Corrosion prediction method
The method enhances corrosion prediction accuracy by integrating vehicle and weather data through machine learning, addressing information gaps and utilizing reliable data changes to improve prediction models.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- TOYOTA JIDOSHA KK
- Filing Date
- 2024-12-20
- Publication Date
- 2026-07-02
AI Technical Summary
Existing corrosion prediction methods lack sufficient information, particularly vehicle-specific data and environmental factors like snowfall and humidity, leading to decreased prediction accuracy, and require extensive data sets with short intervals, further compromising accuracy.
A method involving acquiring vehicle corrosion environment data, constructing a corrosion prediction model using machine learning with correlated explanatory variables, and integrating vehicle and weather data to predict corrosion based on location-specific information.
Improves corrosion prediction accuracy by associating vehicle and weather data that are highly correlated with corrosion, addressing information gaps and utilizing reliable data changes.
Smart Images
Figure 2026110023000001_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a corrosion prediction method, and particularly to a corrosion prediction method for predicting the amount of corrosion of a vehicle.
Background Art
[0002] Conventionally, in such a technical field, for example, there is one described in Patent Document 1. The corrosion prediction method described in Patent Document 1 includes steps of constructing training data that associates vehicle corrosion amount data with vehicle data and weather data, training a machine learning model on the training data with the vehicle corrosion amount data and the weather data to associate them, inputting weather data into the machine learning model to output a predicted value of the corrosion amount, and estimating the cumulative corrosion amount received by the vehicle based on the output predicted value of the corrosion amount.
Prior Art Documents
Patent Documents
[0003]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0004] [[ID=�5]]However, the corrosion prediction method described in Patent Document 1 above has a problem that it is difficult to accurately predict the amount of corrosion for the following reasons. First, in the corrosion prediction method described in this document, a machine learning model is trained with the vehicle's driving speed, driving direction, and weather data obtained from a weather observation station near the vehicle to construct a corrosion prediction model, but there is a lack of information contributing to the prediction of the amount of corrosion, such as vehicle information other than the driving speed and driving direction, snowfall amount, and humidity. For example, a snow melting agent sprayed as a road surface freezing prevention measure in a snowfall area mainly contains sodium chloride, which greatly contributes to the amount of corrosion of the vehicle. Also, changes in humidity greatly contribute to the amount of corrosion. The lack of this information leads to a decrease in the prediction accuracy of the amount of corrosion.
[0005] Furthermore, the corrosion prediction method described in the literature employs a training method that includes an artificial neural network, requiring a large amount of data. Therefore, it is necessary to shorten the data interval to increase the number of data points. However, in this corrosion prediction method, the data time interval is set to 1 to 120 minutes, which is relatively short and contradicts the fact that it takes time for the value of the change in the data to reach a reliable size. These reasons lead to a decrease in the accuracy of corrosion prediction.
[0006] This invention was made to solve these technical problems and aims to provide a corrosion prediction method that can improve the accuracy of predicting the amount of corrosion. [Means for solving the problem]
[0007] The corrosion prediction method according to the present invention is a corrosion prediction method for predicting the amount of corrosion of a vehicle, and is characterized by comprising: a first step of acquiring vehicle corrosion environment data including at least temperature, humidity, salt concentration, and driving time for each road surface condition, and vehicle corrosion amount data; a second step of constructing a corrosion prediction model by performing machine learning with the corrosion environment data acquired in the first step as explanatory variables and the corrosion amount data as the objective variable, and defining explanatory variables among the corrosion environment data that are highly correlated with the corrosion amount as corrosion accelerating factors; a third step of acquiring vehicle data including vehicle location information and meteorological data based on the vehicle's location information; a fourth step of constructing a corrosion accelerating factor selection model by performing machine learning with the vehicle data and meteorological data acquired in the third step as explanatory variables and the corrosion accelerating factors defined in the second step as the objective variable; and a fifth step of predicting the amount of corrosion of a vehicle based on vehicle data of the vehicle to be predicted, meteorological data based on the location information of the vehicle to be predicted, and the corrosion accelerating factor selection model constructed in the fourth step.
