A method and system for predicting automobile quality claims based on survival analysis model

Abstract

The present invention discloses a method and system for predicting automobile quality claims based on a survival analysis model, the method comprising the following steps: obtaining scenarios that affect the occurrence of quality failures of automobile parts through a failure scenario module; obtaining factors that affect the occurrence of quality failures of automobile parts through a failure factor module; establishing a claim model through a report module according to the information obtained by a failure acquisition module and a failure factor module; establishing a claim prediction model through a regression coefficient algorithm and a survival analysis model algorithm according to the data of the claim model; comparing the prediction results of the claim prediction model with the actual number of claims, and re-establishing the claim prediction model if the average prediction accuracy of the claim prediction model is lower than a preset threshold. The present invention can predict the scale of claims for automobile parts, help OEMs prepare for large-scale claims in advance, prepare spare parts for repair in advance, ensure efficiency, and save time.

Description

Technical Field

[0001] The present invention relates to the field of automobile technology, and in particular to a method and system for predicting automobile quality claims based on a survival analysis model. Background Art

[0002] With the vigorous development of the automobile industry, automobile quality issues have become the focus of common concern of automobile companies and consumers. Quality problems often occur when the vehicle's driving time or mileage has reached a certain level. When a large number of problems are collected and fed back to the quality department of the automobile company, some problems that had signs in the early stage are already on the verge of breaking out. The long monitoring cycle will result in the problem cannot be discovered and solved early, which can only improve the reliability of subsequent products, and it will have to bear high maintenance costs, and even cause large-scale recalls.

[0003] After a car is sold, if a maintenance event (such as replacement of parts) occurs during the warranty period, a warranty will be generated. The dealer will provide the warranty to the car manufacturer. After the car manufacturer compensates the dealer based on factors such as whether it is within the warranty period, it will recover the cost from the parts supplier. Generally, the car manufacturer will extract the warranty, manually analyze each item in the warranty to identify the type and cause of the fault, and determine whether to recover the cost (i.e. claim) from the parts supplier based on the responsibility determination. The entire process is completely dependent on manual labor, the data is not transparent, the efficiency is low, and the processing accuracy is low.

[0004] For car companies, technology research and development, product quality, and quality services have become important means for car manufacturers to enhance brand competitiveness. Once a large number of quality problems occur during the warranty period, the following technical problems will be faced:

[0005] 1. Reduce vehicle repair time: Sudden large-scale quality problems will lead to insufficient spare parts inventory, affecting the timeliness of user repairs;

[0006] 2. Direct economic losses: In addition to the cost of the claim, the cost of parts, labor hours, auxiliary materials and rescue costs caused by vehicle repairs and road failures shall also be borne;

[0007] 3. Impact on brand value: Quality is the lifeline of an enterprise’s development strategy. Quality issues will damage brand credibility and directly affect customer loyalty and product sales.

[0008] Therefore, how to design an accurate prediction method for claims caused by automobile parts failure so that automakers can make predictions in advance is a technical problem that needs to be solved urgently. Summary of the invention

[0009] The first object of the present invention is to provide a method for predicting automobile quality claims based on a survival analysis model to solve the technical problem of how to predict the probability of claims occurring for automobile parts.

[0010] The first object of the present invention is achieved by adopting the following technical solution: a method for predicting automobile quality claims based on a survival analysis model, comprising the following steps:

[0011] Obtaining fault scenarios: Obtaining scenarios that may affect the quality failure of automotive parts through the fault scenario module;

[0012] Obtaining fault factors: Obtaining factors that affect the quality failure of automotive parts through the fault factor module;

[0013] Establish a claim model: Based on the information obtained by the fault collection module and the fault factor module, establish a claim model through the report module;

[0014] Establish a claim prediction model: Based on the data of the claim model, establish a claim prediction model through regression coefficient algorithm and survival analysis model algorithm;

[0015] Forecast data correction: Compare the prediction results of the claim prediction model with the actual number of claims. If the average prediction accuracy of the claim prediction model is lower than the preset threshold, the claim prediction model will be re-established.

