改进灰色BP神经网络模型在建筑物沉降监测中的应用

Application of Grey BP Neural Network Model in Building Settlement Monitoring

建筑物的变形是由于多种复杂因素的影响,文章针对该影响以及单一预测模型精度不高的问题,建立了基于灰色模型和BP神经网络模型的组合预测模型。并采用复化梯形求积法对灰色模型的背景值改进,构建基于改进的组合预测模型。结合改进的灰色GM(1,1)BP神经网络组合模型对马鞍山市的某建筑物的沉降进行预测。通过对比可以看出,改进的灰色BP神经网络预测精度最高,其拟合程度更加接于实测值,可更好地适用于实际工程的中长期预测。

The deformation of buildings is due to the influence of many complex factors.Aiming at this influence and the low accuracy of single prediction model,this paper establishes a combined prediction model based on grey model and BP neural network model.The background value of the grey model was improved by the complex trapezoidal quadrature method,and the combined prediction model based on the improvement was constructed.Combined with the modified GM(1,1)BP neural network model,the settlement of a building in Maanshan city is predicted.Through comparison,it can be seen that the improved grey BP neural network has the highest prediction accuracy,and its fitting degree is more in line with the measured value,which can be better applied to the long-term prediction of practical projects.