[0008] In the corrosion prediction method according to the present invention, in the second step, a corrosion prediction model is constructed by performing machine learning with corrosion environment data as explanatory variables and corrosion amount data as the objective variable, and explanatory variables from the corrosion environment data that are highly correlated with corrosion amount are defined as corrosion accelerating factors. In the fourth step, a corrosion accelerating factor selection model is constructed by performing machine learning with vehicle data and weather data as explanatory variables and the corrosion accelerating factors as the objective variable. In this way, vehicle data and weather data can be associated with corrosion environment data, that is, corrosion environment data can be replaced with vehicle data and weather data. Therefore, since the amount of corrosion of a vehicle can be predicted based on vehicle data and weather data that are highly correlated with corrosion amount, it is possible to prevent the lack of information that contributes to corrosion amount prediction as in the conventional method and to improve the accuracy of corrosion amount prediction.
[0009] In addition, in the fifth step, the amount of corrosion of the vehicle to be predicted is predicted based on vehicle data and weather data of the vehicle to be predicted, and a corrosion acceleration factor selection model. In this way, the amount of corrosion of the vehicle can be predicted using data that is highly correlated with the amount of corrosion, thereby improving the accuracy of the corrosion prediction.
[0010] In the corrosion prediction method according to the present invention, the fifth step is preferably a step in which vehicle data of the vehicle to be predicted and weather data based on the location information of the vehicle to be predicted are input into the corrosion acceleration factor selection model constructed in the fourth step to determine the corrosion acceleration factors, and the determined corrosion acceleration factors are input into the corrosion prediction model constructed in the second step to predict the amount of corrosion of the vehicle. In this way, the prediction accuracy can be further improved by using a corrosion prediction model in which corrosion environment data replaced with vehicle data and weather data are used as explanatory variables to predict the amount of corrosion of the vehicle.
[0011] In the corrosion prediction method according to the present invention, the fifth step preferably involves extracting data from the vehicle data and weather data that are highly correlated with the amount of corrosion, based on the vehicle data and weather data acquired in the third step and the corrosion acceleration factor selection model constructed in the fourth step, constructing a model by machine learning using the extracted data as explanatory variables and the amount of corrosion data acquired in the first step as the objective variable, inputting vehicle data of the vehicle to be predicted and weather data based on the location information of the vehicle to be predicted that are highly correlated with the amount of corrosion, into the constructed model, and predicting the amount of corrosion of the vehicle. In this way, vehicle data and weather data that are narrowed down to those related to the corrosive environment can be directly associated with the amount of corrosion data, so that the amount of corrosion of the vehicle can be predicted with even higher accuracy.
[0012] In the corrosion prediction method according to the present invention, it is preferable that the corrosion environment data further include the amount of salt deposition and the driving time for each type of road. By doing so, the amount of corrosion environment data can be increased, thereby improving the accuracy of the constructed corrosion prediction model and the accuracy of the defined corrosion acceleration factors. As a result, the accuracy of corrosion amount prediction can be further improved. [Effects of the Invention]
[0013] According to the present invention, the accuracy of predicting the amount of corrosion can be improved. [Brief explanation of the drawing]
[0014] [Figure 1] This is a flowchart illustrating the corrosion prediction method according to the first embodiment. [Figure 2] This is a schematic diagram illustrating a corrosion prediction method. [Figure 3] This is a flowchart illustrating the corrosion prediction method according to the second embodiment. [Modes for carrying out the invention]
[0015] Hereinafter, embodiments of the corrosion prediction method according to the present invention will be described with reference to the drawings. The corrosion prediction method described in the following embodiments is a method for predicting the amount of corrosion in a vehicle, and more specifically, it predicts the amount of corrosion caused by salt and / or moisture adhering to metal parts of a vehicle (e.g., fender panels, door outer panels, wheels, etc.).
[0016] [First Embodiment] Figure 1 is a flowchart showing a corrosion prediction method according to the first embodiment, and Figure 2 is a schematic diagram for explaining the corrosion prediction method. As shown in Figure 1, the corrosion prediction method of this embodiment comprises a step S11 for acquiring vehicle corrosion environment data and corrosion amount data, a step S12 for constructing a corrosion prediction model, a step S13 for defining corrosion accelerating factors, a step S14 for acquiring vehicle data and meteorological data, a step S15 for constructing a corrosion accelerating factor selection model, a step S16 for determining corrosion accelerating factors, and a step S17 for predicting the amount of corrosion of the vehicle.
