A Comprehensive Sensing Method for Environmental Parameters in a Smart Campus Based on AIoT

Abstract

This application relates to the technical field of video engines, and specifically relates to a method for comprehensively perceiving environmental parameters of a smart campus based on AIoT. The method includes: obtaining the network bandwidth, alarm identifier, position coordinates, and historical alarm vector of each monitoring probe in the smart campus system at each moment, and obtaining the actual cache length of each monitoring probe in real time through the video engine; obtaining the alarm correlation degree between each monitoring probe and the rest of the monitoring probes; calculating the query urgency of each monitoring probe at the current moment; determining the cache urgency of each monitoring probe at the current moment; determining the video cache length of each monitoring probe at the current moment; comparing the actual cache length and the video cache length, and adjusting the video transmission bit rate of different monitoring probes in real time. This application ensures that there is sufficient video cache data when different monitoring probes are called to provide smooth videos for management personnel.

Description

Technical Field

[0001] This application relates to the technical field of video engines, and particularly to a method for comprehensively perceiving environmental parameters of a smart campus based on AIoT. Background Art

[0002] In a smart campus system, a large number of video surveillance devices are deployed to collect campus surveillance data to achieve comprehensive perception of the campus environment. These data are not only used for monitoring, but also can identify potential dangerous situations through AI technology and issue alarms in a timely manner to assist campus security management. After an AI alarm is triggered, managers need to review the alarm to confirm whether it is a false alarm and formulate corresponding countermeasures accordingly. In this process, it is crucial to provide smooth video services for managers so that they can view surveillance information in real time, make accurate judgments and responses. Due to the huge amount of video data involved in the smart campus system, video information is usually stored in the local storage of video surveillance devices and is called by the video engine of the smart campus system to provide online video services. This service needs to adjust the bitrate adaptively according to the network status to ensure the smoothness of video playback.

[0003] However, when the network condition is poor and managers need to switch multiple monitoring probes to judge the alarm situation in an area, due to the lack of an effective video caching mechanism, it is easy for managers to experience video stuttering when switching monitoring probes, resulting in poor robustness of the video engine. Summary of the Invention

[0004] To solve the above technical problems, a method for comprehensively perceiving environmental parameters of a smart campus based on AIoT is provided to solve the existing problems.

[0005] The solution of this application to solve the technical problem is to provide a method for comprehensively perceiving environmental parameters of a smart campus based on AIoT, including the following steps:

[0006] Obtain the network bandwidth of each monitoring probe in the smart campus system at each moment, as well as the position coordinates of each monitoring probe, and obtain the actual cache length of each monitoring probe in real time through the video engine;

[0007] Perform anomaly identification on the video information of each monitoring probe at each moment, judge the alarm state of each monitoring probe at each moment, and obtain the alarm identifier of each monitoring probe at each moment; based on the alarm situation of each monitoring probe in the historical period, obtain the historical alarm vector of each monitoring probe;

[0008] Obtain the alarm correlation degree between each monitoring probe and the rest of the monitoring probes through the spatial distance between each monitoring probe and the rest of the monitoring probes and the similarity degree of the historical alarm vectors;

[0009] Using the alarm correlation degree, weight the average level of alarm identifiers of each monitoring probe at different times before the current time, and calculate the query urgency of each monitoring probe at the current time;

[0010] Analyze the average level of network bandwidth of each monitoring probe at different times before the current time, and combine the query urgency to determine the cache urgency of each monitoring probe at the current time;

[0011] Based on the cache urgency, determine the video cache length of each monitoring probe at the current time; compare the actual cache length with the video cache length, and adjust the video transmission bit rate of different monitoring probes in real time.

[0012] Preferably, determining the alarm state of each monitoring probe at each time and obtaining the alarm identifier of each monitoring probe includes:

[0013] If the video information of each monitoring probe is abnormal at each time, the monitoring probe immediately triggers an alarm. If each monitoring probe is in an alarm state at each time, the alarm identifier is 1; otherwise, the alarm identifier is 0.

[0014] Preferably, the method for obtaining the historical alarm vector is:

[0015] Count the number of times the alarm identifier is 1 for each monitoring probe in each time period, which is recorded as the alarm times; form the historical alarm vectors of each monitoring probe with the alarm times of each monitoring probe in multiple time periods.

