Research On Subspace Identification Method For Time-delay Systems

Modern industrial processes are becoming more and more complex and largescale,and the control requirements for the systems are gradually increasing,accurate control requires a precise process model.System identification,as an effective way to acquire complex models,has been widely used in the industrial field.Among them,the subspace identification algorithm is widely used in the identification of multivariable systems due to its simple structure,simple calculation and good numerical stability.However,for time-delay systems with unknown delay,the subspace identification algorithm has insufficient ability to estimate the time-delay parameters.It is necessary to obtain the process time-delay in advance to accurately identify the system model.In this thesis,a time-delay system with unknown delay and large interval span is taken as the research object,the parameter estimation of time-delay and the parameter identification of process model are studied.For the time-delay systems with unknown time delay,the time delay estimation method based on ARX(auto regressive exogenous)model is studied to obtain the time-delay parameters of the system.The identification principle of the method is expounded from the theoretical level,and the effectiveness and accuracy of the method are verified by simulation experiments.At the same time,the influence of different types of test signals on the identification results and the range of time delays that the method can estimate are studied.In the end,the time-delay interval that this method can accurately estimate is clarified,and it is found that when using this method to deal with the time-delay estimation problem with large delay,the obtained results will be biased.This point has reference significance for the selection of time-delay parameter estimation methods.The time-delay estimation method based on compressed sensing theory is studied for time-delay systems with unknown and large delays.The feasibility of time-delay identification based on the theory is illustrated from the theoretical level.The identification principle is described.Finally,experimental studies are carried out,and the time-delay interval that can be accurately estimated by this method is known.This method and the ARX-based time-delay estimation method complement each other in the estimation interval,which expands the overall time-delay parameter estimation range,and this is of great significance for dealing with the time-delay parameter estimation problem of the actual time-delay system.After estimating the time-delay parameter of the time-delay system,the subspace identification algorithm is studied for the identification of other model parameters of the system.Firstly,the algorithm principle of subspace identification algorithm is analyzed and explained.After the parameter identification problem of time-delay system with unknown time-delay,the identification scheme adopted in this thesis is given.Finally,through simulation experiments,the identification effects of different types of subspace identification are analyzed.In order to test the application effect of the above identification method on the actual system,the variable volume tank liquid level control system,namely the typical first-order and second-order inertia delay systems,was designed and built,and the system identification experiment based on the operational data was carried out.According to the overall identification strategy of this thesis,the system time-delay parameter is estimated and the system process model parameter is identified.The experimental results show that the identification method can accurately identify the time-delay of the experimental system and other model parameters.