带偏差单元BP神经网络土体灌浆压力预测

灌浆压力是土体灌浆加固的重要参数。基于神经网络非线性映射特性，分析土体灌浆压力主要影响因素，建立符合一般工程判断和决策思维的BP网络预测模型，并引入偏差单元对其结构进行改进，实现了快速收敛，较高精度得出灌浆预测压力的具体数值。预测结果与室内灌浆试验压力对比表明，带偏差单元BP神经网络的土体灌浆压力预测结果具有较高准确性和一定的实用意义。

Prediction on Grouting Pressure for Soil Mass by BP Neural Network with Bias Element

Grouting pressure is an important parameter in soil mass consolidation by grouting. Based on the nonlinear mapping characteristics of neural network, the main factors affecting the grouting pressure are analyzed and the BP network prediction model consistent with general engineering judgment and decision- making thoughts is established. In addition, the structure of the model is improved by introduction of bias element; thus fast convergence is realized and the specific value of grouting pressure estimate with high accuracy is obtained. The comparison between predicted result and measured pressure from indoor grouting test shows that the estimate of the grouting pressure by BP neural network with bias element has high accuracy and certain practical significance.