基于改进最小方差无失真响应角度谱算法的 气体泄漏定位方法*

针对噪声环境下微小气体泄漏难以准确定位的问题，提出了一种基于改进最小方差无失真响应角度谱算法的气体泄漏定位方法。该算法通过引入信噪比追踪加权的方式，提取受噪声影响较小且单个声源能量占优的时频支撑域，并通过Softplus激活函数自适应地调整不同频率分量对角度谱函数的贡献，增加泄漏声源占优的时频域权重；此外，引入基于时频稀疏性的分频带处理，使各子频带内存在一个主导声源能量占优，抑制低频段噪声能量的积累同时避免高频混叠现象。通过软件仿真计算以及实验验证算法的性能，结果表明改进最小方差无失真响应角度谱算法可以实现气体泄漏源的精准定位，定位结果的最大误差在3.5°以内。相比传统算法，该方法在低信噪比和低采样点数下有更高的稳定性、抗噪能力及准确率，可为气体泄漏定位的实际应用提供一定的参考价值。

Gas leak location method based on improved minimum variance distortionless response angle spectrum algorithm

It is difficult to accurately locate tiny gas leaks in noisy environments, a gas leak location method based on an improved minimum variance distortionless response (MVDR) algorithm is proposed. The algorithm trackes and weights the signal noise ratio to extract the time-frequency support domain, which is less affected by noise and dominates the energy of a single sound source. Then, the Softplus activation function is used to adaptively adjust the contribution of different frequency components to the angle spectral function, which could increase the time-frequency domain weights dominated by the leaked sound source. In addition, the sub-band processing based on time-frequency sparsity is introduced to ensure that there is a dominant sound source in each sub-band with energy dominance. Therefore, the accumulation of low frequency noise energy is suppressed, while high frequency aliasing is avoided. By simulation and experiments, the results show that the improved MVDR angle spectrum algorithm can realize the precise location of the gas leak source, and the maximum error of the positioning results is within 3.5°. Compared with the traditional algorithm, this method has higher stability, anti-noise ability and accuracy under low signal-to-noise ratio and low sampling points, which can provide a certain reference value for the practical application of gas leak location.