Study On Corrosion Prediction Of Circulating Cooling Water Based On KPCA-CPSO-LSSVM

During the operation of industrial equipment,the economic loss caused by improper corrosion supervision can not be ignored,how to accurately monitor the corrosion condition of equipment in real time has become an urgent problem to be solved in the current production process.In recent years,circulating cooling water has been widely used in industrial production,the water quality information in the circulating water is an important basis for adding the type and quantity of corrosion inhibitors to industrial equipment,therefore,research on the prediction method of circulating cooling water corrosion,realizing the goal of monitoring the corrosion status of the circulating cooling water system in real time,it is of positive significance to prevent corrosion failure of circulating cooling water system caused by monitoring negligence in production and to improve work efficiency and economic benefits.This paper takes the circulating cooling water system of Tianjin Petrochemical Company as the research object,analyze and study the actual measurement data of circulating cooling water in petrochemical enterprises for three years,it is proposed to establish a reasonable and effective prediction model for real-time monitoring of the corrosion rate of circulating cooling water.Through research on the quality of circulating cooling water used in petrochemical equipment,it is found that the water quality parameters affecting the corrosion of circulating cooling water are very complicated,each water quality parameter not only affects corrosion,but also has collinearity between different water quality parameters,using all the parameters as input to the predictive model will increase the computation time of the predictive model.This paper proposes the use of nuclear principal component analysis(KPCA)to reduce the dimensionality of 13 major water quality parameters,eliminate the collinearity between parameters,the four principal components with large cumulative contribution rate are obtained as the input of the prediction model,which effectively reduces the model input,ensures the accuracy,speed and stability of the prediction model,and improves the working efficiency of the model.Choose a machine learning algorithm with strong self-learning ability-Least Squares Support Vector Machine(LSSVM)as the core of the prediction model,different from the traditional quadratic programming method used by SVM,LSSVM uses a least squares linear system as a loss function,thereby simplifying the complexity of model calculations,at the same time,KPCA makes up for the defect that LSSVM does not have sparsity.In order to ensure the diversity of the population,improve the global search capability of the population and the generalization capabilities of the LSSVM model,the Chaos Particle Swarm Optimization(CPSO)optimization algorithm is used to optimize the model parameters in LSSVM.Research shows that the LSSVM algorithm model is optimized compared to theKPCA-LSSVM algorithm model and the particle swarm optimization algorithm(PSO),the KPCA-CPSO-LSSVM combination algorithm has high prediction accuracy and generalization ability,and the convergence speed is also reasonable.The results of this study provide an effective means of predicting corrosion of circulating cooling water.