Theory And Technology Of Intelligent Control For Shearer Steering

The aim of steering control of shearer is high coal recovery and less stone in coal. The most important is that it is benefit to safety of face. In traditional scheme of steering control of shearer, coal-rock interface is identified at first, the information is used to guide the drum directly then. However, coal-rock interface identification (CII) is so hard and zigzag that steering control of shearer can't be realized.A novel scheme is presented in the paper. In the scheme, CII is used firstly, the information of coal-rock interface in lane and historical face are fused at the same time. Based on the fused information, the coal-rock interface of the block is predicted. Then, the path is optimized when the method of mining and the capability of the machine in face are taken into account. At last, the drum is controlled to track the optimized path accurately. The system isconstructed as blackboard structure that is fit to solve complex problem. The structure is composed by blackboard and four knowledge-sources named coal-rock interface identification, coal-rock prediction, path optimization and arm control respectively. The four knowledge-sources work in parallel and asynchronously. In the scheme, the steering control isn't depended on the coal-rock interface identification badly thanks to information fusion. So, the system is more robust; the method of mining and aspect of ledge are taken into account, so the system is practical.In the coal-rock interface identification knowledge-source, the information of lane and historical face are fused by means of artificial neural network. The result of fusion indicates that the neural network can be fit the coal-rock interface if the data is accurate. However, it becomes inaccurate when the data isn't accurate. By means of fusion of the uncertain data and the result of coal-rock interface named the second fusion, the error of prediction is small enough to using practically.The path of drum is optimized by means of genetic arithmetic in path optimization knowledge-source. The result indicates that the arithmetic is fit to the problem.After kinematics and dynamics analyzing of machine of hydraulic servoarm, the feedback-error-learning arithmetic is used to controller that is a hybrid controller of neural network and PID. The result of simulation and experiment on machine show the advantage of the controller compared to the PID.