Claim parameter prediction method based on vehicle first fault
By extracting the vehicle's first failure data and calculating usage rate, and selecting the appropriate distribution function, the problem of incomplete and unclear claims data in the prior art is solved, and effective evaluation and prediction of vehicle reliability and claim parameters is achieved, reducing the claim cost and improving user satisfaction.
Patent Information
- Application Number
- CN202311768467.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-21
- Publication Date
- 2025-06-24
AI Technical Summary
The prior art is difficult to effectively utilize vehicle warranty data, especially due to the incompleteness and unclearity of the claim data information, which makes it difficult to accurately predict the claim parameters, and it is impossible to effectively evaluate the reliable performance of the vehicle and predict the number and cost of subsequent claims.
By obtaining the claim data of the target vehicle model, extracting the first-time failure data, calculating the vehicle usage rate, selecting the appropriate distribution function, estimating the distribution function parameters, and calculating the target claim parameter value based on the preset usage rate and distribution function.
It has achieved the evaluation of vehicle reliability, optimized vehicle design, reduced claim costs, improved user satisfaction, and provided theoretical support for the formulation of reasonable spare parts inventory and claim financial plans.
Smart Images
Figure CN120198233A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of prediction of first-occurrence fault claim parameters of vehicles, and specifically, to a method for predicting claim parameters based on first-occurrence faults of vehicles. Background Art
[0002] Vehicle warranty data contains a large amount of useful information related to product quality and reliability. Especially when the vehicle is used on real roads, the data of faults that occur, due to its real working conditions, is important data for vehicle reliability analysis and prediction. However, due to the incompleteness of claim data information (the data does not contain fault information of out-of-warranty vehicles) and uncleanness (data feedback delay, inaccurate time and mileage data), these characteristics make the analysis of claim data very challenging. Therefore, most current manufacturers focus on after-sales claim data statistics, such as the failure rate per thousand vehicles, the claim cost per vehicle, the annual claim amount, the claim quantity of a single part, etc., and do not evaluate the reliable performance of the whole vehicle, nor effectively predict the subsequent claim quantity and claim cost. The value of its after-sales claim data has not been effectively utilized.
[0003] Currently, regarding the analysis of after-sales claim data, for example, a piecewise statistical distribution combining the Weibull distribution and the exponential distribution is used to analyze past claim data, and it is found that there are obvious differences in claim distribution parameters among product lines, but within the same product family, they still remain approximately the same, thereby improving the overall accuracy of the simulation-based warranty prediction technology. Based on the Weibull distribution probability plot, a rational function fitting method is proposed to estimate the model parameters and compare and analyze the results with the maximum likelihood function. A cumulative usage distribution of censored data is established based on the claim data, and a product reliability model is established based on the censored data and the claim data. There are relatively few domestic scholars' research. For example, Dai Anshu et al. introduced a seasonal index model, adjusted the failure rate function using the usage rate, modeled the two-dimensional quality warranty claim data, and compared and analyzed it with the traditional model. Du Wenchao et al. studied and analyzed the inventory strategy of spare parts based on the usage intensity under two-dimensional quality warranty conditions. At the academic level, most of the research and analysis focus on the research, comparison, and analysis of methods. At the practical application level, there is no relevant report from vehicle manufacturers at home and abroad. Summary of the Invention
[0004] To solve at least one aspect of the above problems, the present invention provides a method for predicting claim parameters of the first failure of a vehicle, including: obtaining claim data of a target vehicle model, where the claim data includes claim information of multiple vehicles, and the claim information of each vehicle includes a sales date, a repair date, and a mileage; extracting first failure data from the claim data, where the first failure data includes first failure information of multiple vehicles, and the first failure information includes a sales date, a first repair date, and a first repair mileage; calculating the usage rates of multiple vehicles respectively according to the first failure data to obtain a usage rate sample set; selecting multiple distribution functions according to the target claim parameters and the usage rate sample set, calculating the compliance test values of the usage rate sample set for the multiple distribution functions respectively, and selecting the distribution function corresponding to the minimum compliance test value as the target distribution function; estimating the parameters of the target distribution function based on the usage rate sample set; calculating the target claim parameter value according to a preset usage rate and the distribution function.
