|
|
|
Elevator traffic pattern fuzzy recognition based on
least squares support vector machine |
| WANG Lu-jun, LV Zheng-yu |
| College of Electrical Engineering,Zhejiang University,Hangzhou 310027,China |
|
|
|
Abstract A method for elevator traffic pattern fuzzy recognition was proposed. The relative membership of two traffic patterns was learned with least squares support vector machine (LSSVM) regression algorithm. Then the memberships of all traffic patterns were obtained by comparing with each other. The LSSVM algorithms of binary classification and traditional multi-class classification were introduced, and the relationship of multi-class classification and function regression was analyzed. Results show that LSSVM function regression algorithm can be used in multiclass classification. If the membership is regarded as class label, then two traffic patterns can be fuzzy recognized by LSSVM regression algorithm. Elevator traffic pattern was recognized in three steps in order to improve the linearity of LSSVM. Experimental results showed that the membership curves of traffic patterns versus time obtained by the method were similar to the curves given by elevator group control experts. The average recognition error using the method was less than that using a neural network, and the fuzzy recognition results can be used as input parameters of elevator group control system.
|
|
Published: 01 July 2012
|
|
|
基于LSSVM的电梯交通模式的模糊识别
提出可以对电梯交通模式进行模糊识别的方法.采用最小二乘支持向量机(LSSVM)的回归算法来学习2种交通模式的相对隶属度,通过相对比较法得到当前时刻所有交通模式的隶属度.介绍了LSSVM二值分类算法及传统的多值分类算法,分析LSSVM多值分类与函数回归的关系.分析结果表明,采用函数回归算法可以进行多值分类.若以交通模式的隶属度作为类标,则可采用LSSVM的回归算法来进行2种交通模式的模糊分类.为了提高LSSVM的线性度,分3步逐步细分电梯客流的交通模式.实验结果表明,采用该方法得到的各交通模式隶属度随时间的变化曲线与依据群控专家经验得到的曲线非常相似,识别结果的平均误差小于应用神经网络识别的平均误差,可将识别结果作为电梯群控系统的输入参数.
|
|
| ［1］ SUN Jin, ZHAO Qianchuan, LUH P B. Optimization of group elevator scheduling with advance information ［J］. IEEE Transactions on Automation Science and Engineering, 2010, 7(2): 352-363．
［2］ LIU Jian, WU Chengdong, WANG Xin, et al. A hybrid control for elevator group system ［C］∥2010 3rd International Workshop on Advanced Computational Intelligence (IWACI). Suzhou: IEEE, 2010: 491-495．
［3］ KIM C B, SEONG K A, KWANG H L. Design and implementation of a fuzzy elevator group control system ［J］.IEEE Transactions on Systems, Man, and Cybernetics, 2003, 11(4): 518-527．
［4］宗群,尚晓光,岳有军, 等. 电梯群控系统的交通模式识别［J］. 控制与决策,2001,16(2):163-166．
ZONG Qun, SHANG Xiaoguang, YUE Youjun, et al. Traffic pattern recognition of elevator group control systems ［J］. Control and Decision, 2001, 16(2): 163 -166．
［5］杨祯山,邵诚.电梯群控技术的现状与发展方向［J］. 控制与决策,2005,20(12): 1321-1331．
YANG Zhenshan, SHAO Cheng. On the present situation and developing trends of elevator group supervisory control technologies ［J］. Control and Decision, 2005, 20(12): 1321-1331．
［6］许玉格,罗飞.新型电梯群控系统交通模式识别方法［J］.控制理论与应用, 2005,22(12):900-904．
XU Yuge, LUO Fei. Traffic pattern recognition method for novel elevator system ［J］. Control Theory and Applications, 2005, 22(12): 900-904．
［7］SCHOKOPF B, SMOLA A J. Learning with kernels ［M］. Cambridge: MIT, 2001: 23-42．
［8］VAPNIK V. Statistical learning theory ［M］. New York：Wiley, 1998: 493-521．
［9］WESTON J, WATKINS C. Support vector machines for multiclass pattern recognition ［C］ ∥ Proceedings of 7th European Symposium on Artificial Neural Networks. Belgium: DFacto Publishing, 1999: 219-224．
［10］姜静清.最小二乘支持向量机算法及应用研究［D］.吉林:吉林大学,2007:36-42．
JIANG Jingqing. Study on least squares support vector machine and its applications ［D］. Jilin: Jilin University,2007:36-42．
［11］孙德山.支持向量机分类与回归方法研究［D］.长沙:中南大学,2004:58-62．
SUN Deshan. Study on least squares support vector machine classification and regression ［D］. Changsha: Zhongnan University,2004:58-62． |
|
Viewed |
|
|
|
Full text
|
|
|
|
|
Abstract
|
|
|
|
|
Cited |
|
|
|
|
| |
Shared |
|
|
|
|
| |
Discussed |
|
|
|
|