[0017] Of the steps described above, step S11, which involves acquiring vehicle corrosion environment data and corrosion amount data, may be performed directly by a computer, or by having workers measure or collect the data and then input it into a computer, or by collaboration between workers and a computer. On the other hand, steps S12 to S17, which involve constructing a corrosion prediction model and predicting the amount of corrosion of the vehicle, are performed by a computer, but they do not necessarily have to be performed by the same computer.
[0018] Although not shown in the diagram, the computer, for example, constitutes a corrosion prediction device and includes an input unit for receiving data input, a CPU (Central Processing Unit) for executing calculations, a ROM (Read Only Memory) as a secondary storage device for storing the calculation program, and a RAM (Random Access Memory) as a temporary storage device for saving the calculation progress and temporary control variables. The computer performs the above-described steps by executing the stored program.
[0019] The step S11 of obtaining the corrosion environment data and the corrosion amount data of the vehicle corresponds to the "first step" described in the claims. In this step S11 of obtaining the corrosion environment data and the corrosion amount data of the vehicle, the corrosion environment data of the vehicle including at least the running time for each of temperature, humidity, salt concentration, and road surface condition, and the corrosion amount data of the vehicle are obtained respectively. In the present embodiment, the corrosion environment data of the vehicle further includes the salt adhesion amount and the running time for each road type in addition to the running time for each of temperature, humidity, salt concentration, and road surface condition.
[0020] The corrosion environment data of the vehicle is obtained, for example, by individual vehicle monitoring shown in FIG. 2. Specifically, the temperature and humidity are obtained, for example, by attaching a thermometer and a hygrometer to the vehicle and measuring the temperature and humidity. The salt concentration is obtained, for example, by measuring the droplets adhering to the vehicle with a salt concentration meter. The salt adhesion amount is obtained, for example, by wiping off the salt adhering to the vehicle and measuring the amount of the wiped-off salt. It is preferable that the salt concentration and the salt adhesion amount are measured after the vehicle runs every day.
[0021] The running time for each road surface condition is obtained, for example, by classifying the road surface conditions during running in the video recorded by a drive recorder attached to the vehicle into DRY, WET, and ICE, and then calculating the running time for each of DRY, WET, and ICE. It is preferable that the running time for each road surface condition is calculated every day.
[0022] The running time for each road type is obtained, for example, by classifying the road types traveled every day into highway, national road, prefectural road, and local road (city street or other general road), and then calculating the running time for each of highway, national road, prefectural road, and local road. The reason for classifying in this way is, for example, that in a snowfall area, snow-melting agents are more frequently sprayed on highways than on local roads.
[0023] On the other hand, the amount of corrosion on a vehicle can be obtained, for example, by attaching a corrosion sensor to the vehicle and measuring it with the corrosion sensor. Examples of corrosion sensors include electrical resistance corrosion sensors (RCM sensors) and wireless corrosion monitoring systems such as Aircorr. It is preferable that the corrosion amount data be divided into daily corrosion change amounts (unit: μm / day).
[0024] Step S12, which involves constructing a corrosion prediction model, and Step S13, which involves defining corrosion accelerating factors, correspond to the "second step" described in the claims. In Step S12, which involves constructing a corrosion prediction model, a corrosion prediction model is constructed by performing machine learning using the corrosion environment data obtained in Step S11, which involves acquiring vehicle corrosion environment data and corrosion amount data, as explanatory variables and the corrosion amount data as the dependent variable.
[0025] Specifically, we first create training data with temperature, humidity, salinity, salt deposit amount, driving time for each road surface condition, and driving time for each road type as explanatory variables, and corrosion amount data as the dependent variable. Then, we build a corrosion prediction model by performing machine learning on the created training data using a random forest (i.e., using a random forest to learn the relationship between each explanatory variable and the dependent variable).
[0026] In step S13, which defines corrosion accelerating factors, explanatory variables from the corrosion environment data that are highly correlated with the amount of corrosion are defined as corrosion accelerating factors. Specifically, using the corrosion prediction model constructed in step S12, items highly correlated with the amount of corrosion are selected from temperature, humidity, salt concentration, salt deposition amount, driving time for each road surface condition, and driving time for each road type, and these selected items are defined as corrosion accelerating factors.
