路灯开关时间的预测算法研究
发布时间:2018-02-05 20:25
本文关键词: 路灯开关控制 预测 小样本 最小二乘支持向量机 出处:《华中科技大学》2015年硕士论文 论文类型:学位论文
【摘要】:路灯照明在市民出行、社会秩序和交通安全中处于特别重要的位置。近年来,城市路灯每年以10%至20%的速度增长。2014年全国城市路灯消耗约2,485亿kWh电能,按每天平均照明12小时,每天若能减少照明1分钟,则每年大约可节约3.45亿kWh,相当于减少排放二氧化碳约34.4万吨。现有的城市路灯开关控制大多是根据模板表定时触发,在正常天气时经常早晨关灯晚,晚上开灯早,造成能源的浪费;在恶劣天气时不能自动提早开灯或延迟关灯,需要工作人员手动下发开关灯命令,从命令下发到路灯完全点亮大约需要10~15分钟,而这时天空已经变得很暗,严重影响市民的正常生活甚至可能造成交通事故。结合现有路灯开关控制的不足,提出将预测应用到路灯开关控制中,提前得到未来的开关灯时间,不仅为路灯控制提供一定的决策支持,也能节约能源。本文研究了路灯开关时间的特点和其影响因素,重点分析了光照度的特性,提出了天气的数值化映射。路灯系统环境复杂,样本数据受到各种影响,根据实验数据的特性介绍了本文使用的数据预处理方法。比较了现有常用的预测算法,因本实验数据的小样本特性,选择用最小二乘支持向量机进行建模。依照训练数据的特性给出了最小二乘支持向量机的模型结构、核函数选择、参数优化等。通过仿真分析可以发现,与BP神经网络模型对比,最小二乘支持向量机在小样本预测中的性能更好。当光照度变化趋势越具体时,预测效果会越好。仅仅根据日落时间、最低温度、最高温度、天气状况,预测的平均绝对误差在2分15秒之内。相比日落时就开灯,根据本文预测的时间进行开灯,每盏灯每天大约能节省7分33秒的照明时间。
[Abstract]:In recent years, urban street lighting is increasing at the rate of 10% to 20% per year. In 2014, the consumption of urban street lights in China was about 2. 48.5 billion kWh power, with an average of 12 hours of lighting per day, can save about 345 million kWh per year if the lighting is reduced by one minute per day. Most of the existing urban street lamp switch controls are triggered according to the template table timing, in normal weather often turn off the lights late in the morning, early in the evening. Waste of energy; In bad weather can not automatically turn on the lights early or late turn off the lights require staff to manually send off the switch light command from the command down to the street lights to be fully turned on about 1015 minutes. By this time, the sky has become very dark, seriously affect the normal life of citizens and even cause traffic accidents. Combined with the deficiency of the existing street lamp switch control, it is proposed that the prediction be applied to the street light switch control. Getting the future switching time in advance can not only provide certain decision support for street lamp control, but also save energy. This paper studies the characteristics of street lamp switching time and its influencing factors, focusing on the characteristics of light intensity. The numerical mapping of weather is presented. The environment of street lamp system is complex and the sample data is affected by various kinds of data. According to the characteristics of experimental data, the data preprocessing method used in this paper is introduced, and the commonly used prediction algorithms are compared. Because of the small sample characteristics of the experimental data, we choose the least squares support vector machine to model. According to the characteristics of the training data, the model structure and kernel function selection of the least squares support vector machine are given. Through the simulation analysis, it can be found that compared with BP neural network model, the least square support vector machine has better performance in small sample prediction. The better the prediction. Only based on sunset time, minimum temperature, maximum temperature, weather conditions, the average absolute error predicted is within 2 minutes 15 seconds. Lights are turned on compared to sunset. According to the time predicted in this paper, each lamp can save about 7 minutes 33 seconds lighting time per day.
【学位授予单位】:华中科技大学
【学位级别】:硕士
【学位授予年份】:2015
【分类号】:TU113.666
【相似文献】
相关期刊论文 前8条
1 洪培青;;γ射线料控仪开关时间分析及选择[J];计量与测试技术;2014年05期
2 袁佳斌;;达林顿管开关时间测试仪设计[J];民营科技;2007年05期
3 刘晓东;徐_";;正弦脉宽调制技术开关时间点精度分析[J];电力自动化设备;2010年12期
4 ;QK6型开关时间测试仪方案原理及其中“时间—电压转换电路”分析[J];电子测量技术;1977年06期
5 付胜杰;彭侠夫;陈国鹏;;采用虚拟开关时间的五相逆变器PWM策略[J];电机与控制学报;2011年07期
6 孙东明,董玮,刘彩霞,郭文滨,张歆东,纪平,徐宝琨,陈维友;MOEMS光开关响应时间的分析计算[J];电子学报;2003年11期
7 赵明;利用示波器测定电器触头的开关时间[J];河北轻化工学院学报;1997年04期
8 ;[J];;年期
相关重要报纸文章 前2条
1 记者 古春婷;路灯节能“精打细算”到每分钟[N];珠海特区报;2009年
2 县报道组 黄擎 记者 陈培华;海盐:精确控制路灯开关时间[N];浙江日报;2005年
相关硕士学位论文 前1条
1 廖明青;路灯开关时间的预测算法研究[D];华中科技大学;2015年
,本文编号:1492698
本文链接:https://www.wllwen.com/jingjilunwen/jianzhujingjilunwen/1492698.html