Abstract The loose parts detection in low signal–noise-ratio (SNR) environments is investigated, which is frequently encountered in safety monitoring of the primary coolant system in nuclear power plant. The objective function method (OFM) based on high-order statistics is firstly applied to the impact response extraction under strong noise background. The parameter optimization of OFM is carried out to improve the solution speed and signal extraction ability. The approach is verified through the steel ball drop experiment. The results show that, compared with the wavelet algorithm, the OFM can suppress the noise more effectively and thus extract the impact response signal submerged in noise better. Secondly, a loose parts detection approach is developed by combining the OFM with the support vector machine. The detection results based on the steel ball drop experiments show that the present method can effectively identify the weak impact response and has good anti-false positive and anti-missing ability.

中文翻译：

---

盲反卷积与支持向量机相结合的松散检测方法

摘要 研究了核电站主冷却系​​统安全监测中经常遇到的低信噪比（SNR）环境下的松动部件检测。基于高阶统计量的目标函数法(OFM)首次应用于强噪声背景下的冲击响应提取。对OFM进行参数优化，提高求解速度和信号提取能力。该方法通过钢球落下实验得到验证。结果表明，与小波算法相比，OFM能够更有效地抑制噪声，从而更好地提取淹没在噪声中的冲击响应信号。其次，通过将 OFM 与支持向量机相结合，开发了一种松散部件检测方法。