Study On Operation Optimization Of Boiler And Turbine In Power Plant Based On Deviation Analysis And Artificial Intelligence

Deviation Analysis is an effective method for power unit performance monitor by analyzing key operation parameters and reducing controllable loss according to the deviation. However, unreasonable reference value of operation parameter is the mean point of providing a limitation for wide application of deviation analysis.The Reference value of operation indices on energy-loss diagnosis and energy-saving analysis of power unit is discussed based on deviation analysis, and a method of Artificial Neural Network and Genetic Algorithms is put forward, which uses the real operation case as comparison standard rather than a conceived ideal case and avoids the dependence on suspicious off-design model.The Self Organization Feature Map and BP neural network models for boiler efficiency, coal consumption of electricity and heat consumption rate of steam turbine under steady operation are adopted for simulation based on Huaneng Dalian Power Plant collected by DCS system and most errors between the predicted and actual values are less than 2%. With the aid of SOFM's ability in clustering, the limitation of conventional collection of training samples are addressed. With the established models, operation parameters under different conditions are simulated and associated with Genetic Algorithms to search the optimized values and obtain the energy loss caused by the deviation of operation parameters.The neural network models for steam turbine power output and power needed in cycle cooling water system are established for the reference values of vacuum value of condenser and have a high precision and reliability. With the established model, the optimization model of condenser is proposed to determine the best vacuum value of condenser and energy loss resulted by deviation of some main operation parameters, reduce the energy loss by optimize the controllable parameters. Compared with conventional models, the established models have the advantage of pertinence to the actual equipment and take nonlinear-in-the-parameter and coupling into account, provide the credible foundations for parameters on-line monitoring, energy consumption analysis, operation advice and equipment management, which benefit to energy saving and management level increasing in power plant.