Method of improving the accuracy of thermal power desulfurization model

This paper addresses the challenge of predicting the efficiency of desulfurization systems in thermal power plants by proposing an efficient prediction method based on an improved Support Vector Machine (SVM). Through a thorough analysis of the factors affecting the efficiency of desulfurization systems, principal component analysis (PCA) is employed to reduce data dimensionality, leading to the construction of a desulfurization system efficiency prediction model based on SVM. Additionally, a hybrid strategy is introduced to improve the Seagull Optimization Algorithm (SOA), which is used for parameter optimization of the SVM, resulting in the ISOA-SVM prediction model. Experimental results show that this model has high prediction accuracy, with an average absolute percentage error below 0.3%, providing an effective new method for predicting the efficiency of desulfurization systems in thermal power plants, and demonstrating significant practical value.