Vehicle performance prediction method and device and vehicle

Abstract

The invention provides a vehicle performance prediction method and device and a vehicle. The vehicle performance prediction method comprises the following steps: acquiring a historical performance prediction model and real-time performance monitoring data of a vehicle; the historical performance prediction model is used for representing the relationship between the historical performance monitoring data and the performance prediction result of the vehicle; the historical performance prediction model comprises at least one parameter; updating parameters according to the real-time performance monitoring data; substituting the updated parameters into a historical performance prediction model to obtain a real-time performance prediction model; and obtaining a real-time performance prediction result of the vehicle according to the real-time performance prediction model. The parameters of the performance prediction model are dynamically adjusted according to the real-time performance monitoring data, so that the performance prediction model is more suitable for the real-time performance monitoring data, the adaptability and robustness of the performance prediction model can be improved, and the reliability of a performance prediction result is improved.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle performance prediction, and in particular to a vehicle performance prediction method, device and vehicle. Background Art

[0002] In the actual use of the vehicle, it is very important to predict the various performance of the vehicle. The performance prediction results help car owners and drivers understand the vehicle status, and can also provide an important basis for vehicle maintenance, fault warning and performance optimization.

[0003] In the field of vehicle performance prediction technology, the vehicle's performance can be continuously predicted through the historical data of the vehicle's performance. Although this static prediction method can reveal some trends and laws based on a large amount of historical data, its accuracy and practicality are often limited in the face of complex and changeable actual operating environments.

[0004] Since the real-time changes in vehicle performance are affected by a variety of dynamic factors, including but not limited to road conditions, weather conditions, driver behavior, vehicle maintenance status, and the real-time status of the vehicle system, these factors are highly uncertain and time-varying, and it is difficult to fully capture and accurately predict them through a simple historical data regression model.

[0005] Therefore, it is necessary to propose a prediction method to achieve a more real-time and dynamic prediction of vehicle performance. Summary of the invention

[0006] The present application provides a vehicle performance prediction method, device and vehicle that can adapt to dynamic data.

[0007] The present application provides a vehicle performance prediction method, comprising:

[0008] Acquire a historical performance prediction model and real-time performance monitoring data of the vehicle; the historical performance prediction model is used to characterize the relationship between the historical performance monitoring data of the vehicle and the performance prediction result; the historical performance prediction model includes at least one parameter;

[0009] updating the parameters according to the real-time performance monitoring data;

[0010] Substituting the updated parameters into the historical performance prediction model to obtain a real-time performance prediction model;

[0011] According to the real-time performance prediction model, a real-time performance prediction result of the vehicle is obtained.

[0012] Optionally, updating the parameter according to the real-time performance monitoring data includes:

[0013] Determining a likelihood function of the parameter based on the real-time performance monitoring data;

[0014] The parameters are updated by Bayesian theorem according to the real-time performance monitoring data and the likelihood function.

[0015] Optionally, updating the parameter by Bayesian theorem according to the real-time performance monitoring data and the likelihood function includes:

[0016] Determining the posterior distribution of the parameter by Bayes' theorem according to the prior distribution of the parameter, the real-time performance monitoring data and the likelihood function; the prior distribution of the parameter is determined according to the historical performance monitoring data;

[0017] The parameters are updated according to the posterior distribution of the parameters.

[0018] Optionally, the prior distribution of the parameter includes a normal distribution, and the mean and variance of the normal distribution are determined by the historical performance monitoring data.

[0019] Optionally, updating the parameter according to the real-time performance monitoring data includes:

[0020] Determining the posterior distribution of the parameter by variational inference or MCMC method according to the prior distribution of the parameter, the real-time performance monitoring data and the likelihood function; the prior distribution of the parameter is determined according to the historical performance monitoring data;

[0021] The parameters are updated according to the posterior distribution of the parameters.

