In-Vehicle Failure Prediction Using DTC and Operating Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automotive Original Equipment Manufacturers (OEMs) face challenges in predicting and preventing vehicle failures during the warranty period, leading to increased repair frequencies and potential large-scale recalls, which affect consumer perception and costs.
Innovation Solution
An in-vehicle predictive failure detection system using a statistical model and machine learning algorithms to analyze Diagnostic Trouble Codes (DTCs) and operating conditions, such as odometer readings and battery voltage, to predict vehicle failures and provide early warnings to operators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If predictive analytics models are implemented to detect vehicle failures early, then warranty expenses are reduced and consumer confidence is improved, but device complexity and data processing requirements increase
Solution Approach 1:
The system performs preliminary analysis by training machine learning models on historical DTC data before deployment. The trained models are then used to predict future failures by analyzing current DTC patterns, enabling early detection and preventive action before actual failures occur.
Solution Approach 2:
The patent introduces trained model objects as intermediaries between raw DTC data and failure predictions. These model objects serve as a bridge that translates complex diagnostic code patterns into interpretable failure probabilities, simplifying the overall system architecture.
2Measurement precision
If machine learning algorithms are used to analyze historical DTC data, then prediction accuracy is improved, but loss of time for data processing and model training increases
Solution Approach 1:
The system performs model training in advance using historical DTC data, creating pre-trained model objects that can be deployed for real-time predictions. This preliminary action separates the time-consuming training phase from the operational prediction phase, reducing real-time processing delays.
Solution Approach 2:
The patent creates trained model objects that capture the learned patterns from historical data. These model objects serve as reusable copies of the training results, allowing the system to make multiple predictions without re-processing the entire training dataset each time.
3Reliability
If the system monitors multiple operating conditions including odometer reading and battery voltage, then prediction reliability is improved, but device complexity and data collection requirements increase
Solution Approach 1:
The system uses a universal predictive model that can handle multiple types of input data (DTCs, odometer readings, battery voltage) through a unified analysis framework. The trained model objects are designed to process diverse data types consistently, reducing the need for separate analysis pipelines for each parameter.
4Productivity
If early warning notifications are provided to operators with recommended service intervals, then productivity is improved by preventing breakdowns, but loss of time for service scheduling increases
Solution Approach 1:
The system calculates and communicates recommended service intervals in advance based on predicted failure probabilities. By providing this information beforehand, operators can proactively schedule maintenance during convenient time windows before failures occur, minimizing disruption to vehicle operation.
Data Source
AI summary
Systems and methods for predictively detecting vehicle failure based on diagnostic trouble codes are provided. In one example, a method is provided, comprising determining a probability of failure of a vehicle based on one or more diagnostic trouble codes (DTCs); and indicating to an operator of the vehicle that failure is likely in response to the probability exceeding a threshold.