Diagnostic pattern prediction method, vehicle diagnostic method, and related devices

Abstract

Embodiments of the present application disclose a diagnosis mode prediction method, a vehicle diagnosis method and related equipment, which are used to improve the accuracy of diagnosis mode prediction. The method of the embodiments of the present application comprises: acquiring current multi-dimensional data affecting diagnosis mode selection, wherein the diagnosis mode selection refers to a data processing strategy of a diagnosis connector for data sent by a diagnosis device and a vehicle to be diagnosed; inputting the current multi-dimensional data into a pre-trained target prediction model to obtain a prediction probability corresponding to each selectable diagnosis mode output by the target prediction model, wherein a diagnosis success rate of the diagnosis device and the vehicle to be diagnosed using each selectable diagnosis mode is positively correlated with the prediction probability corresponding to each selectable diagnosis mode, and the target prediction model is trained based on a plurality of historical diagnosis samples; and determining a target diagnosis mode from a plurality of selectable diagnosis modes according to the prediction probability corresponding to each selectable diagnosis mode output by the target prediction model.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the field of vehicle diagnosis, and in particular to a diagnosis mode prediction method, a vehicle diagnosis method, and related equipment. BACKGROUND

[0002] Automobile remote diagnosis has high requirements for network latency. If the diagnosis times out, it is likely to cause diagnosis failure. As is known to all, the network latency is affected by various conditions and is uncontrollable.

[0003] To solve this problem, remote diagnosis develops different diagnosis modes for users to choose. The existing method is to determine the current network latency when the to-be-diagnosed vehicle and the diagnosis equipment handshake, and after the handshake succeeds, according to any preset latency interval in which the current network latency is located, to select a unique diagnosis mode corresponding to the preset latency interval as the default diagnosis mode, and to use the default diagnosis mode in the first diagnosis process. If the first remote diagnosis fails, the user will try to manually switch to other modes after the failure.

[0004] However, in actual use, different vehicles, different diagnosis equipment, and different diagnosis projects have different requirements for latency. Different countries have different network facilities, and the network latency and fluctuation are also different. Therefore, the existing diagnosis mode prediction method based on the current network latency as a single dimension cannot determine the most suitable default diagnosis mode for the to-be-diagnosed vehicle and the diagnosis equipment under the current conditions. That is, the default diagnosis mode obtained by the existing diagnosis mode prediction method is usually not accurate (that is, it is usually not the optimal diagnosis mode), and the user still needs to switch the mode and retry after the default diagnosis mode fails, which affects the user experience. SUMMARY

[0005] Embodiments of the present application provide a diagnosis mode prediction method, a vehicle diagnosis method, and related equipment to improve the accuracy of diagnosis mode prediction.

[0006] A first aspect of embodiments of the present application provides a diagnosis mode prediction method applied to a diagnosis joint, comprising:

[0007] Obtaining current multi-dimensional data affecting diagnosis mode selection, the diagnosis mode selection referring to a data processing strategy of the diagnosis joint for data sent by a diagnosis equipment and a to-be-diagnosed vehicle;

[0008] inputting the current multidimensional data into a pre-trained target prediction model to obtain a prediction probability corresponding to each selectable diagnosis mode output by the target prediction model, wherein a diagnosis success rate of the diagnosis device and the vehicle to be diagnosed using each selectable diagnosis mode is positively correlated with the prediction probability corresponding to each selectable diagnosis mode, and the target prediction model is trained based on a plurality of historical diagnosis samples;

[0009] determining a target diagnosis mode from the plurality of selectable diagnosis modes according to the prediction probability corresponding to each selectable diagnosis mode output by the target prediction model.

[0010] In an implementation manner, the method further includes:

[0011] obtaining the plurality of historical diagnosis samples, wherein each historical diagnosis sample includes historical multidimensional data affecting a historical communication delay of each historical diagnosis, and a historical diagnosis mode adopted by the each historical diagnosis;

[0012] determining each historical diagnosis sample as a target diagnosis sample in turn;

[0013] inputting the historical multidimensional data and the historical diagnosis mode included in the target diagnosis sample into an initial prediction model to obtain a training probability corresponding to each selectable diagnosis mode output by the prediction model;

[0014] calculating a loss value according to the training probability corresponding to each selectable diagnosis mode output by the prediction model and the historical diagnosis mode corresponding to the target diagnosis sample;

[0015] if the loss value satisfies a preset convergence condition, determining that the initial prediction model is the target prediction model.

[0016] In an implementation manner, the calculating a loss value according to the training probability corresponding to each selectable diagnosis mode output by the prediction model and the historical diagnosis mode corresponding to the target diagnosis sample includes:

[0017] determining a selectable diagnosis mode with the highest training probability in the each selectable diagnosis mode as a prediction diagnosis mode;

[0018] calculating a prediction mode encoding of the prediction diagnosis mode according to a preset encoding mode, and calculating a real mode encoding of the historical diagnosis mode corresponding to the target diagnosis sample according to the preset encoding mode;

[0019] taking a difference between the prediction mode encoding and the real mode encoding as the loss value.

[0020] In an implementation manner, the current multidimensional data includes a current network delay, and the method further includes:

[0021] sending a latency test request to an associated diagnostic connector and receiving a latency test reply sent by the associated diagnostic connector, the diagnostic connector and the associated diagnostic connector being configured to transceive data between the vehicle to be diagnosed and the diagnostic device;

[0022] determining the current network latency as a time duration from sending the latency test request to receiving the latency test reply.