[0022] Optionally, the method for determining the historical performance prediction model includes:

[0023] Obtaining historical performance monitoring data of the vehicle;

[0024] Mapping the historical performance monitoring data to a quantum state to obtain quantum state data;

[0025] The historical performance prediction model is obtained by processing the quantum state data through quantum entanglement.

[0026] Optionally, the real-time performance monitoring data includes noise; and before updating the parameter according to the real-time performance monitoring data, the performance prediction method further includes:

[0027] determining a prior distribution of the noise;

[0028] Determining a posterior distribution of the noise according to the prior distribution of the noise and the real-time performance monitoring data;

[0029] Determining a noise reduction threshold according to the posterior distribution and mapping function of the noise;

[0030] The real-time performance monitoring data is denoised according to the denoising threshold.

[0031] The present application provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, the vehicle performance prediction method described in any one of the above items is implemented.

[0032] The present application provides a vehicle performance prediction device, comprising one or more processors, for implementing any of the vehicle performance prediction methods described above.

[0033] The present application also provides a vehicle, comprising:

[0034] A sensor for detecting the performance of the vehicle and generating the real-time performance monitoring data; and

[0035] The performance prediction device as described above is electrically connected to the sensor.

[0036] In some embodiments, the historical performance prediction model is used to characterize the relationship between the historical performance monitoring data of the vehicle and the performance prediction results. The historical performance prediction model includes at least one parameter. The parameter is updated according to the real-time performance monitoring data, and the updated parameter is substituted into the historical performance prediction model to obtain the real-time performance prediction model; according to the real-time performance prediction model, the real-time performance prediction results of the vehicle are obtained. According to the real-time performance monitoring data, the parameters of the performance prediction model are dynamically adjusted so that the performance prediction model is more suitable for the real-time performance monitoring data, which can improve the adaptability and robustness of the performance prediction model and improve the reliability of the performance prediction results.

[0037] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0039] Figure 1 Shown is a flow chart of an embodiment of a vehicle performance prediction method of the present application.

[0040] Figure 2 Shown is a structural block diagram of an embodiment of a vehicle performance prediction device of the present application. DETAILED DESCRIPTION

[0041] The present application provides a vehicle performance prediction method, device and vehicle. The vehicle performance prediction method, device and vehicle of the present application are described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the features of the following embodiments and implementations can be combined with each other.

[0042] The present application provides a vehicle, comprising: a sensor and a performance prediction device provided by the present application.

[0043] Sensors are used to detect the performance of the vehicle and generate real-time performance monitoring data.

[0044] Vehicles include a variety of sensors, such as engine status sensors, fuel consumption sensors, temperature sensors, vibration sensors, etc., which are used to detect various performances of the vehicle and generate real-time performance monitoring data that reflects the real-time performance of the vehicle. By comprehensively processing and separating the data collected by various sensors, more comprehensive vehicle performance information can be obtained.

[0045] Engine status sensors include: crankshaft position sensor, used to detect the angular position of the crankshaft to control ignition and injection timing; camshaft position sensor, used to determine valve timing to ensure that the opening and closing of the valve is synchronized with the operating status of the engine; oxygen sensor, installed in the exhaust system, used to detect the oxygen content in the exhaust gas to adjust the air-fuel ratio to achieve efficient combustion; knock sensor, used to monitor whether the engine has knocked, so as to adjust the ignition advance angle when necessary to prevent engine damage. Fuel consumption sensors include: fuel gauge sensor, used to monitor the fuel level in the fuel tank to estimate the remaining fuel. Temperature sensors include: coolant temperature sensor, installed in the cooling system, used to monitor the temperature of the coolant to ensure that the engine runs within the appropriate temperature range; oil temperature sensor, used to detect the temperature of the oil, to help evaluate the performance of the lubrication system and the cleanliness of the oil; intake temperature sensor, installed in the intake system, used to detect the temperature of the air entering the engine to adjust parameters such as the fuel injection amount. Vibration sensors include: acceleration sensors, which are installed in key parts of the vehicle, such as the engine, transmission, suspension system, etc., for real-time monitoring of vibration acceleration and for identifying the source of faults; laser displacement sensors, which are used to measure the vibration displacement of vehicle components to provide vibration information of specific components.

