to the preliminary result and display them;

[0191] A category determination unit 121 is configured to respond to the selection of the phenomenon description entry by the staff, and determine the fault category of this diagnosis according to the selected phenomenon description entry;

[0192] A case analysis model selection unit 122 is configured to retrieve in a case library according to the fault category and obtain a plurality of diagnostic analysis case analysis models corresponding to the fault category.

[0193] As Figure 7 shown, in a specific example, the fault event decomposition unit 13 further includes:

[0194] A matching analysis unit 130 is configured to adopt a traversal matching combined with a bubble analysis method to match and analyze the feature data of the plurality of diagnostic analysis case analysis models with the fault data of this diagnosis, and decompose this diagnosis into at least one independent fault event;

[0195] A function analysis model determination unit 131 is configured to obtain a function analysis model corresponding to each unmatched fault data when there is unmatched fault data in the matching result.

[0196] As Figure 8 shown, in a specific example, the fault cause induction unit 14 further includes:

[0197] A fault cause obtaining unit 140 is configured to obtain each fault cause in the case analysis model corresponding to each independent fault event, and obtain each fault cause in the function analysis model corresponding to each unmatched fault data;

[0198] The induction processing unit 141 is configured to classify the various fault causes, and at least determine the fault causes with intersections as the fault causes of this diagnosis.

[0199] Furthermore, the vehicle remote intelligent diagnosis system 1 provided by the present invention further includes:

[0200] The fault trend analysis unit 15 is configured to, when periodically monitoring the vehicle for faults, analyze through the description information matrix of the vehicle. When it is considered that the vehicle has a fault trend, according to the preset function analysis model, determine that the change trend of the precondition is most relevant to the change trend of the description signal, and obtain potential fault causes;

[0201] Among them, according to the preset function analysis model, determining that the change trend of the precondition is most relevant to the change trend of the description signal and obtaining potential fault causes specifically includes:

[0202] Analyze the data of the description matrix, confirm the vehicle signals with fault trends, and calculate the covariance result between the vehicle signals and the corresponding preconditions according to the function analysis model corresponding to the vehicle signals. If the covariance result is a first predetermined positive number (such as 1) or a first predetermined negative number (such as -1), it is determined that the vehicle signal is relevant to the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted; or

[0203] Calculate the correlation coefficient of the covariance between the vehicle signal and the corresponding precondition; if the correlation coefficient is greater than a second predetermined positive number (such as 0.5), it is determined that the vehicle signal is relevant to the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted;

[0204] Among them, the covariance Cov(X,Y) obtained by using the following formula:

[0205] Cov(X,Y) = E[(X - E[X])(Y - E[Y])]

[0206] = E[XY] - 2E[Y]E[X] combined with E[X]E[Y]

[0207] = E[XY] - E[X]E[Y];

[0208] The correlation coefficient Corr(X,Y) of the covariance obtained by using the following formula:

[0209] Among them, X and Y are two random signal variables to be detected, namely the vehicle signal to be detected and the preconditions; E[X] is the expected value of the random signal variable; E[Y] is the expected value of the random signal variable; E[XY] represents the product of the expected values of the random signal variables X and Y, that is, the average of the products of all possible results; D[X] is the variance of a random signal variable; E[Y] is the variance of another random signal variable; Cov(X, Y) is the covariance of the two random signal variables.

[0210] For more details, reference can be made to the foregoing description of Figures 1 to 3 which will not be elaborated here.

[0211] As another aspect of the present invention, there is also provided a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described as above Figures 1 to 3 are implemented. For more details, reference can be made to the foregoing description of Figures 1 to 3 which will not be elaborated here.

[0212] As another aspect of the present invention, there is also provided a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method described as above Figures 1 to 3 are implemented. For more details, reference can be made to the foregoing description of Figures 1 to 3 which will not be elaborated here.

[0213] Implementing the embodiments of the present invention has the following beneficial effects:

[0214] The present invention provides a vehicle remote intelligent diagnosis method, system, storage medium and device. By establishing a description matrix based on the vehicle's overall vehicle state data and vehicle-end diagnosis data to identify and judge vehicle fault conditions, analyze possible fault independent events, and then distinguish between calling a case diagnosis model or a function analysis model according to the matching degree of the description matrix, so as to analyze possible causes and help after-sales personnel provide intelligent diagnosis services; it can reduce the maintenance difficulty of after-sales personnel for vehicles and improve the accuracy of analyzing complex faults.

