The Research On Capability Analysis Of Co-Generation Unit Based On Error Back-Propagation Algorithm

In China, with the deepening of the power industry reform, the market mechanism has gradually formed. How to minimize the energy consuming and how to increase economic profits are very important for the power plant that is in marketing operation. Coal-consuming in power industry is one of the main elements that influences production cost of power plant. However,because of the backwardness of measuring and maintenance, many generating units, especially the ones under 200MW are not able to accurately calculate the coal consuming and economic profits in time and to guide operation and maintenance according to the economic conditions either, which make them fail to meet the needs of market economy.Currently, the Heat-balance Method is wildly used in capability analysis of power plant but the great numbers of original parameters, the complication of calculation and the calculation errors under mal-maintenance become the main problems of this method. This paper introduces the principle of Error Back-propagation Algorithm and how it is used in the emulation process of calculating the coal consumption of the 125MW Co-generation unit in a power plant. According to research on the parameter numbers, the sample groups and the numbers of the hidden node, a preferable BP model will be got and the coal consumption of the whole generating unit can be calculated by few parameters and the error will be controlled within 0.5%. At the end of the paper, an emulation calculation is carried out by the actual statistics and a satisfactory result in practical engineering project can be got. The Error Back-propagation Algorithm is not only able to guarantee the accuracy of coal-consumption calculation, but also makes up for the shortcoming of too many original parameters by Heat-balance Method. It provides new approach for the power plant to carry out its capability analysis.