Practical implementation of the SCADA+AGC/ED system of the hunan power pool in the central China power network

algorithms have been developed to determine the weights using existing knowledge of how the outputs have previously depended (or should have best depended) on the inputs. Since there is no clear mathematical basis on which to formulate the relationship between decision factors such as those discussed previously with the contractual parameters in the incentive power contract, a neural network approach is very useful. A three layer neural network is designed to work out the contractual parameters while bid price, cost, profit, the degree of risk aversion and decision factors contributing to uncertainty are used as the inputs. Each decision factor is grouped into three input neurons (inconsiderable, considerable, and dominant) in a simple example presented in the paper. The example shows how knowledge (known information) is presented as a Training Facts File, how a weight matrix is obtained and how a Test File is used to verify the trained network. The network may then be used as a guide to contracting.