[0016] Preferably, the spatial distance is the Euclidean distance between the position coordinates of each monitoring probe and the position coordinates of the other monitoring probes.

[0017] Preferably, the similarity degree is the reciprocal of the Euclidean distance between the historical alarm vectors of each monitoring probe and the historical alarm vectors of the other monitoring probes.

[0018] Preferably, the alarm correlation degree is the ratio of the similarity degree to the spatial distance.

[0019] Preferably, calculating the query urgency of each monitoring probe at the current time includes:

[0020] Normalize the alarm correlation degrees of each monitoring probe with all the other monitoring probes, and use the normalization result as the alarm influence weight of each monitoring probe with the other monitoring probes;

[0021] Take the mean value of the alarm identifiers of each monitoring probe at multiple times before the current time as the alarm value of each monitoring probe at the current time;

[0022] Based on the alarm influence weights of each monitoring probe and the other monitoring probes, the alarm values of the other monitoring probes at the current moment are weighted and summed, and the sum of the weighted sum result and the alarm value of each monitoring probe at the current moment is used as the query urgency of each monitoring probe at the current moment.

[0023] Preferably, the determining the cache urgency of each monitoring probe at the current moment includes:

[0024] Taking the average value of the network bandwidth of each monitoring probe at multiple moments before the current moment as the bandwidth eigenvalue of each monitoring probe at the current moment;

[0025] Taking the ratio of the query urgency to the bandwidth eigenvalue as the cache urgency of each monitoring probe at the current moment.

[0026] Preferably, the determining the video cache length of each monitoring probe at the current moment includes:

[0027] Normalizing the cache urgency of all monitoring probes at the current moment, and taking the normalized value as the cache weight of each monitoring probe at the current moment;

[0028] Allocating a video cache space with a preset duration for each monitoring probe; taking the product of the cache weight and the preset duration as the video cache length of each monitoring probe at the current moment.

[0029] Preferably, the real-time adjustment of the video transmission bit rate of different monitoring probes includes:

[0030] For the video information corresponding to the monitoring probe whose actual cache length is less than the video cache length, compressing and transmitting it at a preset minimum transmission bit rate, and vice versa, compressing and transmitting it at a preset maximum transmission bit rate.

[0031] This application has at least the following beneficial effects:

[0032] By calculating the alarm correlation degree, the beneficial effect of the present application is that it takes into account the positional distribution of different monitoring probes and the similarity of alarm information occurring in the historical period, reflects whether the monitoring areas between different monitoring probes are directly connected, and shows the possibility of continuous invocation of the monitoring probes with connected monitoring areas; calculating the query urgency of each monitoring probe at the current moment, the beneficial effect of which is that it takes into account the situation where each monitoring probe may be called for video after an alarm occurs, and filters out the monitoring probes that are easily called and queried due to the alarm, so that the subsequent video engine can preferentially perform video caching on these monitoring probes; determining the caching urgency of each monitoring probe at the current moment, determining the video caching length of each monitoring probe at the current moment; comparing the actual caching length and the video caching length, and adjusting the video transmission rate of different monitoring probes in real time, the beneficial effect of which is that it takes into account the network conditions of the monitoring probes, and then allocates more video caching space to the monitoring probes that are more likely to be called and have a worse network status, ensuring that there is sufficient video caching data for the management personnel to provide smooth video when different monitoring probes are called, and adjusting the video transmission rate in real time according to the network conditions, avoiding the video stuttering phenomenon caused by poor network conditions and uneven distribution of video caching resources in the traditional video engine. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The following further describes in detail a method for comprehensively perceiving environmental parameters of an AIoT-based smart campus according to the present application with reference to the accompanying drawings.

[0034] Figure 1 It is a flowchart of the steps of a method for comprehensively perceiving environmental parameters of an AIoT-based smart campus provided by an embodiment of the present application;

[0035] Figure 2 It is a flowchart of the steps of a method for obtaining the video caching length provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] In order to make the objectives, technical solutions and advantages of the present application clearer, the following further describes in detail a method for comprehensively perceiving environmental parameters of an AIoT-based smart campus proposed by the present application with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present application belongs.

[0038] Please refer to Figure 1 , which shows a flowchart of the steps of a method for comprehensively perceiving environmental parameters of an AIoT-based smart campus provided by an embodiment of the present application. The method includes the following steps:

[0039] Step 1: Obtain the network bandwidth, alarm flag of each monitoring probe in the intelligent campus system at each moment, as well as the position coordinates and historical alarm vector of each monitoring probe. The actual cache length of each monitoring probe is obtained in real time through the video engine.

