Research Of A Soft Sensor For Cobalt Ion Concentration In The Inlet Of Zinc Smelt Purification Process

The purification-cobalt removal with antimony trioxide and zinc powder is an important part of zinc hydrometallurgy process. The cobalt ion concentration in the entrance of paragraph II purification process had an important role in guiding optimal control in real time, but it was difficult to detect, affecting the development of production. The development status of zinc smelting and purification technology and soft sensor technology were studied. A soft sensor model was proposed based on method of time series phase space reconstruction, the hybrid genetic algorithm with iterative local search (ILS-GA algorithm) and neural network.Firstly, the historical laboratory data was pretreated to form a time series on the principles of same sampling interval. The attractor of time series was reconstructed through the introduction of chaos theory and phase space reconstruction theory to reproduce the dynamics behavior of cobalt ion concentration variation in the entrance of paragraph II purification process. Saturated correlation dimension and the largest Lyapunov exponent of the attractor were calculated, which showed the time series of cobalt ion concentration are chaotic.Secondly, on the basis of the identification of characteristics of cobalt ion concentration time series, a soft sensor modeling of cobalt ion concentration based on neural network was deeply analyzed. Because of easily getting into local dinky value, being difficult to determine network structure, initial weights having great arbitrariness. The train samples of model were obtained with the delay time and optimal embedding dimension of attractor. A hybrid genetic algorithm with iterative local search (ILS-GA algorithm) was proposed to optimize the network structure and learn neural network weights instead of the traditional BP algorithm. A soft sensor model based on a hybrid genetic algorithm with iterative local search and neural network was formed to predict the current cobalt ion concentration at the entrance of paragraphâ…¡purification process.Finally, take the paragraphâ…¡purification process of zinc hydrometallurgical in a smelting enterprise for example, a soft-sensor model based on a hybrid genetic algorithm(ILS-GA) and neural network for cobalt ion concentration in the entrance of paragraph II purification process had been formed. Two simulations of soft sensor model based on neural network method were done with the input data collected in the field by ILS-GA and BP algorithm methods. Simulation results showed that this method combined ILS-GA algorithm with neural network avoided the shortcoming of slow convergence and easily falling into local extreme point, and improved the model's estimation accuracy and generalization performance. The research of the paper had an important significance of study on the measurement method in real-time of cobalt ion concentration.