[0023] In an implementation, the current multi-dimensional data includes a current network latency, and the method further comprises:

[0024] determining an average network latency as an arithmetic mean of historical network latencies contained in the historical multi-dimensional data in each group of historical diagnostic samples;

[0025] calculating a standardized network latency according to the average network latency and the current network latency;

[0026] the inputting the current multi-dimensional data into a pre-trained target prediction model comprises:

[0027] the inputting the standardized network latency and the current other data except the current network latency in the current multi-dimensional data into the pre-trained target prediction model.

[0028] In an implementation, the current multi-dimensional diagnostic data includes at least two of a current network latency, a region where the vehicle to be diagnosed is located, a network mode of the vehicle to be diagnosed, vehicle information of the vehicle to be diagnosed, a region where the diagnostic device is located, a network mode of the diagnostic device, and an expected diagnosis item.

[0029] A second aspect of the embodiments of the present application provides a vehicle diagnosis method applied to a diagnostic connector, comprising:

[0030] obtaining a target diagnosis mode, the target diagnosis mode being determined according to the method of any one of the first aspect;

[0031] sending a vehicle diagnosis request to a vehicle to be diagnosed in response to a vehicle diagnosis request sent by a diagnostic device, the vehicle diagnosis request being initiated by the diagnostic device to the vehicle to be diagnosed, the diagnostic connector being configured to transceive data between the diagnostic device and the vehicle to be diagnosed;

[0032] processing the vehicle diagnosis request and / or a vehicle diagnosis reply sent by the vehicle to be diagnosed according to the target diagnosis mode, the vehicle diagnosis request and the vehicle diagnosis reply corresponding to each other.

[0033] A third aspect of the embodiments of the present application provides a diagnostic connector, comprising:

[0034] an acquisition unit, configured to acquire current multidimensional data that affects selection of a diagnostic mode, the diagnostic mode being a data processing strategy of the diagnostic joint for data sent by a diagnostic device and a vehicle to be diagnosed;

[0035] a prediction unit, configured to input the current multidimensional data into a target prediction model that is pre-trained, to obtain a prediction probability corresponding to each selectable diagnostic mode output by the target prediction model, and to determine that a diagnostic success rate of the diagnostic device and the vehicle to be diagnosed using the each selectable diagnostic mode is positively correlated with the prediction probability corresponding to the each selectable diagnostic mode;

[0036] a determination unit, configured to determine a target diagnostic mode from a plurality of selectable diagnostic modes according to the prediction probability corresponding to each selectable diagnostic mode output by the target prediction model.

[0037] In a specific implementation manner, the device further includes a calculation unit;

[0038] The acquisition unit is further configured to acquire a plurality of historical diagnostic samples, where each historical diagnostic sample includes historical multidimensional data that affects a historical communication delay of each historical diagnosis, and a historical diagnostic mode used in the each historical diagnosis.

[0039] The determination unit is further configured to determine each historical diagnostic sample as a target diagnostic sample in turn.

[0040] The prediction unit is further configured to input the historical multidimensional data and the historical diagnostic mode included in the target diagnostic sample into an initial prediction model, to obtain a training probability corresponding to each selectable diagnostic mode output by the prediction model.

[0041] The calculation unit is configured to calculate a loss value according to the training probability corresponding to each selectable diagnostic mode output by the prediction model and the historical diagnostic mode corresponding to the target diagnostic sample.

[0042] The determination unit is further configured to determine that the initial prediction model is the target prediction model if the loss value satisfies a preset convergence condition.

[0043] In a specific implementation manner, the calculation unit is specifically configured to determine a selectable diagnostic mode with a highest training probability in the each selectable diagnostic mode as a prediction diagnostic mode.

[0044] The prediction mode encoding of the prediction diagnostic mode is calculated according to a preset encoding mode, and real mode encoding of the historical diagnostic mode corresponding to the target diagnostic sample is calculated according to the preset encoding mode.

[0045] The difference between the predicted mode encoding and the real mode encoding is taken as the loss value.

[0046] In an implementation, the current multi-dimensional data includes a current network latency, and the device further comprises a testing unit and a determining unit.

[0047] The testing unit is configured to send a latency testing request to an associated diagnostic connector of the diagnostic connector, and receive a latency testing reply sent by the associated diagnostic connector, the diagnostic connector and the associated diagnostic connector being configured to transparently transmit data between the vehicle to be diagnosed and the diagnostic device.

[0048] The determining unit is configured to determine the current network latency as a time length from sending the latency testing request to receiving the latency testing reply.

[0049] In an implementation, the current multi-dimensional data includes a current network latency, and the device further comprises a calculating unit.

[0050] The determining unit is further configured to determine an average network latency as an arithmetic mean of historical network latencies contained in the historical multi-dimensional data in each group of historical diagnosis samples.

[0051] The calculating unit is configured to calculate a standardized network latency according to the average network latency and the current network latency.

[0052] The predicting unit is specifically configured to input the standardized network latency and current other data except the current network latency in the current multi-dimensional data into a target prediction model pre-trained.

[0053] In an implementation, the current multi-dimensional diagnosis data includes at least two of a current network latency, a region where the vehicle to be diagnosed is located, a network mode of the vehicle to be diagnosed, vehicle information of the vehicle to be diagnosed, a region where the diagnostic device is located, a network mode of the diagnostic device, and an expected diagnosis item.

[0054] The fourth aspect of the embodiments of the present application provides a diagnostic connector, comprising:

[0055] An obtaining unit is configured to obtain a target diagnosis mode, the target diagnosis mode being determined according to the method of any one of the first aspect.

[0056] A sending unit is configured to send a vehicle diagnosis request to a vehicle to be diagnosed in response to a vehicle diagnosis request sent by a diagnostic device, the vehicle diagnosis request being initiated by the diagnostic device to the vehicle to be diagnosed, the diagnostic connector being configured to transparently transmit data between the diagnostic device and the vehicle to be diagnosed.