[0040] In the intelligent campus system, a comprehensive perception of the campus environment is achieved by deploying multiple monitoring probes. Among them, each monitoring probe stores real-time monitoring video information in an IOT device. Each monitoring probe corresponds to an IOT device. The IOT device, that is, the local system, is responsible for storing the video information of its corresponding monitoring probe, identifying the monitoring video information in the IOT device, and alarming in case of dangerous situations. Secondly, the geographical location information of each monitoring probe and the monitoring video information with alarms occurring at historical moments are stored in the intelligent campus system for the management staff to view in real time.

[0041] After the intelligent campus system issues an alarm, the management staff can view the monitoring information by continuously switching the monitoring probes. However, due to poor network conditions, the video caching mechanism will not be smooth. As a result, when the management staff switches the monitoring probes to view the monitoring videos, video stuttering is likely to occur.

[0042] Based on the above analysis, the network bandwidth of N monitoring probes at each moment is obtained in real time through a network traffic monitoring tool, and the video information monitored by each monitoring probe in real time is identified. Once an abnormal situation is detected, the monitoring probe immediately triggers an alarm and generates and stores alarm information. If each monitoring probe is in an alarm state at each moment, the alarm flag is 1; if each monitoring probe is in a non-alarm state at each moment, the alarm flag is 0.

[0043] It should be noted that the method for abnormal identification of video information is a well-known technology and will not be elaborated here.

[0044] In this embodiment, data is recorded every 1 s. As other implementation manners, the implementer can set it according to the actual situation. Secondly, the network bandwidth represents the network state and reflects the data transmission efficiency and capacity of the network. Secondly, there are 30 monitoring probes in the intelligent campus system. As other implementation manners, the implementer can set it according to the actual situation.

[0045] Secondly, obtain the position coordinates of each monitoring probe. Among them, the position coordinates are a coordinate point in a three-dimensional space, reflecting the coordinate position of the monitoring probe in the intelligent campus system.

[0046] Furthermore, according to the stored alarm information, count the number of times when the alarm flag is 1 for each monitoring probe in each time period, and record it as the alarm times; the alarm times of each monitoring probe in multiple time periods are combined to form the historical alarm vector of each monitoring probe.

[0047] In this embodiment, the number of alarm flags with a value of 1 for each monitoring probe within every 1 minute is counted, that is, how many seconds are in the alarm state within every 1 minute. Secondly, the length of the historical alarm vector is 14,400, corresponding to the alarm information of the past 10 days. When the alarm information is less than 10 days, the length of the corresponding historical alarm vector is less than 14,400. At this time, the historical alarm vector is filled with element value 0.

[0048] The actual cache length of each monitoring probe at the current moment is obtained through the API (Application Programming Interface) provided by the video engine in the smart campus system.

[0049] Thus, the network bandwidth, alarm flag of each monitoring probe at each moment, as well as the position coordinates and historical alarm vector of each monitoring probe are obtained, and the actual cache length of each monitoring probe at the current moment is obtained.

[0050] Step 2, the alarm correlation degree between each monitoring probe and the rest of the monitoring probes is obtained through the spatial distance between each monitoring probe and the rest of the monitoring probes and the similarity degree of the historical alarm vectors.

[0051] During the monitoring process of the entire campus, there are usually obvious correlation characteristics in the alarm situations of the monitoring probes, mainly because the monitoring areas between the monitoring probes are directly connected, and even there is monitoring overlap. When the campus management personnel use the video engine of the smart campus system to verify alarms, usually in addition to sending a video acquisition and viewing request for the monitoring video of the monitoring probe where the alarm occurs, they will also send a video acquisition request to other alarmed monitoring probes with connected spatial areas, so as to judge the alarm situation from multiple angles and form a comprehensive perception of the campus environment.

[0052] Therefore, the more obvious the correlation characteristics of the alarm situations of two monitoring probes are, the more likely they are the monitoring probes that the management personnel will continuously request. When performing online video caching, when the management personnel retrieve the video of one of the monitoring probes, the video of the other monitoring probe can be cached synchronously to reduce the lag phenomenon during video switching.

