摘要
考虑到电梯交通流本身所存在的非线性、复杂性和随机性,提出了一种基于小波支持向量机的电梯交通流预测模型。该方法采用某大厦实测的4周交通流数据,以前三周统计的交通流时间序列构成训练样本对预测模型进行训练,后一周的交通流时间序列作为测试样本。仿真实例验证了该模型在精度、训练时间、泛化能力、最优性等方面取得了较好的效果。
Considering the nonlinearity, complexity and randomicity of elevator traffic flow, the prediction model of elevator traffic flow based on wavelet support vector machines was proposed. The method utilized the 4-week traffic flow data of a building surveyed. The former 3-week traffic flow series as training samples was used to train the prediction model and the latter one-week traffic flow series as testing samples was used to validate the prediction model. The simulation verifies the proposed model has the better effects in terms of the prediction preciseness, learnt time, genability and optimal possibility.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2009年第19期6321-6324,共4页
Journal of System Simulation
关键词
电梯群控系统
电梯交通流预测
小波支持向量机
BP神经网络
小波神经网络
高斯核
elevator group control system
elevator traffic network
wavelet neural network
Gaussian kernel flow prediction
wavelet support vector machine
BP neural
