A method and device for determining whether a room is occupied or unoccupied.

Abstract

This invention relates to the field of smart home technology, and in particular to a method and device for determining the occupancy status of a room. This solution collects historical data related to occupancy determination, trains a decision tree model using the collected data according to predefined decision conditions, and uses the decision tree model to determine the adjustment method for the occupancy determination duration of different rooms in different time periods. Then, it uses a predefined occupancy determination duration adjustment logic to correct the output of the decision tree model, obtaining the occupancy determination duration for each room in each time period. Finally, based on the occupancy determination duration of each room in each time period, the occupancy status of each room in each time period is determined. This solution not only adjusts the occupancy determination duration but also improves the accuracy of the occupancy determination duration, thus making the judgment of the occupancy status of rooms in different scenarios more precise and closer to practical applications, thereby improving the user experience.

Description

Technical Field

[0001] This invention relates to the field of smart home technology, and in particular to a method and device for determining whether a room is occupied or unoccupied. Background Technology

[0002] Existing occupancy / vacancy algorithms are often accurate in determining occupancy because it can be done manually by the user or reported by security devices, thus updating the room status to occupied. However, for determining vacancy, the main logic of existing technologies is to update the status to vacant only after a period of time (such as 30 minutes or 15 minutes) without any manual operation records or reports from security devices.

[0003] However, analysis of occupancy data and consideration of the actual conditions in office building areas reveals that existing technologies that determine occupancy status by setting fixed time intervals for occupancy / vacancy transitions are prone to misjudgments. For example, rooms may become empty and then quickly become occupied again, especially in rooms with occupancy-related devices or scene-linked systems. This frequent status changes not only affect equipment lifespan but also impact user experience. Furthermore, during evening work or meetings, prolonged periods of minimal movement may not trigger infrared motion sensors, leading to a misjudgment of occupancy and subsequent light shutdown. This necessitates manual re-enabling of the lights back on, further impacting user experience. Both of these scenarios easily result in occupancy being mistaken for vacancy; therefore, a longer vacancy period is needed before officially vacating the room.

[0004] Similarly, there are situations where a room remains unoccupied for an extended period after it has been initially designated as unoccupied. For example, even after the last person leaves the office area at night, the system continues to determine whether a room is occupied or unoccupied based on a fixed time frame, resulting in a prolonged period of time being spent determining whether a room is unoccupied and wasting energy. Summary of the Invention

[0005] The purpose of this invention is to provide a method for determining whether a room is occupied or unoccupied, in order to solve the problem that the existing technology, which determines the occupancy status by setting a fixed occupancy / unoccupied transition time, is prone to misjudgment, thereby affecting the user experience; and also to provide a device for determining whether a room is occupied or unoccupied, for implementing the above method.

[0006] To solve the above-mentioned technical problems, the present invention provides a method for determining whether a room is occupied or unoccupied, comprising the following steps:

[0007] 1) Collect historical data related to occupancy and vacancy determination, and use the collected data to train a decision tree model according to the set decision conditions; the data related to occupancy and vacancy determination includes the duration of vacancy and the cumulative number of times occupancy is detected in each room during the collected time periods; the duration of vacancy refers to the time that a room remains vacant after its current state changes from occupied to vacant; the cumulative number of times occupancy is detected refers to the number of times occupancy or vacancy is detected in each time period; the decision conditions refer to the number of times the occupancy or vacancy status changes in each time period.

[0008] 2) Collect data related to occupancy and vacancy determination for different rooms at different time periods, input the collected data into the decision tree model, and obtain the adjustment method for the vacancy determination duration; the adjustment methods for the vacancy determination duration include increasing, keeping it unchanged, and decreasing;

[0009] 3) The output of the decision tree model is corrected by adjusting the set unattended judgment time, so as to obtain the unattended judgment time of each room in each time period; and the unattended judgment time of each room in each time period is used to determine the status of each room being occupied or unattended in each time period.

[0010] Beneficial effects: Compared with existing technologies that use fixed methods to determine whether a room is occupied or unoccupied, this invention, on the one hand, uses collected data related to the determination of occupancy to train a decision tree model according to set decision conditions, and uses the output of the decision tree model to determine the adjustment method for the unoccupied judgment duration, thereby improving the accuracy of the unoccupied judgment duration; on the other hand, by using the unoccupied judgment duration adjustment logic to correct the output of the decision tree model, this invention can further improve the accuracy of the unoccupied judgment duration, thus making the judgment on whether a room is occupied or unoccupied in different scenarios more accurate and more in line with actual applications, which is conducive to improving the user experience.

