Revolutionizing And Optimization Of Industrial Safety With A Deep Learning Framework For Predictive Risk Assessment And Mitigation

Abstract

In recent days, industry growth has been incredible in various sources by providing various technologies and frames in software and hardware, chemical, thermal, agriculture, and soon in multiple fields. Developing infrastructure, the environment, and the nature of work require safety and precautions for all employees. Analyzing the risk of various factors is important to improve safety and employee management by analyzing the risk assessment. At the existing levels, most techniques take features observation from employees to estimate risk factors and provide safety precautions. Analyzing improper features and labels creates more dimension, causing the precision false rate to degrade the prediction on risk assessment due to high er false positives. Addressing the problem and considering developing a new optimized deep learning system will improve prediction accuracy in risk assessment and improve industry safety measures. To address these issues, an Arima index feature analysis abased Deep learning using LSTM-gated RNN to predict the risk assessment and enhance the safety measures. The preprocessing is done by Min-max-Data Normalization Scaling Factor (DNSF). Then, an Automated Integrated Moving Average (ARIMA) is used to identify the variance scaling difference on feature levels. The adaptive XGboost feature section identifies the risk margins based on the stress impacts on reducing the non-relation features. The prediction is done by LSTM gated Recurrent neural network to determine the risk assessment. The proposed system improves the accuracy of prediction in risk assessment to improve safety measures and protect the employee. The accuracy attains higher precision, recall rate, sensitivity, specificity, f1 score, and reduced false rate and time complexity.