[0016] Furthermore, the scenarios that affect the quality failure of auto parts include quality problems of products of auto parts suppliers, quality problems introduced by OEMs during vehicle assembly, quality problems caused by vehicles being stored in dealerships for too long, quality problems that occur when vehicles are used normally and reach their sales lifespan, and quality problems caused by users.

[0017] Furthermore, the factors that affect the quality failure of automobile parts include parts data, production data, inventory data, vehicle after-sales data and user usage data.

[0018] Furthermore, the parts data includes parts and supplier information and production batches; the production data includes production year and month and production quality inspection data; the inventory data includes store entry time, sales time and inventory time; the vehicle after-sales data includes sales date, maintenance date and amount, and whether maintenance is carried out on time; the user usage data includes average daily mileage, vehicle purpose, user gender, user age, user location and seasonal weather.

[0019] Furthermore, the method for establishing a claim model through a reporting module is as follows: based on the scenario information affecting the quality failure of automobile parts obtained by the fault collection module and the factor information affecting the quality failure of automobile parts obtained by the fault factor module, the corresponding claim business-related tables are sorted out and the data is connected, and the data of claims that have occurred and claims that are not yet occurred are distinguished, and positive and negative samples are made respectively, and then based on the positive and negative samples, a wide table of data items that affect the occurrence of claims is sorted out.

[0020] Furthermore, the positive and negative samples include at least 30,000 or more sample data, and the ratio of positive and negative samples cannot be less than 10%.

[0021] Furthermore, the wide table of data items includes the following valid fields: region, vehicle model, year and month of production, sales time, whether regular maintenance is required, time of last maintenance, inventory time, time of last claim, season of occurrence, supplier, gender of owner, age of owner, whether there is any customer complaint, time of last customer complaint, purpose of vehicle and mileage.

[0022] Furthermore, the method of establishing a claim prediction model based on the data of the claim model through the regression coefficient algorithm and the survival analysis model algorithm is: the correlation of events affecting the occurrence of claims is calculated through the regression coefficient algorithm, and the calculation formula of the correlation β is:

[0023]

[0024] In the formula, x is the eigenvalue, is the single column feature mean, y is the target, is the target value mean;

[0025] The probability of a claim event occurring under relevant influencing factors is calculated by the survival analysis model algorithm. The calculation formula for the probability of a claim event occurring is:

[0026] h(t,X)=λ0(t)exp(β*X)

[0027] Where X is the relevant influencing factor, which is composed of multiple influencing factors, including region, season, fuel price and GDP growth rate; t is the unit span from standard capacity demand to increased capacity demand; h(t,X) represents the probability of a claim event when the time is t and the relevant influencing factor is X.

[0028] The second object of the present invention is to provide an automobile quality claim prediction system based on a survival analysis model to solve the technical problem of how to predict the probability of claims occurring for automobile parts.

[0029] The second object of the present invention is achieved by adopting the following technical scheme: a car quality claim prediction system based on a survival analysis model, including a fault scenario module, a fault factor module and a report module, wherein the fault scenario module is used to obtain the scenarios that affect the quality failure of automobile parts; the fault factor module is used to obtain the factors that affect the quality failure of automobile parts; a claim model is established through the report module based on the information obtained by the fault collection module and the fault factor module; a claim prediction model is established through the regression coefficient algorithm and the survival analysis model algorithm based on the data of the claim model; and a prediction data correction module is also included, wherein the prediction data correction module compares the prediction results of the claim prediction model with the actual claim quantity. If the average prediction accuracy of the claim prediction model is lower than the preset threshold, the claim prediction model is re-established.