[0011] Furthermore, the logic for adjusting the unattended duration includes: if the adjustment method for the currently obtained unattended duration is unchanged, then it remains unchanged; if the adjustment method for the currently obtained unattended duration is an increase, and the adjustment method for the unattended duration in the same time period of the previous day was not an increase or was not consecutively an increase for the same time period of the previous several days, then it increases, otherwise it remains unchanged; if the adjustment method for the currently obtained unattended duration is a decrease, and the adjustment method for the unattended duration in the same time period of the previous day was not a decrease or was not consecutively a decrease for the same time period of the previous several days, then it decreases, otherwise it remains unchanged.

[0012] Beneficial effects: By adding restrictions on not making the same adjustment method for the unattended judgment duration in adjacent time periods, the present invention can avoid continuous adjustment causing the unattended judgment duration to be too long or too short, which is conducive to improving the accuracy of adjusting the unattended judgment duration.

[0013] Furthermore, the logic for adjusting the duration without human intervention is as follows: if the decision model output for the current time period is an increase, then the decision model output for the previous time period is adjusted to an increase; if the decision model output for the current time period is a decrease, and the decision model output for the adjacent previous time period remains unchanged, then the decision model output for the previous time period is adjusted to a decrease.

[0014] Beneficial effects: Since the decision tree model assumes that the time period in which the output result is located is the current processing time period, in reality, the output result of the current time period corresponds to the judgment result of the undetermined judgment duration of the previous time period. This invention improves the accuracy of adjusting the undetermined judgment duration of each time period by adding this constraint, thereby improving the rationality of the algorithm in adjusting the undetermined judgment duration when outputting results across time periods.

[0015] Furthermore, if the decision tree model outputs an increase, and the original undecided duration for the current time period is the first undecided duration, then it is adjusted to the second undecided duration; if it is the second undecided duration, it is adjusted to the third undecided duration; if it is the third undecided duration, it remains unchanged. If the decision tree model outputs no change, then the undecided duration for the current time period remains unchanged. If the decision tree model outputs a decrease, the original undecided duration for the current time period is the first undecided duration, remaining unchanged; if it is the second undecided duration, it is adjusted to the first undecided duration; if it is the third undecided duration, it is adjusted to the second undecided duration. The first undecided duration, the second undecided duration, and the third undecided duration increase sequentially.

[0016] Beneficial effects: This invention can increase or decrease the unoccupied judgment time within a corresponding time period based on the output results of the decision tree model, or keep the unoccupied judgment time within a corresponding time period unchanged, which can make the judgment result of whether the room is occupied or not more humanized, thereby improving the user experience.

[0017] Furthermore, the unoccupied time of each room in step 3) is periodically stored in the cloud at set time intervals. When the cloud does not receive data within the corresponding time interval, the previous unoccupied time is retained.

[0018] Beneficial effects: By periodically storing the unoccupied duration of each room in different time periods to the cloud according to the set time intervals, the unoccupied duration can be accurately obtained based on the records of changes in the occupancy status, which facilitates data statistics and maintenance.

[0019] Furthermore, the decision tree model is repeatedly trained using the collected data related to determining whether someone is present or not, in order to obtain an optimized decision tree model.

[0020] Beneficial effects: By repeatedly training the decision tree model with continuously updated data related to whether someone is present or not according to a set time period, the decision tree model becomes closer to reality, which helps to improve the accuracy of the algorithm in adjusting the duration of the no-person judgment.

[0021] The present invention also provides a room occupancy determination device, the device comprising a processor and a memory, wherein the memory is used to store executable instructions of the processor; the processor is configured to execute the room occupancy determination method described above by executing the executable instructions. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the method for determining whether a room is occupied or unoccupied according to the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical principles and practical applications of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0024] Example of a method for determining whether a room is occupied:

[0025] To address the aforementioned problems, this invention primarily uses a decision tree model to calculate the time required for room occupancy / vacancy transitions at various time periods throughout the day, based on historical records of room occupancy / vacancy status changes for each room.

[0026] The method for determining whether a room is occupied or unoccupied in this embodiment is generally divided into three stages: model training stage, cloud computing stage, and host application stage, such as... Figure 1 As shown. The implementation process of each stage is as follows:

[0027] 1. Model Training Phase

[0028] The purpose of the model training phase is to obtain a decision tree model, which includes three steps: data collection, data labeling, and training the decision tree model. This phase only needs to be run once to obtain a reasonably effective model. Subsequent phases can be re-executed periodically based on the algorithm's performance to obtain a more suitable model.

