论文部分内容阅读
化工过程的数据中常常含有较多的随机误差和粗差干扰,导致传统的稳态检测方法无法得到准确的结果,从而降低系统故障诊断的可靠性。针对实际的工业过程数据,提出一种融合自适应平滑技术的稳态检测方法,该方法首先以基于导数分析的自适应平滑算法进行降噪处理,消除随机误差的影响,然后引入阈值拟合技术进一步抑制粗差干扰,以多项式滤波方法对数据进行稳态检测,根据测量信号的趋势特征确定过程是否处于稳态。仿真实验研究表明:融合自适应平滑技术的稳态检测方法能够克服传统稳态检测方法中随机误差和粗差干扰对检测结果的影响,进而显著提高稳态检测处理的准确性,检测结果明显优于传统的基于多项式滤波的检测方法。
Chemical process data often contain more random error and gross error interference, resulting in the traditional steady-state detection methods can not get accurate results, thereby reducing the reliability of system fault diagnosis. Aiming at the actual industrial process data, a steady-state detection method based on adaptive smoothing technique is proposed. Firstly, the adaptive smoothing algorithm based on derivative analysis is used to reduce the noise and eliminate the influence of random error. Then, the threshold fitting technique To further suppress the interference of gross errors, polynomial filtering method for steady-state detection of the data, according to the trend of the measured signal to determine whether the process is in steady state. The simulation experiments show that the steady-state detection method based on adaptive smoothing technique can overcome the influence of random error and gross error on the detection results in traditional steady-state detection method, and then significantly improve the accuracy of steady-state detection and the detection results are obviously superior In the traditional polynomial filtering based detection method.