The Application Of Neuron Self-adapted PID Controller In Electric Stoves Of Mechanical Test Machine

The technique of PID's control is one kind of control technique that the application is very universal , and all has the extensive application at a lot of aspect at present . This dissertation according to the sequence of " puttingforward problems------theory analysis------simulation experiment------pointingout the shortcoming------putting forward improving the measure " , leadsrespectively integral separation and Smith estimating ware and PSD's control algorithm in advance with following in order and advancing step by step , and systematically analyses and studies on the foundation expounding traditional PID's regulator and neuron PID's controller.The most before all others the following tradition PID's control theory of this text is started with , and with the electric stove of the mechanical test machine is the controlled device , uses PID to control and leads the control algorithm of Smith-Estimator in advance , and does not gain satisfied control effect .Then this dissertation has discussed respectively to single neuron PID's controller based on the regular of Delta and Hebb's study .For the use of self-learning and the selfadaption ability of neuro to improve afterwards to PID's regulator , the controller can certainly traditional PID's controller of settlement in the level that does not change the online real time and puts in order the shortcoming to decide the parameter.Meanwhile,it effectively improved the unstable aspect owing to the system architecture and the variation of the parametric.But the simulation object in the electricity heating furnace indicates that when the object skew or the controller is increaseed greatly , the output from a system can appear that the overshoot is bigger and the vibration aggravates even to the appearance of the unstable.Finally ,this dissertation has deeply study the algorithms baseds on integral separations and Smith-Estimator and PSD etc. Then it organiclyly combine them together with neuron PID's control respectively. By way of a large number of simulation experiments , the object indicates that the neuron controller under the algorithm of integral separation can effectively surmount the overshoot and vibration of export when the system input changes suddenly. The neuron controller of Smith-Estimator can avoid the circumstances that time lag of object is too great to stabilize. And the neuron PID controller based on PSD's algorithm can effectively make up the shortcoming that the increase of ordinary neuron controller can't realize the online adjustment.