[0027] Step S14, which involves acquiring vehicle data and weather data, corresponds to the "third step" described in the claims. In step S14, vehicle data, including vehicle location information, and weather data based on vehicle location information are acquired. As shown in Figure 2, the vehicle data is, for example, part of vehicle big data and includes vehicle location information, speed information, engine speed, accelerator input, wheel rotation speed, brake input, ABS activation, hazard signals, steering angle, turn signals, wipers, headlights, fog lights, etc.
[0028] These vehicle data include connected data collected from connected cars, for example, and include navigation probe data and CAN (Controller Area Network) data. Navigation probe data is data that includes location information, speed information, etc., generated by the navigation system installed in the vehicle. CAN data is data obtained and stored through communication (wired and / or wireless) between many ECUs (Electronic Control Units) installed in the vehicle, and includes data on vehicle behavior such as driving, turning, and stopping, as well as data on vehicle status such as wiper and light operation.
[0029] This vehicle data is acquired by a computer via communication (wired and / or wireless) and stored in a storage device. In this embodiment, the vehicle data is acquired and stored from the cloud (e.g., a data center) via, for example, a DCM (Data Communication Module) of a wireless communication module.
[0030] Furthermore, as shown in Figure 2, vehicle big data includes not only vehicle data but also weather data and map data. Weather data is obtained from the weather station closest to the vehicle, based on the vehicle's location information. Weather data includes, for example, temperature, humidity, precipitation, and snowfall.
[0031] Map data includes, for example, road classifications or road names, and includes, for example, expressways (national expressways and motorways), national roads, prefectural roads, and other general roads. Therefore, the type of road a vehicle is traveling on can also be obtained from map data based on the vehicle's location information.
[0032] Step S15, which involves constructing a corrosion-accelerating factor selection model, corresponds to the "fourth step" described in the claims. In this corrosion-accelerating factor selection model construction step S15, the vehicle data and weather data acquired in step S14 are used as explanatory variables, and the corrosion-accelerating factors defined in step S13 are used as the objective variable. A corrosion-accelerating factor selection model is constructed by performing machine learning.
[0033] Specifically, training data is created using vehicle data and weather data as explanatory variables and corrosion-promoting factors as the dependent variable. A corrosion-promoting factor selection model is then constructed by performing machine learning on this training data using a random forest. This allows vehicle data and weather data to be correlated with corrosion environment data.
[0034] Step S16 for determining corrosion accelerating factors and Step S17 for predicting the amount of corrosion of a vehicle correspond to the "fifth step" described in the claims. In Step S16 for determining corrosion accelerating factors, the amount of corrosion of a vehicle is predicted based on vehicle data of the vehicle to be predicted, weather data based on the location information of the vehicle to be predicted, and the corrosion accelerating factor selection model constructed in Step S15 for constructing the corrosion accelerating factor selection model. More specifically, first, vehicle data of any vehicle to be predicted and weather data based on the location information of the vehicle are obtained. Next, the obtained vehicle data and weather data are input into the constructed corrosion accelerating factor selection model to determine the corrosion accelerating factors among the corrosion environment data.
[0035] Next, in the vehicle corrosion amount prediction step S17, the corrosion accelerating factors obtained in the corrosion accelerating factors determination step S16 are input into the corrosion prediction model constructed in the corrosion prediction model construction step S12, and a predicted value for the vehicle corrosion amount is calculated.
[0036] In the corrosion prediction method of this embodiment, in step S12 of constructing a corrosion prediction model, a corrosion prediction model is constructed by performing machine learning with corrosion environment data as explanatory variables and corrosion amount data as the objective variable. In step S13 of defining corrosion accelerating factors, explanatory variables from the corrosion environment data that are highly correlated with corrosion amount are defined as corrosion accelerating factors. Furthermore, in step S15 of constructing a corrosion accelerating factor selection model, a corrosion accelerating factor selection model is constructed by performing machine learning with vehicle data and weather data as explanatory variables and corrosion accelerating factors as the objective variable. In this way, vehicle data and weather data can be associated with corrosion environment data, that is, corrosion environment data can be replaced with vehicle data and weather data. This makes it possible to predict the amount of corrosion of a vehicle based on vehicle data and weather data that are highly correlated with corrosion amount, thereby preventing the lack of information that contributes to corrosion amount prediction as in the conventional method and improving the accuracy of corrosion amount prediction. Moreover, since the amount of corrosion of a vehicle can be predicted by selecting data that is highly correlated with corrosion amount (in other words, data that contributes highly to corrosion amount prediction), the accuracy of corrosion amount prediction can be improved.
