摘要
针对动车(EMU)经过时引起的系统误报,提出了一种新颖的识别算法。采用将信号的时域峰值与频域内固定频段内能量积分的方法以提高系统抗干扰性能。利用时频分析方法中的短时傅立叶变换(STFT),采用Hanning窗窗长为128,窗滑动点数为64,将时域峰值所在时间段对应的STFT 1~5Hz频率范围内的能量进行积分,以此辨别入侵与非入侵事件。根据数据分析结果以及实际应用情况,所提算法可以在保证人为入侵漏报率FNR为零的情况下,有效地将EMU经过引起的误报率FAR降低80%,使得光纤布拉格光栅(FBG)周界安防系统的抗环境干扰性得到大幅提高。
Perimeter intrusion defense system based on fiber Bragg grating (FBG) is a new generation fiber optic perimeter security safety system. The FBG intrusion detection system can intelligently determine location with high performance. The false alarm rate (FAR) is one of the most fundamental parameters of intrusion system. It is related to a lot of factors such as the sensitivity of FBG sensor, the installation environment and installation quality. In this paper, in a fence with FBG intrusion system, using FBG wavelength peak in time domain together with frequency domain band energy integral method to improve the anti-jamming performance algorithm is presented for recognition of real intrusion and nonintrusion,especially for EMU-induced false alarm events. We use time-frequency analysis method of short-time Fourier.transform (STFT), with Hanning window length of 128, the window sliding points of 64, So the integration of the energy within the frequency range of the wavelength peak corresponding to the STFT of 1~5 Hz is obtained,in order to identify the invasion of non-invasion events. According to the results of data analysis and practical application of this method,assuming the premise of a false negative rate (FNR) of zero,the EMU caused FAR is reduced by 80%. This algorithm greatly improves the system ability of resistance to environmental disturbance.
出处
《光电子.激光》
EI
CAS
CSCD
北大核心
2013年第6期1138-1144,共7页
Journal of Optoelectronics·Laser
基金
国家自然科学面上基金(51175393)
博士点专项基金(20100143110008)资助项目