基于时间序列法的胶带机火灾信号分析与合成

Analyzing and synthesizing of conveyor belt fire signal by using time sequence method

胶带机火灾信号属于非平稳的随机过程,根据实际遇到的随机信号可以由一个白噪声信号通过一个有理传输函数系统逼近的原理,利用自回归模型分析、合成出了胶带机火灾和非火灾信号。将格式滤波器结构引入了单通道自回归模型的参数求解中,避免了应用横向滤波器时对数据窗两侧信号做人为假定为零的假设,降低了数据窗口较短时产生的误差。结果表明,通过该方法合成出的胶带机火灾和非火灾信号可有效扩充测试火灾探测算法的数据集、验证算法的准确性和灵敏度。

Conveyor-belt fire signal belongs to non-stable random process. A method of synthesizing fire and non-fire signals using auto-regressive model is put forward, which is based on the theory that stable random signal can be obtained by a white noise passing a rational function system. The usage of lattice filter avoid artificial hypothesis that considers the data of two sides of data window is zero, and so the error is reduced when the data window is very short. The result states that the simulating signals have the same trends in the whole and have a matching power spectrum with the original true signals. Using this method that provides favorable data set to test the algorithm of detecting conveyor belt fire extends the samples set.