A Modified Predictive Maintenance Approach Based on ARIMA Techniques and LSTM Architectures for IoT Industrial Robots Environments

Abstract

In today's Industry 4.0 environments, predictive maintenance is very important to assure that the industrial robots are efficient and reliable on its own with less human interference. This paper is showing and explaining the entire structure of predictive maintenance that combines IoT -enabled sensor data with hybrid ARIMA-LSTM models and machine learning methods to achieve predictive maintenance. The proposed system predicts potential failures by analyzing real-time data that is collected from the industrial robots. ARIMA is great for short-term forecasts. On the other hand, LSTM is used with long-term data. With the hybrid implementation of LSTM and ARIMA, the system has better prediction accuracy in fault prediction. Adding ioT and machine learning into these systems can help with prediction maintenance. As a result, the system has improvement in productivity and reduction in costs.