Fleet Vehicle Component Life Prediction for Prescriptive Maintenance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Transportation systems face challenges in predicting the remaining useful life of components, such as tires, and unforeseen incidents in vehicle fleets, leading to increased operational costs and customer dissatisfaction due to inadequate maintenance planning and lack of data on component conditions.
Innovation Solution
A system and method that utilize operating conditions, survival analysis algorithms, and deep neural networks to predict the remaining useful life of components and anticipate future incidents, allowing for prescriptive maintenance and optimal component allocation, leveraging both proprietary and public databases to model life expectancy and criticality of faults.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensors are installed on components to provide physical data about component condition, then measurement precision of component condition is improved, but device complexity and cost increase
Solution Approach 1:
The patent uses data from existing vehicle sensors (engine sensors, telematics data) to create proxy measurements of component conditions rather than installing dedicated sensors on each component. For example, engine temperature and vibration data are used to infer tire and transmission component states, creating virtual copies of direct measurement data through correlation models.
Solution Approach 2:
The system introduces an intermediary layer of predictive analytics and machine learning models that translate readily available vehicle operational data into component condition indicators. These intermediary algorithms process engine data, telematics, and maintenance records to derive component health metrics without requiring direct component sensors.
2Reliability
If preventive maintenance is performed based on predicted remaining useful life, then reliability of component operation is improved, but loss of time for maintenance operations increases
Solution Approach 1:
The system performs preliminary predictive analysis to determine remaining useful life of components before failures occur. By calculating RUL (Remaining Useful Life) predictions using survival analysis models and historical data, the system schedules maintenance activities in advance during optimal time windows, preventing both premature maintenance and unexpected failures.
Solution Approach 2:
The maintenance scheduling system dynamically adjusts maintenance timing based on real-time component condition predictions and vehicle operational patterns. Rather than fixed schedules, the system continuously updates RUL predictions and optimizes maintenance windows to minimize fleet downtime while ensuring component reliability, adapting to actual component degradation rates.
3Reliability
If fleet operations respond to unforeseen incidents, then service level quality is maintained, but productivity is reduced due to capacity loss
Solution Approach 1:
The system performs preliminary predictive analysis to forecast potential vehicle failures and incidents before they occur. By predicting component failures, accidents, or breakdowns in advance using historical data and machine learning models, the system enables proactive fleet reconfiguration and resource allocation to maintain service levels without losing productivity to unexpected downtime.
Data Source
AI summary
Systems, methods, and processing nodes determine and perform preventive maintenance on a transport vehicle in a transportation system. The method includes extracting features for previous incidents that have occurred to a plurality of transport vehicles in the transportation system. The method also includes determining a criticality of types of incidents based on the features extracted. The method includes predicting, based on the criticality of types of incidents and the features extracted, details of at least one future incident for a first transport vehicle from the plurality of transport vehicles. The details include a predicted type of the at least one future incident, a predicted time of the at least one future incident, and a predicted criticality of the at least one future incident. Additionally, the method includes performing a prescriptive action for the first transport vehicle to mitigate the at least one future incident in the first transport vehicle.