Research On Intelligent Lighting System Of College Classroom Based On Deep Learning

In recent years,with the improvement of big data and computing power,artificial intelligence technology has been well developed and applied.In 2018,the Ministry of Education issued the education informatization 2.0 action plan,which emphasized that colleges and universities should use emerging technologies such as artificial intelligence,big data and cloud computing to help colleges and universities in teaching,management and service.In the future,the application of these high and new technologies in college classroom lighting system will have far-reaching and important significance for the power saving,energy saving and intelligent reform of college classroom lighting.With the rapid development of our economy and society,the problem of energy shortage is becoming more and more serious.Reducing unnecessary energy consumption is an effective way to solve the problem of energy shortage.As a densely populated area,colleges and universities are the major consumers of energy,especially electric energy,in which lighting power accounts for a large proportion.Due to the weak awareness of energy conservation,the phenomenon of "long light" can be seen everywhere in Colleges and universities.The operation and maintenance management of lighting facilities mostly adopts the way of manual inspection,which is high cost and low efficiency,resulting in a large number of power resources waste.This topic mainly aims at how to improve and improve the energy-saving and intelligent level of college classroom lighting,designs a kind of college classroom intelligent lighting system based on in-depth learning,solves the problems of College lighting power waste,high operation cost of artificial maintenance in management,and great difficulty in management,mainly from the following aspects:(1)Overall scheme design of the system.The overall architecture of intelligent lighting system is discussed,and the software algorithm module,control module,upper computer monitoring module and communication mode are designed and introduced.(2)Design and train human detection model in classroom.After introducing the related theoretical knowledge of deep learning,according to the application scenario,fast RCNN neural network is selected as the classroom human detection model,and then the model is adapted to use crowd The human open data set pre trained the model and manually labeled the classroom crowd data set.Finally,a good classroom human detection model was obtained by using the self-made data set for deep migration learning of the pre training model.(3)Design and training the human body area positioning model.According to the characteristics of computer vision,a scheme of using deep learning classification network to locate the human body area in the classroom is designed.In the experiment,a variety of classification networks(AlexNet,VGG16,ResNet18,ResNet50)are used to train the model.After the evaluation of the trained network model,ResNet18 is selected as the best classification network to locate the human body area in the classroom to complete the human body map like the mapping of coordinates to lighting areas.(4)Monitoring platform and hardware module building and system testing.Firstly,according to the functional requirements of the upper computer monitoring module,the upper computer monitoring platform with simple operation and friendly interface is realized.Then the hardware of the control module(microcontroller,Zigbee module,human infrared detection module and illumination detection module)is introduced and selected,and the hardware circuit logic is designed.Finally,a single classroom hardware model is built,and the intelligent lighting system is simulated and tested.