The Research Of Iterative Learning Algorithm Based On The Network Delay

Networked Control System is a kind of distributed feedback control system with a closed control loops through the network. In NCS, sensors, controllers and actuators are communicated through the network. However, the performance of NCS may be decreased by the time delay and even the system will be unstable caused by it. In order to keep the system stable, an algorithm will be researched to compensate the network delay.Firstly, the concept of NCS is presented. The advantaged and the concomitant emergence of new problems are described, such as network time delay and dropout. The algorithms dealing with network delay which were used by predecessors are also reviewed.Secondly, some algorithms which are used to compensate the time delay are classified. The advantages and weaknesses of each type time delay compensation algorithm are analyzed, leading to the iterative learning algorithm.Thirdly, the effect of the network time delay to the linear system is analyzed and the time delay composition model is built. For a class of NCS with uncertain time delay, a PD-type learning algorithm is studied to compensate it. Based on the strictly repetition of the initial state, the sufficient conditions which guarantee the uniform convergence of the learning control is given and the limit output trajectories generated by the action of the learning control are also presented. Then, comparing with the efficiency of the P-type iterative learning control algorithm, it is shown that the effectiveness of the PD-type learning algorithm is more satisfied to compensate the delay system. Moreover, the output trajectories converge faster than the one of the P-type iterative learning control algorithm. When the range of the time delay becomes small, by the PD-type learning method, the state trajectories can be precisely tracked under the same number of iteration.Furthermore, base on the algorithm, the fuzzy adaptive PD, BP neural network PD, genetic PD and particle swarm optimization iterative learning algorithm are proposed. By analyzing the result of the simulation, it is shown that the above control algorithms are all effective.In order to verify the actual feasibility of the algorithm, in this paper, the water tank in the laboratory is controlled by the above time delay compensation algorithm through network. The circuit board is designed to acquire and control the water level. Specific features include:voltage scaling, voltage current conversion, serial transmission, analog to digital conversion, digital to analog conversion. And design the PC for the network control. Specific features include:serial transmission, network communication base on the TCP/IP, data storage. Finally, the water tank level is controlled by the algorithm, and the result is analyzed.At last, the conclusions and the prospects of the paper are proposed.