Research On Spreader Model And Neural Network Control Technology Of Leveling System

Concrete spreader is one of the machines that have been widely used for road pavement nowadays, and the Automated Leveling Device (ALD) takes an important role in the performance of the whole machine, to get the flatness of the road and to improve the performance of the pavement. With the Equalization Beam and traditional controlling methods, traditional ALD has got a good performance. But to adapt to the requirement of the high order road, the ALD still needs to be improved. Here we try to test the advanced sensors and Artificial Neural Network (ANN) controller, using the computer for a simulation, to get a better performance.In order to achieve this object, this text mainly study lectotype design of spreader leveling hydraulic control system. On the basis of reference and analysis massive existing products parameter. Reference the existing spreader, this text designed spreader experiment model on the basis of lectotype design, established leveling device approximate mathematical model. And analyzing the stability and the dynamic characteristic of transverse and longitudinal slope control system. According to this control model this text has already carried on the direct feedback control, conventional digitized PID control system analog simulation. Determined the convention control mode controlled parameter and its the control effect. And uses the neural network inversion control method based on this, recognizing and controlling leveling system and has carried on the training and the computer simulation to it.In the research we found the traditional PID control system may obtain the good control effect. But needs to trial and error many times. But the parameters adjustment must especially careful, is extremely sensitive, sometimes has bad control effect vibration, and even loses steadily. But using approaches ability of the BP neural network function, may obtain the better control effect in a wider scope. In the control process uses two BP network. One is used as NNI recognizing the model, another as neural network control device (NNC).But first Off-line recognizes controlled device, make sure NNC initial weights. Then carries on the neural network on-line control again. Thus can further enhance the...