Search

This paper studies a fault isolation method for an optical fiber vibration source detection and early warning system. We regard the vibration sources in the system as faults and then detect and isolate the faults of the system based on a two-step neural network. Firstly, the square root B-spline expansion method is used to approximate the output probability density functions. Secondly, the nonlinear weight dynamic model is established through a dynamic neural network. Thirdly, the nonlinear filter and residual generator are constructed to estimate the weight, analyze the residual, and estimate the threshold, so as to detect, diagnose, and isolate the faults. The feasibility criterion of fault detection and isolation is given by using some linear matrix inequalities, and the stability of the estimation error system is proven according to the Lyapunov theorem. Finally, simulation experiments based on a optical fiber vibration source system are given to verify the effectiveness of this method. [ABSTRACT FROM AUTHOR]

Accurate and fast recognition of fiber intrusion signals has always been a fundamental task in the Optical Fiber Pre-warning System (OFPS). However, currently existing recognition models tend to focus on one aspect and lack a comprehensive approach. In this paper, a dropout-based Stochastic Configuration Network (SCN) optical fiber intrusion signal recognition model is first proposed, which is named DropoutSCN. By combining dropout with randomized algorithm models, it not only enhances the fast learning ability, but also improves the generalization performance of the recognition model. In the experiment, compared with traditional Artificial Neural Network (ANN), Random Vector Functional Link (RVFL), and original SCN models, the DropoutSCN model proposed in this paper has the lowest root-mean-square error (RMSE). In terms of time efficiency, it reduces the time delay by about 2.5 times compared with the traditional ANN. In addition, this paper applies dropout to SCN, which provides a feasible thinking and reference for the study of randomized algorithm models. [ABSTRACT FROM AUTHOR]

Optical fiber pre-warning system (OFPS) is widely used in several fields. But there are massive vibration data which is varied in complicated surroundings of OFPS. This will be a challenge to locate and identify the vibrations accurately. A new detection and recognition method is established in this article. The method consists of two parts, detection and recognition. First, the presented detection method is based on the theory of constant false alarm rate (CFAR) and dilation and erosion (DE). The former can detect the harmful intrusion signals and the latter can eliminated some isolated interferences. Harmful intrusions can be located and the data quantity for further recognition is reduced by using the detection method. Second, a multi-feature recognition method is established in this article to determine the type of the intrusions. Typical signal features, such as average magnitude difference function (AMDF), pitch period (PP) and duty cycle (DC), are used to identify the vibrations generated by vehicles, machine and artificial intrusions. In order to check out the feasibility and validity of the proposed method, several tests were carried out in Rushan of Shan Dong, China. The results show that the proposed detection and recognition method can locate the harmful intrusions and identify the type of vibrations effectively. [ABSTRACT FROM AUTHOR]

Optical fiber pre-warning systems (OFPS) based on Φ-OTDR are applied to many different scenarios such as oil and gas pipeline protection. The recognition of fiber vibration signals is one of the most important parts of this system. According to the characteristics of small sample set, we choose stochastic configuration network (SCN) for recognition. However, due to the interference of environmental and mechanical noise, the recognition effect of vibration signals will be affected. In order to study the effect of noise on signal recognition performance, we recognize noisy optical fiber vibration signals, which superimposed analog white Gaussian noise, white uniform noise, Rayleigh distributed noise, and exponentially distributed noise. Meanwhile, bootstrap sampling (bagging) and AdaBoost ensemble learning methods are combined with original SCN, and Bootstrap-SCN, AdaBoost-SCN, and AdaBoost-Bootstrap-SCN are proposed and compared for noisy signals recognition. Results show that: (1) the recognition rates of two classifiers combined with AdaBoost are higher than the other two methods over the entire noise range; (2) the recognition for noisy signals of AdaBoost-Bootstrap-SCN is better than other methods in recognition of noisy signals. [ABSTRACT FROM AUTHOR]

