遗传神经网络方法预测地下工程位移演化规律

提出并分析了一种用于隧道和地下结构位移演变预测预报的遗传神经网络方法·应用遗传算法优化传统的神经网络结构,避免了人为选择网络结构的盲目性,较好地解决了神经网络结构选择问题,同时提高了网络学习的效率和推广预测能力;利用这种非线性智能识别新方法,预测下步施工位移变形量,并与工程中监测到的历史数据进行对比分析,以便工程技术人员据此及时调整和优化施工步序,维护地下结构的稳定性·工程实例分析表明,该方法随着样本的积累,预测精度不断提高,并具有实时性的优点,具有广泛的应用前景

Displacement Evolution Forecast by Genetic-Neural Network Method in Underground Engineering

The genetic neural network method was used in forecasting the displacement evolution history in tunnel and underground structure. It is shown that the neural network structure in addition of the genetic algorithm optimized system can escape from the blindness which is the common phenomenon in man made choice to the network structure. The neural structure enhances the efficiency of the network study and the capability of the forecasting application. It allows the engineering staff convenient to adjust and optimize the construction step in time by non linear intelligent recognition to forecast the displace distortion. Measurement to make a contrast analysis with the historical data in order to maintain the stability of the underground structure. The engineering case analyses indicate that there is an extensive prospect for this real time prediction method and the forecasted precision can be continuously better with the accumulating of the samples.