Stochastic Load Modeling and Optimal Dynamic Generation Control in Power Systems

摘要

A statistical model is described for the dynamic generation control of a power boiler operating in a variable pressure mode to reduce thermally-generated stress levels in cycling units. Under current practices, a boiler is operated at the maximum design throttle pressure, resulting in a maximum amount of stored energy in the boiler. Generation control is achieved by manipulating the turbine valves in response to changing load while the boiler is fired to maintain maximum design throttle pressure. This type of operation causes large fluctuations in turbine metal temperatures over the load range and large throttling losses, both of which are undesirable. The proposed stochastic load model overcomes these difficulties by allowing boilers to operate in a variable pressure mode. Load control is achieved by calculating the optimal feed-forward control sequence so that the fuel consumption is minimized, and the forecasted MW is generated at the final time of control. This method combines the economic dispatch of real power with load frequency control. Since valve setting policy can be based on valve point loading, temperature changes within the turbine from throttle losses are eliminated. This in turn reduces thermal fatigue and increases efficiency of the operation. Optimal control problem formulation places bounds on state variables, input variables, and allowable rates of change of state variables for reasons of boiler and combustion safety.