[0046] The performance prediction device is electrically connected to the sensor and is used to predict the performance of the vehicle according to the real-time performance monitoring data generated by the sensor. The performance prediction device is used to execute the vehicle performance prediction method of the present application, thereby predicting the performance of the vehicle.

[0047] Figure 1 FIG. 1 is a flow chart of an embodiment of a vehicle performance prediction method 10 of the present application. Figure 1As shown, the vehicle performance prediction method 10 includes: steps 11 to 14.

[0048] Step 11, obtaining the vehicle's historical performance prediction model and real-time performance monitoring data.

[0049] The historical performance prediction model is used to characterize the relationship between the historical performance monitoring data of the vehicle and the performance prediction result. The historical performance prediction model includes at least one parameter.

[0050] The historical performance monitoring data represents the various performance conditions of the vehicle before the current performance prediction method 10 is executed. The historical performance monitoring data includes but is not limited to the vehicle's fuel efficiency, mileage, maintenance records, fault codes, acceleration performance, braking distance, battery life and other data. These data reflect the performance of the vehicle under different conditions of use. The historical performance monitoring data is collected by the vehicle's sensors and stored in the performance prediction device. Based on the historical performance monitoring data, the performance prediction results can be inferred through the historical performance prediction model. For example, based on the data detected by the engine status sensor, the vehicle's engine efficiency can be predicted; based on the fuel consumption sensor, the future fuel consumption can be predicted; based on the temperature sensor and vibration sensor, the vehicle's component wear rate can be predicted.

[0051] The historical performance prediction model includes multiple models and at least one parameter. For example, the historical performance prediction model includes a linear regression model, as shown in formula (1).

[0052] y=β0+β1x+∈ ​​Formula (1)

[0053] Among them, x is the historical performance monitoring data, y is the performance prediction result, β0, β1 are parameters, and ∈ is noise.

[0054] When the parameters β0, β1, and noise ∈ are determined, the acquired historical performance monitoring data is substituted into the historical performance prediction model to obtain the performance prediction result.

[0055] In some embodiments, a method for determining a historical performance prediction model includes: obtaining historical performance monitoring data of a vehicle; mapping the historical performance monitoring data to a quantum state to obtain quantum state data; and processing the quantum state data through quantum entanglement to obtain a historical performance prediction model.

[0056] Quantum states can represent more information than classical bits. By converting the characteristics or attributes of historical performance monitoring data into quantum states of quantum bits and encoding historical performance monitoring data into quantum states, we can take advantage of quantum computing and process and analyze these data through quantum entanglement, thereby accelerating data processing speed and obtaining a more accurate historical performance prediction model. Quantum entanglement allows multiple quantum bits to operate at the same time, thereby achieving parallel data processing and increasing data processing speed. By using quantum entanglement for encoding, the entangled state that is inseparable during data transmission can be used to detect and prevent eavesdropping and tampering, thereby improving data processing security.

[0057] Real-time performance monitoring data can be obtained through the vehicle's sensors. Real-time performance monitoring data records various performance indicators of the vehicle in real time, reflecting the current actual operating status of the vehicle.

[0058] Step 12: Update parameters based on real-time performance monitoring data.

[0059] Based on the real-time performance monitoring data, the parameters in the historical performance prediction model are adjusted and optimized. When the real-time performance monitoring data changes, the parameters change accordingly, so the parameters can adapt to the changes in the vehicle's performance.

[0060] Step 13, substituting the updated parameters into the historical performance prediction model to obtain a real-time performance prediction model.

[0061] Substituting the updated parameters into the historical performance prediction model, a new and more accurate real-time performance prediction model is obtained. The real-time performance model can better reflect the current actual state of the vehicle and possible performance changes in the future. The real-time performance prediction model adjusts the prediction results as the vehicle's operating state changes, which can ensure the accuracy and reliability of the prediction.