[0053] Based on the above analysis, the alarm correlation degree is determined through the position distribution and alarm flag situations between different monitoring probes to reflect the alarm correlation situation between the monitoring probes, specifically including:

[0054] Calculate the spatial distance between the position coordinates of each monitoring probe and the rest of the monitoring probes.

[0055] In this embodiment, the Euclidean distance between the position coordinates of each monitoring probe and those of the other monitoring probes is calculated. The calculation of the Euclidean distance is a well-known technique and will not be elaborated here.

[0056] Among them, the smaller the spatial distance, the closer the monitoring areas of the two monitoring probes, and the more likely they are the two monitoring probes that managers will continuously call. The spatial distance only considers the positional relationship of the monitoring probes in the three-dimensional space and does not consider whether the monitored areas of the monitoring probes are connected. If the monitored areas of two monitoring probes are very close in space but not directly connected and are blocked by buildings such as walls, then these two monitoring probes may not be the two probes that managers continuously call. Therefore, it is necessary to analyze the connectivity of the monitored areas of the two monitoring probes based on historical alarm information.

[0057] Calculate the similarity degree between the historical alarm vectors of each monitoring probe and those of the other monitoring probes;

[0058] In this embodiment, the reciprocal of the Euclidean distance between the historical alarm vectors of each monitoring probe and those of the other monitoring probes is calculated. As other implementation manners, implementers can adopt other methods of the prior art, such as the reciprocal of the DTW distance, cosine similarity, etc. This embodiment does not make special restrictions on this.

[0059] Take the ratio of the similarity degree to the spatial distance as the alarm correlation degree between each monitoring probe and the other monitoring probes;

[0060] It should be noted that the greater the similarity degree, the more similar the historical alarm information of the two monitoring probes, and the more likely the monitored areas of the two monitoring probes are directly connected; the greater the alarm correlation degree, the closer the spatial distance between the two monitoring probes, and the more frequently the two monitoring probes issue alarms simultaneously during the historical monitoring period. The monitored areas of the two monitoring probes are more directly connected and closer areas, and managers are more likely to continuously call the two monitoring probes for alarm review.

[0061] Thus, the alarm correlation degree between each monitoring probe and the other monitoring probes is obtained.

[0062] Step 3: Use the alarm correlation degree to weight the average level of the alarm flags of each monitoring probe at different times before the current time, and calculate the query urgency of each monitoring probe at the current time.

[0063] Furthermore, when managers conduct alarm review, they usually prefer to select the monitoring probe at the alarm location for video call first. Therefore, when the video engine performs video caching, it should first select the monitoring probe at the alarm location and the surrounding monitoring probes for video caching to reduce the lag situation when managers switch monitoring probes.

[0064] Based on the above analysis, calculate the query urgency, which reflects the degree to which a monitoring probe may be called for video, so that the video engine preferentially caches the monitoring videos of the monitoring probes, and assigns a higher cache length to the monitoring probes at the alarm location and the surrounding locations. The video engine will preferentially cache the monitoring videos of these monitoring probes.

[0065] Normalize the alarm correlation degree of each monitoring probe with all the other monitoring probes, and use the normalization result as the alarm influence weight of each monitoring probe with the other monitoring probes.

[0066] In this embodiment, the normalization method adopted is: calculate the sum value of the alarm correlation degrees of each monitoring probe with all the other monitoring probes, denoted as the first sum value, and use the ratio of the alarm correlation degree to the first sum value as the alarm influence weight of each monitoring probe with the other monitoring probes.

[0067] It should be noted that the greater the alarm influence weight, the more likely it is that each monitoring probe will become the next monitoring probe to be called due to the alarms of other monitoring probes.

[0068] Take the mean value of the alarm flags of each monitoring probe at multiple moments before the current moment as the alarm value of each monitoring probe at the current moment.

[0069] In this embodiment, take the mean value of the alarm flags of each monitoring probe at all moments within 10 minutes before the current moment as the alarm value of each monitoring probe at the current moment. As other implementation manners, the implementer can set it according to the actual situation.

[0070] Based on the alarm influence weight of each monitoring probe with the other monitoring probes, perform a weighted sum of the alarm values of the other monitoring probes at the current moment, and use the sum of the weighted sum result and the alarm value of each monitoring probe at the current moment as the query urgency of each monitoring probe at the current moment.

[0071] It should be noted that the method of weighted sum is a well-known technology and will not be elaborated here.

