Research On The Key Issues And Control Strategy Of Activated Sludge Process

The activated sludge process is a dynamic biochemical reaction process accompanied by the conversion and transfer of substance and energy.Due to the influence of multivariable,large time delay,strong disturbance and dynamic nonlinearity in the process of sewage treatment,the traditional control strategy is often difficult to obtain satisfactory control effect.In addition,there are some important parameters that are difficult to measure directly in the process of sewage treatment,which seriously restrict the improvement of the automation level of sewage treatment control system.Therefore,the study of new intelligent optimal control algorithm can not only enrich the theoretical significance of intelligent optimal control method for complex processes,but also have the certain practical value in modeling of sewage treatment system,the design of intelligent controller and intelligent sensor.The thesis is based on the benchmark simulation model BSM1,which is developed jointly by International Water Association?IWA?and the European Union Organization for Scientific and technological Cooperation?COST?and studied several key problems in intelligent control of sewage treatment process.The main researches and innovation points of this thesis are summarized as follows:Firstly,the biochemical reaction characteristics of activated sludge process?APS?were studied by elaborating and analyzing the activated sludge model No.1?ASM1?and the secondary clarifier tank model.Next,the activated sludge benchmark simulation model?BSM1?,which is jointly proposed by COST and IWA,is studied in depth,and its evaluation standard is introduced in detail.The accuracy and stability of the model are verified by establishing BSM1 in MATLAB environment and using the data provided by COST.It lays a foundation for the further study of ASM1 which includes parameter estimation,intelligent controller design and BOD₅ soft sensor technology research.Secondly,in view of the effect of the running environment of ASM1 on the dynamic parameters of the model,this paper first studied the improved strategy of the standard cuckoo search algorithm.Then,the objective function is established by the least square relation between the water value of the model and the actual measured value.Finally,this paper estimated and corrected the dynamic parameters of ASM1model based on an Improved Cuckoo Search?ICS?algorithm on MATLAB platform,which could estimate the dynamic parameters quickly and accurately,compared with other algorithms.Thirdly,the traditional PID controller is difficult to maintain the DO concentration at the expected level quickly and accurately with the unknown disturbance.To solve this problem,an adaptive PID dissolved oxygen controller based on RBF neural network was designed.The controller uses the powerful learning ability and adaptive ability of RBF neural network to dynamically change the three parameters of PID controller by gradient descent method.The control strategy is implemented on BSM1 and compared with the control effect of traditional PID.Experimental results show that RBFNNPID has a good tracking performance,anti-interference and strong robustness.Finally,in order to solve the problem of on-line measurement of BOD₅ in the real time in the activated sludge,in this paper builds a soft sensor model based on ICS algorithm optimize extreme learning machine?ICS-ELM?is established in this paper.This method encodes the input weight matrix and the hidden layer threshold of the extreme learning machine to the nest position of the cuckoo.Through powerful global search capabilities of ICS,the optimal extreme learning machine parameters are obtained,which can reduce the hidden layer nodes and improve the prediction accuracy.At last,the prediction accuracy and training speed of the proposed soft sensor model are obviously superior to those of other algorithms compared with other soft sensor models.