Design of Bao-steel coal moisture control system based on neural network method

Abstract

Coal moisture control process (CMC) in Bao-Steel is an open-loop nonlinear system. An effective control method for CMC is provided in this paper. First, principal parameters of the system are selected by statistical analysis of history data and summary work experience; Second, BP Neural Network is specified and its structure is determined by repeated experiments. Then, calculation of heat balance is used to verify the efficiency of the method; At last, the system is put into operation and produces a good effect.Coal moisture control [1][2] (CMC) is a new energy saving technology developed in recent years. It is a technology by heating directly or indirectly to reduce and stabilize the water of coal into the furnace. By this step, it can reduce coking energy consumption, improve the production capacity, and improve the quality and stable operation of coke oven.In 2008, Bao-steel finished building a coal moisture control technology workshop. Because this device is completed by Bao-steel independently and nothing can be learned from the experience, some difficulties were encountered during the operation. The difficulties existing are that coal moisture control system is a multi-parameter, strong coupling, nonlinear delay system. So the system model is difficult to establish. In the daily production process, the coal-moisture tested in laboratory is lagged behind the production about 15 hours. So it is difficult to guide production operations in time. Therefore, analyzing the operation theories and modeling the existing coal moisture control system, realizing automatic online coal moisture control has important practical significance.To model for the coal-moisture, Ergun [3] proposed a linear partial differential equation. But the assumption is that the water evaporation only occurs in solid surface. So the results are not satisfactory. Gao [4] established a water mass balance and heat balance mathematical model, but an actual coal-moisture control model was not given. Xie [5] proposed coal-moisture process is in line with page model under certain conditions, but the coal-moisture control system in Bao-steel is far away from their qualification. PID control method is the most commonly used method for general process control system, but the existing coal-moisture control system cannot measure online. So it is not a closed-loop control. In this paper, the experience of workers is summarized, and neural network method is used to model and control the system. Then, heat balance calculations are used to prove its validity. Finally, the feasibility and effectiveness of this method is verified by laboratory and practice data.