Application Research Of Iterative Learning Control For Urban Drainage System

In recent years,with the prosperity of the process of urbanization,a large number of industrial wastewater,daily sewage and rainfall runoff have been testing the capacity of city drainage and storage system.On the other hand,there is no denying fact that that a high percentage of drainage pumps have already extended services,which means that the task of extension and reconstruction of drainage system is to some extend a quiet heavy burden.There is an urgent demand for making use of the current limited resource,finding a way out of relying on too much on manual operation and traditional control method.It should be noted that more intelligent and advanced control strategy need to be involved and implemented in the designing of the real-time optimization strategy for urban drainage system.The occurrence of extreme climate events in urban areas is relatively rare.When analyzing the features of regional rainfall events,the rules of the intensity of the precipitation based on the time scale have been revealed.it should be mentioned that intermitted rainfall is similar to the intermitted procedure in the industry,namely,we can depose the precipitation process into two dimensions which are time and batch.This principle of deposition and method of searching better solution with former experience iteratively agrees with the idea that learning from repetitive,which is the core idea of learning algorithm.At the same time,the model of urban drainage system built according to the SaintVenant equations mainly designed for the purpose of meeting the practical need can barely satisfy the command for mechanism studying.The deficiencies can be compensated by iterative learning control which depends more on the experience accumulation.The structure of the article is shown as below:1)Application of centralized iterative learning predictive control in urban drainage network.First of all,we have a brief introduction about the elements composing of the simplifid model of the urban drainage system;what is more,Collect and analyze the features of the storm and conclude the storm into three phases.Then we proposed the centralized iterative learning predictive control algorithm.The simulation result shows that the algorithm functions well.2)Application of distributed iterative learning predictive control in urban drainage network.Considering some most known cities in china,we may see that the drainage pipe network system has a wide geographical distribution,which belongs to the multi-objective and multivariable coupling system where the centralized control may not be satisfactory.In this paper,we have proposed the distributed iterative learning predictive control algorithm which ignored the interactive relationship between coupling sub-systems.The results shows that the dismissed part has been compensated well by iterative learning control.3)Application of distributed iterative learning predictive control in urban drainage network.When taking account of the difference of the climate condition in different areas,the intermittent process may lead to the difference of the phases of different area,which will undoubtedly affect the control results.With hypothesis of exchanging states information in the intermission,we have Proposed distributed iterative learning predictive control to reduce the expense on the information exchange while not sacrificing the robust stability of the subsystems.