[0029] A decision tree is a tree-like structure where each node represents a test on an attribute, each branch represents a test output, and each leaf node represents a classification category. Decision trees are built from training data to classify unknown data.

[0030] By collecting and processing data related to whether the person is present or not, a decision tree model is trained according to the decision conditions to obtain the adjustment method for the unattended judgment time. The decision conditions refer to the number of times the person is present or not in each time period. The adjustment methods for the unattended judgment time include increasing, keeping it unchanged, and decreasing.

[0031] This algorithm primarily uses data on room occupancy changes, including specific times and occupancy values. It analyzes data characteristics based on these changes, labeling and processing the data by the duration of vacancy to obtain data related to occupancy determination. Specifically, it identifies: the number of times vacancy lasted more than 30 minutes, the number of times vacancy lasted less than 15 minutes, the number of times vacancy lasted between 15 and 30 minutes, and the cumulative number of occupancy counts. The duration of vacancy, measured in minutes, refers to the time from when a room becomes empty until it becomes occupied again. For example, in the second data point, if a room becomes empty at 7:41 and occupied at 8:58, the duration of vacancy is 77 minutes. The cumulative number of occupancy counts refers to the number of times a room is occupied within each time period during the data collection and occupancy determination process.

[0032] Then, the data related to determining whether a room is occupied or unoccupied is input into the decision tree model for training. The output results are -1, 0, and 1, and the corresponding unoccupied judgment duration is adjusted to 15 minutes, 30 minutes, and 60 minutes respectively based on the output results. Here, -1 means shortening the unoccupied judgment duration, i.e., 15 minutes without any operation to change the room status to unoccupied; 0 means keeping the current unoccupied judgment duration unchanged, i.e., 30 minutes without any operation to change the room status to unoccupied; and 1 means increasing the unoccupied judgment duration, i.e., 60 minutes without any operation to change the room status to unoccupied.

[0033] For example, Table 1 shows the specific times and status of whether a certain meeting room was occupied or not at different times on August 3, 2022.

[0034] Table 1

[0035] Is the meeting room occupied? 1 2022 / 8 / 3 7:15 Is the meeting room occupied? 0 2022 / 8 / 3 7:46 Is the meeting room occupied? 1 2022 / 8 / 3 8:02 Is the meeting room occupied? 0 2022 / 8 / 3 12:35 Is the meeting room occupied? 1 2022 / 8 / 3 13:46 Is the meeting room occupied? 0 2022 / 8 / 3 18:41 Is the meeting room occupied? 1 2022 / 8 / 3 19:20 Is the meeting room occupied? 0 2022 / 8 / 3 22:12

[0036] Based on the characteristics of the data, the data is labeled and processed according to the duration of no human presence, as follows:

[0037] 1. If the room status is changed to occupied within 15 minutes after being changed to unoccupied, the timer duration should be increased at this point.

[0038] 2. If the room status is not changed within 15 minutes after being changed to "unoccupied", it means that the timer duration remains unchanged at this point in time.

[0039] 3. If the room status is not changed within 30 minutes after being changed to "unoccupied", it means that the timer duration should be reduced at this point.

[0040] The historical occupancy data of different rooms at different time periods were statistically analyzed and processed. Here, a 7-day period was used as a cycle. Taking the occupancy data of a certain negotiation room from July 25th to July 31st as an example, the data after annotation and processing is shown in Table 2.

[0041] Table 2

[0042] 0 0 0 0 0 1 0 0 0 0 2 0 0 0 0 3 0 0 0 0 4 0 0 0 0 5 0 0 0 0 6 0 0 0 0 7 5 0 0 5 8 0 0 1 3 9 0 0 1 6 10 0 0 0 6 11 0 0 0 6 12 3 0 3 6 13 1 1 0 4 14 1 0 1 6 15 1 1 0 5 16 1 0 0 4 17 1 0 0 3 18 1 0 0 4 19 0 2 0 4 20 1 0 3 5 21 3 0 0 4 22 1 0 1 1 23 0 0 0 0

[0043] Among them, time refers to a specific moment, such as 1 refers to 1:00-2:00; unoccupied time refers to the duration of unoccupied time, that is, the time that the room status changes from occupied to unoccupied and remains unoccupied at the current time, and is counted according to the unoccupied time range of less than 15 minutes, 15 to 30 minutes, and more than 30 minutes; cumulative occupied days refers to the number of days in the past 7 days when the occupied time at this moment exceeded 0 minutes.