[0057] a processing unit configured to process the vehicle diagnosis request and / or a vehicle diagnosis reply sent by the vehicle to be diagnosed according to the target diagnosis mode, the vehicle diagnosis request corresponding to the vehicle diagnosis reply.

[0058] The fifth aspect of the embodiments of the present application provides a diagnosis connector, comprising:

[0059] a central processing unit, a memory and an input / output interface;

[0060] The memory is a volatile memory or a persistent memory.

[0061] The central processing unit is configured to communicate with the memory and execute instruction operations in the memory to perform the method of the first aspect or the second aspect.

[0062] The sixth aspect of the embodiments of the present application provides a computer program product comprising instructions, which, when executed on a computer, cause the computer to perform the method of the first aspect or the second aspect.

[0063] The seventh aspect of the embodiments of the present application provides a computer storage medium, which stores instructions, and the instructions, when executed on a computer, cause the computer to perform the method of the first aspect or the second aspect.

[0064] From the above technical solutions, it can be seen that the embodiments of the present application have the following advantages: current multi-dimensional data affecting diagnosis mode selection is obtained, the diagnosis mode selection refers to a data processing strategy of the diagnosis connector for a diagnosis device and data sent by a vehicle to be diagnosed; the current multi-dimensional data is input into a target prediction model pre-trained, a prediction probability corresponding to each selectable diagnosis mode output by the target prediction model is obtained, a diagnosis success rate of the diagnosis device and the vehicle to be diagnosed using each selectable diagnosis mode is positively correlated with the prediction probability corresponding to each selectable diagnosis mode; and the target diagnosis mode is determined from the plurality of selectable diagnosis modes according to the prediction probability corresponding to each selectable diagnosis mode output by the target prediction model. Therefore, by training a plurality of historical diagnosis samples, the target prediction model is obtained, then the current multi-dimensional data is input into the target prediction model, and the prediction effect obtained is obviously more accurate than simply predicting the diagnosis mode by the current network delay. At the same time, the prediction accuracy of the target diagnosis mode is improved, the number of diagnosis mode adjustments required by the user due to diagnosis failure can be effectively reduced, and the user experience can be effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0065] Figure 1 A system architecture diagram of the remote diagnosis system disclosed by the embodiments of the present application;

[0066] Figure 2A flowchart of a diagnostic mode prediction method disclosed in embodiments of the present application;

[0067] Figure 3 A structural diagram of a neural network model disclosed in embodiments of the present application;

[0068] Figure 4 A flowchart of model training disclosed in embodiments of the present application;

[0069] Figure 5 A structural diagram of a diagnostic connector disclosed in embodiments of the present application;

[0070] Figure 6 Another structural diagram of a diagnostic connector disclosed in embodiments of the present application;

[0071] Figure 7 Another structural diagram of a diagnostic connector disclosed in embodiments of the present application. DETAILED DESCRIPTION

[0072] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0073] The embodiments of the present application provide a diagnostic mode prediction method, a vehicle diagnostic method, and related equipment, to improve the accuracy of diagnostic mode prediction.

[0074] Please refer to Figure 1 To better implement the diagnostic prediction method and the vehicle diagnostic method of the embodiments of the present application, the embodiments of the present application provide the following system architecture of a remote diagnostic system, which includes a diagnostic device 105, a first diagnostic connector 104, a server 103 (or platform), a second diagnostic connector 102, and a vehicle to be diagnosed 101. The data generated by the diagnostic device 105 is transmitted by the first diagnostic connector 104, the server 103, and the second diagnostic connector 102 to the vehicle to be diagnosed 101, and the data generated by the vehicle 101 can also be transmitted by the second diagnostic connector 102, the server 103, and the first diagnostic connector 104 to the diagnostic device 105. Among them, the first diagnostic connector 104 and the second diagnostic connector 102 can respectively interact with the vehicle to be diagnosed 101 and the diagnostic device 105 through the OBD interface to exchange data.

[0075] The first diagnostic adapter 104 and the second diagnostic adapter 102 can be common diagnostic adapters or diagnostic adapters with display modules, which are not specifically limited herein. In addition, the embodiments of the present application do not limit the communication protocol between the first diagnostic adapter 104 and the second diagnostic adapter 102. For example, under the system architecture described above, the first diagnostic adapter 104 and the second diagnostic adapter 102 can communicate using the transmission control protocol (TCP). In the scenario where the first diagnostic adapter 104 and the second diagnostic adapter 102 communicate using the user datagram protocol (UDP), the first diagnostic adapter and the second diagnostic adapter can directly communicate without relying on the server 103 to forward, that is, in the scenario of using UDP communication, the system architecture of the remote diagnosis system can only include the diagnostic device 105, the first diagnostic adapter 104, the second diagnostic adapter 102, and the vehicle to be diagnosed 101.

[0076] Please refer to the following Figure 2 On the basis of the foregoing remote diagnosis system architecture, the embodiments of the present application provide a diagnosis mode prediction method, which is executed by the foregoing first diagnostic adapter 104 or second diagnostic adapter 102, and includes the following steps:

[0077] 201. Obtain current multi-dimensional data that affects diagnosis mode selection, where diagnosis mode selection refers to a data processing strategy of the diagnostic adapter for data sent by the diagnostic device and the vehicle to be diagnosed.

