Research On Automatic Control System Of Hump Push Peak Speed Based On Fuzzy Neural Network

Hump is the core equipments of the hump disintegration operations and hump marshalling operations. The efficiency of hump peak of operation has a very big effect on the efficiency of the whole marshalling station. Generally speaking, the modem system of automatic hump yard should include remote control for humping speed, automatic switching control of humping yard by route, automatic rolling down speed control, and the processing system of synthetic information in marshalling. Under today's technical factors, the wireless remote control of hump engines is used successfully, and it is connected with the system of automatic rolling down speed control. If we do the research of the hump push peak operation and train's rolling operation as a whole operation, then we can realize the optimization of the whole hump operations.The traditional push peak speed is given by the operator using his artificial experience, but due to the subjectivity and uncertainty of the artificial experience, and the hump push peak speed control process is influenced by many factors and is a nonlinear and time-varying process, the traditional push peak speed will make big error sometimes. With the increasing of the hump marshalling operations, in order to increase the hump efficiency in the exiting basis, we must research of the push speed and the real-time push speed of the hump push peak operations.In the hump yard, on the basis of field practical operation experience and the actual situation, operator usually can produce a more appropriate, more precise push peak speed through the brain's sketchy judgment. Hump push peak process is a multivariable nonlinear time-varying system; traditional control method needs to establish mathematical model of the controlled objects, while the model of hump push peak process cannot exactly is established. So, we only to research the hump push peak speed and real-time push peak speed by using fuzzy controller and neural network. The main contents are as follows:Firstly, introduce the hump automation system and the status of the research of hump push peak, secondly analyze the existing hump push peak system, and points out the deficiency and improving direction; thirdly the various factors which affect hump push peak process are analyzed and summed up, then put forward a research hump push peaks theory based on fuzzy neural network. After determining the idea, the first step is established the fuzzy distribution of the variables which influence hump push peak optimal push peak speed and real-time push peak speed; the second step respectively on optimal push peak velocity model and real-time push peak velocity model, design a neural network to replace the fuzzy reasoning machine, the third step selected hump yard actual operation data as the training sample, based on neural network of sample training, automatically generate fuzzy control rules. These two models combine the respective advantages of neural network and fuzzy control, the expert's or skilled operator's control strategies are converted to control functions, then can timely adjust and modify the parameters such as fuzzy control rules and membership functions by using the neural network's self-learning and self-revised function. In this way the adaptive of the model can be achieved, then can achieve the better control of the controlled objects.The computer simulation shows that after study these two fuzzy neural networks could accurately reflects the situation of the hump yard.