Flexible manufacturing processes, which are considered as being crucial to the fourth industrial revolution, are playing more and more essential role in modern industry nowaday. Almost all hardware and software manufacturers made every effort to ensure optimal performance with the lowest defection as possible. However, beside the high-accuracy operation of actuators, chemical processing industry specifically call for more stringent conditions. Even a tiny difference in the outcome of chemical reactions can result in subpar quality. As a result, monitoring operations must be quick and accurate enough to identify and isolate errors whenever system issues arise. This research investigates a data-driven estimation-based technique for process fault detection and diagnostic. According to this method, the discrepancy between the process responses and the estimated model responses is used to identify all process defects. The Kernel Principal Component Analysis (KPCA) is used to identify issues for fault detection in conjunction with an Artificial Neural Network (ANN) fault classifier model. The suggested approach has been validated using the stirred-tank heating process. The simulation results demonstrate how successful the suggested approach is.