Search

CN-122029552-A - Method for verifying and permitting a predictive program, in particular an onboard predictive program, for vehicle data analysis

CN122029552ACN 122029552 ACN122029552 ACN 122029552ACN-122029552-A

Abstract

A method for verifying and approving a prediction program for vehicle data analysis of a new generation vehicle, in particular an on-board prediction program, is proposed, comprising the steps of creating a prediction program for creating a prediction based on vehicle data, providing a first data record with real vehicle data collected by one or more new generation vehicles, providing a second data record with real vehicle data collected by a plurality of old generation vehicles, providing a third data record with simulated vehicle data generated by a computer simulation of a new generation vehicle, creating a prediction based on the first data record by means of the prediction program, creating a prediction based on the specific event by means of the prediction program, based on the second data record, creating a prediction based on the specific event by means of the prediction program, based on the third data record, comparing the prediction based on the first data record with the prediction based on the second data record and the third data record, and confirming the validity of the prediction program when the prediction based on the first data record has a prediction based on the second data record and the third data record and the prediction based on the second data record and the prediction based on the third data record.

Inventors

  • T. L. ABT
  • O. Gruber

Assignees

  • 宝马股份公司

Dates

Publication Date
20260512
Application Date
20240902
Priority Date
20231120

Claims (14)

  1. 1. A method for verifying and approving a predictive program for a new generation vehicle for vehicle data analysis, in particular an on-board predictive program, the method comprising the steps of: Creating a prediction program (12) for creating a prediction based on vehicle data; Providing a first data record (14) having real vehicle data collected by one or more new generation vehicles; providing a second data record (16) having real vehicle data collected by a plurality of older generation vehicles; Providing a third data record (18) having simulated vehicle data generated by computer simulation of a new generation vehicle; Creating a prediction (20) about the particular event based on the first data record by means of the prediction program; creating a prediction (22) about the particular event based on the second data record by means of the prediction program; Creating a prediction (24) about the particular event based on the third data record by means of the prediction program; Comparing a prediction (20) based on the first data record with a prediction (22, 24) based on the second data record and the third data record; When a prediction (20) based on the first data record fluctuates within predetermined limits around predictions (22, 24) based on the second and third data records, validating the prediction program (12) and licensing the prediction program (40).
  2. 2. The method of claim 1, wherein the step of determining the position of the substrate comprises, To compare the predictions (20, 22, 24) to each other, a performance index (26, 28, 30) is calculated for each prediction and the performance indexes are compared to each other, either directly or indirectly.
  3. 3. The method of claim 2, wherein the step of determining the position of the substrate comprises, For comparing the predictions (20, 22, 24) with each other, a comparison value (32) is determined from performance indicators (28, 30) of the predictions (22, 24) based on the second and third data records (16, 18), and the performance indicator (26) based on the prediction of the first data record is compared with the comparison value (32).
  4. 4. A method according to any one of claim 1 to 3, wherein, The amount of data in the second data record (16) that is present in relation to the specific event is greater than the amount of data in the first data record (14) that is present in relation to the specific event by a factor between 101 and 10 5 .
  5. 5. The method according to any one of claim 1 to 4, wherein, The amount of data present in the third data record (18) in relation to the specific event is greater than the amount of data present in the second data record (16) in a multiple between 101 and 10 5 .
  6. 6. The method according to any one of claim 1 to 5, wherein, A prediction (22, 24) based on the second data record and the third data record is created outside the vehicle.
  7. 7. The method according to any one of claim 1 to 6, wherein, A prediction (20) based on the first data record is created on board of a new generation vehicle.
  8. 8. The method of claim 7, wherein the step of determining the position of the probe is performed, Predictions (20) created on board the vehicle are wirelessly transmitted to a computing center external to the vehicle and automatically compared at the computing center with predictions (22, 24) based on the second and third data records.
  9. 9. The method according to any one of claims 1 to 8, further comprising the step of: An error notification is generated when a prediction (20) based on the first data record (14) does not fluctuate within predetermined limits around predictions (22, 24) based on the second and third data records (16, 18).
  10. 10. The method according to any one of claims 1 to 9, wherein, The predictive program (12) is created using a learning algorithm and a training data record.
  11. 11. The method of claim 10, wherein, If the validity of the prediction program (12) is not confirmed and not permitted, the prediction program is re-input to the learning algorithm.
  12. 12. The method according to any one of claims 1 to 11, wherein, The licensing (42) of the predictive program includes the step of partially licensing the predictive program (12) for only a predetermined portion of the new generation vehicles.
  13. 13. The method of claim 12, wherein, After partial approval of the predictive program (12), the first data record (14) is supplemented with real vehicle data of a new generation vehicle using the partially approved predictive program (12).
  14. 14. A computer program product for performing at least the following steps in a method according to any of claims 1 to 13: comparing a prediction (20) based on the first data record (14) with a prediction (22, 24) based on the second and third data records (16, 18), and Confirming the validity of the prediction program (12).

