摘要
针对路灯的夜间智能化水平低,耗电量大的问题,提出了一种基于深度学习与模糊控制的智能调光路灯控制系统。使用引入新碰撞机制的飞蛾扑火(MFO)算法对门控循环单元(GRU)的超参数进行优化,经过超参数优化后的GRU模型对当前环境的能见度进行预测。相较于其他神经网络模型,在减少模型大小的同时提高了预测精度。控制系统将传感器收集到的环境信息作为模型输入参数,对能见度进行实时预测。使用一种新的路灯调光方法,将预测能见度与当前亮度值作为模糊控制系统输入参数,输出调光值对灯光亮度进行调节。实验结果表明,该系统可实现根据环境自适应调节亮度,能够有效减少电量消耗,具有较强的现实意义。
Aiming at the problems of low intelligence level and high power consumption of street lights at night, an intelligent dimming street light control system based on deep learning and fuzzy control is proposed. The hyperparameters of the gate recurrent unit(GRU) are optimized using the moth-flame optimization(MFO) algorithm introducing a new collision mechanism. The optimized GRU model is used to predict the brightness of the current environment. Compared with other neural network models, the prediction accuracy is improved while the model size is reduced. The control system uses the environmental information collected by the sensors as model input parameters to predict the visibility in real time. A new method of street lamp dimming is used. The predicted visibility and the current brightness are used as the input parameters of the fuzzy control system, and the output dimming value is used to adjust the light brightness.The experimental results show that the system can adjust the brightness adaptively according to the environment, and can effectively reduce the power consumption, which has strong practical significance.
作者
张基延
秦会斌
Zhang Jiyan;Qin Huibin(Institute of New Electronic Devices and Application,Hangzhou Dianzi University,Hangzhou 310018,China)
出处
《国外电子测量技术》
北大核心
2022年第10期148-154,共7页
Foreign Electronic Measurement Technology
基金
浙江省科技计划项目(2017C01027)资助。