基于多层感知器的异常数据实时检测方法

上海交通大学学报（自然版） ›› 2011, Vol. 45 ›› Issue (08): 1226-1229.

基于多层感知器的异常数据实时检测方法

潘轶彪1,袁景淇1,朱凯1,陈宇2,张锐锋2

（1. 上海交通大学自动化系，系统控制与信息处理教育部重点实验室，上海 200240; 2. 贵州电力试验研究院，贵阳 550002）

收稿日期:2011-01-22 出版日期:2011-08-30 发布日期:2011-08-30
基金资助:
国家高技术研究发展计划（863）项目（2009AA04Z162）；国家自然科学基金资助项目（60974068）

Online Detection of Abnormal Data Based on Multilayer Perceptron

PAN Yi-Biao-1, YUAN Jing-Qi-1, ZHU Kai-1, CHEN Yu-2, ZHANG Rui-Feng-2

（1. Department of Automation, Key Laboratory of System Control and Information Processing of Ministry of Education of China, Shanghai Jiaotong University, Shanghai 200240, China； 2. Guizhou Electric Power Research Institute, Guiyang 550002, China）

Received:2011-01-22 Online:2011-08-30 Published:2011-08-30

摘要/Abstract

摘要： 基于神经网络的多层感知器模型，结合滚动学习预报机制，提出了一种异常数据实时检测方法.该方法在每个当前时刻通过最近的固定长度的历史数据训练神经网络，完成下一时刻的预报.通过神经网络模型残差，确定概率为P的置信区间.当下一时刻数据落入置信区间内，则该数据被判为正常；反之，则为异常.被判为异常的数据不再用作更新历史数据，而以相应的预报值代替.通过某300 MW燃煤火力电站实际过程数据的在线验证，结果证明了所提出方法的有效性.

关键词: 人工神经网络, 多层感知器, 滚动学习预报, 异常数据, 实时监测

Abstract: This work proposed an online detection method of abnormal data based on multilayer perceptron (MLP) and rolling-learning prediction mechanism. In this method, latest historical data with fixed length of data window is used to train an MLP model, and then a one-step-ahead prediction is obtained with the trained MLP model. Secondly, a confidence interval with probability p is calculated with the help of the one-step-ahead prediction and the model residual. New measurement is identified as normal one, if it falls inside the prediction interval; or an abnormal record when it is located out of the prediction interval. Instead of the real measurement, the prediction value is used to update the historical data if abnormal data occurs. Furthermore, through on-line test of real process data collected from a 300 MW coal-fired power generation unit, the effectiveness of the proposed method was verified.

Key words: artificial neural network (ANN), multilayer perceptron (MLP), rolling-learning prediction, abnormal data, online detection

中图分类号:

TB61
TK31

潘轶彪1, 袁景淇1, 朱凯1, 陈宇2, 张锐锋2. 基于多层感知器的异常数据实时检测方法[J]. 上海交通大学学报（自然版）, 2011, 45(08): 1226-1229.

PAN Yi-Biao-1, YUAN Jing-Qi-1, ZHU Kai-1, CHEN Yu-2, ZHANG Rui-Feng-2. Online Detection of Abnormal Data Based on Multilayer Perceptron[J]. Journal of Shanghai Jiaotong University, 2011, 45(08): 1226-1229.

参考文献

［1］Hodge V J, Austin J. A survey of outlier detection methodologies［J］. Artificial Intelligence Review, 2004, 22: 85126.

［2］Rousseeuw P, Leroy A. Robust regression and outlier detection ［M］. 3rd ed. New York: John Wiley & Sons, 1996.

［3］Fauconnier C, Haesbroeck G . Outliers detection with the minimum covariance determinant estimator in practice ［J］. Statistical Methodology, 2009, 6: 363379.

［4］Verboven S, Hubert M. MATLAB library LIBRA ［M］. New York: John Wiley & Sons, Inc, 2010.

［5］Bulut A, Singh A K, Shin P, et al. Realtime nondestructive structural health monitoring using support vector machines and wavelets ［J］. Proceedings of the SPIE Advanced Sensor Technologies for Nondestructive Evaluation and Structural Health Monitoring, 2005, 5770: 180189.

［6］Nairac A, Townsend N, Carr R, et al. A system for the analysis of jet engine vibration data ［J］. Integrated ComputerAided Engineering, 1999, 6(1): 5365.

［7］Upadhyaya B, Glockler O, Eklund J. Multivariate statistical signal processing for fault detection and diagnostics［J］. ISA Transactions, 1990, 29 (4): 7995.

［8］李运锋，袁景淇. 生物发酵过程效益函数的在线预报［J］. 上海交通大学学报，2005, 39（8）：341348.

LI Yunfeng, YUAN Jingqi. The online prediction of profit function for fedbatch bioprocesses ［J］. Journal of Shanghai Jiaotong University, 2005, 39（8）：341348.

［9］李运锋，袁景淇. 头孢菌素C发酵过程状态变量及效益函数预报方法［J］. 化工学报，2005，56（7）：12811283.

LI Yunfeng, YUAN Jingqi. Prediction of state variables and profit function for cephalosporin C fedbatch fermentation ［J］. Journal of Chemical Industry and Engineering, 2005，56（7）：12811283.

［10］Bishop C. Neural networks for pattern recognition ［M］. New York: Oxford University Press, 1995.

[1]	秦艺超, 黄礼敏, 王骁, 马学文, 段文洋, 郝伟. 基于人工神经网络的自航浮标测波方法可行性[J]. 上海交通大学学报, 2022, 56(4): 498-505.
[2]	黄伟恺, 马行, 刘进, 李宇平, 郑丹力. 基于改进K均值聚类的战术导弹试验数据分析[J]. 空天防御, 2019, 2(4): 69-74.
[3]	严正1，孔祥瑞1，徐潇源1，谢伟2. 微型同步相量测量单元在智能配电网运行状态估计中的应用[J]. 上海交通大学学报（自然版）, 2018, 52(10): 1195-1205.
[4]	李伟，何其昌，范秀敏，. 基于汽车操纵信号的驾驶员疲劳状态检测[J]. 上海交通大学学报（自然版）, 2010, 44(02): 292-0296.