[0030] The beneficial effect of the present invention is that the present invention integrates and mines historical claim data, maintenance data, component information, Internet of Vehicles data, etc., and based on the application of the survival analysis model algorithm, calculates the probability of each component from production to failure under the influence of highly correlated factors, thereby predicting the scale of claims for the component, helping OEMs to prepare for large-scale claims in advance, prepare maintenance spare parts in advance, ensure efficiency, and gain time. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0032] Figure 1 It is a flow chart of the present invention;

[0033] Figure 2 To establish a flow chart for the claims prediction model;

[0034] Figure 3 This is a system block diagram of the present invention. DETAILED DESCRIPTION

[0035] Taking the claims business of a certain brand as an example, there is currently no forecast of transportation demand and no prejudgment for the claims business. The procurement and forecast of claim spare parts are based on the MAD model, that is, the weighted average of the number of claims shipped out in the last three months, and the weights of N-3, N-2 and N-1 are 20%, 30% and 50% respectively; the MAD model is inaccurate in ordering maintenance and accident appearance parts; at least 5% of the parts (without rules) each month have insufficient inventory, resulting in delays in dealer orders, affecting the index of inventory satisfaction rate>95%. To achieve the index of inventory satisfaction rate, only the factors affecting quality claims from a business perspective can be re-examined, and the probability of claim time can be more accurately predicted through multi-dimensional data analysis. Therefore, the present invention integrates and mines historical claim data, maintenance data, parts information, Internet of Vehicles data, etc., and applies the survival analysis model (cox regression) algorithm to calculate the probability of each component from production to failure under the influence of highly correlated factors, thereby predicting the scale of claims for the component. Please refer to the embodiments of the present invention for details.

[0036] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0037] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0038] Some embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0039] Embodiment 1,

[0040] See also Figure 1-Figure 2 , a method for predicting automobile quality claims based on a survival analysis model, comprising the following steps:

[0041] Obtaining fault scenarios: Obtaining scenarios that may affect the quality failure of automotive parts through the fault scenario module;

[0042] Obtaining fault factors: Obtaining factors that affect the quality failure of automotive parts through the fault factor module;

[0043] Establish a claim model: Based on the information obtained by the fault collection module and the fault factor module, establish a claim model through the report module;

[0044] Establish a claim prediction model: Based on the data of the claim model, establish a claim prediction model through regression coefficient algorithm and survival analysis model algorithm;

[0045] Forecast data correction: Compare the prediction results of the claim prediction model with the actual number of claims. If the average prediction accuracy of the claim prediction model is lower than the preset threshold, the claim prediction model will be re-established.

[0046] In this embodiment, the scenarios that affect the quality failure of auto parts include quality problems of products of auto parts suppliers, quality problems introduced by OEMs during vehicle assembly, quality problems caused by vehicles being stored in dealerships for too long, quality problems that occur when vehicles are used normally and reach their sales lifespan, and quality problems caused by users.

[0047] In this embodiment, the factors that affect the quality failure of automobile parts include parts data, production data, inventory data, vehicle after-sales data and user usage data.

[0048] In this embodiment, the parts data includes parts and supplier information and production batches; the production data includes production year and month and production quality inspection data; the inventory data includes store entry time, sales time and inventory time; the vehicle after-sales data includes sales date, maintenance date and amount, and whether maintenance is carried out on time; the user usage data includes average daily mileage, vehicle purpose, user gender, user age, user location and seasonal weather.

[0049] In this embodiment, the method for establishing a claim model through a report module is as follows: based on the scenario information that affects the quality failure of automobile parts obtained by the fault collection module and the factor information that affects the quality failure of automobile parts obtained by the fault factor module, the corresponding claim business-related tables are sorted out and the data is connected, and the data of claims that have occurred and claims that are not yet occurred are distinguished, and positive and negative samples are made respectively. Then, based on the positive and negative samples, a wide table of data items that affect the occurrence of claims is sorted out.

[0050] In this embodiment, the positive and negative samples include at least 30,000 or more sample data, and the ratio of positive and negative samples cannot be less than 10%.

[0051] In this embodiment, the wide table of data items includes the following valid fields: region, vehicle model, year and month of production, sales time, whether regular maintenance is required, time of last maintenance, inventory time, time of last claim, season of occurrence, supplier, gender of owner, age of owner, whether there is any customer complaint, time of last customer complaint, purpose of vehicle and mileage.

