基于振动信号的推移质输沙率监测研究

Research on monitoring bedload transport rate based on vibration signals

推移质输沙率精确测量是河流动力学的研究难点之一,传统的推移质直接测量方法受限于推移质运动的复杂性和测量仪器的局限性,无法对推移质进行长时间的连续监测。为了连续监测推移质运动,采用推移质间接测量方法,利用安装振动传感器的冲击板系统对推移质运动产生的振动信号进行高分辨率采集,提取振动信号特征值,并建立与推移质输沙率及流量之间的关系,进一步采用人工神经网络算法对推移质输沙率进行有效预测。结果表明:推移质振动信号的特征值均值与推移质输沙率有良好的相关性;中等流量条件的神经网络预测效果最佳,小流量条件的预测效果次之,大流量条件的预测效果相对较差,且其最优输入参数与流量及河床变化均有关系。

The accurate measurement of bedload transport rate presents a significant challenge in river dynamics due to the complex nature of bedload movement and the limitations of existing measuring instruments.Conventional direct measurement techniques fail to provide continuous long-term monitoring.To address this,an indirect bedload measurement method,capturing high-resolution vibration signals from bedload movement using an impact plate system equipped with vibration sensors was employed.After extracting the characteristic values of these vibration signals their relationship with bedload transport rate and flow rate was established.Artificial neural network algorithm was also incorporated to predict bedload transport rates effectively.The results reveal a strong correlation between the mean eigenvalue of the vibration signal and the bedload sediment transport rate.Meanwhile,the neural network performed optimally under medium flow conditions,with decreased performance under low and high flow conditions.The optimal input parameters have been found to be related to flow rate and changes in the riverbed.