[0037] In addition, in step S16, the step for determining corrosion acceleration factors, vehicle data and weather data of the vehicle to be predicted are input into the corrosion acceleration factor selection model to determine the corrosion acceleration factors among the corrosion environment data. Then, in step S17, the vehicle corrosion amount prediction is performed by inputting the determined corrosion acceleration factors into the corrosion prediction model to predict the vehicle corrosion amount. In this way, the prediction accuracy can be improved by using a corrosion prediction model that uses corrosion environment data replaced with vehicle data as explanatory variables to predict the vehicle corrosion amount.
[0038] Furthermore, the corrosion prediction method of this embodiment uses vehicle data, weather data, corrosive environment data, and corrosion amount data, which are numerous and whose data change values are reliable, thus further improving the accuracy of corrosion amount prediction compared to conventional methods.
[0039] [Second Embodiment] A second embodiment of the corrosion prediction method will be described below with reference to Figure 3. The corrosion prediction method according to the second embodiment differs from the first embodiment described above in that, instead of the step S16 for determining corrosion acceleration factors and the step S17 for predicting the amount of corrosion of the vehicle, it includes a step S26 for extracting data highly related to the amount of corrosion, a step S27 for constructing a second corrosion prediction model, and a step S28 for predicting the amount of corrosion of the vehicle. Other aspects are the same as in the first embodiment, so a redundant explanation will be omitted.
[0040] As shown in Figure 3, the corrosion prediction method of this embodiment comprises the steps of acquiring vehicle corrosion environment data and corrosion amount data (S21), constructing a first corrosion prediction model (S22), defining corrosion accelerating factors (S23), acquiring vehicle data and weather data (S24), constructing a corrosion accelerating factor selection model (S25), extracting data highly correlated with corrosion amount (S26), constructing a second corrosion prediction model (S27), and predicting the amount of corrosion of the vehicle (S28).
[0041] Steps S21 to S25, from acquiring vehicle corrosion environment data and corrosion amount data to constructing a corrosion acceleration factor selection model, are the same as steps S11 to S15 described in the first embodiment. However, to distinguish it from the corrosion prediction model described in step S27 of constructing the second corrosion prediction model, the corrosion prediction model in step S22 of constructing the first corrosion prediction model is referred to as the first corrosion prediction model. This first corrosion prediction model is the same as the corrosion prediction model in the first embodiment.
[0042] The steps S26 to S28, which involve extracting data highly correlated with the amount of corrosion, correspond to the "fifth step" described in the claims. First, in step S26, which involves extracting data highly correlated with the amount of corrosion, data highly correlated with the amount of corrosion is extracted from the vehicle data and weather data based on the vehicle data and weather data acquired in step S24 and the corrosion acceleration factor selection model constructed in step S25.
[0043] For example, among the vehicle data, vehicle speed, accelerator input, and brake input are related to water exposure when driving on wet roads, so the data for vehicle speed, accelerator input, and brake input are extracted. On the other hand, among the meteorological data, temperature, humidity, and precipitation are related to the progression of corrosion, and snowfall is related to the application of de-icing agents, so the data for temperature, humidity, precipitation, and snowfall are extracted.
[0044] In step S27, the construction of the second corrosion prediction model, the data extracted in step S26, which is highly correlated with the amount of corrosion, is used as explanatory variables, and the corrosion amount data obtained in step S21, which is used as the vehicle corrosion environment data and corrosion amount data acquisition step, is used as the objective variable. By performing machine learning, the second corrosion prediction model is constructed.
[0045] Specifically, training data is created using vehicle data and weather data highly correlated with corrosion amount, extracted in step S26 (data extraction step), as explanatory variables, and corrosion amount data obtained in step S21 (acquisition step of vehicle corrosion environment data and corrosion amount data), as the dependent variable. A second corrosion prediction model is then constructed by performing machine learning on the created training data.