[0078] To better illustrate the technical solutions of the embodiments of the present application, the diagnosis mode referred to by the embodiments of the present application will be described first. In fact, the diagnosis mode is the processing of the first diagnostic adapter 104 for the vehicle diagnosis request and / or the vehicle diagnosis reply after receiving the vehicle diagnosis request, which is to ensure the normal operation of vehicle diagnosis, that is, to avoid vehicle diagnosis failure. Specifically, diagnosis mode A can be that the first diagnostic adapter 104 sends a response to the diagnostic device after receiving the vehicle diagnosis request sent by the diagnostic device, so as to make the diagnostic device prolong the waiting time and avoid diagnosis failure; or diagnosis mode B can be that the first diagnostic adapter 104 sends multiple vehicle diagnosis replies to the diagnostic device in turn after receiving the multiple vehicle diagnosis replies sent by the diagnostic device, so as to make the diagnostic device receive the vehicle diagnosis reply within a specified time and avoid diagnosis failure.

[0079] Based on existing diagnostic pattern prediction methods, there is a wealth of data (i.e., current multidimensional data) that can influence the communication latency between the diagnostic device and the vehicle under diagnosis. Therefore, to better determine the optimal diagnostic pattern (i.e., the default diagnostic pattern between the two) for both the diagnostic device and the vehicle under diagnosis given the current network latency, diagnostic pattern prediction should be based on multidimensional rather than single-dimensional data as much as possible. Details of the multidimensional data that affects communication latency will be discussed later.

[0080] The existing technical solutions have the following shortcomings: 1. As mentioned earlier, different vehicles and different original equipment manufacturers (OEMs) have different definitions of timeout. For example, some manufacturers consider 55 milliseconds to be a timeout, while others consider 100 milliseconds to be a timeout. Obviously, it is difficult to accurately determine which mode to use based solely on timeout. 2. Network conditions vary from country to country. For example, in China, when the network latency is 80 milliseconds, most vehicle models can meet the requirements by using proxy flow control. However, in some sparsely populated countries with outdated network equipment, although the latency is still 80 seconds, the network fluctuations are greater. Therefore, proxy flow control may not be able to solve the problem, and other diagnostic modes need to be tried. 3. Different service items have different latency requirements. For example, programming and reading fault codes have different latency requirements, with programming having higher requirements for network latency.

[0081] Therefore, in some specific implementations, the current multidimensional diagnostic data includes, but is not limited to, the current network latency, the region where the vehicle to be diagnosed is located, the network mode of the vehicle to be diagnosed, the vehicle information of the vehicle to be diagnosed, the region where the diagnostic equipment is located, the network mode of the diagnostic equipment, and at least two of the expected diagnostic items. Among them, the network mode includes, but is not limited to, WiFi mode or mobile network mode; the vehicle information includes, but is not limited to, vehicle brand, model, and year.

[0082] 202. Input the current multidimensional data into the pre-trained target prediction model to obtain the prediction probability corresponding to each optional diagnostic mode output by the target prediction model. The diagnostic success rate of the diagnostic equipment and the vehicle to be diagnosed using each optional diagnostic mode is positively correlated with the prediction probability corresponding to each optional diagnostic mode. The target prediction model is trained based on multiple sets of historical diagnostic samples.

[0083] The target prediction model is trained based on multiple sets of historical diagnostic samples. The input to the target prediction model is multidimensional data (such as the current multidimensional data), and the output is the predicted probability of each optional diagnostic mode. Therefore, by inputting the current multidimensional data into the pre-trained target prediction model, the predicted probability corresponding to each optional diagnostic mode output by the target prediction model can be obtained.

[0084] It should be noted that the target prediction model considers that the prediction probability corresponding to any optional diagnosis mode is the diagnosis success rate of the diagnosis equipment and the vehicle to be diagnosed when selecting the optional diagnosis mode as the target diagnosis mode (or the default diagnosis mode) under the current multi-dimensional data. That is, the prediction probability of each optional diagnosis mode output by the target prediction model can be understood as a recommended index of the target prediction model for using different optional diagnosis modes by the diagnosis equipment and the vehicle to be diagnosed under the current multi-dimensional data. Therefore, the higher the prediction probability is, the more reliable the optional diagnosis mode corresponding to the target prediction model is considered to be.

[0085] It can be understood that the target prediction model can be trained by the server or any diagnosis joint and saved locally in the diagnosis joint, and is not specifically limited here.

[0086] 203. Determine the target diagnosis mode from the plurality of optional diagnosis modes according to the prediction probability corresponding to each optional diagnosis mode output by the target prediction model.

[0087] According to the embodiments of the foregoing step 202, the higher the prediction probability is, the more reliable the optional diagnosis mode corresponding to the target prediction model is considered to be. Generally, the embodiments of the present application will generally determine the optional diagnosis mode corresponding to the highest prediction probability from a plurality of prediction probabilities as the target diagnosis mode (i.e., the default diagnosis mode) of the diagnosis equipment and the vehicle to be diagnosed under the current multi-dimensional data.

[0088] In the embodiments of the present application, therefore, the target prediction model is obtained by training a plurality of historical diagnosis samples, and then the current multi-dimensional data is input into the target prediction model, and the prediction effect obtained is obviously more accurate than simply predicting the diagnosis mode by the current network delay. At the same time, the prediction accuracy of the target diagnosis mode is improved, the number of diagnosis mode adjustments required by the user due to diagnosis failure can be effectively reduced, and the user experience can be effectively improved.

[0089] In some specific implementations, before step 202 of the present application, the initial prediction model can also be trained by historical diagnosis samples to obtain a target prediction model with high accuracy through the following steps: obtaining a plurality of sets of historical diagnosis samples, wherein each set of historical diagnosis samples includes historical multi-dimensional data affecting the historical communication time delay of each historical diagnosis, and a historical diagnosis mode adopted in each historical diagnosis; determining each set of historical diagnosis samples as a target diagnosis sample in turn; inputting the historical multi-dimensional data and the historical diagnosis mode included in the target diagnosis sample into the initial prediction model to obtain a training probability corresponding to each selectable diagnosis mode output by the prediction model; calculating a loss value according to the training probability corresponding to each selectable diagnosis mode output by the prediction model and the historical diagnosis mode corresponding to the target diagnosis sample; and if the loss value meets a preset convergence condition, determining the initial prediction model as the target prediction model. The historical diagnosis refers to a diagnosis operation that is successfully completed before step 202 of the present application. The embodiments related to model training of the present application can be executed by any device with training capability, such as a diagnosis joint that executes diagnosis mode prediction or a server, which is not limited here.