Description

Method for verifying and permitting a predictive program, in particular an onboard predictive program, for vehicle data analysis Technical Field The present invention relates to a method for verifying and permitting a predictive program for a new generation vehicle for vehicle data analysis, in particular an on-board predictive program. Background Today vehicles are equipped with high-performance on-board computers which can perform a number of completely different auxiliary functions, on the one hand advantageously making the driving more comfortable and more convenient, but also particularly considerably safer, and on the other hand making the maintenance process more efficient and environmentally friendly in that, for example, the maintenance is not carried out simply in terms of a specific driving distance, but only if this is justified based on the actual vehicle state, which is derived in particular from the individual driving behavior of the user. The corresponding auxiliary functions generally use so-called "machine learning models", which are hereinafter referred to simply as "prediction programs" according to their function, which are able to analyze the vehicle data recorded by the vehicle itself by means of the corresponding sensors, which are usually supplemented with other data, in particular in the case of a specific problem, in particular on board the vehicle, but also in other locations, for example on a cloud-based basis, and thus to predict specific events, for example the degree of wear of specific components, which can be relevant at completely different levels, for example for the driving or maintenance of the vehicle. The prediction programs of the type referred to herein are generated using so-called "machine learning algorithms" and corresponding training data, and the more real vehicle data is provided for training, the better the prediction program generally works, i.e. the more accurate its prediction. Since modern vehicles typically collect a large amount of vehicle data and transmit it to the vehicle manufacturer in a wireless or wired manner (for example, at the time of service entry) according to the license of the customer side, the manufacturer side can collect a large amount of data (so-called fleet data) about the corresponding vehicle model of a specific generation and its behavior, which data enable the creation of a predictive program with extremely high operation accuracy. Before such a predictive program is put into use, it must be verified, i.e. checked in terms of its reliability of prediction, and licensed for use, since it not only involves safety-related aspects, but can also contribute significantly to customer satisfaction and brand loyalty, depending on the type. When, for example, an accurately running program forces a customer to enter a drive house for repair, which later proves to be reasonable, the customer will often have an impressive impression of the reliability of his vehicle. Disclosure of Invention When new-generation vehicles are to be put on the market, there is the problem that on the one hand no fleet data, i.e. a large data record about the actual daily behavior of the new vehicles, is present, since only a small number of new-generation vehicles are put on trial run, and on the other hand assistance functions that are as comprehensive as possible should be provided to the customer. For auxiliary functions (e.g. control of a radio) using rarely changed software, although a corresponding admission flow has been established, this is not the case for predictive programs that are often optimized and retrained based on new data. On the basis of this, the invention is based on the task of providing a method for verifying and permitting a prediction program for a new generation vehicle for vehicle data analysis, in particular an on-board prediction program, which method enables the prediction program to be checked in a standardized, understandable and unified manner in terms of its validity and, if applicable, permitted even in the event that no comprehensive fleet data is yet present. This object is achieved by a method having the features of claim 1. Advantageous embodiments and improvements are the subject matter of the dependent claims. The apposition claim 13 relates to a computer program product for performing certain steps of the method according to the present invention. This object is achieved in particular by a method for verifying and permitting a prediction program, in particular an on-board prediction program, for a first group of vehicles, in particular new-generation vehicles, for vehicle data analysis, wherein the method comprises the following steps: creating a prediction program for creating a prediction based on vehicle data; Providing a first data record having real vehicle data collected by one or more new generation vehicles; providing a second data record having real vehicle data collected by a plurality of second group vehicl