Study On Corrosion Prediction Of Industrial Circulating Cooling Water System

Cooling water system is an important lifeline of enterprise production.Corrosion problems seriously restrict the improvement of efficient utilization of circulating water.At present,the monitoring methods for the problem of circulating water corrosion are relatively simple and the input cost is large.Based on the industrial circulating cooling water system of a chemical company in Dezhou,this paper studies the corrosion problem and aims to provide a scientific basis for predicting the corrosion trend of the circulating cooling water system.The process indicators involved in the corrosion parameters in the actual production process mainly include water quality index data and equipment operating parameters.Due to the standard normative operability of equipment parameters,the article only analyzes the water quality index data.Aiming at the complexity and diversity of water quality index data and the nonlinearity between parameters,this paper adopts principal component analysis(PCA)and nuclear principal component analysis(KPCA)to reduce the dimensionality of sample data.Judging the data obtained by the KPCA method for dimensionality reduction by the cumulative contribution rate not only simplifies the data complexity,but also preserves the internal relationship of the original data.The principal component parameters obtained by the KPCA method are used as input parameters of the subsequent model,and are reduced at the data source to improve the overall computational efficiency of the model.Based on the least squares support vector machine(LS-SVM)algorithm,the corrosion prediction model is constructed to predict the corrosion trend of circulating water,and the prediction results are less accurate.The article optimizes the model in two ways to improve prediction accuracy.Aiming at the problem caused by the error data to the prediction result,different sample data are allocated by weights to occupy different proportioning coefficients according to the proportion of the error,so as to reduce the influence of the error sample data on the prediction result.By optimizing the internal structural parameters of the model:regularization parameters and kernel width parameters;chaotic particle swarm optimization-compression annealing(CPSO-SA)optimization combination algorithm is used to optimize the two parameters to obtain the best match with the working condition model.Operating parameters.Finally,the CPSO-SA-WLSSVM combination algorithm is obtained.The MATLAB simulation prediction is carried out on the principal component variables obtained by combining KPCA.The prediction accuracy is over 90%.Experiments show that the circulating cooling water corrosion prediction has important guidance for the treatment of on-site anti-corrosion problems significance.