[0215] In an embodiment of the present invention, the described matrix is not simply established based on a knowledge graph, but is established from both forward and reverse perspectives. Moreover, it supports technicians to optimize and iterate the described matrix according to the actual combat cases, optimize and expand the elements of the described matrix, and optimize the description coverage of the described matrix for fault events. At the same time, the solution of the present invention does not simply establish a diagnostic model by the knowledge graph, but allows technicians to establish a case diagnostic model according to diagnostic experience respectively and establish a functional analysis model according to the knowledge graph and forward design, and fuse these two analysis models to comprehensively analyze and judge problems, so as to improve the accuracy of diagnosis.

[0216] At the same time, in an embodiment of the present invention, for a vehicle without clear fault codes and fault characteristic data, the intelligent diagnostic analysis system can help after-sales personnel analyze and discover potential fault correlation signals by using the described matrix in combination with the functional analysis model, so as to find out possible fault causes.

[0217] In summary, implementing the embodiments of the present invention provides an efficient and convenient data processing solution, which well realizes the problem of extracting independent fault events from a large amount of vehicle fault data, efficiently calling a diagnostic model to diagnose and analyze vehicle fault events, greatly reducing the diagnostic pressure on after-sales personnel and improving the diagnostic analysis work efficiency; at the same time, it can also realize fault trend analysis, improve brand competitiveness, increase user viscosity, and improve product quality. Implementing the embodiments of the present invention can greatly reduce the maintenance cost of users and improve vehicle use safety.

[0218] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, apparatus, or computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0219] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0220] The above-disclosed is only a preferred embodiment of the present invention. Of course, the scope of the rights of the present invention cannot be limited thereby. Therefore, equivalent changes made according to the claims of the present invention still fall within the scope covered by the present invention.

Claims

1. A vehicle remote intelligent diagnosis method, characterized in that, At least include the following steps: Receive a vehicle fault diagnosis request or obtain the vehicle's overall vehicle status data and vehicle-end diagnosis data according to periodic triggering; Screen and extract description signals from the overall vehicle status data to form a description signal matrix; and analyze the description signal matrix to obtain a preliminary result; Form multiple phenomenon description entries respectively associated with each description information according to the preliminary result, respond to the staff's selection of the phenomenon description entries, determine the fault category of this diagnosis, and determine the scope of the diagnostic analysis case analysis model corresponding to the fault category in the case library; Match and analyze the characteristic data of each case analysis model of the fault category with the vehicle-end diagnosis data of this diagnosis, and decompose this diagnosis into at least one independent fault event; Conduct a classification analysis according to the various fault causes associated with the case analysis model corresponding to each independent fault event to determine the fault cause of this diagnosis.

2. The method according to claim 1, characterized in that, Receive a vehicle fault diagnosis request or obtain the vehicle's overall vehicle status data and vehicle-end diagnosis data according to periodic triggering, which further includes: Receive the uploaded repair work order, and obtain the corresponding overall vehicle status data and vehicle-end diagnosis data of the vehicle from the vehicle or from the cloud according to the vehicle VIN code in the repair work order; or Periodically obtain the corresponding overall vehicle status data and vehicle-end diagnosis data of the vehicle from the vehicle or from the cloud according to the pre-set vehicle VIN code.

3. The method according to claim 2, wherein The screening and extracting partial description signals from the overall vehicle status data to form a description signal matrix; and analyzing the description signal matrix to obtain a preliminary result, which further includes: According to the preset extraction rules, screen and extract partial key description signals that can express each electrical system and functional module of the vehicle from the overall vehicle status data, and fill them into the general description signal matrix to form the description signal matrix for this diagnosis; Compare and verify each signal of the description signal matrix according to the preset analysis rules and standard thresholds, and obtain a preliminary result according to each description signal that exceeds the standard value.

