基于盲源分离的电站锅炉多源泄漏定位研究

Passive Location in Power Plant Boiler with BSS Under Multi-source Environment

为解决电站锅炉多源泄漏问题,对多源泄漏混叠信号分离及定位进行研究。采用快速独立分量算法对混叠信号进行分离,将分离后的各源信号进行广义互相关得出时延估计,通过最小二乘定位算法得出泄漏点位置。仿真结果表明,快速独立分量算法能够将压力为0.8 MPa的5 mm、8 mm口径混叠喷流信号进行分离,不同位置分离后的同一源信号保持了原有的相关性,计算所得时延估计与真实值误差在0.029 ms;最小二乘定位算法在-0.15 ms至0.15 ms时延误差范围内时,定位误差保持在0.2 m以内,相对模型尺寸精度为2%。通过基于盲源分离的电站锅炉多源泄漏定位方法能够有效地将多源信号分离并进行定位。

To solve the problem of multi-source leakage in power plant boiler,this paper studies the separation of aliasing signal in multi-source leakage and the location.It uses the FastICA algorithm to separate aliasing signal,gets the TDOA of each signal separated by GCC,and obtains the location by LS algorithm.The simulation results show that the FastICA algorithm can separate the mixing flow signals of 5mm and 8mm nozzles with pressure of0.8 MPa,the separation signal of the same source in different positions keeps the original correlation,the error is 0.029 ms between calculation TDOA and the real value.The location error is kept within 0.2 m by LS algorithm while the TDOA error is-0.15 ms to 0.15 ms,relative precision of model dimension is 2％.Therefore,the location method of power plant in multi-source leakage based on BSS can effectively separate mixing signals.