Static Gain Estimation for Automatic Generation Control Systems From Historical Ramp Responses

This brief proposes an approach to estimate static gains of dynamic models for automatic generation control systems in coal-fired power generation units. The main idea is to infer static gains of dynamic models from amplitude changes of model inputs and outputs in finite-time ramp responses. First, input and output time sequences are approximated by a few straight lines as piecewise linear representations. Second, finite-time ramp responses are detected as the data segments where inputs and outputs simultaneously move along straight lines with significant amplitude changes. Third, estimated static gains are obtained by solving linear equations of significant amplitude changes in the ramp responses. One technical challenge of selecting the thresholds of significant amplitude changes is resolved by exploiting a relationship among the amplitude change, noise variance, and signal-to-noise ratio. Comparing with a standard system identification approach, the obtained results from the proposed approach can be validated in a convincing manner: ramp responses being found from historical data samples can be confirmed by checking whether data segments of inputs and outputs are actually in straight lines; estimated static gains can be verified by comparing output amplitude changes in the ramp responses with their estimated values. Numerical and industrial examples are provided to support the proposed approach.