4. The method according to claim 2, wherein The forming multiple phenomenon description entries according to the preliminary result, responding to the staff's selection of the phenomenon description entries, determining the fault category of this diagnosis, and determining the scope of the diagnostic analysis case analysis model corresponding to the fault category in the case library, which further includes: Obtain multiple pre-associated phenomenon description entries according to the preliminary result and display them; Respond to the staff's selection of the phenomenon description entries, and determine the fault category of this diagnosis according to the selected phenomenon description entry; Conduct a search in the case library according to the fault category to obtain multiple diagnostic analysis case analysis models corresponding to the fault category.

5. The method according to claim 4, characterized in that, The matching and analyzing the characteristic data of each case analysis model of the fault category with the vehicle-end diagnosis data of this diagnosis, and decomposing this diagnosis into at least one independent fault event, which further includes: Adopt a traversal matching combined with a bubble analysis method to match and analyze the characteristic data of the multiple diagnostic analysis case analysis models with the fault data of this diagnosis, and decompose this diagnosis into at least one independent fault event; If there is unmatched fault data, obtain the function analysis model corresponding to each piece of unmatched fault data.

6. The method according to claim 5, wherein According to the various fault causes associated with the case analysis model corresponding to each independent fault event, conduct a classification analysis to determine the fault cause of this diagnosis, which further includes: Obtain the various fault causes associated with the case analysis model corresponding to each independent fault event, and obtain the various fault causes in the function analysis model corresponding to each piece of unmatched fault data; Classify the various fault causes, and at least determine the fault causes with an intersection as the fault causes of this diagnosis.

7. The method according to claim 6, wherein Further include: If a preliminary result, fault classification, or fault cause cannot be obtained, add a label to this diagnosis and submit it to a human for analysis and judgment.

8. The method according to claim 1 or 2, characterized in that, Further include: When periodically monitoring the vehicle for faults, through the analysis of the vehicle's description information matrix, when it is considered that the vehicle has a fault trend, according to the preset function analysis model, determine that the change trend of the precondition is most relevant to the change trend of the description signal, and obtain potential fault causes.

9. The method according to claim 8, wherein According to the preset function analysis model, determine that the change trend of the precondition is most relevant to the change trend of the description signal, and obtain potential fault causes, which at least includes: Analyze the data in the description matrix, confirm the vehicle signals with a fault trend, and according to the function analysis model corresponding to the vehicle signals, calculate the covariance result between the vehicle signals and the corresponding preconditions. If the covariance result is the first predetermined positive number or the first predetermined negative number, it is determined that the vehicle signal is correlated with the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted; Among them, the covariance Cov(X,Y) calculated by the following formula: Cov(X,Y) = E[XY] - E[X]E[Y]; Among them, X and Y are two random signal variables to be detected, that is, the vehicle signal to be detected and the precondition; E[X] is the expected value of a random signal variable; E[Y] is the expected value of another random signal variable; E[XY] represents the product of the expected values of the random signal variables X and Y.

10. The method according to claim 9, wherein According to the preset function analysis model, determine that the change trend of the precondition is most relevant to the change trend of the description signal, and obtain potential fault causes, which further includes: Calculate the correlation coefficient of the covariance between the vehicle signal and the corresponding precondition; if the correlation coefficient is greater than the second predetermined positive number, it is determined that the vehicle signal is correlated with the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted; Among them, the correlation coefficient Corr(X,Y) of the covariance calculated by the following formula: Among them, X and Y are two random signal variables to be detected, that is, the vehicle signal to be detected and the precondition; D[X] is the variance of a random signal variable; E[Y] is the variance of another random signal variable; Cov(X,Y) is the covariance of the two random signal variables.

11. A vehicle remote intelligent diagnosis system, characterized in that, At least include the following steps: A vehicle data acquisition unit, configured to receive a vehicle fault diagnosis request or obtain the vehicle's overall vehicle status data and vehicle-end diagnosis data according to periodic triggering; A preliminary result analysis unit, configured to screen and extract description signals from the overall vehicle status data to form a description signal matrix; and analyze the description signal matrix to obtain a preliminary result; A fault category determination unit, configured to form multiple phenomenon description entries respectively associated with each description information according to the preliminary result, respond to the selection of the phenomenon description entries by the staff, determine the fault category of this diagnosis, and determine the range of the diagnostic analysis case analysis model corresponding to the fault category in the case library; A fault event decomposition unit, configured to perform a matching analysis on the characteristic data of each case analysis model of the fault category and the vehicle-end diagnosis data of this diagnosis, and decompose this diagnosis into at least one independent fault event; A fault cause induction unit, configured to perform a classification analysis according to the various fault causes associated with the case analysis model corresponding to each independent fault event, and determine the fault cause of this diagnosis.