[0044] Taking the data in the 8th row of Table 2 as an example, let's explain the physical meaning of each data point in each row. This data is [7, 5, 0, 0, 5], which means that in the time period of 7:00-8:00, there were 5 times in the past 7 days when the room status changed from occupied to unoccupied and the unoccupied period lasted for more than 30 minutes. The number of times the unoccupied status changed from less than 15 minutes to 15 minutes to 30 minutes was 0. The number of days in the past 7 days when the room status was occupied in this time period was 5 days.

[0045] 2. Cloud computing stage

[0046] The purpose of the cloud computing phase is to obtain the unoccupied time duration for each host and room at each time period based on data and a decision tree model, and to store the results in the cloud. This includes three steps: collecting and processing data related to whether a room is occupied or not; inputting the collected data into the decision tree model to obtain the unoccupied time duration for each room at each time period; and storing the results in the cloud. The entire phase needs to be run once a day to obtain the unoccupied time duration for each room at each time period that day.

[0047] The main logic of cloud computing is as follows:

[0048] (1) Based on the changes in whether there are people or not and the decision tree, determine whether the duration of the judgment for no one is increased, unchanged, or decreased per hour.

[0049] Historical occupancy data was statistically analyzed and processed by room and time period, using a 7-day cycle. The number of times an empty room remained unoccupied for less than 15 minutes, 15 to 30 minutes, and more than 30 minutes was recorded, along with the number of days in the past 7 days when an empty room remained unoccupied for more than 0 minutes during that time period. Generally, the unoccupied time should be increased for rooms unoccupied for less than 15 minutes, and decreased for rooms unoccupied for more than 30 minutes.

[0050] For example, if there's a high probability that someone is in a meeting room after 9 AM, but the people in the meeting might be moving very little, making it difficult to detect via infrared sensors. Assuming the original unoccupied period is 30 minutes, then 9:30 AM would be considered unoccupied. If the meeting is detected by infrared sensors at 9:35 AM, and the room remains unoccupied for 5 minutes (within the 15-minute unoccupied period range), then the unoccupied period should be increased.

[0051] For example, if the statistics show that there are many instances of no one being online for more than 30 minutes after 6 PM, the time period for judging no one being online should be reduced.

[0052] The specific unattended time for decision-making is determined based on the algorithm results and the current unattended time. For example, if the unattended time is set to three levels: 15, 30, and 60 minutes, and room A's unattended time at 8:00 is 15 minutes, while at 9:00 it is 30 minutes, the algorithm's decision for that day will increase the time for 8:00 from 15 minutes to 30 minutes, and for 9:00 from 30 minutes to 60 minutes. As another implementation method, the set time and / or level can be adjusted based on the current unattended time of the room and the decision tree optimization results to meet personalized needs in different scenarios.

[0053] (2) Correct the current result based on the algorithm results of the previous week.

[0054] Because the calculation results are based on data from the past week, there might be instances where the unattended time has already been increased or decreased, yet the algorithm still determines whether the unattended time should be increased or decreased. Therefore, a restriction needs to be added: the unattended time should not be increased or decreased consecutively. If the algorithm determines that the time remains unchanged, then the result remains unchanged; if the determination is to increase, and the previous day's result was not an increase, or the result for the past 7 consecutive days has been an increase, then the time is increased; otherwise, it remains unchanged; if the determination is to decrease, and the previous day's result was not a decrease, or the result for the past 7 consecutive days has been a decrease, then the time is decreased; otherwise, it remains unchanged.

[0055] If the duration is increased on January 1st, the increase will be made on January 1st. On the second day, January 2nd, the algorithm will still judge an increase. Because January 1st saw an increase, based on the principle of non-continuous increases or decreases, January 2nd will not see an increase, remaining unchanged. If the algorithm's judgment on January 2nd is unchanged, January 2nd will remain unchanged. If the algorithm's judgment on January 2nd is a decrease, no one will judge the duration to decrease on January 2nd. If January 1st shows no change, it does not affect the judgment on January 2nd. If January 1st shows a decrease, January 2nd will also show a decrease; based on the principle of non-continuous increases or decreases, the result on January 2nd will be changed to no change. If January 2nd shows an increase, the duration will be increased; if January 2nd shows no change, the duration will remain unchanged.

