Industrial IOT Framework Design For Predictive Maintenance In Intelligent Manufacturing Environments

Abstract

In recent years, predictive maintenance is the best solution in many commercial and individual industries because it predicts the machine status before the failure occurs. But it contains some issues like high error, delay, and processing time also less prediction accuracy because of data complexity. So designed Honeybee-based Random Forest (HbRF) scheme to enhance the performance of predictive maintenance and is implemented using MATLAB tool. Initially, data are collected from the welding machine using IoT sensors and they are stored in the cloud. Then the collected data were sampled per second that are collected for one month or one week. Then preprocessing and decorrelator is performed to eliminate the errors, noise, missing values, and correlation in the input dataset. After that, feature extraction is utilized to extract the relevant information from the dataset. Moreover, update the honey bee fitness for continuous monitoring of machine features at a certain time. Finally, predict the machine status before the fault occurs. Consequently, the developed framework attained results are compared with other states of the techniques such as processing time, error, delay, prediction accuracy, precision, and recall.