Neural network based controller for Cr 6+–Fe 2+ batch reduction process

Abstract

An automated pilot plant has been designed and commissioned to carry out online/real-time data acquisition and control for the Cr 6+–Fe 2+ reduction process. Simulated data from the Cr 6+–Fe 2+ model derived are validated with online data and laboratory analysis using ICP-AES analysis method. The distinctive trend or patterns exhibited in the ORP profiles for the non-equilibrium model derived have been utilized to train neural network-based controllers for the process. The implementation of this process control is to ensure sufficient Fe 2+ solution is dosed into the wastewater sample in order to reduce all Cr 6+–Cr 3+. The neural network controller has been utilized to compare the capability of set-point tracking with a PID controller in this process. For this process neural network-based controller dosed in less Fe 2+ solution compared to the PID controller which hence reduces wastage of chemicals. Industrial Cr 6+ wastewater samples obtained from an electro-plating factory has also been tested on the pilot plant using the neural network-based controller to determine its effectiveness to control the reduction process for a real plant. The results indicate the proposed controller is capable of fully reducing the Cr 6+–Cr 3+ in the batch treatment process with minimal dosage of Fe 2+.