Reliability Calculation Of Vertical Breakwater Based On Artificial Neural Network Theory

This paper mainly uses three artificial neural network-based structure reliabilityanalysis methods, which are first-order second moment method, second-order secondmoment method, and Monte Carlo simulation, to do some research on the anti-slidingand anti-overturning reliability of the vertical breakwater.Safety coefficient method is frequently used in early structure design. Long-termpractice shows that this method is not scientific and has many shortcomings. Scholarsdevelop the structural reliability theory to improve the structure design methods. Theperformance functions of complex structures are usually implicit in terms of basicvariables in structure reliability analysis. After getting the basic variables and thestructure response data by numerical simulation or experiments, the performancefunction can be approximated by using artificial neural network method, and then thereliability index of complex structures can be calculated. Therefore, the artificialneural network-based structure reliability analysis methods can easily cope with theproblems whose performance functions are implicit in terms of basic variables. Theyneed smaller number of numerical simulation or experiment than other methods, sothe calculation efficiency could be increased.Firstly, all the parameters in the sliding failure limit state function and theoverturning limit state function are regarded as variables when carrying out thereliability analysis of the horizontal sliding and overturning stability of the verticalbreakwater. Then first-order second moment method, second-order second momentmethod, and Monte Carlo simulation which are based on artificial neural network areutilized to calculate the reliability indexes of the anti-sliding and anti-overturningabout a certain vertical breakwater in Qinghuangdao, and compare with the resultswhich are solved by JC method, Breitung method, and Monte Carlo simulation(including direct sampling method and importance sampling method of Monte Carlosimulation). The comparison shows that the artificial neural network-based structurereliability analysis methods are similar to those calculated by the correspondingstructure reliability analysis methods, but the former’s results are slightly lower thanthe latter’s. It can be concluded that the artificial neural network-based structurereliability analysis methods, whose accuracy is higher, can be used to calculate the anti-sliding and anti-overturning reliability of the vertical breakwater. And thesemethods are unlimited in the complexity of the performance function and the numberof variables of the limit state equation. In the end, the weight of the verticalbreakwater in the water and the stable moment of force created by the weight, andtheir impacts to anti-sliding and anti-overturning reliability of the vertical breakwaterwhen the friction coefficient between the caisson and the foundation bed are analyzed,and the results give these conclusion: the results are more accurate and realistic whenall parameters in the sliding failure limit state function and the overturning limit statefunction are regarded as variables; and the impacts of the structure self weight, themoment of stability induced by structure self weight and the friction coefficient can’tbe ignored, especially to the anti-overturning reliability index.In addition, the reliability index is introduced to control the overtoppingdischarge of the sea dikes, and the limit state function of the overtopping dischargeunder permissive overtopping conditions is established. The overtopping failureprobability and reliability index of one sea dike are calculated using artificial neuralnetwork-based Monte Carlo simulation and Monte Carlo simulation. It’s concludedthat the sea dike overtopping reliability index is more reasonable and realistic than thestandard of the tolerable overtopping limits.In one word, the application of artificial neural network-based structurereliability analysis methods, which are feasible, have great theoretical significanceand project reference value in vertical breakwater reliability analysis and the sea dikeovertopping reliability analysis.