The Prediction Study Of The Industrial Heat Exchanger Fouling State Based On Real-Time Database

Heat exchanger (exchanger heat), is a kind of exchanger which can exchange energy in several material to achieve the purpose of energy saving. In industry, fluid is heated by this method, so that its temperature can meet the needs of process, energy saving and improvement of economic efficiency. In the current industrial environment, the use of heat exchangers is very extensive, involving the industrial chain, which includes water treatment, chemical industry, heating, petroleum, petrochemical and other fields of industries. In the current chemical production, the heat exchanger is one of the indispensable chemical process equipment. Assessed according to tonnage of equipment, proportion of heat exchanger accounts for 20%, some even up to 30%. Its position in the production is essential. In continuous production, when the heat exchanger problems occurs, it will not only affect the operation of the entire process, and even cause production system paralysis crash, cause incalculable economic loss of property.Heat exchanger in after long-term use, will be in the heat transfer tube wall between the accumulation of fouling, the fouling will lead to change heat exchanger heat transfer efficiency of a sharp decline in, due to the chemical composition of fouling, and even erosion heat exchanger equipment and reduce the maintenance cycle of the equipment, but also cause accidents.The real-time database platform to collect relevant production environment around the heat exchanger data exchange. According to the change of the actual working condition of the heat exchanger and the data through the analysis and processing that can reflect the thermal efficiency of the data of the heat exchanger, the use of support vector machine algorithm to the data of the training that can heat exchanger node scale state classification model. In addition, through the linear fitting of the relevant data, the two can predict the heat exchanger can predict the two fitting formula.Aiming at the scene for thermal sensor missing problem proposed by support vector machine modeling method and heat exchanger surrounding environment of data modeling, using the model of heat exchanger fouling state prediction. At the same time, according to the real data, the heat exchanger node scale state by the above three methods to predict, results and the actual fouling condition carries on the contrast analysis, comparison of three methods for predicting the accuracy, finally comparing the support vector machine model to predict accuracy the most high to verify the practicability and accuracy of the algorithm can guide the production and improve the economic efficiency of company.