Adaptive soft measurement method for the furnace exit gas temperature based data-driven method

The furnace exit gas temperature (FEGT) is one of the significant variables for the control of power plant boiler. It is a great challenge to obtain the FEGT online since the high temperature, high pressure and high noise. To solve this problem, a soft measurement method is proposed by fully use of the historical data based on the differential evolution algorithm and least square support vector machine (LSSVM). In this algorithm, the relative variables are selected by the correlation analysis and mechanism analysis. Then the data model is constructed by LSSVM. The DE algorithm is utilized to improve the stability and accurate by optimize the parameters of LSSVM. The experimental results illustrate the effectiveness of proposed algorithm.