Research And Dynamic Simulation Of Anti - Surge Control Strategy For Ion Compressor

Centrifugal compressor is the heart of the power equipment in the industrial sectors such as petrochemical and oil and gas storage and transportation department, however a special problem of surge phenomenon always appears with the running of this kind of compressors. Not only will the surge phenomenon significantly deteriorate the performance of compressor decreasing the life and efficiency of the compressor units, but also may cause explosions creating a huge economic loss. Therefore it is high important to prevent the occurrence of surge phenomenon. Currently, the control technology of compressors has been monopolized by foreign companies. What is worse, a variety of advanced control methods have not yet been applied to the control of anti-surge. Based on this background, the subject of this assay is raised to develop the advanced control methods applying in the anti-surge of compressor.Firstly, the basic theory of centrifugal compressor is introduced, its surge characteristics and influencing factors are analyzed. Then the existing published anti-surge control literatures are summarized, and based on this summary the anti-surge control strategies including the basic control strategy and intelligent control strategy are proposed. According to the original control strategy, the function of differential control is added to the basic control strategy; whereas, the RBF network is introduced to the PI control through the intelligent control strategy to achieve the goal of the online self-tuning of the control parameters. While the realization of the PI control parameters are adjusted by the RBF network, the traditional gradient descent learning algorithm is improved and the algorithm of "k-means, RLS, gradient descent" is developed. The hybrid simulation platform based on HYSYS and Matlab is established, the dynamic model of centrifugal compressor is built in HYSYS, and the anti-surge controlling program is written by Matlab, thus getting the achievement of simulating the anti-surge control strategy. The simulation results show that the proposed strategy can not only effectively prevent the surge phenomenon of compressor, but also short the control margin line, and improve the economics of the compressor operation. What is more, according to the proposed RBF network learning algorithm of "k-means, RLS, gradient descent", the selection problem of initial value is solved, the fail of initial operation of the controlling system is prevented and to some extent the adaptive capability of RBF network is increased.