Vehicle Corrosion Estimation Using Weather-Linked Machine Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for predicting vehicle corrosion are unreliable and inefficient due to the lack of a clear correlation between corrosion, weather conditions, salt exposure, and vehicle usage, relying heavily on visual inspections by non-technical personnel or numerous manual checks.

Innovation Solution

A machine learning model trained using a database of vehicle usage and weather data, including corrosion sensor readings, to estimate corrosion based on real-time environmental conditions, reducing the need for frequent inspections.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If regular visual inspections are carried out by non-technical personnel, then corrosion detection is performed, but the accuracy and reliability of corrosion assessment is low

Engineering Contradiction:
Improvecorrosion detection reliabilityVSAvoidcorrosion severity assessment precision
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent replaces manual visual inspection with a machine learning-based automated assessment system. The system uses sensors to collect data about environmental conditions (temperature, humidity, salt exposure) and vehicle usage patterns, then applies trained machine learning models to predict corrosion risk and severity, eliminating the need for human visual assessment.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces machine learning models as intermediaries between raw sensor data and corrosion assessment results. These models process environmental and usage data to generate predictions about corrosion risk, serving as a mediator that translates complex sensor readings into actionable corrosion assessments.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If numerous manual inspections are conducted by skilled engineers, then corrosion data is collected, but the time and resources required are excessive

Engineering Contradiction:
Improvecorrosion prediction reliabilityVSAvoidinspection time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent enables the vehicle to self-monitor its own corrosion risk through integrated sensors and onboard processing. The system automatically collects data from environmental sensors and usage sensors, processes this data through machine learning models, and generates corrosion predictions without requiring external inspection personnel.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces the manual inspection process with an automated electronic system that continuously monitors environmental conditions and vehicle usage, using machine learning to predict corrosion risk in real-time, thereby eliminating the time-consuming nature of manual inspections.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If corrosion sensors are installed on test vehicles to collect training data, then accurate corrosion measurements are obtained, but the complexity of the vehicle system increases

Engineering Contradiction:
Improvecorrosion measurement precisionVSAvoidvehicle system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts the corrosion sensing and measurement functionality from the main vehicle system into a separate, dedicated corrosion estimation device. This modular approach allows corrosion monitoring to be added as an optional feature without fundamentally altering the core vehicle architecture.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent integrates multiple sensor types (environmental sensors for temperature and humidity, usage sensors for vehicle operation data) into a unified corrosion estimation system that serves multiple functions: environmental monitoring, usage tracking, and corrosion prediction, thereby justifying the added complexity through multi-functionality.

Inventive Principle:
Principle #6Universality (Multi-functionality)

4Measurement precision

If a machine learning model is trained with extensive weather and usage data, then corrosion prediction accuracy is improved, but the data processing and model training complexity increases

Engineering Contradiction:
Improvecorrosion prediction accuracyVSAvoidmodel training complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs machine learning model training in advance using historical weather data, salt exposure data, and vehicle usage data collected during a test phase. The trained model is then deployed for production use, separating the complex training process from the actual corrosion prediction operations.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent divides the machine learning system into distinct components: data collection from sensors, data preprocessing, model training using historical data, and deployment of the trained model for real-time predictions. This segmentation allows each component to be optimized independently.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4257951B1Corrosion estimating method for a vehicle
Publication Date: 2025.07.09 TOYOTA JIDOSHA KK
  • EP4257951B1 patent drawingFigure 1~2
  • EP4257951B1 patent drawingFigure 3
  • EP4257951B1 patent drawingFigure 4~6

AI summary

A training method for a machine learning model, and a corrosion estimating method using such a machine learning model, the training method comprising equipping at least one vehicle with at least one corrosion sensor, determining the location of the vehicle thanks to a GPS device provided in the vehicle, determining, in a weather station database, which weather station is the closest to the location of the vehicle, retrieving at least some weather data from said closest weather station, retrieving at least some corrosion data from said at least one corrosion sensor, building a training database (20) associating, for several moments over at least ten days, said corrosion data with said weather data, and training the machine learning model (30) on the training database (20) thus built.