In the realm of using acoustic methods for locating leakages in water supply pipelines, existing research predominantly focuses on single leak localization, with limited exploration into the challenges posed by multiple leak scenarios. To address this gap, a genetic algorithm‐optimized frequency‐domain convolutional blind source separation algorithm is proposed for the precise localization of multiple leaks. This algorithm effectively separates mixed leak sources and accurately estimates the delays of source propagation. Signal simulations confirm the algorithm's effectiveness, revealing that the distribution of leak positions, signal‐to‐noise ratio, and frequency characteristics of the leakage source all influence the algorithm's performance. Comparative analysis demonstrates the algorithm's capability to eliminate signal interactions, facilitating the localization of multiple leaks. The algorithm's efficacy is further validated through extensive full‐scale experiments, underscoring its potential as a novel and practical solution to the complex challenge of multiple leakage localization in water supply pipelines.

中文翻译：

---

遗传算法优化了供水管道中多个泄漏位置的频域卷积盲源分离


在使用声学方法定位供水管道中的泄漏领域，现有研究主要集中在单个泄漏定位上，对多种泄漏场景带来的挑战的探索有限。为了解决这一差距，提出了一种遗传算法优化的频域卷积盲源分离算法，用于多个泄漏的精确定位。该算法有效地分离了混合泄漏源，并准确估计了源传播的延迟。信号仿真证实了该算法的有效性，揭示了泄漏位置的分布、信噪比和泄漏源的频率特性都会影响该算法的性能。比较分析表明，该算法能够消除信号相互作用，从而促进多个泄漏的定位。该算法的有效性通过广泛的全面实验得到进一步验证，强调了它作为一种新颖实用的解决方案的潜力，以应对供水管道中多次泄漏定位的复杂挑战。