ABSTRACT
This paper presents a ridge regression model for the prediction of output power in a steam turbine. The operating data above the 95% load level from the plant's past one year records were collected and validated to serve as a baseline performance data set. The threshold range determined from the baseline data set was used to detect signal errors in new operating data. Finally, the variables most strongly influence turbine-generator output were selected as inputs for the ridge regression based turbine model. After training and validation of key parameters, including main steam pressure, main steam temperature and main steam flow were used to estimate turbine-generator output. The effectiveness of the proposed ridge regression based turbine model was demonstrated using plant operating data obtained from a 40 MW thermal power plant. Results show that the ridge regression based turbine model is more reliable with the good estimation efficiency and the trend. This paper investigates the ridge regression problem in multi-collinear data. Regular and standardised ridge coefficients are compared with regular and standardised least squares coefficients. Ridge standard error is compared with least squares standard error. The eigen-vectors associated with each eigen value, incremental percentage, cumulative percentage, condition number of the regressors and variance inflation factors associated with the variables are discussed. Methods to choose biasing parameter k are also presented.