[0062] Step 14, obtaining a real-time performance prediction result of the vehicle according to the real-time performance prediction model.

[0063] Based on the real-time performance prediction model, the real-time performance prediction results of the vehicle are generated, such as remaining mileage, estimated maintenance time, potential fault warning, etc. The performance prediction results can help users understand the vehicle status in a timely manner and make reasonable decisions and arrangements.

[0064] In some embodiments, the parameters of the performance prediction model are dynamically adjusted according to the real-time performance monitoring data, so that the performance prediction model is more suitable for the real-time performance monitoring data, which can improve the adaptability and robustness of the performance prediction model and improve the reliability of the performance prediction results.

[0065] In some embodiments, after step 11, the performance prediction method 10 further includes: filtering the real-time performance monitoring data to achieve preliminary noise reduction.

[0066] In some embodiments, step 12 includes:

[0067] Determine the likelihood function of the parameters based on real-time performance monitoring data;

[0068] Based on the real-time performance monitoring data and the likelihood function, the parameters are updated using the Bayesian theorem.

[0069] The likelihood function represents the probability that the parameters of the historical performance prediction model take a specific value under the condition of given real-time performance monitoring data. In the vehicle performance prediction model, the likelihood function describes the relationship between the real-time performance monitoring data and the parameters of the model. By constructing the likelihood function, the impact of the real-time performance monitoring data on the parameters of the historical performance prediction model can be quantified.

[0070] Bayes' theorem is a method of updating parameter values ​​by combining prior information and observed data. In this embodiment, the prior information is the initial value of the parameter of the historical performance prediction model, which is derived based on the historical performance monitoring data. The observed data is the real-time performance monitoring data. Substituting the real-time performance monitoring data into the likelihood function, the probability estimate of the parameter can be obtained through Bayes' theorem, thereby obtaining the updated parameters. The parameters obtained in this way refer to the real-time performance monitoring data, are more in line with the real-time operation of the vehicle, and are conducive to processing the dynamic data of the vehicle.

[0071] In some embodiments, the updating of parameters by Bayesian theorem based on real-time performance monitoring data and likelihood function includes: determining the posterior distribution of parameters by Bayesian theorem based on the prior distribution of parameters, real-time performance monitoring data and likelihood function; the prior distribution of parameters is determined based on historical performance monitoring data; and updating the parameters based on the posterior distribution of parameters.

[0072] The prior distribution of the parameters can be determined based on the historical performance monitoring data. In some embodiments, the prior distribution of the parameters includes a normal distribution, and the mean and variance of the normal distribution are determined by the historical performance monitoring data. By calculating the mean and variance of the historical performance monitoring data, the specific parameters of the normal distribution can be determined, thereby constructing the prior distribution.

[0073] Through Bayes' theorem, the prior distribution can be combined with the likelihood function to obtain the posterior distribution of the parameters. The posterior distribution is a new probability estimate of the parameters after observing the real-time performance monitoring data. Once the posterior distribution of the parameters is obtained, the posterior distribution can be used to update the parameters in the historical performance monitoring model. Specifically, samples can be drawn from the posterior distribution or its posterior mean, median and other statistics can be used as new parameter values. In this way, the historical performance monitoring model is updated to a real-time performance monitoring model, which can adapt to changes in vehicle performance based on the latest real-time performance monitoring data and more accurately predict future performance.

[0074] In some embodiments, step 12 includes: determining the posterior distribution of the parameter by variational inference or MCMC method based on the prior distribution of the parameter, real-time performance monitoring data and likelihood function; the prior distribution of the parameter is determined based on historical performance monitoring data; and updating the parameter based on the posterior distribution of the parameter.

[0075] In some embodiments, directly applying Bayes' theorem to calculate the posterior distribution involves a large number of complex operations, such as high-dimensional integration. Therefore, an approximate method can be used to estimate the posterior distribution.