[0072] In this embodiment, taking the nth monitoring probe at the current moment t as an example, the calculation formula for the query urgency of the nth monitoring probe at the current moment t is:

[0073]

[0074] Where, U n,t is the query urgency of the nth monitoring probe at the current moment t, Av n,t is the alarm value of the nth monitoring probe at the current moment t, α n,ris the alarm impact weight of the nth monitoring probe and the rth monitoring probe, Av r,t is the alarm value of the rth monitoring probe at the current moment t, and N is the number of all monitoring probes.

[0075] It should be noted that the larger the alarm value of the nth monitoring probe, the more serious the alarm situation of the nth monitoring probe itself in the past period of time, the more likely the monitoring video of the nth monitoring probe will be called; the larger the associated alarm value of the nth monitoring probe, the greater the alarm impact weight, the more serious the alarm situation of the monitoring probes around it that monitor the same area, and the more likely the nth monitoring probe will be called for video; finally, the greater the query urgency, the more likely the nth monitoring probe will be called for video, and the video engine should give priority to video caching of the nth monitoring probe.

[0076] It should be noted that since the value of the alarm flag is 0 and 1, the value range of the alarm value of the nth monitoring probe is [0, 1]; after weighted summation of multiple alarm values, the value range of the result is also [0, 1]. Therefore, the value range of the query urgency is [0, 2].

[0077] Thus, the query urgency of each monitoring probe at the current moment is obtained.

[0078] Step 4, analyze the average level of the network bandwidth of each monitoring probe at different moments before the current moment, and combine the query urgency to determine the caching urgency of each monitoring probe at the current moment; based on the caching urgency, determine the video caching length of each monitoring probe at the current moment; compare the actual caching length and the video caching length, and adjust the video transmission bit rate of different monitoring probes in real time.

[0079] Furthermore, the purpose of the video engine to perform video caching is to cope with the situation of poor network bandwidth. When the network bandwidth status is poor, the video engine needs to increase the video caching length to ensure that the video does not freeze. Therefore, analyze the network bandwidth status of each monitoring probe to adjust the video caching scheme of the monitoring probe, specifically:

[0080] Take the average value of the network bandwidth of each monitoring probe at multiple moments before the current moment as the bandwidth eigenvalue of each monitoring probe at the current moment;

[0081] In this embodiment, take the average value of the network bandwidth of each monitoring probe at all moments within 10 minutes before the current moment as the bandwidth eigenvalue of each monitoring probe at the current moment. As other implementation manners, the implementer can set it according to the actual situation.

[0082] It should be noted that the larger the bandwidth eigenvalue is, the better the network state of the monitoring probe in the recent period of time, and the faster the data transmission rate. When performing video caching, a shorter video caching length can be set for it, and at the same time, the transmission bit rate can be increased to improve the video caching quality, so that the videos called by the management personnel have better visual effects.

[0083] Furthermore, different video caching lengths are calculated for different monitoring probes. For monitoring probes that are more likely to be called and have a worse network condition, a longer video caching length is set to avoid stuttering problems when the management personnel switch the monitoring probes. The video caching length calculation method is as follows:

[0084] Take the ratio of the query urgency to the bandwidth eigenvalue as the caching urgency of each monitoring probe at the current moment;

[0085] In this embodiment, when calculating the ratio of the query urgency to the bandwidth eigenvalue, to avoid the denominator being 0, a preset value greater than 0 is added to the denominator. Among them, the value of the preset value greater than 0 is 0.01. As other implementation manners, the implementer can set it according to the actual situation.

[0086] It should be noted that the greater the query urgency and the smaller the bandwidth eigenvalue, the more likely the monitoring probe is to be called, and the worse the network state. The video engine needs to allocate more video caching space for it.

[0087] Furthermore, based on the caching urgency, calculate the video caching length, specifically:

[0088] Normalize the caching urgency of all monitoring probes at the current moment, and take the normalized value as the caching weight of each monitoring probe at the current moment;

[0089] In this embodiment, the normalization method adopted is: calculate the sum value of the caching urgency of all monitoring probes at the current moment, denoted as the second sum value, and take the ratio of the caching urgency to the second sum value as the caching weight of each monitoring probe at the current moment.