• An optical fiber intrusion signal recognition method based on randomized algorithm. • Universal approximation property of SCN can be guaranteed by inequality constraint. • This paper carries out fiber signal feature extraction from multiple dimensions. • The ill-posed problem tends to cause a poor generalization performance. • The hidden layer matrix is reconstructed by TSVD method to improve the accuracy. Stochastic Configuration Networks (SCN) introduce the inequality constraint of a supervised mechanism to ensure the universal approximation property of learner model. However, in the processing of building SCN, due to the properties of used activation function and the way of assigning the random input weights and biases of the hidden nodes, the hidden output matrix is often ill-posed, i.e., the matrix can be of rank deficient or demonstrate multicollinearity. Thus, the least squares method for evaluating the output weights may result in poor generalization performance for data modelling problems. This paper aims to overcome this drawback through modifying the computation of the generalized pseudo inverse of the output matrix by a Truncated Singular Value Decomposition (TSVD) method with an adaptively chosen truncation threshold. The improved SCN model is then applied for recognizing intrusion signals in Optical Fiber Pre-warning System. Experimental results show that the proposed improved algorithm can achieve higher recognition rate compared to the original SCN classifier. [ABSTRACT FROM AUTHOR]

Optical fiber intrusion signal detection is an effective long-distance sensing technology for perimeter intrusion behavior. The existing optical fiber pre-warning system (OFPS) has been able to detect and identify a variety of intrusion behaviors. However, when various intrusion behaviors occur at the same time, identification performance in OFPS will decrease. In this paper, we model the mixed intrusion signal from a geometric perspective at first, and then the unmixing method is studied under the model constraint condition. We set the model error function as an objective function, and transform the unmixing problem into a constrained quadratic programming problem. Through iteration, the pure intrusion signals and their proportion are obtained. This method does not require prior knowledge such as the number of pure components and corresponding characteristics. Compared with the conventional methods, the proposed algorithm can obtain more accurate features under unsupervised conditions. [ABSTRACT FROM AUTHOR]

Feature extraction of the optical fiber intrusion signal is one of the most basic problems in the optical fiber pre-warning system (OFPS). This paper proposes a spectrum analysis method in the OFPS that adopts the time-frequency analysis technique to realize multiband imaging of the intrusion signal at first. Subsequently, the characteristic spectrum of the intrusion signal can be obtained as the intrusion feature from the imaging results. Since the practical intrusion spectrum is usually a mixture of multiple intrusion types, in order to extract the spectrum corresponding to each intrusion type, the mixed spectrum is characterized by a linear mixing mode. Then, we exploit the nonnegative matrix factorization (NMF) in the mixture analysis. Moreover, we add physical constraints to the NMF and solve it by gradient descent algorithm. Finally, we extract the characteristic spectrums from the mixed spectrum resulting to the spectrum unmixing. The real data experiments verify the effectiveness of this method. [ABSTRACT FROM AUTHOR]

This paper presents detection and recognition method to locate and identify harmful intrusions in the optical fiber pre-warning system (OFPS). Inspired by visual attention architecture (VAA), the process flow is divided into two parts, i.e., data-driven process and task-driven process. At first, data-driven process takes all the measurements collected by the system as input signals, which is handled by detection method to locate the harmful intrusion in both spatial domain and time domain. Then, these detected intrusion signals are taken over by task-driven process. Specifically, we get pitch period (PP) and duty cycle (DC) of the intrusion signals to identify the mechanical and manual digging (MD) intrusions respectively. For the passing vehicle (PV) intrusions, their strong low frequency component can be used as good feature. In generally, since the harmful intrusion signals only account for a small part of whole measurements, the data-driven process reduces the amount of input data for subsequent task-driven process considerably. Furthermore, the task-driven process determines the harmful intrusions orderly according to their severity, which makes a priority mechanism for the system as well as targeted processing for different harmful intrusion. At last, real experiments are performed to validate the effectiveness of this method. [ABSTRACT FROM AUTHOR]