| In the evening,most classrooms without courses and events will be used by students for self-study.In general,classrooms used for evening self-study account for a relatively high proportion of all classrooms,but the average seat occupancy rate of each evening self-study classroom is low,which is likely to cause electricity waste.Therefore,it is necessary to schedule self-study classrooms in the evenings to improve the utilization of seat resources.For the evening self-study,students pay more attention to whether the space comfort can be satisfied,that is,whether there is enough personal space.Currently,there is a lack of classroom energy-saving scheduling research that considers the evening study space required by students.By collecting and analyzing data related to evening self-study,this work simulates the behavior of students in micro seat selection,and proposes a new scheduling scheme,which can save electricity energy and ensure that most students have enough space for self-study.Main contributions are as follows.(1)Simulation research on students’ micro seat selection behavior: When choosing a seat,students will not only consider their own preferences,but also consider the seating situation of other students,and maintain a comfortable seating distance to meet the comfort of the self-study space.The repulsive force in the Social Force Model(SFM)can reflect the distance that different students expect to maintain.From the four aspects of personal preference,gender differences,peer relationship and obstacle factors,SFM was improved to obtain a Modified Social Force Model(MSFM).Through the simulation of some students choosing seats in the classroom,combining the students’ preferences,gender,and peer relationship,this study compares the seat distance between the students and verifyies that the improved method is practical.Then,we discuss the influence of peer ratio,male-to-female ratio,and preference ratio on the number of people in a classroom under the condition of meeting the self-study space required by students in the classroom.The experimental results show that the change in the ratio of peer relationships has the greatest impact,and the change in the preference ratio has the least impact.In actual scenarios,the minimum,average,and maximum capacity of each type of classroom is used to schedule evening self-study classrooms,and compare the student comfort rate and classroom power consumption under three fixed-value scheduling strategies.Under the condition that the comfort rate is greater than or equal to 93%,the average scheduling strategy outperforms others in saving energy.(2)Research on the scheduling strategy of evening self-study classrooms: Based on the simulation model of students’ micro seat selection,this study consideres the space for evening study required by students and optimizes the scheduling of evening study classrooms.According to the data collection results,the arrival and departure of students are simulated,and popular seats are defined,which can meet the needs of most students for self-study space.This work uses the remaining number of popular seats in power-on classroom as a threshold to determine whether an additional classroom needs to be electrified.Under different flow of people and size of classrooms,four heuristic algorithms such as Improved Cuckoo Search(ICS)are used to search for this threshold.Each scheduling strategy was verified 20 times,the average value and standard deviation of the power consumption were calculated,and the stability of each scheduling strategy was compared.In addition,we combine the situation of students using and leaving their seats in the evening self-study,and use Long Short-Term Memory(LSTM)to predict short-term student flow and assist in making exact decisions on whether to power on an additional classroom or not.Experimental results show that compared with a fixed number of people in the evening self-study classroom,the ICS+LSTM hybrid scheduling strategy can further save energy while ensuring that the comfort rate of the student’s self-study space is higher than 98%. |