为及时可靠地检测出气动失稳现象，保障压气机的安全工作，采用 Laplace小波构建原子函数库，引入稀疏分解算法提取阻尼比小于 0的信号成份，通过与预设阈值比较内积值的大小，进行气动失稳信号的检测。结合在线检测系统的特点，将 Laplace小波蜕化为衰减正弦函数，限定了参数取值范围；并仅进行一步匹配追踪，由此减少了稀疏分解的计算量。采用压气机试验台实测数据验证，该方法可以在压气机进入喘振前 49～154ms准确发出报警信号。

A new approach is introduced to detect the aerodynamic instability phenomenon timely and reli. ably，to ensure the safety of compressors. Firstly，the atom library is constructed with Laplace wavelet. The sparsedecomposition is introduced to extract the signal component with damping ratio being less than 0. The instabilityis detected if the inner product value is larger than a preset threshold. Then，taking into account the characteris. tics of the online detecting system，the Laplace wavelets are degraded into decay sine functions，and parameters are limited into a small range. Only one step of matching pursuit is calculated. In this way，the computing of sparse decomposition is reduced greatly. Finally，examples are given with data from compressor test rig. It is dem.onstrated that alarm signals can be sent out 49-154ms before the compressor entering surge state.

V263.3

国家自然科学基金项目( 51205311)。