Study On Corrosion Scaling Prediction Of Industrial Circulating Cooling Water System

The circulating cooling water system of Tianjin Petrochemical Company as the research object,in reducing failure rate,improve economic efficiency,improve the safety factor requirements under the background of the common quality fault of the corrosion and scaling of circulating cooling water system,in order to accurately predict the corrosion and scaling trend for enterprises to take timely measures to provide the scientific basis for prevention and control.Because the corrosion and scaling of circulating cooling water is influenced by many factors and is not regular and irregular,it is a typical nonlinear problem.Therefore,there is no unified prediction method at home and abroad.Therefore,looking for a most accurate and suitable prediction algorithm is the key of this paper.In the actual operation of petrochemical industry,the main parameters affecting corrosion and scaling are water quality parameters and equipment parameters,but the water quality parameters are only considered in this paper.As the water quality parameters are more than 15,as the modeling prediction,too much input will lead to the slow calculation and take up a large amount of resources and so on.However,each parameter has its influencing factors and therefore can not be ignored.Therefore,principal component analysis(PCA)is applied to reduce the dimension of 15 water quality parameters,and six principal components are finally obtained.The six principal components cover the information of the original data to the greatest extent,so the six principal components not only ensure the integrity of data,but also solve the problems caused by too large input samples.After reducing the input variables,the self supervised and highly self-learning machine algorithm support vector machine(SVM)is used as the core prediction algorithm.Due to the insensitivity of the internal parameters,the regularization parameter and the width of the kernel function of support vector machine(SVM),the different values will affect the accuracy of the model.Therefore,the least square method is used to optimize the insensitivity coefficient,and the scatter search algorithm(SS)is used to optimize the regularization parameter and the kernel function width,so as to get the best combination of prediction parameters.At the same time,the algorithm of this paper,genetic algorithm and support vector machine(GA),particle swarm optimization(PSO)after the optimized support vector machine(SVM)were compared.The results show that the algorithm studied in this paper in terms of the experimental results accuracy and convergence speed are better than the other algorithm optimized support vector machine.The accurate corrosion scaling rate model obtained in this paper provides a theoretical basis and guidance for the prevention and treatment of corrosion and scaling in petrochemical industry.