[0076] Variational inference approximates the posterior distribution by optimizing the variational distribution. By finding a variational distribution that minimizes the difference between it and the posterior distribution. The difference can be measured by the KL divergence. Specifically, a simple distribution family (such as the Gaussian distribution family, the exponential distribution family, etc.) is selected as a candidate set for the variational distribution. Then, by calculating the expectation, gradient and other information of the variational distribution, an optimization algorithm is used to adjust the parameters of the variational distribution to minimize the KL divergence between the variational distribution and the posterior distribution of the target. Finally, the evidence lower bound (ELBO) is used as the optimization target. ELBO is a lower bound of the log-likelihood function of the posterior distribution of the target. KL divergence can be indirectly minimized by maximizing ELBO.

[0077] The MCMC method approximates the posterior distribution of the target by constructing a Markov chain. Starting from the initial state, a series of transitions are performed according to the transition probability of the Markov chain, and finally a sample that obeys the posterior distribution of the target can be generated.

[0078] In some embodiments, the real-time performance monitoring data includes noise. Before step 12, the performance prediction method 10 further includes: determining a priori distribution of noise; determining a posterior distribution of noise according to the priori distribution of noise and the real-time performance monitoring data; determining a noise reduction threshold according to the posterior distribution of noise and the mapping function; and reducing noise on the real-time performance monitoring data according to the noise reduction threshold.

[0079] Noise may come from many factors such as sensor inaccuracy, environmental interference, measurement errors, etc. The presence of noise will seriously affect the accuracy and reliability of real-time performance monitoring data analysis. Therefore, before performance prediction, noise reduction processing of real-time performance monitoring data can improve the accuracy of performance prediction results.

[0080] The prior distribution of noise is an assumption about the probability distribution of possible noise values ​​before real-time performance monitoring data is observed. The prior distribution of noise can be based on historical experience, domain knowledge, or expert judgment. For example, through historical performance monitoring data, it is concluded that the distribution of noise follows a Gaussian distribution, so a Gaussian distribution can be set as the prior distribution of noise, where the mean and variance can be set based on the mean and variance of the historical performance monitoring data.

[0081] After obtaining the real-time performance monitoring data, the posterior distribution of noise is calculated by combining the prior distribution of noise and the real-time performance monitoring data using Bayes' theorem. The posterior distribution of noise reflects the latest understanding of the possible values ​​of noise after observing the real-time performance monitoring data. Numerical methods or approximate algorithms can be used to solve the posterior distribution of noise.

[0082] The noise reduction threshold is used to distinguish between signals and noise. The noise reduction threshold determines which data points are considered noise and removed, and which data points are considered valid signals and retained. The noise reduction threshold can be determined based on the posterior distribution of noise. In addition, a mapping function can be introduced to further adjust the threshold to adapt to different application scenarios and data characteristics.

[0083] The noise reduction threshold can be determined by a noise estimation-based method. By analyzing the statistical characteristics of the real-time performance monitoring data, such as calculating the mean or variance, a relatively high noise reduction threshold is set according to the noise level to ensure that most of the noise is suppressed while minimizing interference with useful signals.

[0084] The noise reduction threshold can also be determined by using certain characteristics of the real-time performance monitoring data (such as spectrum characteristics, time characteristics, etc.). An automated algorithm can also be used to dynamically adjust the noise reduction threshold. The threshold is automatically adjusted according to the real-time characteristics of the real-time performance monitoring data or the change in the noise level to achieve a better noise reduction effect.

[0085] After determining the noise reduction threshold, the real-time performance monitoring data can be subjected to noise reduction processing. The original data can be filtered or smoothed to remove data points that are considered to be noise. Noise reduction methods include median filtering, mean filtering, Kalman filtering, etc.

[0086] Figure 2 Shown is a structural block diagram of an embodiment of a vehicle performance prediction device of the present application.

[0087] like Figure 2 As shown, the vehicle performance prediction device includes one or more processors 21 for implementing the vehicle performance prediction method 10 as described above.

