The Application of Random Neural Networks in Fiber Optic Vibration Signal Recognition

With the prosperity of information technology, fiber optic sensing technology is popularly applied in scenarios such as disturbance detection and security issues. In order to enhance the recognition effect of detection systems on fiber optic vibration signals, a signal detection method grounded on random neural networks is proposed. The research method collects vibration signal data through fiber optic detectors and converts the vibration signal using Fourier transform. At the same time, the collected signals are denoised using ensemble empirical mode decomposition to obtain processed real-time vibration signals. Random neural networks are utilized to identify fiber optic vibration signals and achieve the resolution and warning of abnormal signals. The research results show that the convergence effect of the research model has been effectively improved, and the convergence speed is faster compared to other algorithms. At the same time, the research model has a high recognition accuracy for fiber optic signals, with a signal recognition accuracy of 91.06% -95.41%, which is 2.65% - 23.12% higher than other detection models. The research model can also maintain a low signal recognition error in low signal-to-noise ratio environments, with a reduction of 4.97% -23.64% compared to other algorithms, indicating that the signal detection coverage and accuracy of the research model remain at a high level and can effectively detect the trend of signal changes. Therefore, the design of the research model can validly raise the precision of signal detection and provide a highly accurate detection model for signal recognition of fiber optic sensors.