Construction of performance degradation model and development of performance estimation system for stock management of irrigation tank

本研究旨在建立功能诊断方法，并开发用于灌溉水箱库存管理的性能评估系统。采用电阻率法和表面波法两种物探方法，对3个超龄灌罐进行了野外调查。两种方法分别获得了电阻率和横波速度的空间分布，并利用自组织映射（SOM）将两者有机地结合起来。SOM是一种人工神经网络，它使用无监督学习进行训练，以产生称为地图的二维离散表示。这使得SOM对于可视化高维数据的低维视图非常有用。以横波速度、满水位和低水位时测得的电阻率及电阻率变化率为输入矢量，对土壤类型进行自动客观分类。这个程序也使我们能够估计注浆面积。提出了一种将有限元模型与蒙特卡罗滤波相结合的计算方法，利用实测沉降数据对灌池路堤内的劣化面积进行估计。

This research aimed to establish the method of functional diagnosis and to develop the performance estimation system for stock management of irrigation tanks. The field investigations by two types of the geophysical exploration methods, that is, the electrical resistivity method and the surface wave method, are carried out in three overage irrigation tanks. The spatial distributions of the electric resistivity and the S-wave velocity are individually obtained by the two methods, and the self-organizing map(SOM) is employed to combine the two results organically. SOM is a type of artificial neural network that is trained using unsupervised learning to produce a two dimensional discretized representation called a map. This makes SOM useful for visualizing low-dimensional views of high dimensional data. The soil types are classified automatically and objectively with the S-wave velocity, the electric resistivity and the change ratio of electric resistivity measured in full water level and low water level adopted as the input vectors. This procedure also enables us to estimate the grouting area. A computational method, incorporating the finite element model into identification using the Monte Carlo filter, is suggested in order to estimate deteriorated area within embankment of irrigation tank based on measured data on the settlement.