Investigation On Anticorrosive Coatings' Life Prediction Of Pre-buried Channel Based On The Performance Evaluation

Life prediction and reliability evaluation of coatings,as a means of evaluating the service status of coatings,are far behind the materials forming and modification technologies such as casting,forging,welding and heat treatment.At present,there are few literatures on life prediction and reliability evaluation of coatings.The research on life evaluation methods and life prediction models of coatings on materials is of great significance for the development and application of surface engineering technology.Rational classification of coatings system is an important basis for life prediction of coatings.There are many kinds of anticorrosive coating systems for buried channel in rail transit tunnel,but there are few comparative studies on the anticorrosive performance of different anticorrosive systems,and the research on life prediction of anticorrosive coating for buried channel is more blank.In this paper,the characteristics and classification of material surface coating system are analyzed and summarized,and the life evaluation method of material surface coating is preliminarily discussed.Combining with the corrosion resistance of surface coating of buried channel in rail transit tunnel,a life prediction model of anticorrosive coating is established based on BP neural network theory.The results show that:(1)Based on the failure modes and mechanisms of coatings,the corresponding contents can be found according to the protective mechanism of coatings,and the relationship between different coatings can be found;based on the service environment,different coating processes can be selected according to the service environment level,and the factors affecting the life of coatings can be found and used as parameters of life prediction model.(2)From the perspective of overall thinking,development path,boundary conditions,key factors and expression of conclusions,the idea of establishing life prediction model of coatings is preliminarily explored.The idea is verified from single coatings such as thermal spraying coatings,thermal barrier coatings,organic coatings,hot-dip coatings to composite coatings.The feasibility of empirical formula is proved by available data,and composite coatings are preliminarily proposed.Empirical formula for life prediction.(3)According to the neutral salt spray test,the composite coating system basically meets the anticorrosion requirements of the buried channel.The anticorrosion performance is closely related to the properties of the composite coating material and the preparation process.The salt spray test performance of single hot dip galvanizing process is poor.The corrosion resistance of single hot dip galvanizing process can be greatly improved by strictly controlling the quality of hot dip galvanizing process,improving the compactness of galvanizing layer and increasing the surface sealing layer(such as appropriate passivation).(4)The greater the thickness of the coating on the surface of the embedded channel,the higher the corrosion resistance of the composite coating,the better the corrosion resistance of the composite coating than that of the single coating.The CASS time of 70-170 um thickness of single hot-dip galvanized coating is 80 h-110 h,the CASS time of 80-180 um thickness of single organic coating is 80 h-130 h,and the CASS time of 160-320 um total thickness of composite coating is 280 H-600 H.The empirical formula of composite coatings is satisfied when the thickness of coatings is small,but not when the thickness of coatings is large.(5)The BP neural network model in this paper chooses 30 sets of experimental data,25 of which are training data and 5 are validation data.The network structure is an input layer(including two nodes,zinc layer thickness and organic coating thickness,500 neurons,three hidden layers(100,80,20 neurons in turn),and one output layer(copper salt).Accelerated acid test endpoint time),the number of iterations of the model is 5000,the predetermined error is 0.0001,and the error value is 6.2511e-4 after 5000 iterations.The error between predicted value and real value of five groups of validated data is 0.04%,0.53%,0.93%,0.59%,0.18%,respectively.The results show that the BP neural network life prediction model of preembedded channel coating is reliable and can predict the corrosion resistance of samples with arbitrary thickness of pre-embedded channel under certain conditions.