IoT Sensor RUL Prediction Using Wavelet-BiLSTM Autoencoders

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Accurately predicting the remaining useful life (RUL) of complex industrial machines is challenging due to varying operational modes and types of failures in different sub-components, making it difficult to use IoT sensor data effectively.

Innovation Solution

A system and method utilizing multilevel discrete wavelet data transformation in conjunction with a bidirectional long short-term memory (BiLSTM) based autoencoder deep learning architecture to improve the accuracy of RUL prediction from IoT sensor data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If a regression model is employed to estimate RUL parameters from time series data, then the prediction process is simplified, but the prediction accuracy deteriorates when machines operate in various operational modes or conditions

Engineering Contradiction:

Improveprediction process simplicityVSAvoidRUL prediction accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent transforms the input time series data through feature extraction and selection processes, changing the parameters from raw sensor readings to meaningful operational features. This allows the regression model to maintain simplicity while improving accuracy by operating on transformed data that better represents the underlying degradation patterns

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces feature extraction and selection as intermediary steps between data collection and RUL prediction. These intermediaries process the raw time series data to extract relevant patterns and characteristics, enabling the simple regression model to achieve better accuracy by working with processed rather than raw data

Inventive Principle:

Principle #24Intermediary (Mediator)

2Ease of manufacture

If traditional data-driven approaches are used for RUL prediction, then the implementation is straightforward, but the ability to handle varying operational modes and failure types deteriorates

Engineering Contradiction:

Improveimplementation straightforwardnessVSAvoidhandling of varying operational modes

Core Design Contradiction:

Ease of manufactureVSAdaptability or versatility

Solution Approach 1:

The patent segments the time series data into meaningful features and patterns through feature extraction. This segmentation allows the model to handle different operational modes and failure types separately, improving adaptability while maintaining implementation simplicity through modular processing steps

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent transforms the data from one-dimensional time series into multi-dimensional feature space through feature extraction and selection. This dimensional transformation enables the model to capture complex patterns across different operational modes and failure types, enhancing versatility without complicating the overall implementation approach

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

Data Source

PatentUS12321251B2System and method for predicting remaining useful life of a machine component

Publication Date: 2025.06.03 EUGENIE AI INC

AI summary

Some embodiments are associated with a system and method for deep learning unsupervised remaining useful life (RUL) prediction in Internet of Things (IoT) sensor networks or manufacturing execution systems. The system and method use multilevel discrete wavelet for raw data transformation and a bidirectional long short-term memory (BiLSTM) based autoencoder neural network for RUL prediction.