Study On Safety Risk Early Warning Of Construction Of Hydroelectric Underground Cavern Based On BP Neural Network

As China’s hydropower development progresses toward the southwest and Tibet,the underground water diversion system became the best choice because of the constraints of high altitude terrain and environmental protection.Due to the objective reasons such as long construction period of underground engineering,complex geological conditions,and multiple risk factors,coupled with some subjective reasons such as the lack of comprehensive understanding of underground construction risks by some technicians,construction safety accidents of hydropower underground cavern frequently occurred.Therefore,real-time and accurate warning of safety risks is of great significance to the construction of hydropower underground cavern.In view of this,this paper referred the BP neural network to the safety risk warning field of hydropower underground cavern construction.BP neural network model for early warning of construction safety risk of hydropower underground cavern was established.From the perspective of the construction side,we conducted an early warning study on the construction safety risks of hydroelectric underground cavern.It is expected to increase the risk early warning capability and safety management level of hydropower underground cavern.Firstly,the connotation of construction safety risk for hydropower underground cavern was clarified,and its risk characteristics were analyzed.At the same time,the concept of risk early warning was elaborated,and BP neural network was selected as an early warning method.In combination with the basic process of risk early warning,the construction safety risk warning process for hydropower underground cavern was determined.Secondly,based on the principle of selection of early warning indicators,the hazard sources were analyzed,and then an early warning index system for construction safety risk of hydropower underground cavern was constructed,which included four aspects: human,material,environment,and management.The data collection criteria for each early warning indicator were given and the warning interval was divided.Then,after collected 1036 sets of effective training samples,the sample data were normalized and input into the designed BP neural network.Through repeated training and debugging,a BP neural network model for construction safety risk warning of hydropower underground cavern was successfully established.Finally,an example of a pumped-storage power station under construction at a certain location was used to verify the early warning model.Through early warning analysis,it was found that the early warning results were not precise enough.Therefore,the early warning model was improved,and the improved early warning model was used again for early warning.Again,the early warning results were basically consistent with the actual construction conditions,thus verifying that the improved warning model had certain practicality and accuracy.Through this early-warning model,the construction safety grades of subsequent underground cavern can be quickly known,and at the same time,dynamic early warning of changes in risk can be achieved,which was conducived to adopting effective countermeasures in advance and improving construction safety management levels.