The efficiency of technological processes particularly depends on the efficiency of control system used. The construction of control system on the base of traditional technology for complicated processes characterizing with non-linearity and uncertainty is not enough satisfy such characteristics as high speed, reliability, adequacy, accuracy of the model. In this condition one of perspective way of construction of control system is the use softcomputing element s such as neural technology and genetic algorithm that satisfy above characteristics of the system.
In the paper using genetic algorithm the developments of PID- and recurrent neural controllers for technological processes control by are considered. The synthesis of the PID controller by the traditional methods requires to posses a great deal of control system knowledge, tuning experience, full information about control object. The use of genetic algorithm(GA) allows to automate the tuning process, and does not require to have much domain knowledge. Using genetic operators – selection, crossover and mutation operator the tuning of the PID controllers coefficients is carried out. The synthesis procedure and result of simulation of control system with PID controller are described.
The development of control system based on recurrent neural network is described. The learning of the recurrent neural controller is performed on the base of GA. Using this algorithm and desired time response characteristics of the system the synthesis of neural controller for technological process control is carried out. The results of simulation of the neural control system are described.