The goal of this paper is to present an approach to detect structural changes by using nonparametric models. The impulse response functions (IRFs) of mechanical systems in healthy conditions are identified through the sum of convolution expanded on an orthonormal basis. The Kautz filters with multiple poles optimized are implemented as an efficient orthonormal basis to identify nonparametric reference models. Identifying the IRFs of a smart beam with PZTs actuators/sensors coupled is used to illustrate the necessary procedures. The damages are simulated as a loss of bolts, nuts, and washers attached near a PZT sensor. A statistical procedure of a hypothesis test is also used to confirm the smart beam’s structural state. The experimental results achieved show that it is possible to detect the structural changes with statistical confidence.

本文的目的是提出一种通过使用非参数模型来检测结构变化的方法。通过在正交基础上展开的卷积和，可以确定机械系统在健康状态下的脉冲响应函数（IRF）。具有优化过的多极点的Kautz滤波器被用作有效的正交基础，以识别非参数参考模型。识别耦合了PZT致动器/传感器的智能光束的IRF，以说明必要的过程。损坏模拟为PZT传感器附近的螺栓，螺母和垫圈的损失。假设检验的统计过程也用于确认智能梁的结构状态。获得的实验结果表明，有可能以统计的置信度检测结构变化。