Design And Realization Of Forecasting System For Industrial Recirculation Corrosion And Fouling

Industrial cooling water is an indispensable link in petrochemical and metallurgical industries.Due to the long technological process and continuous operation time in petrochemical and metallurgical industries,a large number of pipes and pressure vessels are used to produce high temperature liquids.Cooling water is generally recycled to ensure continuity of production.However,circulating cooling water in the enterprise to bring benefits at the same time also caused huge problems to the enterprise.Chilled water continuously reused in the equipment causes the growth of microorganisms,leading to deterioration of water body and then lead to corrosion and scaling problems.Corrosion easily lead to leakage of equipment pipeline pollution and fouling is blocking the pipeline affect the heat exchange efficiency,these two cases will result in serious business unplanned parking,causing significant economic losses.At present,common corrosion and scale detection methods in industrial sites are mainly divided into the following categories: First,the use of fouling resistance meter and corrosion on-line monitor for the detection of circulating water corrosion and scaling,but the probe of these devices are more expensive,Greatly affected by the water quality is extremely easy to wear and tear,equipment maintenance and installation costs are higher;the second is the use of coupons method,a month test,a long sampling period.Third,by virtue of the experience of on-site technicians on the trend of corrosion and scaling judgment,anthropogenic factors.Based on the theory of soft sensor technology,combined with the research of circulating water corrosion and scaling at home and abroad,a set of online prediction system for circulating water corrosion and scaling has been designed and developed.The system combined with the field production process,the design function is comprehensive,the operation is simple,meet the needs of the on-site staff.The main components and characteristics of the system are:(1)For complex data on the scene to provide data integration approach.Such as the error data using 3? law,missing data completion,timing matching,numerical conversion to ensure complete and accurate data samples.(2)Flexible screening of auxiliary variables.The auxiliary variables required for the system are combined with the combination of intelligent screening and artificial screening.Intelligent screening is based on weighted grey correlation analysis and principal component analysis,and the variable filter is completed automatically through the algorithm.Artificial screening is for on-site operators to select flexible variables according to the needs of the field.(3)In this paper,the method of model performance monitoring and correction based on off-line measurement data and working condition discrimination is studied.The default condition monitoring center of industrial process changes,when the evaluation index model and the confidence limit of performance over the trigger system model correction function;take corrective measures corresponding condition change type condition: if the center based on the industrial process characteristics of gradient,adopt model prediction correction method for pre valuation model of the value of the correction;if the industrial process characteristics by mutation,model updating reconstruction correction method to update the model.(4)Implement the interaction between C# and MATLAB.The system combined with C# has rich front-end interface effect and fast program execution ability,as well as MATLAB's powerful computing power.Therefore,the designed prediction system provides good operation and accurate prediction results.