[0005] Preferably, the step of extracting first failure data of multiple vehicles from the claim data further includes: when a vehicle includes multiple pieces of claim information, the first failure information is the claim information with the earliest repair date among the multiple pieces of claim information.
[0006] Preferably, the step of extracting first failure data of multiple vehicles from the claim data further includes: when a vehicle includes multiple pieces of claim information, the first failure information is the claim information with the shortest repair mileage among the multiple pieces of claim information.
[0007] Preferably, the step of calculating the usage rates of multiple vehicles respectively according to the first failure data includes: calculating the number of usage days according to the sales date and the first repair date; calculating the number of usage months according to the number of usage days, the average number of days in a month, and the usage time parameter, and then the vehicle usage rate is equal to the ratio of the first repair mileage to the number of usage months.
[0008] Preferably, the target claim parameter includes the first failure time, and the step of determining the target distribution function according to the target claim parameter and the usage rate sample set includes: determining the conditional distribution function of the first failure time according to the first failure time and the usage rate sample set; calculating the parameters of the conditional distribution function of the first failure time according to the usage rate sample set by using the linear fitting method.
[0009] Preferably, the target claim parameter includes the usage rate failure probability, and the step of estimating the parameters of the target distribution function based on the usage rate sample set includes: estimating the parameters of the target distribution function by using the maximum likelihood function method according to the usage rate sample set.
[0010] Preferably, it further includes: dividing the usage rate sample set into multiple usage rate subsets according to a preset tolerance; establishing a lognormal distribution function of usage rate and usage months, calculating the maximum likelihood estimate value of the usage months of each usage rate subset, and determining the parameters of the distribution function according to the median usage rate and the maximum likelihood function value of each usage rate subset; calculating the number of failures corresponding to a preset usage rate and a preset time according to the preset time, the preset usage rate, and the distribution function.
[0011] Preferably, it further includes: calculating the usage months of multiple vehicles respectively according to the first-occurrence failure data to obtain a usage month sample set; fitting a usage time distribution function according to the usage month sample set, and estimating the coefficients of the usage time distribution function by using the maximum likelihood function or the least square method according to the usage month sample set; calculating the failure probability corresponding to the preset usage time according to the preset usage time and the usage time distribution function.
[0012] Preferably, it further includes: generating a usage mileage sample set according to the first-occurrence failure data; fitting the usage mileage sample set according to an approximate distribution function, and estimating the parameters of the usage mileage distribution function by using the maximum likelihood estimate or the least square method according to the usage mileage sample set; calculating the failure probability corresponding to the preset usage mileage according to the preset usage mileage and the usage mileage distribution function.
[0013] The claim parameter prediction method based on the first-occurrence failure of vehicles of the present invention has the following beneficial effects: evaluating the reliability of vehicles by using claim data, optimizing and improving vehicles in the design stage, improving product quality, which is an effective method for reducing claim costs and improving user satisfaction. Predicting the subsequent claim quantity by using the existing claim data provides effective theoretical support for formulating a reasonable spare parts inventory, claim financial plan, and designing a product warranty strategy. Description of the Drawings
[0014] In order to better understand the above and other objects, features, advantages, and functions of the present invention, reference may be made to the embodiments shown in the drawings. The same reference numerals in the drawings refer to the same components. Those skilled in the art should understand that the drawings are intended to schematically illustrate the preferred embodiments of the present invention and have no limiting effect on the scope of the present invention. The components in the drawings are not drawn to scale.
[0015] Figure 1 Shows a flowchart of a claim parameter prediction method based on the first-occurrence failure of vehicles according to an embodiment of the present invention;
[0016] Figure 2 Shows a schematic flowchart of the usage rate failure probability of a claim parameter prediction method based on the first-occurrence failure of vehicles according to an embodiment of the present invention. Detailed Embodiments
[0017] The exemplary embodiments of the present disclosure will be described below with reference to the accompanying drawings. Various details of the embodiments of the present disclosure are included to facilitate understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, descriptions of well-known functions and structures are omitted in the following description for clarity and conciseness.
