Identification The Operating Characteristics And Optimization Load Based On Data Mining For Power Plant Unit

As the further development of the socialist market economy and the power system in the electricity market environment, the competition of the thermal power plants is based on development of power generation share of the plan, to determine the optimal distribution of load between units to achieve the best economic benefits, and it is very important. Zhejiang Province, into a"local power plant management informantion system in Zhejiang Province", the accumulation of a large number of power plant operation data, which contains the actual power state of the operating units. The mining of the data in power plant can be favorable to economic mentoring program run.Firstly, being based on the analysis of power plant thermal system, the operating mechanism and the establishment of a mathematical model of each device, the use of statistic and uncerntaintu analysis methods, repectively, through a single parameter error, and based on thermal power plants, electrical systems the overall balance between the three runs as a system error, and the collected reliable historical data were selected for follow-up basis.Secondly, the problem is that it needs fitting the unit stream demand curve. For large thermal power plant, power plants on a regular basis the working conditions of unit testing, and update the condition chart. But for small and medium sized power plant, because of financial, technical and other reasons, the regular test of the unit features is very difficult. Therefore, the small thermal power generating units only by everyday experience and the theoretical working conditions chart to understand the working conditions of unit operation. Unit maintenance or modification will change the thermodynamic properties, resulting in pratical graph theory very poor condition. If the conditions don't update the map units, the unit will be unable to accurately carry out economic operations, resulting in waste of resources, therefore, the reliable data that has been selected is a necessary choice for indentifying the characteristic curve of thermal units. A focus of this study is that by selecting the M5P model tree piecewise linear data mining algorithms, the establishment of the unit steam consumption rate of the exact model. Compared with the traditional multiple linear regression equations its reliability and accuracy is verified. Further, the paper gives the exact condition of Extraction Methods of Drawing Unit, and roll recognition, to identify the state unit operation conditions.Finally the optimization distribution of heat and electric load is essentially a lot of variables and constrains, the objective function of complex large-scale nonlinear programming problems. In this paper, M5P model tree built based on piecewise linear model and genetic units-Annealing optimization for load distribution model for power plants. The genetic-annealing algorithm converges faster, and better optimization results. The data validation appears that the optimization model for power plant steam consumption save 2.2% of the load, and provided a strong guiding line for decition-makers.