[0052] In this embodiment, the method of establishing a claim prediction model based on the data of the claim model through the regression coefficient (coef) algorithm and the survival analysis model algorithm is: the correlation of events affecting the occurrence of claims is calculated through the regression coefficient algorithm, and the calculation formula of the correlation β is:

[0053]

[0054] In the formula, x is the eigenvalue, is the single column feature mean, y is the target, is the target value mean;

[0055] The probability of a claim event occurring under relevant influencing factors is calculated by the survival analysis model (Cox regression model) algorithm. The calculation formula for the probability of a claim event occurring is:

[0056] h(t,X)=λ0(t)exp(β*X)

[0057] Where X is the relevant influencing factor, which is composed of multiple influencing factors, including region, season, fuel price and GDP growth rate; t is the unit span from standard capacity demand to increased capacity demand; h(t,X) represents the probability of a claim event when the time is t and the relevant influencing factor is X.

[0058] In this embodiment, the method for correcting the predicted data is: comparing the predicted results of the claim prediction model with the actual number of claims. If the average prediction accuracy of the claim prediction model is lower than the preset threshold, the claim prediction model is re-established by adding or reducing the relevant influencing factor X.

[0059] In this embodiment, the preset threshold is preferably 85%. When the average prediction accuracy of the claim prediction model is lower than 85%, it is necessary to re-establish the claim prediction model by adding or reducing the relevant influencing factor X; if the average prediction accuracy of the claim prediction model is higher than 85%, the claim prediction model can be used to predict the probability of claims for each component of the automobile.

[0060] See also Figure 3A system for predicting automobile quality claims based on a survival analysis model includes a fault scenario module, a fault factor module and a report module. The fault scenario module is used to obtain scenarios that affect the occurrence of quality failures of automobile parts; the fault factor module is used to obtain factors that affect the occurrence of quality failures of automobile parts; a claim model is established through the report module based on the information obtained by the fault collection module and the fault factor module; a claim prediction model is established through the regression coefficient algorithm and the survival analysis model algorithm based on the data of the claim model; and a prediction data correction module is also included. The prediction data correction module compares the prediction results of the claim prediction model with the actual claim quantity. If the average prediction accuracy of the claim prediction model is lower than a preset threshold, the claim prediction model is re-established.

[0061] Further, according to the data of the claim model, the method of establishing the claim prediction model through the regression coefficient algorithm and the survival analysis model algorithm is: the correlation of the events affecting the occurrence of the claim is calculated through the regression coefficient algorithm, and the calculation formula of the correlation β is:

[0062]

[0063] In the formula, x is the eigenvalue, is the single column feature mean, y is the target, is the target value mean;

[0064] The probability of a claim event occurring under relevant influencing factors is calculated by the survival analysis model (Cox regression model) algorithm. The calculation formula for the probability of a claim event occurring is:

[0065] h(t,X)=λ0(t)exp(β*X)

[0066] Where X is the relevant influencing factor, which is composed of multiple influencing factors, including region, season, fuel price and GDP growth rate; t is the unit span from standard capacity demand to increased capacity demand; h(t,X) represents the probability of a claim event when the time is t and the relevant influencing factor is X.

[0067] The beneficial technical effects that can be achieved by the above embodiments include: the present invention integrates and mines historical claim data, maintenance data, parts information, Internet of Vehicles data, etc., and based on the application of the survival analysis model algorithm, calculates the probability of each part from production to failure under the influence of highly correlated factors, thereby predicting the scale of claims for the part. According to the above processing logic, the probability of claim occurrence for each type of part can be calculated, and the scale of claims can be predicted in combination with the total number of vehicles, helping OEMs to prepare for large-scale claims in advance, prepare maintenance spare parts in advance, ensure efficiency, and save time.

[0068] It should be noted that, for the above embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification belong to preferred embodiments, and the actions involved are not necessarily required by the present application.