[0090] Specifically, when each historical diagnosis should record the historical multi-dimensional data that will affect the historical communication time delay of the diagnosis and the historical diagnosis mode used when the diagnosis is finally successful. Then, the initial prediction model constructed in advance is trained through these historical diagnosis samples until the model converges (that is, when the loss value calculated according to the training probability corresponding to each selectable diagnosis mode output by the prediction model and the historical diagnosis mode corresponding to the target diagnosis sample meets the preset convergence condition), it is considered that the initial prediction model at this time meets the requirements, that is, the initial prediction model at this time is determined as the target prediction model. If the model has not converged, a set of historical diagnosis samples that have not been used for training is selected from the plurality of sets of historical diagnosis samples, and the model is trained for a new round.

[0091] It should be noted that the network structure of the neural network model (i.e., the initial network model) used in the embodiments of the present application can refer to Figure 3The initial network model provided in the application is composed of a first fully connected layer, a second fully connected layer, and a third fully connected layer, which are used for feature extraction with different precisions and different channels. Specifically, the first fully connected layer is composed of an activation function Relu, the second fully connected layer is composed of an activation function Relu, and the third fully connected layer is composed of an activation function sotfmax. The model input is multi-dimensional diagnostic data (such as historical diagnostic data and / or current multi-dimensional data), and the output is the training probability and / or prediction probability of different selectable diagnostic modes corresponding to the input. In addition, the model loss function (or classifier loss function) adopts Categorical Cross-entropy, and the optimizer of the model adopts Stochastic Gradient Descent (SDG) or its further derived optimization algorithms such as Momentum, AdaGrad or RMSProp, and / or Adam algorithm, etc.

[0092] Further, the aforementioned loss value calculation method can refer to the following steps: determining the selectable diagnostic mode with the highest training probability in each selectable diagnostic mode as the predicted diagnostic mode; calculating the prediction mode code of the predicted diagnostic mode according to the preset encoding mode, and calculating the true mode code of the historical diagnostic mode corresponding to the target diagnostic sample according to the preset encoding mode; taking the difference between the prediction mode code and the true mode code as the loss value. It can be understood that the loss value calculation method here is the calculation method of the loss function, which will not be described here.

[0093] Specifically, in order to better enable the model to learn the differences between different diagnostic modes and make better predictions, in addition to directly taking the difference between the predicted diagnostic mode and the historical diagnostic mode corresponding to the target diagnostic sample as the loss value, the predicted diagnostic mode and the corresponding historical diagnostic mode can also be processed using a preset encoding mode, and the difference between the corresponding encoding values of the two is calculated as the loss value. Compared with the method of directly calculating the difference between the two text, the method of calculating the difference between the corresponding encoding values of the two is a numerical variable, which is more conducive to model learning. The preset encoding mode includes but is not limited to one-hot encoding and target encoding, and different encoding modes have their own characteristics and can be selected as needed. Using one-hot encoding can make different selectable diagnostic modes in an "equal position" and not be affected by the size of the numerical value when predicting different diagnostic modes.

[0094] It can be understood that in other implementations, when there are a large number of historical diagnosis samples, in order to ensure the training mode of the model, a plurality of groups of historical diagnosis samples can be respectively divided into a training set, a validation set and a test set according to a preset grouping ratio for cross-validation learning. For example, if the number of samples is more than ten thousand, the data can be randomly divided into three parts according to the preset grouping ratio, one part is a training set (Training Set), one part is a validation set (Validation Set), and the last part is a test set (Test Set). The training set is used to train the model, and the validation set is used to evaluate the prediction of the model and select the model and the corresponding parameters. The final obtained model is used for the test set again, and finally it is determined which model and corresponding parameters are used. The preset grouping ratio indicates the proportion of the number of historical diagnosis samples in the training set, the validation set and the test set to the total number of historical diagnosis samples. For details, refer to the training process shown in Figure 4

[0095] The foregoing describes the model training and model application and other related embodiments of the present application. The following describes a specific determination method of the current network delay when the current multi-dimensional data includes the current network delay. For details, see the following steps: sending a delay test request to an associated diagnosis connector of the diagnosis connector, and receiving a delay test reply sent by the associated diagnosis connector, the diagnosis connector and the associated diagnosis connector being used for transparently transmitting data between the vehicle to be diagnosed and the diagnosis equipment; determining the current network delay as the time length from sending the delay test request to receiving the delay test reply.

[0096] Specifically, in order to realize diagnosis mode prediction without the awareness of the diagnosis equipment, the vehicle and the user, any diagnosis connector (the first diagnosis connector 104 and the second diagnosis connector 102) can initiate the above-mentioned delay test request after receiving the vehicle diagnosis reply or establishing a communication connection with the associated diagnosis connector, which is not limited here. Each diagnosis connector under the same diagnosis system architecture is an associated diagnosis connector.

[0097] The diagnosis equipment and the vehicle to be diagnosed can agree on a delay test instruction (i.e. delay test request). Whenever the diagnosis equipment sends a delay test instruction, the vehicle to be diagnosed directly sends a corresponding reply instruction (i.e. delay test reply) back to the diagnosis equipment. In this way, the diagnosis equipment can quickly determine the current network delay according to the time length from sending the delay test instruction to receiving the corresponding reply instruction.

