The power cable plays a very important role in today's social development and urban operation. With the increase of people's electricity consumption，the amount of cable laying is also increasing synchronously.In order to prevent the cable from being damaged，the environmental vibration signals around the cable duct gallery are monitored，and potential threats are found in time to ensure the normal operation of the cable，a vibration monitoring system for underground cable damage prevention based on Matlab wavelet noise reduction algorithm and EMD-AR spectrum analysis was proposed. The system was composed of solar power supply module，six axis gyroscope attitude sensing module，vibration sensor module and internet of things communication module.The main control chip uses STM32F4 microcontroller. The wavelet threshold de-noising algorithm was used to remove the complex high-frequency noise and improve the signal identifiability. Then the de-noised signal was subjected to empirical mode decomposition（EMD）and AR spectrum analysis of the first six intrinsic mode function（IMF）components.Finally，the AR spectrum of the first six IMF components was accumulated to obtain the energy contrast map of the signal. The research shows that the application of this method can accurately analyze the type of vibration，which can provide a more simple and convenient method for the monitoring and analysis of vibration near the cable duct gallery，and effectively protect the safe and stable operation of the cable.