Study On Dry Control Of Slime Roller Based On Neural Network Identification

With the improvement of production and the proportion of washing coal industry, coal production increases accordingly. To realize the large-scale utilization of mud slurry, It had to be on the moisture content of coal slime and have no quality put forward higher requirements and specifications, in addition to some other way, such as burning coal-water slurry preparation、mixing and so on, Reusing is a slime after drum drying is worth considering, Not only can obtain higher returns in terms of economy, but also for resource recycling and conservation are the long-term practical significance.Drum drying is the most widely used technology. However, drum drying process has some complex features of highly nonlinear, uncertain, time-varying, big lag and multivariable coupling. At present, because of the interior structure of drum drying and the influence of random disturbance on the control system, it has lead that precise mathematical model in theory cannot get better control effect. Therefore, outlet temperature of the drum drying has not been get obvious improvement during the development of the automatic control level.In this paper, introduced the feedback and feedback-negative feedback control principle, analyzed the neural network on discernible model reference adaptive control, under the environment of MATLAB simulation, it can be seen that based on neural network identification of drum dryer outlet temperature of the single neuron model reference adaptive control system to control the outlet temperature model accord well with those of tracking, and the actual output is small and adjusting time of the error is small, the system response time is short and relatively stable control process.Based on the rising curve method, the paper gets the transfer function of the outlet temperature of the roller drier. The paper designs the single neuron controller and neural network RBF neural network controller, and simulates the traditional PID control and the above two kinds of controller. After pair wise comparison and analysis, it designs the single neuron PID model reference adaptive control system based on RBF neural network identification. The simulation results show that the control system has a better effect on model identification and tracking. It also has the characteristics of small controlled error, good robustness, quick system response, etc.