Research On Optimal Operation Of The Steam Turboset Based On Immune-Tabu Hybrid Algorithm

Optimal operation of the steam turboset is an important research field of modernization of electricity industry in our country. With the supervisory information system developing dramatically, this paper focused on the selection principle of real-time database and the function of SIS, forecasting and validation of real-time data, the optimal operation technique, the Immune-Tabu hybrid algorithm and the optimized distribution of unit commitment. The research results will be useful for performance monitoring and optimal operation of the steam turboset.Firstly the construction of SIS was discussed, including the network, connection mode between SIS and MIS, the selection principle of real-time database, the functions of high-level application, maters need attention and so on. The real-time database should be selected according to managerial requirement, real-time performance and security based on analysis of MIS and exiting control systems. Proper real-time data should be gathered considering the expansion requirement. The applications will be developed and improved step by step on the safe database platform.A new forecasting model, double neural network combination model was proposed. It was based on regressive and time delay neural network, and the best combination of the regressive and time delay neural networks was combined by neural network. The forecasting model has better accuracy for it can make the best of the history trend of measuring data and other correlative parameters information. Then the model was trained with main steam flow of 660MW unit. The average absolute relative error of checking samples using this combinative forecasting model was 1.5%. While the average absolute relative errors using regressive neural network and time delay neural network were respectively 2.7% and 1.9%. The results showed the double neural network can improve the forecasting accuracy. The validity of real-time data was demonstrated by sequence probability ratio testing method based on calculation of residual error and probability ratio. This method was effective for validation of abnormal data and online validation of measuring data.Next, the model of optimal operation of the steam turboset was analyzed. The calculation of exhaust steam enthalpy is complicated. A method based on extrapolation and power equation was proposed and tested by some thermal-design data of one unit. The most relative error between calculation results and the thermal-design data was 0.9%, which showed this method has high precision and could be used for online performance calculation. The optimal exhaust steam pressure could be obtained according to maximum net income in optimization of condenser system. Thus the extra income of recycled water and the effect of time-sharing price...