[0018] As used herein, the term "comprising" and its variations mean open inclusion, i.e., "including but not limited to". Unless otherwise specified, the term "or" means "and / or". The term "based on" means "at least partially based on". The terms "an example embodiment" and "an embodiment" mean "at least one example embodiment". The term "another embodiment" means "at least one additional embodiment". The terms "first", "second", etc. may refer to different or the same objects. There may also be other explicit and implicit definitions below.
[0019] To at least partially solve one or more of the above problems and other potential problems, embodiments of the present disclosure propose a method for predicting claim parameters for the first-occurring failure of a vehicle, including: obtaining claim data of a target vehicle model, where the claim data includes claim information of multiple vehicles, and the claim information of each vehicle includes a sales date, a repair date, and a mileage; extracting first-occurring failure data from the claim data, where the first-occurring failure data includes first-occurring failure information of multiple vehicles, and the first-occurring failure information includes a sales date, a first repair date, and a first repair mileage; calculating the usage rates of multiple vehicles respectively based on the first-occurring failure data to obtain a usage rate sample set; selecting multiple distribution functions according to the target claim parameter and the usage rate sample set, calculating the compliance test values of the usage rate sample set for the multiple distribution functions respectively, and selecting the distribution function corresponding to the minimum compliance test value as the target distribution function; and estimating the parameters of the target distribution function based on the usage sample set; calculating the target claim parameter value according to a preset usage rate and the target distribution function.
[0020] Specifically, the target vehicle model is a vehicle model specified by the user, the claim data includes claim information of multiple vehicles, and the multiple vehicles are all corresponding to the vehicle model of the target vehicle model. Among them, the claim information of each vehicle includes a sales date, a repair date, and the mileage corresponding to the repair date. In some embodiments, the claim data includes claim data of factory vehicles in the consecutive twelve months.
[0021] Extract the initial failure data based on the claim data. Specifically, extract the initial failure information according to the claim information of each vehicle. When a vehicle includes only one claim information, the repair date of this claim information is the first repair time, and the mileage at the time of this claim information is the first repair mileage. In some embodiments, the step of extracting the initial failure data of multiple vehicles based on the claim data further includes: when a vehicle includes multiple claim information, the initial failure information is the claim information with the earliest repair date among the multiple claim information. Or in other embodiments, when a vehicle includes multiple claim information, the initial failure information is the claim information with the shortest repair mileage among the multiple claim information.
[0022] The usage rate sample set is the set of the usage rates of multiple vehicles in the initial failure data. In some embodiments, the step of calculating the usage rates of multiple vehicles respectively based on the initial failure data includes: calculating the number of usage days according to the sales date and the first repair date; calculating the number of usage months according to the number of usage days, the average number of days per month, and the usage time parameter, then the vehicle usage rate is equal to the ratio of the first repair mileage and the number of usage months.
[0023]
[0024] Where, MIS is the number of usage months, d r represents the first repair date, d s represents the sales date, the usage time parameter is set to 0.99, and the average number of days per month is set to 30.4.
[0025] Assume that the usage rate of each vehicle is different, and use (MIS i , l i ) to represent the number of usage months and the first repair mileage (KM) when user i makes the first claim. Calculate the vehicle usage rate of each claim data.
[0026]
[0027] Where, l i represents the first repair mileage (KM) when the vehicle fails; MIS i represents the number of usage months when the vehicle fails.
[0028] Use an appropriate distribution function to fit the usage rate sample set, and estimate the coefficients of the distribution function based on the usage rate sample set; calculate the target claim parameter value corresponding to the usage rate according to the preset usage rate and the distribution function. In some embodiments, the multiple distribution functions selected according to the target claim parameter and the usage rate sample set include Weibull distribution function, lognormal distribution function, exponential distribution function, normal distribution function, and Logistic distribution function. The goodness-of-fit test value (i.e., the AD statistic, Anderson-Daeling) is the test value for the data in the usage rate sample set to follow the distribution function. For example, when the goodness-of-fit test value of the data in the usage rate sample set for the normal distribution is the smallest, the target distribution function is the normal distribution function. Further, the parameters of the target distribution function are estimated using the maximum likelihood function method according to the usage rate sample set. Or in other embodiments, the parameters of the target distribution function are estimated using the least squares method according to the usage rate sample set. In some embodiments, the target claim parameter includes the usage rate failure probability.
