The monitoring of environmental vibration is very important for the safe operation of the equipment. The perception of environmental vibration information can be realized by using piezoelectric vibration energy harvester, and then the safe operation status of the equipment can be monitored wireless through intelligent information processing methods. Combining the wireless sensing with deep learning, on the basis of fully studying the output signal characteristics of piezoelectric vibration energy harvester, an optimized convolutional neural network model is proposed to identify the abnormal vibration patterns of the environment. The intelligent sensing wireless monitoring sensor node is designed and implemented. When the system works, the sensor node can monitor the environmental vibration and temperature information and alarm the abnormal events. The test results show that the wireless transmission distance exceeds 100 m,the sensor node can realize real-time monitoring of the environmental vibration events, the alarm time is less than 5 s, and the accuracy of environmental vibration pattern recognition can reach 95.7%. The node can monitor the ambient temperature and give an accurate alarm for abnormal combustion events in less than 3.7 s. The node has a wide application prospect in fields target monitoring and other occasions.