地下厂房锚杆支护的反向传播神经网络智能化设计模型

水电站地下厂房的支护设计一般采用工程类比法,而西南地区高地应力、大跨度地下洞室的修建难度往往超出以往的工程,难以完全适应。亟待建立考虑复杂地质条件影响的围岩支护定量设计理论和方法。在29个跨度为18.0～34.0 m,强度应力比为2.00～80.8的地下厂房支护参数的基础上,采用反向传播(back propagration, BP)神经网络方法,训练出了地下厂房系统锚杆支护的智能化设计模型。该模型通过输入洞室跨度值和强度应力比对系统锚杆直径、间排距进行优化设计。采用对神经网络权重分析的方法探讨了洞室跨度和强度应力比对系统锚杆支护方案选择的影响程度。研究结果表明:基于BP神经网络的水电站地下厂房系统锚杆支护的智能化设计模型具有良好适用性和可靠性;在仅考虑强度应力比和洞室跨度的情况下,水电站地下厂房系统锚杆的支护设计应该优先考虑强度应力比的大小。

BP Neural Network Intelligent Design Model of Bolt Support for Underground Powerhouse

The engineering analogy method is generally used in the support design of the underground powerhouses of hydropower stations. However, the construction of high geo-stress and large-span underground caverns in the southwest China is more difficult than previous engineering projects, so the method can not be fully applicable. It is urgent to establish quantitative design theory and methods for surrounding rock support considering complex geological conditions. In this paper, based on 29 supporting parameters of underground powerhouses with span of 18.0~34.0m and strength-to-stress ratio of 2.00~80.8, an intelligent design model for the anchor support of the underground powerhouse system was trained by BP neural network method. The model optimized the design of the system anchor diameter and row spacing by inputting the cavern span and strength-stress ratio. The influence degree of cavern span and strength-stress ratio on the system anchor support scheme was discussed by using the method of neural network weight analysis. The results show that the intelligent design model based on BP neural network is applicable and reliable; for the strength-stress ratio and the span of the cavern, the magnitude of the strength-stress ratio should be given priority in the support design of the underground powerhouse system of the hydropower station.