[0056] For example, using occupancy data from July 1st to July 7th to determine the unoccupied status duration on July 8th, if the decision tree model outputs 1, then the unoccupied status duration needs to be increased, i.e., 60 minutes of inactivity to change the room status to unoccupied. Then, using occupancy data from July 2nd to July 8th to determine the unoccupied status duration on July 9th, if the decision tree model outputs 1 again, since the occupancy data from July 8th includes data with increased unoccupied status duration, continuously increasing the unoccupied status duration in adjacent time periods might affect the accuracy of the decision tree model's output. Therefore, a constraint is added to prevent consecutive increases or decreases in the unoccupied status duration to ensure the accuracy of the decision tree model's output.

[0057] (3) Correct the results of the previous hour based on the current hour's decision tree results.

[0058] If the decision model output for the current time period is an increase, then the decision model output for the previous time period will be adjusted to an increase; if the decision model output for the current time period is a decrease, and the decision model output for the adjacent previous time period remains unchanged, then the decision model output for the previous time period will be adjusted to a decrease.

[0059] For cases where the algorithm obtains results from data in the previous time period and these results occur in the next time period, this embodiment adds the following constraints: if the current hour is increased, the previous hour should also be increased; if the current hour is decreased, the previous hour remains unchanged, and the previous hour is changed to decrease. This improves the rationality of adjusting the duration for unattended judgments when the algorithm outputs results across time periods.

[0060] Note that in step (2), the time unit for correcting the current result based on the previous week's algorithm result is days, that is, the time unit for non-continuous increase or decrease judgment is days; in step (3), the time unit for correcting the previous hour's result based on the current hour's decision tree result is hours, that is, the time unit for continuous increase or decrease of time period is hours.

[0061] The output of the decision tree model refers to the judgment result of increase, unchanged, or decrease obtained after inputting historical data on whether the user is present or not. The judgment result of the algorithm refers to the judgment result of increase, unchanged, or decrease obtained by combining the decision tree model and logical judgment.

[0062] The following examples illustrate the applicable scenarios for the above constraints, specifically the situation where an increase should be applied not only in the current hour but also in the previous hour. For instance, suppose a room is occupied from 7:00 to 9:00, but the person's movements are subtle, potentially causing infrared sensors to miss them. If 7:50 is the last manual operation or infrared sensor record during the 7:00-8:00 period, and the unattended period is 30 minutes, the room would become unattended at 8:20. However, if the room is actually occupied and a manual operation record is made at 8:25, causing the room to become occupied again, meaning it became occupied again within 15 minutes of becoming unattended during the 8:00-9:00 period, the decision tree model might determine that an increase in the unattended period is needed at 8:00. However, this determination is actually based on the last manual operation or infrared sensor record within 7:00, therefore the unattended period should be increased at 7:00.

[0063] This scenario involves a decrease in the current hour, no change in the previous hour, and a decrease in the previous hour. For example, if the decision tree starts at 7:50 and the undecided time is 30 minutes, then at 8:20 it will be determined that there is no action. If no one takes action for a long period afterward, the undecided time for the 8 o'clock period will be reduced from 30 minutes to 15 minutes. However, this result was actually obtained starting at 7 o'clock, so the undecided time for 7 o'clock should be reduced from 30 minutes to 1 minute. This assumes that the decision tree result at 7 o'clock remains unchanged.

[0064] 3. Host-side application stage

[0065] The purpose of the host application phase is to set up an algorithm based on cloud data to determine the occupancy period for each room at each time period. This requires receiving cloud data daily for configuration. Specifically, if no data is received, the occupancy period will remain as previously set.

[0066] Example of a room occupancy / vacancy determination device:

[0067] The room occupancy determination device of this embodiment includes a processor and a memory. The processor executes a computer program stored in the memory to enable the present invention to implement the method of the above-described method embodiment. That is, the method in the above method embodiment should be understood as a process based on the above-described room occupancy determination method that can be implemented by computer program instructions. These computer programs can be provided to issue instructions to the processor, causing the processor to execute these instructions to generate the functions specified for implementing the above-described method flow.