[0088] In some embodiments, the performance prediction device of the vehicle may include a computer-readable storage medium 22, which may store a program that can be called by the processor 21 and may include a non-volatile storage medium. In some embodiments, the performance prediction device of the vehicle may include a memory 23 and an interface 24. In some embodiments, the performance prediction device of the vehicle may also include other hardware according to actual applications.

[0089] The computer-readable storage medium 22 of the embodiment of the present application stores a program thereon, and when the program is executed by the processor 21, it is used to implement the vehicle performance prediction method 10 described above.

[0090] The present application may take the form of a computer program product implemented on one or more computer-readable storage media 22 (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. The computer-readable storage medium 22 includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. The information can be a computer-readable instruction, a data structure, a module of a program, or other data. Examples of computer-readable storage media 22 include but are not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassette, tape disk storage or other magnetic storage device or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

Claims

1. A vehicle performance prediction method, characterized in that: include: Acquire a historical performance prediction model and real-time performance monitoring data of the vehicle; the historical performance prediction model is used to characterize the relationship between the historical performance monitoring data of the vehicle and the performance prediction result; the historical performance prediction model includes at least one parameter; updating the parameters according to the real-time performance monitoring data; Substituting the updated parameters into the historical performance prediction model to obtain a real-time performance prediction model; According to the real-time performance prediction model, a real-time performance prediction result of the vehicle is obtained.

2. The vehicle performance prediction method according to claim 1, characterized in that: The updating of the parameters according to the real-time performance monitoring data includes: Determining a likelihood function of the parameter based on the real-time performance monitoring data; The parameters are updated by Bayesian theorem according to the real-time performance monitoring data and the likelihood function.

3. The vehicle performance prediction method according to claim 2, characterized in that: The updating of the parameters by Bayesian theorem according to the real-time performance monitoring data and the likelihood function includes: Determining the posterior distribution of the parameter by Bayes' theorem according to the prior distribution of the parameter, the real-time performance monitoring data and the likelihood function; the prior distribution of the parameter is determined according to the historical performance monitoring data; The parameters are updated according to the posterior distribution of the parameters.

4. The vehicle performance prediction method according to claim 3, characterized in that: The prior distribution of the parameter includes a normal distribution, and the mean and variance of the normal distribution are determined by the historical performance monitoring data.

5. The vehicle performance prediction method according to claim 1, characterized in that: The updating of the parameters according to the real-time performance monitoring data includes: Determining the posterior distribution of the parameter by variational inference or MCMC method according to the prior distribution of the parameter, the real-time performance monitoring data and the likelihood function; the prior distribution of the parameter is determined according to the historical performance monitoring data; The parameters are updated according to the posterior distribution of the parameters.

6. The vehicle performance prediction method according to claim 1, characterized in that: The method for determining the historical performance prediction model includes: Obtaining historical performance monitoring data of the vehicle; Mapping the historical performance monitoring data to a quantum state to obtain quantum state data; The historical performance prediction model is obtained by processing the quantum state data through quantum entanglement.

7. The vehicle performance prediction method according to claim 1, characterized in that: The real-time performance monitoring data includes noise; before updating the parameter according to the real-time performance monitoring data, the performance prediction method further includes: determining a prior distribution of the noise; Determining a posterior distribution of the noise according to the prior distribution of the noise and the real-time performance monitoring data; Determining a noise reduction threshold according to the posterior distribution and mapping function of the noise; The real-time performance monitoring data is denoised according to the denoising threshold.

8. A computer-readable storage medium, characterized in that: A program is stored thereon, and when the program is executed by a processor, the vehicle performance prediction method described in any one of claims 1 to 7 is implemented.

9. A vehicle performance prediction device, characterized in that: The method comprises one or more processors for implementing the vehicle performance prediction method according to any one of claims 1 to 7.

10. A vehicle, characterized in that: include: A sensor for detecting the performance of the vehicle and generating the real-time performance monitoring data; and The performance prediction device according to claim 9 is electrically connected to the sensor.

Images