12. The system according to claim 11, wherein, The preliminary result analysis unit further includes: A description signal matrix establishment unit, configured to screen and extract some key description signals that can express each electrical system and functional module of the vehicle from the overall vehicle status data according to a preset extraction rule, and fill them into a general description signal matrix to form the description signal matrix of this diagnosis; A comparison and analysis unit, configured to perform a comparison and verification on each signal of the description signal matrix according to a preset analysis rule and a standard threshold, and obtain a preliminary result according to each description signal that exceeds the standard value.

13. The system according to claim 12, wherein, The fault category determination unit further includes: A phenomenon description entry association unit, configured to obtain multiple pre-associated phenomenon description entries according to the preliminary result and display them; A category determination unit, configured to respond to the selection of the phenomenon description entries by the staff, and determine the fault category of this diagnosis according to the selected phenomenon description entries; A case analysis model selection unit, configured to retrieve in the case library according to the fault category and obtain multiple diagnostic analysis case analysis models corresponding to the fault category.

14. The system according to claim 13, wherein The fault event decomposition unit further includes: A matching analysis unit, configured to adopt a traversal matching combined with a bubble analysis method to perform a matching analysis on the characteristic data of the multiple diagnostic analysis case analysis models and the fault data of this diagnosis, and decompose this diagnosis into at least one independent fault event; A function analysis model determination unit, configured to obtain the function analysis model corresponding to each unmatched fault data when there is unmatched fault data in the matching result.

15. The system according to claim 14, wherein The fault cause induction unit further includes: A fault cause acquisition unit, configured to acquire the various fault causes in the case analysis model corresponding to each independent fault event, and acquire the various fault causes in the function analysis model corresponding to each unmatched fault data; An induction processing unit, configured to classify the various fault causes, and at least determine the fault causes with an intersection as the fault causes of this diagnosis.

16. The system according to any one of claims 11 to 15, characterized in that, Further includes: A fault trend analysis unit, which is used to analyze the description information matrix of the vehicle when periodically monitoring the faults of the vehicle. When it is considered that the vehicle has a fault trend, according to a preset function analysis model, it determines that the change trend of the precondition has the highest correlation with the change trend of the description signal, and obtains potential fault causes; Among them, according to the preset function analysis model, determining that the change trend of the precondition has the highest correlation with the change trend of the description signal and obtaining potential fault causes specifically includes: Analyze the data in the description matrix to confirm the vehicle signals with fault trends. According to the function analysis model corresponding to the vehicle signals, calculate the covariance result between the vehicle signals and the corresponding preconditions. If the covariance result is a first predetermined positive number or a first predetermined negative number, it is determined that there is a correlation between the vehicle signal and the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted; or Calculate the correlation coefficient of the covariance between the vehicle signal and the corresponding precondition; if the correlation coefficient is greater than a second predetermined positive number, it is determined that there is a correlation between the vehicle signal and the precondition, and the precondition is determined as a potential fault cause, and a corresponding inspection is prompted; Among them, the covariance Cov(X,Y) obtained by using the following formula: Cov(X,Y) = E[XY] - E[X]E[Y]; The correlation coefficient Corr(X,Y) of the covariance obtained by using the following formula: Among them, X and Y are two random signal variables to be detected, that is, the vehicle signal to be detected and the precondition; E[X] is the expected value of the random signal variable; E[Y] is the expected value of the random signal variable; E[XY] represents the product of the expected values of the random signal variables X and Y; D[X] is the variance of a random signal variable; E[Y] is the variance of another random signal variable; Cov(X,Y) is the covariance of the two random signal variables.

17. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 10.

18. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 10.

Images