[0046] Next, in the vehicle corrosion prediction step S28, first, vehicle data of the vehicle to be predicted and weather data based on the location information of the vehicle to be predicted are acquired. Next, from the acquired vehicle data and weather data of the vehicle to be predicted, data highly related to corrosion are extracted from this data using the same method as in the data extraction step S26. Then, the extracted data (i.e., vehicle data and weather data highly related to the corrosion amount of the vehicle to be predicted) are input into the second corrosion prediction model constructed in the second corrosion prediction model construction step S27, and the predicted value of the corrosion amount is calculated.
[0047] The corrosion prediction method according to this embodiment provides the same effects as the first embodiment described above, and further includes steps S26 to S28 for extracting data highly correlated with the amount of corrosion, thus providing the following additional effects.
[0048] Specifically, in step S26, which involves extracting data highly correlated with corrosion amount, data highly correlated with corrosion amount is extracted from the vehicle data and weather data obtained in step S24, and the corrosion acceleration factor selection model. In step S27, which involves constructing a second corrosion prediction model, the extracted data is used as explanatory variables and the corrosion amount data as the dependent variable, and machine learning is performed to construct a second corrosion prediction model. In step S28, which involves predicting the corrosion amount of a vehicle, the vehicle data of the vehicle to be predicted and weather data based on the location information of the vehicle to be predicted, which are highly correlated with corrosion amount, are input into the second corrosion prediction model to predict the corrosion amount of the vehicle. In this way, vehicle data and weather data that are narrowed down to those related to the corrosive environment can be directly associated with corrosion amount data, making it possible to predict the corrosion amount of a vehicle with even greater accuracy.
[0049] Although embodiments of the present invention have been described in detail above, the present invention is not limited to the embodiments described above, and various design modifications can be made without departing from the spirit of the invention as described in the claims. [Explanation of symbols]
[0050] S11: Acquisition of vehicle corrosion environment data and corrosion amount data; S12: Construction of corrosion prediction model; S13: Definition of corrosion accelerating factors; S14: Acquisition of vehicle data and meteorological data; S15: Construction of corrosion accelerating factor selection model; S16: Determination of corrosion accelerating factors; S17: Prediction of vehicle corrosion amount
Claims
1. A corrosion prediction method for predicting the amount of corrosion in a vehicle, The first step involves acquiring vehicle corrosion environment data, which includes at least temperature, humidity, salinity, and driving time for each road surface condition, and vehicle corrosion amount data, respectively. The second step involves constructing a corrosion prediction model by performing machine learning using the corrosion environment data obtained in the first step as explanatory variables and the corrosion amount data as the objective variable, and defining explanatory variables from the corrosion environment data that are highly correlated with the corrosion amount as corrosion accelerating factors. A third step involves acquiring vehicle data including the vehicle's location information and weather data based on the vehicle's location information, The fourth step involves constructing a corrosion acceleration factor selection model by performing machine learning using the vehicle data and weather data acquired in the third step as explanatory variables and the corrosion acceleration factor defined in the second step as the dependent variable. A fifth step in which the amount of corrosion of a vehicle is predicted based on vehicle data of the vehicle to be predicted, weather data based on the location information of the said vehicle to be predicted, and the corrosion acceleration factor selection model constructed in the fourth step, A corrosion prediction method characterized by comprising the following features.
2. The corrosion prediction method according to claim 1, wherein the fifth step is to input vehicle data of the vehicle to be predicted and weather data based on the location information of the vehicle to be predicted into the corrosion acceleration factor selection model constructed in the fourth step to determine the corrosion acceleration factors, input the determined corrosion acceleration factors into the corrosion prediction model constructed in the second step to predict the amount of corrosion of the vehicle.
3. The corrosion prediction method according to claim 1, wherein the fifth step is to extract data from the vehicle data and weather data that are highly correlated with the amount of corrosion, based on the vehicle data and weather data acquired in the third step and the corrosion acceleration factor selection model constructed in the fourth step, construct a model by performing machine learning with the extracted data as explanatory variables and the amount of corrosion data acquired in the first step as the objective variable, and input data from the vehicle data of the vehicle to be predicted and weather data based on the location information of the vehicle to be predicted that are highly correlated with the amount of corrosion, into the constructed model, and predict the amount of corrosion of the vehicle.
4. The corrosion prediction method according to claim 1, further comprising salt deposition amount and driving time for each type of road as the corrosion environment data.