[0098] ​Further, considering that the multi-dimensional data affecting the communication delay at least includes one other data in addition to the network delay, which can reflect the communication delay, in order to ensure that the influence of each dimension data on the prediction probability is balanced, the current network delay needs to be standardized here. This standardization can be completed before the historical diagnosis data is input during the model training process (processing the historical network delay in the historical diagnosis data), or before the current multi-dimensional data is input during the model application process (processing the current network delay in the current multi-dimensional data), which is not limited here. Taking processing the current network delay in the current multi-dimensional data as an example, see the following steps: the arithmetic mean of the historical network delay contained in the historical multi-dimensional data in each group of historical diagnosis samples is determined as the average network delay; the standardized network delay is calculated according to the average network delay and the current network delay; after the standardized network delay is determined, the input of the model becomes: the standardized network delay and the current other data in the current multi-dimensional data except the current network delay.

[0099] Specifically, the calculation formula is:

[0100] The foregoing describes various embodiments of the diagnosis mode prediction method of the present application. The following describes a specific way to implement vehicle diagnosis after determining the target diagnosis mode according to the various embodiments of the diagnosis mode prediction method described above, including the following steps: obtaining the target diagnosis mode, which is determined according to the diagnosis mode prediction method described above; in response to the vehicle diagnosis request sent by the diagnosis device, sending a vehicle diagnosis request to the vehicle to be diagnosed, the vehicle diagnosis request being initiated by the diagnosis device to the vehicle to be diagnosed, and the diagnosis connector being used to transmit data between the diagnosis device and the vehicle to be diagnosed; processing the vehicle diagnosis request and / or the vehicle diagnosis reply sent by the vehicle to be diagnosed according to the target diagnosis mode, the vehicle diagnosis request corresponding to the vehicle diagnosis reply.

[0101] That is, after determining the target diagnosis mode, each diagnosis connector needs to be informed so that each diagnosis connector can respond in a timely manner according to the corresponding diagnosis model after receiving the vehicle diagnosis request sent by the diagnosis device and / or the vehicle diagnosis reply sent by the vehicle to be diagnosed.

[0102] The following describes the diagnosis mode prediction method and the vehicle diagnosis method of the present application in a specific scenario.

[0103] 1. Data collection

[0104] Considering that a successfully completed historical diagnosis usually has a longer diagnosis duration and a larger amount of data transmitted and received than an uncompleted diagnosis, the platform (or server) extracts historical data with a diagnosis duration greater than a preset duration threshold and an amount of data transmitted and received greater than a preset data amount threshold from the historical data as historical diagnosis samples.

[0105] The historical diagnosis data of each set of historical diagnosis samples includes a current network delay, a region where the first diagnosis connector is located, a region where the second diagnosis connector is located, a network mode of the diagnosis device, a network mode of the vehicle to be diagnosed, an expected diagnosis item, and vehicle information of the vehicle to be diagnosed; and a historical diagnosis mode of each set of historical diagnosis samples.

[0106] 2. Data preprocessing

[0107] 2-1 The non-continuous data such as a region where the first diagnosis connector (that is, a region where the diagnosis device is located) is located, a region where the second diagnosis connector (that is, a region where the vehicle to be diagnosed is located) is located, a network mode of the diagnosis device (that is, a network mode of the first diagnosis connector), a network mode of the vehicle to be diagnosed (that is, a network mode of the second diagnosis connector), an expected diagnosis item, vehicle information of the vehicle to be diagnosed, and a historical diagnosis mode are converted into one-hot encoding.

[0108] 2-2 The current network delay is subjected to standardization processing, and the calculation formula is referred to the foregoing related embodiments.

[0109] 2-3 The plurality of sets of historical diagnosis samples are divided into a training set, a verification set, and a test set according to a preset grouping ratio.

[0110] 3. Construction of an initial prediction model

[0111] For details, refer to the foregoing related embodiments, which are not described herein again.

[0112] 4. Model training

[0113] For details, refer to the foregoing related embodiments, which are not described herein again.

[0114] 5. Diagnosis mode prediction using the trained model

[0115] After each diagnosis device and the vehicle to be diagnosed successfully handshake, the platform inputs the current multi-dimensional data to the target prediction model for prediction, and sends the target diagnosis mode to the diagnosis connector (that is, the associated diagnosis connector, that is, it is necessary to inform each diagnosis connector under the remote diagnosis system architecture that the target diagnosis mode needs to be known), and the diagnosis connector takes the target diagnosis mode recommended by the platform as the default diagnosis mode. That is, the diagnosis connector will first use the default diagnosis mode to process the vehicle diagnosis request sent by the diagnosis device and / or the vehicle diagnosis reply sent by the vehicle to be diagnosed.

[0116] Please refer to Figure 5 The embodiment of the present application provides a diagnosis connector, which comprises:

[0117] An acquisition unit 501 is configured to acquire current multi-dimensional data affecting diagnosis mode selection, and the diagnosis mode selection refers to a data processing strategy of the diagnosis connector for data sent by a diagnosis device and a vehicle to be diagnosed.

[0118] The prediction unit 502 is configured to input the current multidimensional data into the pre-trained target prediction model to obtain a prediction probability corresponding to each selectable diagnosis mode output by the target prediction model, and the diagnosis success rate of each selectable diagnosis mode used by the diagnosis equipment and the vehicle to be diagnosed is positively correlated with the prediction probability corresponding to each selectable diagnosis mode.

[0119] The determination unit 503 is configured to determine a target diagnosis mode from the plurality of selectable diagnosis modes according to the prediction probability corresponding to each selectable diagnosis mode output by the target prediction model.

