RUL Prediction with Multi-Task LSTM Under Sensor Faults
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In industrial RUL prediction scenarios, the presence of random missing values in monitoring data due to interference factors like vibration, dust, and electromagnetic interference poses a challenge for accurate Remaining Useful Life (RUL) prediction.
Innovation Solution
A multi-task learning-based method that preprocesses data with missing values using a sliding window, fully fuses spatio-temporal information with a deep LSTM module, and performs missing value imputation and RUL prediction in parallel using a multi-task learning approach, ensuring complete information is used for accurate predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional RUL prediction methods are used with missing sensor data, then the prediction process can proceed without complex processing, but the prediction accuracy deteriorates due to random missing values from sensor faults
Solution Approach 1:
The patent introduces an intermediary imputation model that acts as a mediator between the incomplete sensor data and the RUL prediction model. This intermediary component fills missing values by learning from available data patterns and correlations, thereby restoring data completeness without requiring the actual missing sensor readings. The imputation model serves as a bridge that transforms unreliable incomplete data into reliable complete data for accurate prediction
Solution Approach 2:
The patent applies preliminary action by performing data imputation before the RUL prediction process. The missing values are filled in advance using the imputation model, which learns from historical complete data patterns. This preliminary restoration of data completeness ensures that the subsequent prediction model receives high-quality input data, thereby improving prediction accuracy without requiring changes to the prediction algorithm itself
2Loss of information
If multi-sensor networks are deployed to collect monitoring data, then more comprehensive information is obtained for prediction, but the data quality deteriorates due to interference factors affecting multiple sensors
Solution Approach 1:
The patent converts the harmful effect of sensor interference and missing data into a benefit by using the imputation model to learn from the patterns in available data. Instead of treating missing values as harmful data loss, the system leverages the correlations between different sensors and time points to infer and fill missing information. The interference-induced missing data becomes an opportunity to demonstrate the robustness of the imputation approach
Solution Approach 2:
The patent applies universality by creating a unified imputation framework that handles multiple types of sensor failures and interference patterns simultaneously. The imputation model is designed to work with various sensor types and failure modes, providing a universal solution that restores data quality across different sensing conditions. This multi-functional approach allows the system to maintain information completeness regardless of which specific sensors are affected by interference
Data Source
AI summary
A multi-task learning-based remaining useful life prediction method under a sensor fault condition, including the following steps: firstly, preprocessing data with missing values by a sliding window to construct the data into data samples in a sequential pattern; then, fully fusing spatio-temporal information in the data by a deep long short-term memory (LSTM) module to extract implicit representations containing complete degradation information; next, inputting the implicit representations extracted from the deep LSTM module into a missing value imputation module and an RUL prediction task module by a multi-task learning method in parallel, thereby ensuring that the implicit representations contain as complete degradation information as possible with the aid of a missing value imputation task to obtain accurate RUL prediction results.