[0068] In this embodiment, the processor refers to a processing device such as a microprocessor (MCU) or a programmable logic device (FPGA); the memory refers to a physical device used to store information, which typically involves digitizing the information and then storing it using media that utilizes electrical, magnetic, or optical methods. Examples include: various types of memory that store information using electrical energy, such as RAM and ROM; various types of memory that store information using magnetic energy, such as hard disks, floppy disks, magnetic tapes, magnetic core memory, bubble memory, and USB flash drives; and various types of memory that store information using optical methods, such as CDs or DVDs. Of course, there are other types of memory, such as quantum memories and graphene memories.

[0069] The device consisting of the aforementioned memory, processor, and computer program is implemented in a computer by the processor executing the corresponding program instructions. The processor can run various operating systems, such as Windows, Linux, Android, and iOS.

[0070] As an alternative implementation, the device may also include a display for showing the classification results for staff reference.

Claims

1. A method for determining whether a room is occupied or unoccupied, characterized in that, Includes the following steps: 1) Collect historical data related to occupancy and vacancy determination, and train a decision tree model using the collected data according to the set decision conditions; the data related to occupancy and vacancy determination includes the duration of vacancy and the cumulative number of occupancy times for each room in each time period collected; the duration of vacancy refers to the time that a room remains vacant after its current state changes from occupied to vacant; the cumulative number of occupancy times refers to the number of times occupancy and vacancy status are collected in each time period; the decision conditions refer to the number of times the occupancy and vacancy status changes in each time period. 2) Collect data related to the determination of whether a room is occupied or not at different time periods in different rooms, and input the collected data into the decision tree model to obtain the adjustment method for the unoccupied judgment time; the adjustment method for the unoccupied judgment time includes increasing, keeping it unchanged, and decreasing; 3) The output of the decision tree model is corrected by adjusting the set unattended judgment time, so as to obtain the unattended judgment time of each room in each time period; and the unattended judgment time of each room in each time period is used to determine the status of each room being occupied or unattended in each time period.

2. The method for determining whether a room is occupied or unoccupied according to claim 1, characterized in that, The logic for adjusting the unattended judgment duration includes: if the adjustment method for the currently obtained unattended judgment duration is unchanged, then it remains unchanged; if the adjustment method for the currently obtained unattended judgment duration is an increase, and the adjustment method for the unattended judgment duration in the same time period of the previous day was not an increase or was not consecutively an increase for the same time period of the previous several days, then it increases, otherwise it remains unchanged; if the adjustment method for the currently obtained unattended judgment duration is a decrease, and the adjustment method for the unattended judgment duration in the same time period of the previous day was not a decrease or was not consecutively a decrease for the same time period of the previous several days, then it decreases, otherwise it remains unchanged.

3. The method for determining whether a room is occupied or unoccupied according to claim 1 or 2, characterized in that, The logic for adjusting the unmanned judgment duration is as follows: if the decision model output result for the current time period is an increase, then the decision model output result for the previous time period is adjusted to an increase; if the decision model output result for the current time period is a decrease, and the decision model output result for the adjacent previous time period is unchanged, then the decision model output result for the previous time period is adjusted to a decrease.

4. The method for determining whether a room is occupied or unoccupied according to claim 3, characterized in that, If the decision tree model outputs an increase, and the original undecided duration for the current time period is the first undecided duration, then it is adjusted to the second undecided duration; if it is the second undecided duration, it is adjusted to the third undecided duration; if it is the third undecided duration, it remains unchanged. If the decision tree model outputs no change, then the undecided duration for the current time period remains unchanged. If the decision tree model outputs a decrease, the original undecided duration for the current time period is the first undecided duration, remaining unchanged; if it is the second undecided duration, it is adjusted to the first undecided duration; if it is the third undecided duration, it is adjusted to the second undecided duration. The first undecided duration, the second undecided duration, and the third undecided duration increase sequentially.

5. The method for determining whether a room is occupied or unoccupied according to claim 4, characterized in that, The unoccupied time of each room in each time period described in step 3) is periodically stored in the cloud at set time intervals. When the cloud does not receive data within the corresponding time interval, the previous unoccupied time is retained.

6. The method for determining whether a room is occupied or unoccupied according to claim 5, characterized in that, Using the collected data related to determining whether someone is present or not, the decision tree model is repeatedly trained to obtain an optimized decision tree model.

7. A device for determining whether a room is occupied or unoccupied, characterized in that, The device includes a processor and a memory, wherein the memory is used to store executable instructions of the processor; the processor is configured to execute the room occupancy determination method according to any one of claims 1-6 by executing the executable instructions.

Images