[0120] In a specific implementation, the device further includes a calculation unit.

[0121] The acquisition unit 501 is further configured to acquire a plurality of groups of historical diagnosis samples, where each group of historical diagnosis samples includes historical multidimensional data affecting a historical communication delay of each historical diagnosis, and a historical diagnosis mode used in each historical diagnosis.

[0122] The determination unit 503 is further configured to determine each group of historical diagnosis samples as a target diagnosis sample in turn.

[0123] The prediction unit 502 is further configured to input the historical multidimensional data and the historical diagnosis mode included in the target diagnosis sample into the initial prediction model to obtain a training probability corresponding to each selectable diagnosis mode output by the prediction model.

[0124] The calculation unit is configured to calculate a loss value according to the training probability corresponding to each selectable diagnosis mode output by the prediction model and the historical diagnosis mode corresponding to the target diagnosis sample.

[0125] The determination unit 503 is further configured to determine the initial prediction model as the target prediction model if the loss value meets a preset convergence condition.

[0126] In a specific implementation, the calculation unit is specifically configured to determine a selectable diagnosis mode with the highest training probability in each selectable diagnosis mode as the prediction diagnosis mode.

[0127] The prediction mode encoding of the prediction diagnosis mode is calculated according to the preset encoding mode, and the real mode encoding of the historical diagnosis mode corresponding to the target diagnosis sample is calculated according to the preset encoding mode.

[0128] The difference between the prediction mode encoding and the real mode encoding is taken as the loss value.

[0129] In a specific implementation, the current multidimensional data includes a current network delay, and the device further includes a test unit and a determination unit.

[0130] The test unit is configured to send a time delay test request to an associated diagnosis connector of the diagnosis connector, and receive a time delay test reply sent by the associated diagnosis connector, the diagnosis connector and the associated diagnosis connector are configured to transmit data between the vehicle to be diagnosed and the diagnosis device.

[0131] The determination unit is configured to determine the current network delay as a time length from sending the time delay test request to receiving the time delay test reply.

[0132] In a specific implementation, the current multi-dimensional data includes the current network delay, and the device further includes a calculation unit.

[0133] The determination unit 503 is further configured to determine the average network delay as an arithmetic mean value of the historical network delays contained in the historical multi-dimensional data in each group of historical diagnosis samples.

[0134] The calculation unit is configured to calculate the standardized network delay according to the average network delay and the current network delay.

[0135] The prediction unit 502 is specifically configured to input the standardized network delay and other current data in the current multi-dimensional data except the current network delay into the pre-trained target prediction model.

[0136] In a specific implementation, the current multi-dimensional diagnosis data includes at least two of the current network delay, a region where the vehicle to be diagnosed is located, a network mode of the vehicle to be diagnosed, vehicle information of the vehicle to be diagnosed, a region where the diagnosis device is located, a network mode of the diagnosis device, and an expected diagnosis item.

[0137] For details, please refer to Figure 6 The diagnosis connector provided by the embodiment of the present application comprises:

[0138] The acquisition unit 601 is configured to acquire a target diagnosis mode, the target diagnosis mode being determined according to any one of the foregoing diagnosis mode prediction methods.

[0139] The sending unit 602 is configured to send a vehicle diagnosis request to the vehicle to be diagnosed in response to a vehicle diagnosis request sent by the diagnosis device, the vehicle diagnosis request being initiated by the diagnosis device to the vehicle to be diagnosed, and the diagnosis connector being configured to transmit data between the diagnosis device and the vehicle to be diagnosed.

[0140] The processing unit 603 is configured to process the vehicle diagnosis request and / or a vehicle diagnosis reply sent by the vehicle to be diagnosed according to the target diagnosis mode, the vehicle diagnosis request corresponding to the vehicle diagnosis reply.

[0141] Figure 7is a diagnostic adapter structure schematic diagram provided by an embodiment of the present application. The diagnostic adapter 700 can include one or more central processing units (CPU) 701 and a memory 705, which stores one or more application programs or data.

[0142] The memory 705 can be volatile storage or persistent storage. The programs stored in the memory 705 can include one or more modules, each of which can include a series of instruction operations in the diagnostic adapter. Further, the central processing unit 701 can be configured to communicate with the memory 705 and execute the series of instruction operations in the memory 705 on the diagnostic adapter 700.

[0143] The diagnostic adapter 700 can also include one or more power supplies 702, one or more wired or wireless network interfaces 703, one or more input / output interfaces 704, and / or one or more operating systems, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

[0144] The central processing unit 701 can execute the series of instruction operations described above Figures 1 to 6 The diagnostic adapter performs the operations described in the embodiments shown, which will not be described here.

[0145] It should be noted that although the steps in the flowcharts involved in the embodiments are drawn in sequence according to the arrows, unless otherwise stated herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowcharts involved in the embodiments can include multiple steps or stages, which do not necessarily be executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with other steps or steps or stages in other steps.

[0146] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0147] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the described device embodiments are merely schematic. The division of the units is merely logical function division. There can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0148] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0149] In addition, each functional unit in the various embodiments of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0150] The integrated unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application, essentially or the part that contributes to the prior art, or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, read-only memory), a random access memory (RAM, random access memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0151] The embodiments of the present application also provide a computer program product containing instructions, which, when the computer program product is run on a computer, causes the computer to execute the diagnostic mode prediction method and the vehicle diagnostic method as described above.

