Adaptive Psd Algorithm For Multi-temperature Measurement And Control System

With the continuous development of science and technology of productive forces, objects in the actual process of industrial production become more and more complex, and some obvious characters were shown, such as MIMO,time-variability,coupling,large-delay,high-indicators needed and so on. For such complex systems and controlled plants, it is very difficult to establish its accurate module and get a good control result with traditional control algorithms. So model free and self-tuning on line was required to realize real-time control.Among many plants having characters as above, Multi-temperature zone measurement&control system is a common type. Not only the plant have parameter uncertainty, but also it have strong coupling factor because of its multi-temperature zones structure. So because of large-delay,strong coupling,model uncertainty, especially this paper study one kind of self-tuning PSD algorithm, which doesn't need the plant's model and can overcome the plant's time-delay. The paper validate the self-tuning PSD algorithm in the temperature plant with time-delay, meanwhile some improved methods was prosposed. Then, the simulation results have proved that the algorithm is effective in applications.Meanwhile, the PSD algorithm was generalized in a multi-temperature zone measurement&control system. That the easy realization in applications of traditional PSD algorithm combine with the advantages of neuron network for model free control, is a good method for the plant having uncertainty model. By combining the advantages of those two methods, and a new structure of PSD network was proposed finally. Furthermore, various kinds of PSD network controller were compared in this paper, and the advantage of PSD network algorithm was analyzed in theory.Lastly, according to the multi-temperature zone plant with time delay and strong coupling, a kind of PSD network algorithm was applicated, and the approach of algorithm is given, too. The simulation results have proved that the algorithm is availability and adaptability.