[0029] In some embodiments, the target claim parameter includes the time to first failure. Then, the steps of determining the target distribution function according to the target claim parameter and the usage rate sample set include: determining the conditional distribution function of the time to first failure (i.e., the target distribution function) according to the time to first failure and the usage rate; calculating the coefficients of the conditional distribution function of the time to first failure according to the usage rate sample set using the linear fitting method.
[0030] Specifically, taking the usage rate as a function of the working time, simplifying the two-dimensional warranty data into one-dimensional warranty data, assuming that the user's usage rate remains unchanged within the warranty period, and the usage rate Z is a random variable. The conditional distribution function F of the time to first failure is:
[0031]
[0032] where the conditional hazard function h(t|z) ≥ 0 and is a non-decreasing function for both t and z.
[0033] h(t|z) = θ0 + θ1t + θ2z + θ3zt,
[0034] In the formula, θ0, θ1, θ2, and θ3 are non-negative coefficients.
[0035] Substitute the preset usage rate into the conditional distribution function F of the time to first failure, and substitute the time to first failure value into the conditional distribution function of the time to first failure, then the probability of the time to first failure value occurring under the preset usage rate condition can be obtained.
[0036] In some embodiments, it further includes: dividing the usage rate sample set into multiple usage rate subsets according to a preset tolerance; establishing a relationship equation between the usage rate and the parameters of the distribution function, and determining the parameter values of the distribution function corresponding to different usage rates according to the relationship equation between the median usage rate and the maximum likelihood function value of each usage rate subset; calculating the number of failures corresponding to the preset usage rate and preset time according to the preset time, preset usage rate, and the distribution function.
[0037] Specifically, the usage rates are grouped into an arithmetic progression according to the preset tolerance G i , i = 1, 2, 3…, and calculate G i The maximum likelihood estimate of the usage months of the group, that is, the maximum likelihood estimate of the parameters of different usage rate groups; establish an equation between the maximum likelihood estimate of the parameters and the median usage rate, which can be fitted by a linear regression model, as follows:
[0038]
[0039]
[0040] is the coefficient, where the number of parameters is related to the type of distribution function. Multiple usage rate subsets are combined with the location parameter μ z and the scale parameter σ z to calculate the coefficient value, which is a one-dimensional model of the service life T with the usage rate z as the condition; in some embodiments, the service life T of the usage rate z based on the working time is a lognormal distribution (i.e., the target distribution function), and the parameters μ z and the parameter σ z are respectively defined by the above formula, so there is
[0041]
[0042] When a z value is given, substitute the z value into the location parameter μ z and the scale parameter σ z formula to calculate μ z and σ z , and further the cumulative distribution function F(t|z) corresponding to the z value can be obtained. Further, the preset time can be set to the value of the usage months to obtain the corresponding number of failures. Those skilled in the art can understand that in other embodiments, the distribution function of the service life T of the usage rate z based on the working time is any one of the Weibull distribution function, lognormal distribution function, exponential distribution function, normal distribution function, and Logistic distribution function.
[0043] In some embodiments, it further includes: calculating the usage months of multiple vehicles respectively according to the initial failure data to obtain a usage month sample set; determining a usage time distribution function according to the usage month sample set, and estimating the parameters of the usage time distribution function by using the maximum likelihood function or the least squares method according to the usage month sample set; calculating the failure probability corresponding to a preset usage time according to the preset usage time and the usage time distribution function, where the preset usage time is a set usage month value. In other embodiments, the usage time distribution function adopts any one of the Boolean distribution, the lognormal distribution, the exponential distribution, the normal distribution, and the Logistic distribution. And estimating the coefficients of the usage time distribution function by using the maximum likelihood function according to the usage month sample set.
[0044] In some embodiments, it further includes: generating a usage mileage sample set according to the initial failure data, where the usage mileage sample set is a set of the first repair mileages of multiple vehicles in the initial failure data; selecting a distribution function to fit the mileage sample set, and estimating the parameters of the usage mileage distribution function by using the maximum likelihood function or the least squares method according to the usage mileage sample set; calculating the failure probability corresponding to a preset usage mileage according to the preset usage mileage and the usage mileage distribution function, where the preset usage mileage is a set driving mileage value. In other embodiments, the usage mileage distribution function adopts any one of the Boolean distribution, the lognormal distribution, the exponential distribution, the normal distribution, and the Logistic distribution. And estimating the coefficients of the usage time distribution function by using the maximum likelihood function according to the usage mileage sample set.
