高压釜泄漏声音的高频高阶空间交互识别算法研究
2024,32(10):169-174
摘要：高压釜是湿法冶金领域常用的重要设备,存在危险气体泄漏的风险。同时,泄漏会导致高压釜内压不稳,严重时甚至引起爆炸,威胁设备和生产安全。高压釜泄漏发生时的声音监测是比较常规的手段,文章提出了一种高压釜泄漏声音的高频高阶空间交互识别算法。首先通过高通滤波器消除低频噪声对于识别结果的干扰,然后利用递归门控卷积块实现高频分量在高阶空间的交互,最后使用全卷积层识别高压釜泄漏的声音。实验结果表明,所提算法具有较好的高压釜泄漏识别效果,平均置信度达到0.93,以0.65作为置信度阈值时,识别准确率可达到99.5%。
关键词：湿法冶金;高压釜;泄漏声音识别;递归门控卷积
High-Frequency Higher-Order Spatial Interaction Algorithm for Autoclave Leaking Voice Recognition
Abstract：The autoclave is a critical piece of equipment often used in the field of hydrometallurgy, presenting the risk of hazardous gas leaks. Additionally, such leaks could destabilize the pressure within the autoclave, potentially causing explosions that threaten both the equipment and production safety. Voice monitoring during autoclave gas leaks is a standard procedure. This paper proposes a high-frequency, high-order spatial interaction recognition algorithm for the voice of autoclave leaks. Firstly, low-frequency noise interference is eliminated from the recognition results using a high-pass filter. Next, a recursive gated convolutional block is employed to enable high-frequency components to interact in high-order spatial dimensions. Finally, a fully convolutional layer is utilized to recognize the sound of autoclave leaks. Experimental results demonstrate that the proposed algorithm achieves good recognition results for autoclave leaks, with an average confidence level of 0.93. When the confidence threshold is set at 0.65, the recognition accuracy can reach up to 99.5%.
Key words：hydrometallurgy; autoclave; leak voice recognition; recursive gated convolution
收稿日期：2023-08-30
基金项目：重点研发计划(工业软件)(2022YFB3304901)；金属冶炼重大事故防控技术支撑基地项目。
