Density Control Of Dense Medium Suspension Based On The Model-Free Adaptive Control

During the process of coal preparation, the heavy medium separation has become the main sorting method of the coal preparation because of its wide separation size range, high precision, strong adaptability and easy to realize automation and so on. As the most important factor in heavy medium coal preparation process, medium suspension density directly influence the coal products quality and separation efficiency. Therefore, it is particularly important to control the density of dense medium suspension.Suspension density control system is a highly nonlinear, time-varying, strong coupling, large time delay process. So the precise model of the controlled object is difficult to be established. The traditional control algorithm has gradually shown insufficient in control speed and precision. Meanwhile, with the change of coal property and other parameters, the density given must also be changed to ensure that coal products meet the requirements. Aiming at these problems, the paper is researched separately from two aspects of the suspension density control and the given prediction.Heavy medium coal preparation process is introduced firstly, and then the dense medium cyclone process and its main influencing factors are described in detail. The density control model of dense medium suspension is established through the method which combine mechanism knowledge with the data collected by experiment based on the analysis of the relationship between input and output. The overall scheme of dense medium suspension density control system is designed finally.Model-free adaptive control(MFAC) algorithm is applied to the suspension density control, which is not dependent on the mathematical model in the overall control system scheme. The MFAC controller in the suspension density control system is designed and developed on the basis of the basic theoretical knowledge, and then compared with the PID and Fuzzy-PID algorithm. Anti-delay MFAC(ADMFAC)controller is designed to solve the problems of standard MFAC under the hysteresis,and then compared with the standard MFAC under the circumstances of nominal model and the anti-delay is increased.As the main parameters affecting the system performance, the setting method about the step sequencekr and the weight factor l of MFAC algorithm has still no perfect theory. Therefore, the max min ant system(MMAS) is used to set the parameters of ADMFAC on the basis of analyzing the basic principle of ACO, theparameter influence and the advantages and disadvantages of the algorithm. Thus, the optimal combination of the step sequencekr and weight factorl is obtained to achieve the best control performance of ADMFAC.The extreme learning machine(ELM) algorithm is used to train the predictive model to predict the given of suspension density. The factors affecting the density given are introduced firstly, and then the prediction model of suspension density given is established on the basis of the basic knowledge of ELM algorithm, finally the prediction model is used to predict and then compare with the actual value.