[0045] The embodiments of the present disclosure have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, the practical application, or the improvement of the technology in the market, or to enable other ordinary skill in the art in the technical field to understand the present disclosure.
Claims
1. A method for predicting claim parameters based on the first failure of a vehicle, characterized in that, Including: Obtain the claim data of the target vehicle model, where the claim data includes the claim information of multiple vehicles, and the claim information of each vehicle includes the sales date, repair date, and mileage; Extract the initial failure data according to the claim data, where the initial failure data includes the initial failure information of multiple vehicles, and the initial failure information includes the sales date, first repair date, and first repair mileage; Calculate the utilization rates of multiple vehicles respectively according to the initial failure data to obtain a utilization rate sample set; Select multiple distribution functions according to the target claim parameter and the utilization rate sample set, calculate the compliance test values of the utilization rate sample set for the multiple distribution functions respectively, and select the distribution function corresponding to the minimum compliance test value as the target distribution function; Estimate the parameters of the target distribution function based on the utilization rate sample set; calculate the target claim parameter value according to the preset utilization rate and the target distribution function.
2. The method according to claim 1, wherein The step of extracting the initial failure data of multiple vehicles according to the claim data further includes: When a vehicle includes multiple claim information, the initial failure information is the claim information with the earliest repair date among the multiple claim information.
3. The method according to claim 1, characterized in that, The step of extracting the initial failure data of multiple vehicles according to the claim data further includes: When a vehicle includes multiple claim information, the initial failure information is the claim information with the shortest repair mileage among the multiple claim information.
4. The method according to claim 2 or 3, characterized in that, The step of calculating the utilization rates of multiple vehicles respectively according to the initial failure data includes: Calculate the number of usage days according to the sales date and the first repair date; Calculate the usage months according to the number of usage days, the average number of days per month, and the usage time parameter, and then the vehicle utilization rate is equal to the ratio of the first repair mileage to the usage months.
5. The method according to claim 4, characterized in that When the target claim parameter includes the initial failure time, the step of determining the target distribution function according to the target claim parameter and the utilization rate sample set includes: Determine the initial failure time conditional distribution function according to the initial failure time and the utilization rate sample set; Use the linear fitting method to calculate the parameters of the initial failure time conditional distribution function according to the utilization rate sample set.
6. The method according to claim 4, characterized in that When the target claim parameter includes the utilization rate failure probability, the step of estimating the parameters of the target distribution function based on the utilization rate sample set includes: estimating the parameters of the distribution function according to the utilization rate sample set by using the maximum likelihood function method.
7. The method according to claim 4, wherein Also including: Divide the utilization rate sample set into multiple utilization rate subsets according to the preset tolerance; Establish a relationship equation between the utilization rate and the distribution function parameters, and determine the parameter values of the distribution function corresponding to different utilization rates according to the relationship equation between the utilization rate median and the maximum likelihood function value of each utilization rate subset; Calculate the number of failures corresponding to the preset utilization rate and the preset time according to the preset time, the preset utilization rate, and the distribution function.
8. The method according to claim 4, wherein Also including: Calculate the usage months of multiple vehicles respectively according to the initial failure data to obtain a usage month sample set; Determine the usage time distribution function according to the usage month sample set, and estimate the parameters of the usage time distribution function according to the usage month sample set by using the maximum likelihood function; Calculate the failure probability and the number of failures corresponding to the preset usage time according to the preset usage time and the usage time distribution function.
9. The method according to claim 8, characterized in that Also including: Generate a set of usage mileage samples based on the initial failure data; Determine the usage mileage distribution function according to the set of usage mileage samples, and estimate the parameters of the usage mileage distribution function using the maximum likelihood function based on the set of usage mileage samples; Calculate the failure probability corresponding to the preset usage mileage according to the preset usage mileage and the usage mileage distribution function.