Lifetime Prediction Correction for Incomplete Repair Event Records
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The Kaplan-Meier method for predicting product lifetime is unreliable when events such as repairs are not accurately recorded, particularly in work machines where users may perform repairs with imitation parts, leading to inaccurate event history.
Innovation Solution
A lifetime prediction system that includes a server device with operation and history information databases, calculating an event data acquisition rate to correct lifetime predictions using the Kaplan-Meier method, accounting for accurately recorded events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the Kaplan-Meier method is used to predict product lifetime based on recorded events, then the prediction can be performed using existing survival analysis techniques, but the prediction reliability deteriorates when event records are inaccurate (such as when users perform repairs with imitation parts without accurate recording)
Solution Approach 1:
The patent introduces an intermediary correction mechanism that mediates between the inaccurate event records and the Kaplan-Meier prediction method. A correction coefficient is calculated based on the relationship between actual product lifetimes and predicted lifetimes from historical data, then applied to correct future predictions. This intermediary correction factor compensates for information loss in event records without requiring direct access to accurate event data.
Solution Approach 2:
The system implements feedback by using actual product lifetime outcomes to continuously refine the correction coefficient. The correction coefficient is updated based on the difference between actual lifetimes and predicted lifetimes from historical prediction data, creating a closed-loop system that improves prediction reliability over time despite ongoing inaccuracies in event recording.
2Loss of information
If regular inspections with specialist knowledge are performed to accurately record events, then event recording accuracy is improved, but the system complexity and operational burden increase
Solution Approach 1:
The system enables self-service by allowing the prediction model to automatically compensate for recording inaccuracies through the correction coefficient mechanism. Instead of requiring specialized inspections to ensure accurate recording, the system autonomously corrects prediction errors using historical feedback, reducing the need for complex inspection infrastructure and specialist involvement.
3Ease of repair
If users perform repairs themselves or use imitation parts, then ease of repair is improved, but event recording accuracy deteriorates
Solution Approach 1:
The correction coefficient acts as an intermediary that bridges the gap between user-performed repairs with potentially inaccurate recording and reliable lifetime predictions. The system doesn't require users to maintain accurate records; instead, the correction mechanism compensates for their recording inaccuracies using aggregated historical data from multiple products.
Data Source
AI summary
An objective is to provide a lifetime prediction system that allows a highly reliable lifetime prediction even when record of events having occurred in products are inaccurate. A lifetime prediction system 1 includes a server device 5 that predicts a lifetime of a product as a machine such as a work machine 2 or a machine part. The server device 5 includes an operation information database 52 that stores operation information for each of a plurality of products, a history information database 53 that stores history information indicating a history of event data as a record of an event having occurred in each of the plurality of products and an arithmetic processing device 54 that predicts the lifetime of the product. The arithmetic processing device 54 calculates an event data acquisition rate indicating a rate of product having the event data in which occurrence of the event is accurately recorded to the plurality of products, based on the operation information and history information, and predicts a corrected lifetime of the product based on the calculated event data acquisition rate.