Study On Slope Safety Monitoring Model Of Diversion Project Based On RF And Its Optimization Algorithm

As an important means to solve the problems of uneven distribution of water resources and regional water shortage,water diversion project plays a key role in the process of national water resources allocation and makes great contribution to economic construction and development.As one of the main components of the water diversion project,the slope is usually hundreds of kilometers long or thousands of kilometers long.It passes through many cities and regions.The natural geographical conditions vary greatly.Once it loses stability,it will directly threaten the safety of people’s lives and property along the slope.Therefore,it is particularly important to implement safety monitoring on the water diversion project slope.The safety and stability of the diversion project slope is related to many influencing factors.It is of great theoretical and practical significance to find an algorithm that can accurately reflect the complex internal relationship between the slope state and its influencing factors,so as to establish the corresponding slope safety monitoring model and realize the real-time grasp and accurate prediction of the slope state.As a data mining algorithm,random forest(RF)algorithm has a strong tolerance to noise data and outliers,and is not easy to fall into over fitting.In this paper,RF algorithm is introduced into the field of diversion project slope safety monitoring and control.By analyzing the main influencing factors of diversion project slope state quantity(displacement,seepage pressure),the input variables of the model are determined,the out of bag(OOB)data error is analyzed,and the optimal combination of parameters is determined,and the slope safety monitoring model based on RF is constructed.The results show that RF model has higher fitting and prediction accuracy.In order to improve the adaptability of RF model to slope safety monitoring data set and improve the performance of the model,this paper considers the influence of the randomness of characteristic variables on RF model,and grey relation analysis(GRA)is introduced into the selection of influencing factors of slope state quantity of diversion project.By calculating the grey correlation degree,the correlation of different influencing factors with displacement and seepage pressure is determined,and the influencing factors with high correlation are selected as the input variables of RF prediction model,The GRA-RF safety monitoring model of slope is established;then,the GRA-RF model is optimized from the perspective of sample randomness.The similar days of the date to be predicted are selected from the historical days by using the weighted grey correlation projection method,and the selected similar day sample set is used as the training sample to establish the improved GRA-RF safety monitoring model of similar days.The application results show that the prediction accuracy of RF model,GRA-RF model and similar day improved GRA-RF model is improved in turn,and the improvement is obvious,which fully proves the effectiveness of similar day selection and gray correlation method in improving the performance of RF slope safety monitoring model.