Low Exergy Loss Design Method Of Complex Equipment Process Parameters Based On Reinforcement Learning And Its Application

This paper proposed the design plan exergy loss prediction method based on amended P-R equation, constructed the key process parameters exergy loss response surface model of complex equipment, realized the complex equipment low exergy process parameters design based on reinforcement learning, to solve the complex equipment high energy consumption problem while its function, structure, constraint is complicated and the multivariable is coupled. A complex equipment low exergy process parameters design system is given to provide the theory foundation for low exergy plan design and tools for the equipment on run to set low exergy process parameters.The full paper is organized as follows:The first chapter proceeds from the engineering problem of complex equipment high energy consumption problem, reviews the research status of exergy loss analysis and green design, low exergy design, and process parameters low exergy loss design based on reinforcement learning development and application both domestic and abroad, combined with the973project expands the contents of this paper, and finally introduces the general framework of this paper.The second chapter aims at the difficulty to accurately predict exergy loss for complex equipment at design stage without physical prototype, and uses the plant historical data to modify the P-R equation coefficient and sets the foundation for estimating the material property parameter. Finally, the air separation equipment is simulated uses the modified P-R equation and the exergy of eighty thousand class air separation equipment is calculated.The third chapter extracts the key process parameters of air separation equipment by verifying its material and heat balance, and establishes the sampling point database by simulation and actual air separation equipment monitoring. Finally the response surface model of air separation equipment is constructed based on the response surface model which sets the foundation for low exergy loss process parameter optimization. The fourth chapter proposes prior and real time learning dual framework reinforcement learning based on the basic framework of reinforcement learning, and studies the complex equipment reinforcement learning method and realization based on exergy response surface model. Finally the method is verified in the design of eighty thousand class air separation equipment with low exergy process parameters.The fifth chapter constructs the complex equipment process parameters low exergy design system based on the secondary development of Aspen Plus, the exergy loss prediction method with modified P-R equation under design stage, the complex equipment exergy response surface model, and the complex equipment low exergy process design technology based on reinforcement learning. Finally the system is verified in eighty thousand class air separation equipment.The sixth chapter summaries the work carried out in the paper and points out the direction for further research facing practical engineering application.