Avoiding Environmental Consequences of Equipment Failure via an LSTM-Based Model for Predictive Maintenance
Haiyue Wu, Aihua Huang, John W. Sutherland
Abstract
This paper builds on the previously developed maintenance strategy of predictive maintenance (PdM) that can detect the incipient breakdown of a system and determine the condition of in-service equipment to estimate the maintenance scheduling. Here, a new data-driven model based on Long Short-Term Memory (LSTM) Recurrent Neural Networks (RNN) algorithm is proposed to detect the degradation of a manufacturing system and predict its future health condition for PdM. To validate the accuracy and efficiency of the proposed model, a motor bearing failure process is used to demonstrate the proposed method.
Topics & Concepts
Predictive maintenanceRecurrent neural networkLong short term memoryScheduling (production processes)Computer scienceCondition monitoringReliability engineeringProcess (computing)Condition-based maintenanceEngineeringArtificial neural networkMachine learningArtificial intelligenceOperations managementOperating systemElectrical engineeringMachine Fault Diagnosis TechniquesReliability and Maintenance OptimizationQuality and Safety in Healthcare