A Hybrid Strategy for Monitoring of Measured Critical Parameters for a Steel Ingot Manufacturing Facility

The assurance of quality is the most important parameter that every industry wants to achieve. In order to achieve good quality, a robust process for monitoring of measured critical parameters needs to be devised. The first pre-requisite for the devising of such strategy is to have a plethora of data concerning the critical process parameters. With the advent of modern sensor-based measuring facilities, the accumulation of relevant measurable data is possible in modern industrial scenario. However, to take advantage of the collected of data, it is prudent to devise a strategy for extracting maximum information about the process from the measured data and estimate the process health. The article suggests a feedforward neural network (FFNN) to transform the data pertaining to the phases of the industrial phases, instead of using complicated Kernel-based approaches. The devised monitoring model is further executed to segregate the faults in the process. Once faults are distinguished, an analysis of faults is done to understand the causes of the fault. These faults along with their possible causes may be communicated to the process engineers for remedial action. The practical application of the devised approach is verified via a case study from the steel plant which manufactures ingots of different varieties.