基于遗传神经网络的灌浆单位注灰量预测方法  

Prediction method of grouting unit amount of cement injection based on genetic neural networks

作　　者：闫福根 李子康 钟坤 YAN Fugen;LI Zikang;ZHONG Kun(Changjiang Institute of Survey,Planning,Design and Research Co.,Ltd.,Wuhan 430000,Hubei,China)

机构地区：[1]长江勘测规划设计研究有限责任公司,湖北武汉430010

出　　处：《水利水电技术(中英文)》2023年第S02期224-230,共7页Water Resources and Hydropower Engineering

基　　金：长江设计集团自主创新项目“深水水下灌浆关键技术研究与应用”(CX2021Z21);长江设计集团自主创新项目“智慧灌浆三维可视化分析平台开发与应用”(CX2020Z08)。

摘　　要：灌浆为地下隐蔽工程,合理预测单位注灰量对于控制灌浆工程质量具有重要指导意义。考虑单位注灰量与岩体透水率、岩性、RQD值、破碎程度、灌浆压力和高程等有高度非线性关系,提出一种基于遗传神经网络模型的单位注灰量预测方法,模型由输入层(6个神经元节点)、隐藏层(13个神经元节点)和输出层(1个神经元节点)构成,并采用遗传算法确定初始权重向量,以确保模型获取的最优值为全局最优解。为验证模型精度,以某水电站为例,采用模型对10组样本进行预测,结果表明:10组样本单位注灰量预测值和实际值绝对误差为0.38~78.75 kg/m,相对误差平均值为15.78%,两者基本吻合,模型具有较好的预测精度,灌浆工程师可根据预测值实时调整灌浆工艺,实现灌浆质量过程控制。Grouting is an underground concealed engineering,and reasonable prediction of the unit amount of cement injection has important guiding significance for controlling the quality of grouting engineering.Considering the unit amount of cement injection has a high nonlinear relationship with the water permeability,lithology,RQD,crushing degree,grouting pressure and elevation of rock mass,this paper proposes a prediction method for the unit amount of cement injection based on the genetic neural network model.The model consists of input layer(6 neuron nodes),hidden layer(13 neuron nodes)and output layer(1 neuron node),and the genetic algorithm is used to determine the initial weight vector to ensure that the optimal value obtained by the model is the global optimal solution.In order to verify the accuracy of the model,this paper takes a hydropower station as an example and uses the model to predict 10 sets of samples.The result show that the absolute error between the predicted value and the actual value of the unit amount of cement injection in the 10 groups of samples is 0.38~78.75 kg/m,and the average relative error is 15.78%,which is basically consistent.The model has good prediction accuracy,and the grouting engineer can adjust the grouting process in real time according to the predicted value to realize control of grouting.

关 键 词：遗传神经网络 单位注灰量 预测分析 灌浆工程 

分 类 号：TV72[水利工程—水利水电工程]