Research On False Alarm Control Algorithm For Fiber Vibration Source Detection Under Non-stationary Disturbances

The safe transportation of oil and gas pipelines requires advanced detection technology and control strategies as a guarantee.Under non-stationary disturbances the false alarm control of fiber vibration source detection has important significance for the accuracy of vibration source detection and social security protection.With the development of optical time domain reflection positioning method and constant false alarm detection method the application of optical fiber early warning technology is becoming more and more mature.Since the singlestage detection system will cause a sudden change in the shape of the false alarm distribution and considering that the output false alarm generally does not necessary obey the Gaussian distribution this paper proposes a probability density function(PDF)shape control for the false alarm of the detection signal which enables the system output PDF to track the given target PDF realizes the closed-loop control of the false alarm signal and solves the problem of distortion of distribution shape of the detection result under the two-dimensional disturbances of space and time.In this paper the fiber vibration signal processing and the false alarm control of vibration source detection under non-stationary disturbances are studied.The main contents of the paper are listed as follows:1.Detection Principle and Data Acquisition and Processing MethodIn this part the detection principle of the optical fiber vibration source detection system is studied the composition of the optical fiber sensor hardware circuit is introduced and the hardware system is installed and tested.Then all kinds of the fiber vibration source detection data are collected in the field.After data processing the data is read and the position of the vibration source is detected by the Constant False Alarm(CFAR)detection method of the fiber vibration signal.2.Modeling of False Alarm Stochastic Distribution SystemAccording to the vibration data detected by the fiber vibration source after data processing the PDF of the false alarm data is estimated by the kernel density estimation method.The PDF of the false alarm data is approximated by the linear B-spline modeling method and transformed into the corresponding weighted dynamic model.Using the Numerical Subspace State Space Identification(N4SID)method the model parameters are identified.The weight dynamic modeling of PDF is simulated in the MATLAB compiler environment.3.Output PDF Control of False Alarm Stochastic Distribution SystemAccording to the established weight dynamic model of false alarm data the appropriate performance index function is selected.By minimizing the performance index the control algorithm of tracking target output PDF shape is designed and the corresponding control input is obtained.The control input is used to adjust the threshold of CFAR detection to improve detection performance.The false alarm PDF control system of the vibration data is simulated in the MATLAB compiler environment.