Claims

1. A diagnostic pattern prediction method characterized by, The method applied to a diagnostic joint comprises: obtaining current multi-dimensional data affecting diagnostic mode selection, the diagnostic mode selection indicating a data processing strategy of the diagnostic joint for data sent by a diagnostic device and a vehicle to be diagnosed; the current multi-dimensional data comprising at least two of current network latency, a region where the vehicle to be diagnosed is located, a network mode of the vehicle to be diagnosed, vehicle information of the vehicle to be diagnosed, a region where the diagnostic device is located, a network mode of the diagnostic device, and an expected diagnostic item; inputting the current multi-dimensional data into a target prediction model pre-trained to obtain a prediction probability corresponding to each selectable diagnostic mode output by the target prediction model, the diagnostic success rate of the diagnostic device and the vehicle to be diagnosed using each selectable diagnostic mode being positively correlated with the prediction probability corresponding to each selectable diagnostic mode, the target prediction model being trained based on a plurality of historical diagnostic samples; determining a target diagnostic mode from a plurality of selectable diagnostic modes according to the prediction probability corresponding to each selectable diagnostic mode output by the target prediction model.

2. The method of claim 1, wherein, The method further comprises: obtaining the plurality of historical diagnostic samples, wherein each historical diagnostic sample comprises historical multi-dimensional data affecting historical communication latency of each historical diagnosis, and a historical diagnostic mode adopted in the each historical diagnosis; determining each historical diagnostic sample as a target diagnostic sample in turn; inputting the historical multi-dimensional data and the historical diagnostic mode included in the target diagnostic sample into an initial prediction model to obtain a training probability corresponding to each selectable diagnostic mode output by the prediction model; calculating a loss value according to the training probability corresponding to each selectable diagnostic mode output by the prediction model and the historical diagnostic mode corresponding to the target diagnostic sample; if the loss value satisfies a preset convergence condition, determining the initial prediction model as the target prediction model.

3. The method of claim 2, wherein, The calculating a loss value according to the training probability corresponding to each selectable diagnostic mode output by the prediction model and the historical diagnostic mode corresponding to the target diagnostic sample comprises: determining a selectable diagnostic mode with the highest corresponding training probability in the each selectable diagnostic mode as a prediction diagnostic mode; calculating a prediction mode code of the prediction diagnostic mode according to a preset coding mode, and calculating a real mode code of the historical diagnostic mode corresponding to the target diagnostic sample according to the preset coding mode; taking a difference value between the prediction mode code and the real mode code as the loss value.

4. The method of claim 1, wherein, The current multi-dimensional data comprises current network latency, and the method further comprises: sending a latency test request to an associated diagnostic joint of the diagnostic joint, and receiving a latency test reply sent by the associated diagnostic joint, the diagnostic joint and the associated diagnostic joint being used to transmit data between the vehicle to be diagnosed and the diagnostic device; determining a time length from sending the latency test request to receiving the latency test reply as the current network latency.

5. The method according to any one of claims 1-4, characterized in that, The current multi-dimensional data comprises current network latency, and the method further comprises: determining an arithmetic mean value of historical network latency contained in the historical multi-dimensional data in each historical diagnostic sample as an average network latency; According to the average network delay and the current network delay, a standardized network delay is calculated; The current multi-dimensional data is input into the pre-trained target prediction model, including: The standardized network delay and the current other data in the current multi-dimensional data except the current network delay are input into the pre-trained target prediction model.

6. A vehicle diagnosis method characterized by, The method is applied to a diagnostic adapter, and the method includes: Obtaining a target diagnostic mode determined according to the method in any one of the preceding claims 1-5; In response to a vehicle diagnostic request sent by a diagnostic device, sending the vehicle diagnostic request to a vehicle to be diagnosed, the vehicle diagnostic request being initiated by the diagnostic device to the vehicle to be diagnosed, the diagnostic adapter being used to transparently transmit data between the diagnostic device and the vehicle to be diagnosed; Processing the vehicle diagnostic request and / or a vehicle diagnostic reply sent by the vehicle to be diagnosed according to the target diagnostic mode, the vehicle diagnostic request corresponding to the vehicle diagnostic reply.

7. A diagnostic linker, characterized in that, The method includes: An acquisition unit is configured to acquire current multi-dimensional data affecting diagnostic mode selection, the diagnostic mode selection indicating a data processing strategy of the diagnostic adapter for data sent by a diagnostic device and a vehicle to be diagnosed, and the current multi-dimensional data including at least two of a current network delay, a region where the vehicle to be diagnosed is located, a network mode of the vehicle to be diagnosed, vehicle information of the vehicle to be diagnosed, a region where the diagnostic device is located, a network mode of the diagnostic device, and an expected diagnostic item; A prediction unit is configured to input the current multi-dimensional data into a pre-trained target prediction model to obtain a prediction probability corresponding to each selectable diagnostic mode output by the target prediction model, the diagnostic success rate of the diagnostic device and the vehicle to be diagnosed using the each selectable diagnostic mode being positively correlated with the prediction probability corresponding to the each selectable diagnostic mode; A determination unit is configured to determine a target diagnostic mode from a plurality of selectable diagnostic modes according to the prediction probability corresponding to each selectable diagnostic mode output by the target prediction model.

8. A diagnostic linker, characterized in that, The method includes: A central processing unit, a memory, and an input / output interface; The memory is a volatile memory or a persistent memory; The central processing unit is configured to communicate with the memory and execute instruction operations in the memory to perform the method in any one of claims 1-5 or 6.

9. A computer storage medium, characterized in that The computer storage medium stores instructions, and the instructions make the computer execute the method in any one of claims 1-5 or 6 when executed on the computer.

10. A computer program product containing instructions, which, when the computer program product is run on a computer, make the computer execute the method in any one of claims 1-5 or 6.

10. A computer program product containing instructions, which, when the computer program product is run on a computer, make the computer execute the method in any one of claims 1-5 or 6.

Images