The Application Research Of Dynamic Programming Method In Natural Gas Transmission Networks Operation Optimization And Control

With the continuously increase of natural gas production and market demand, natural gas transmission pipelines system is becoming large-scale, complicated and networked rapidly. Therefore, the operation optimization and control of natural gas network is crucial. A good operation scheme can not only complete the transportation tasks safely and efficiently, but also reduce operating costs and improve economic efficiency.Dynamic programming is a simple and effective optimization technique which is widely used in the study of pipe network operation optimization problems. But it is weak to solve the complicated topologies and transient state optimization problems, particularly, incapable or inefficient. In order to improve the solving ability and efficiency, this thesis presented network cutting technique and time-layers independent method based on the researches of optimization models. The cutting technique could decompose the complicated network into tree structure and make it easy to solve with dynamic programming. Time-layers independent method could transform the multiple time-layers transient state optimal problem into a serial of independently single time-layer optimal problems, which avoided the problems about curse of dimensionality and distinctly improved the solving efficiency.Additionally, this thesis simply discussed the transmission characteristics and the widely used PID control method in gas networks. The PID method exists an obviously time-delay and does not consider the entire networks controlling at the same time. In order to overcome the aforementioned disadvantages and improve the controlling capability of transmission networks, this thesis presented a combining method of segmentation control and optimum control based on dynamic programming, which adopted segmentation strategy in flow and pressure control, and applied optimum control in flow regulating.The according simulation results showed that the proposed cutting decomposition and time-layers independent methods can effectively improved the performance of dynamic programming. Besides, the combining control method considering the entire network and all the compress stations technical constraints at the same time could effectively transform networks from the current state to the expected state.