Study Of The Control Strategy Of Aluminum Electrolytic Production Process

The reaction of aluminum electrolytic tank process is a complex、nonlinear、multivariable coupling、time-varying and big delay industrial process system, it’s extremely difficult to effective control it. With the computer was used in aluminum electrolytic production in the1960s, the level of control had been greatly improved, and the application of fuzzy control and neural network technology application made a big progress of aluminum electrolytic control. But the overall levels of aluminum industry in our country still have a large gap with the international advanced levels.In china, the adaptive control technology is used by most of the electrolytic aluminum manufacturers, with this control method, the alumina concentration can basically be maintained at a certain range, in order to guarantee the stability of the production process and achieve the goal of reducing energy consumption. However, this adaptive control method is not very satisfactory, in the control process,excess/insufficient amount alternate methods is used in feeding, it makes the control segment is too rough, resulting in the alumina concentration fluctuation range, the electrolysis process can not remain in a stable efficient range. Therefore, we need to study more rational control strategy to improve electricity’s efficiency and reduce electricity’s consumption.The paper first introduces the development of the aluminum industry、the mode of production、control process and the lack of control mode in the production of aluminum in China, then it proposed using fuzzy neural network adaptive system to control the production of alumina concentration, design the modular system and research the interface circuit, achieve more precise control of alumina concentration. Specific work as follows:First, the paper discuss how does the experts fuzzy control system be used on the blanking control in the production of aluminum, the system switch in three modes including the concentration workspace check、normal feeding control、non-normal feeding control,to monitor the resistance in electrolytic tank and on the basis of this it control the alumina concentration.Second, the paper study the fuzzy controller and fuzzy neural network controller, and discuss the structure, design methods and algorithms about these two controllers and the fuzzy neural network controller in particular, the paper also studied a self-learning algorithm, it make the system has more excellent robustness and self-adaptability.Third, research the system hardware, the control system project and the interface circuit. There be adopted modularization design concept, the system is divided into four functional modules:analog input channel, analog output channel, switch output channel, switch input channel. Analog input channel input the signal including the series of current、the series of voltage and the tank voltage onto the A/D conversion template, then input onto the computer for processing; analog output channel output control signals to control the alumina feeding device, it output the computer’s control signal through the D/A converter module to control the actuator, thereby to control the alumina feeding; switch input channel and switch output channel are mainly used to check and control the anode. So the main object of study are analog input and analog output channels, and switch input channels switch output channel are not detailed discussion.On the basis of ordinary fuzzy control, this paper adopt the project which uses a fuzzy neural network adaptive controller to control the alumina concentration, when the system dynamics features change, fuzzy neural network controller based on neural network identification can change control rules via online learning, so it’s effect is better than ordinary fuzzy control, the MATLAB simulation experiment has been verified it.