Condition Monitoring System of Electromechanical Equipment Based on Fiber Bragg Grating Sensors and Artificial Neural Network

The use of optical fiber sensor for the purpose of condition monitoring has expanded rapidly due to its unique advantages such as light weight, compact size, electromagnetic transparency and ease of multiplexing. In this work, we realized a fiber Bragg grating (FBG) based vibration sensing system and an artificial neuro network (ANN) for condition monitoring of electromechanical equipment, especially in vivo classification of step signal caused by component detachment and low frequency noise due to normal vibration during the operation, which help alert unexpected breakdown of the equipment. Firstly, the ANN training set is built on the basis of features of data which have been analyzed, and ANN can be used to describe the complex relationship between input and output. Secondly, the error back propagation algorithm is used to optimize the network to minimize the error. Finally, this classification model is implemented on MATLAB. The system shows a good performance and achieves 0.00481 error level after 500 training times. In further research, it can continue to expand on the basis of software experiment, download and debug it on hardware, and put it into practical application.