The traditional fault warning research based on the key components of numerical control （NC）machines have some limitations in engineering applications, such as sensor life problems and tedious testing procedures. By researching deeply the fault frequency of key components, and considering the shock absorption of the special structure and the interaction between the key components, a signal analysis method combining Daubechies filter with Fourier spectrum is proposed. The method of installing the original 〖JP2〗sensors into the specific components was changed, and a macroscopic overall test method was put forward. Finally, through the compared researching of overall test and single point test, the correctness of the overall testing method based on characteristic frequency is verified.