By means of stronger non-linear fitting ability

generalization of network

and fault tolerant ability of the neural network

an on-line calculation model of thermal economic efficiency based on BP neural network has been established.Taking one 320 MW unit as example

directing against the specified thermal economic indices

based on the thermodynamic property of said indices

and fully using the advantages of neural network

a model for calculating the heat consumption rate of steam turbine has been constructed.Comparison with results obtained from calculation by using conventional method shows that the said model has higher accuracy and stability

the parameter's fluctuation resulted from unstable operation has very small effect upon the result of calculation

providing a new way of thinking and a new method for on-line calculation of the thermal economic indices.