Data-based Research On Energy-saving Optimization And Control Of Power Station

Energy-saving optimization and control is an important way to energy saving and emission reduction for power plant. Huge amounts of data stored in real-time and history database of power plant supervisory information system (SIS) contains a large number of potential knowledge and information, which can be utilized for the energy-saving optimization and control. Based on the operation historical data and data-based theories and methods, this paper focuses on the deep analysis of historical data in database and finds the knowledge and means to improve system operation from the thermal system operation data. Around the application of data-based theories and methods in power plant energy-saving optimization and control, the following areas of study are expanded in this paper:1. Power plant energy-saving optimization and control theoretical system is summarized. Were Plant-level and unit-level power plant energy-saving optimization and control are separately summarized, and study in this paper focuses on plant-level load optimal distribution method and unit-level operating parameters optimization based on data-based methods.2. Coal quality changes a lot and unit load varies in a wide range in power plant, a natural classification method of operating conditions is proposed to satisfy the requirements of all conditions energy-saving optimization and control. The relative coal quality coefficient is proposed to characterize coal properties, combined with unit load and ambient temperature, based on historical data and clustering methods operating conditions are classified into different operating condition clusters which are the basis of following optimization study.3. An operation parameters target value optimization approach is presented. Fuzzy rough set attribute reduction method is utilized to reduce operating parameters affecting boiler efficiency, and important adjustable parameters in combustion process are obtained. Based on characteristics of the operating data, a fuzzy association rules mining algorithm combined with fuzzy C means clustering is presented and it is applied to the optimal target values mining of operating parameters.4. Problem in the unit coal consumption characteristic curve of load optimal distribution is analyzed, and a plant-level load optimal distribution approach based on coal consumption real-time prediction is proposed. Affecting factors to unit coal consumption rate for power supply are analyzed and summarized. Multi-factors weight distribution method based on information entropy, fuzzy clustering and rough sets is employed to predict coal consumption rate for power supply. Based on the proposed speediness and economy constraints, global optimal solution for optimal load dispatching was approached by dynamic programming method. The method was applied in a power plant and it was proved to be practical and effective.