[0069] The above embodiments describe the basic principles and main features of the present invention and the advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the changes and modifications made by those skilled in the art shall be within the scope of protection of the appended claims of the present invention without departing from the spirit and scope of the present invention.

Claims

1. A method for predicting automobile quality claims based on a survival analysis model, characterized in that: The steps include: Obtaining fault scenarios: Obtaining scenarios that may affect the quality failure of auto parts through the fault scenario module; the scenarios that may affect the quality failure of auto parts include quality problems of auto parts suppliers' products, quality problems introduced by OEMs during vehicle assembly, quality problems caused by vehicles being stored in dealerships for too long, quality problems that occur when vehicles are used normally and reach their sales lifespan, and quality problems caused by users; Obtaining fault factors: Obtaining factors that affect the quality failure of auto parts through the fault factor module; the factors that affect the quality failure of auto parts include parts data, production data, inventory data, vehicle after-sales data and user usage data; Establishing a claim model: establishing a claim model through a report module according to the information obtained by the fault collection module and the fault factor module; the method of establishing a claim model through the report module is: according to the scenario information affecting the quality failure of automobile parts obtained by the fault collection module and the factor information affecting the quality failure of automobile parts obtained by the fault factor module, sorting out the corresponding claim business related tables and connecting the data, and distinguishing the data of claims that have occurred and those that have not occurred, making positive and negative samples respectively, and then sorting out the data item wide table affecting the claim occurrence event according to the positive and negative samples; Establishing a claim prediction model: According to the data of the claim model, a claim prediction model is established through a regression coefficient algorithm and a survival analysis model algorithm; the method of establishing a claim prediction model according to the data of the claim model through a regression coefficient algorithm and a survival analysis model algorithm is: calculating the correlation of events affecting the occurrence of claims through a regression coefficient algorithm, and the calculation formula of the correlation is: ； In the formula, is the characteristic value, is the single column feature mean, For the goal, is the target value mean; The probability of a claim event occurring under relevant influencing factors is calculated by the survival analysis model algorithm. The calculation formula for the probability of a claim event occurring is: ； Wherein, X is a relevant influencing factor, which is composed of multiple influencing factors, including region, season, fuel price and GDP growth rate; The unit span from standard capacity requirement to increased capacity requirement; It represents the probability of a claim event occurring when the time is t and the relevant influencing factor is X; Forecast data correction: Compare the prediction results of the claim prediction model with the actual number of claims. If the average prediction accuracy of the claim prediction model is lower than the preset threshold, the claim prediction model will be re-established.

2. The method for predicting automobile quality claims based on a survival analysis model as claimed in claim 1, characterized in that: The parts data includes parts and supplier information and production batches; the production data includes production year and month and production quality inspection data; the inventory data includes store entry time, sales time and inventory time; the vehicle after-sales data includes sales date, maintenance date and amount, and whether maintenance is carried out on time; the user usage data includes average daily mileage, vehicle purpose, user gender, user age, user location and seasonal weather.

3. The automobile quality claim prediction method based on the survival analysis model as claimed in claim 1, characterized in that: The positive and negative samples include at least 30,000 sample data, and the ratio of positive and negative samples cannot be less than 10%.

4. A vehicle quality claim prediction system based on a survival analysis model, used to implement the vehicle quality claim prediction method based on a survival analysis model as described in any one of claims 1 to 3, characterized in that: It includes fault scenario module, fault factor module and report module. The fault scenario module is used to obtain the scenarios that may affect the quality failure of automotive parts. Obtain factors that affect the quality failure of automotive parts through the failure factor module; According to the information obtained by the fault collection module and the fault factor module, a claim model is established through the report module; according to the data of the claim model, a claim prediction model is established through the regression coefficient algorithm and the survival analysis model algorithm; it also includes a prediction data correction module, which compares the prediction results of the claim prediction model with the actual number of claims. If the average prediction accuracy of the claim prediction model is lower than the preset threshold, the claim prediction model is re-established.

Images