Improved Wind Power Generation Prediction through Novel Linear Regression over Ridge Regression

The linear regression algorithm is preferred to enhance the accuracy of wind energy production forecast instead of the conventional techniques. In this work two different regression algorithms are compared for their accuracy values and their adaptability for finding the energy production is evaluated. The data is collected from the wind turbine that runs for several months. The collected data set consists of the values of various parameters such as velocity of wind, orientation of wind flow, moisture content present in the wind and other environmental related parameters that are having impact on the measurement of wind production energy. The data collected are made ready by preprocessing to remove the missed data and outliers. From the data set, in proper ratio the training and testing data are chosen to construct and evaluate the proposed model. For every two groups, 10 samples are taken from the data set for analysis in which the first group is considered to train the conventional Ridge regression and the second group is considered for training the proposed LR method. The two different algorithms considered here are analyzed with their training data and they are evaluated with their associated with their testing data. The simulation results are achieving 82.15% accuracy for proposed model and 77.90% for the conventional model. This work is also analyzed with SPSS tool for its better understanding of statistical analysis to ensure its significance 0.8, α = 0.05, CI = 95% confidence interval was also calculated. With the analysis experimented, the results are evident to show that the proposed model is outperforms than the conventional model for calculating the wind energy production.