基于多传感器和RBF神经网络的建筑沉降监测

Building settlement monitoring based on multi-sensor and RBF neural network

针对传统建筑物结构沉降监测存在可靠性不高和预测精度差的问题,提出了一种基于多传感器和RBF神经网络的建筑沉降监测方法。分别通过多种传感器和GPRS通信模块等硬件设备对建筑物沉降信息进行采集和无线传输;对传感器采集的监测数据进行对比分析,以便得出建筑物的沉降情况,并对可能的沉降点构建了RBF神经网络预测模型。此外,采用蛙跳算法对RBF神经网络的结构参数进行优化。实验结果表明,该方法能够在实际环境中对可能的建筑结构沉降做出准确评估,且预测误差较小,最大相对误差为4.83%,具有较好的预警能力。

Aiming at the problems of low reliability and poor prediction accuracy in traditional building structure settlement monitoring,a building settlement monitoring method based on multi-sensor and RBF neural network is proposed.Building settlement information is collected and transmitted through hardware devices such as digital pressure sensor,GPRS communication module and digital dual-axis inclination sensor.The monitoring data collected by the two sensors are compared and analyzed so as to obtain the settlement of the building,and the RBF neural network settlement prediction model is constructed for the possible settlement points.The experimental results show that the method can accurately evaluate the possible building structure settlement in the actual environment,and the prediction error is small,the maximum relative error is 4.83%,and it has good early warning capability.