[0090] Allocate a preset duration of video caching space for each monitoring probe; take the product of the caching weight and the preset duration as the video caching length of each monitoring probe at the current moment;

[0091] In this embodiment, 10 seconds of video caching space is allocated for each monitoring probe. As other implementation manners, the implementer can set it according to the actual situation.

[0092] It should be noted that the greater the cache weight, the more likely the monitoring video of the corresponding monitoring probe will be called, and the worse the network environment of the monitoring probe. Therefore, in order to ensure that the video engine can timely call the cached data and provide smooth video services when the management personnel switch the monitoring probes, a longer video cache length is assigned to the monitoring probes; the step flowchart of the method for obtaining the video cache length provided by the embodiments of the present application is as Figure 2 shown.

[0093] Furthermore, for all monitoring probes, by using the video cache length, it is possible to prevent video stuttering when the management personnel switch the monitoring probes, so as to meet the video cache conditions of the monitoring probes and adjust the video transmission bit rate. Specifically:

[0094] For the video information of the monitoring probes whose actual cache length is less than the video cache length, compression transmission is performed using a preset minimum transmission bit rate, and for the video information of the monitoring probes whose actual cache length is greater than or equal to the video cache length, compression transmission is performed using a preset maximum transmission bit rate;

[0095] In this embodiment, the preset minimum bit rate is 0.5 Mdps, and the preset maximum bit rate is 2 Mdps. As other implementation manners, the implementer can set them according to the actual situation.

[0096] It should be noted that if the actual cache length is less than the video cache length, it indicates that there are problems such as network latency for the corresponding monitoring probe. By reducing the video quality, the smoothness of video transmission is ensured. If the actual cache length is greater than or equal to the video cache length, it indicates that the network state of the corresponding monitoring probe is good and the data transmission is smooth. By increasing the video transmission bit rate, the video quality is ensured on the premise of video smoothness.

[0097] It should be noted that adjusting the video transmission bit rate through the cache length is a well-known technology in the field of video engines and will not be elaborated here.

[0098] It should be understood that although Figure 1 the steps in the flowchart of Figure 1 are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover,

[0099] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0100] The above-described embodiments merely represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation to the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made. Therefore, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present application without departing from the content of the technical solution of the present application all belong to the protection scope of the technical solution of the present application.

Claims

1. A comprehensive perception method for environmental parameters in a smart campus based on AIoT, characterized in that, The method includes the following steps: Obtain the network bandwidth of each monitoring probe in the intelligent campus system at each moment, as well as the position coordinates of each monitoring probe, and obtain the actual cache length of each monitoring probe in real time through the video engine; Perform anomaly recognition on the video information of each monitoring probe at each moment, determine the alarm state of each monitoring probe at each moment, and obtain the alarm identifier of each monitoring probe at each moment; based on the alarm situation of each monitoring probe in the historical period, obtain the historical alarm vector of each monitoring probe; Obtain the alarm correlation degree between each monitoring probe and the rest of the monitoring probes through the spatial distance between each monitoring probe and the rest of the monitoring probes and the similarity degree of the historical alarm vectors; Normalize the alarm correlation degrees between each monitoring probe and all the other monitoring probes, and use the normalized result as the alarm influence weight between each monitoring probe and the rest of the monitoring probes; take the mean value of the alarm identifiers of each monitoring probe at multiple moments before the current moment as the alarm value of each monitoring probe at the current moment; based on the alarm influence weight between each monitoring probe and the rest of the monitoring probes, perform a weighted sum of the alarm values of the rest of the monitoring probes at the current moment, and use the sum of the weighted sum result and the alarm value of each monitoring probe at the current moment as the query urgency of each monitoring probe at the current moment; Analyze the average level of the network bandwidth of each monitoring probe at different moments before the current moment, and combine the query urgency to determine the cache urgency of each monitoring probe at the current moment; Based on the cache urgency, determine the video cache length of each monitoring probe at the current moment; compare the actual cache length and the video cache length, and adjust the video transmission bit rate of different monitoring probes in real time.

2. The comprehensive perception method for environmental parameters of an AIoT-based smart campus according to claim 1, characterized in that The determination of the alarm state of each monitoring probe at each moment and the acquisition of the alarm identifier of each monitoring probe at each moment include: If there is an anomaly in the video information of each monitoring probe at each moment, the monitoring probe immediately triggers an alarm. If each monitoring probe is in the alarm state at each moment, the alarm identifier is 1; otherwise, the alarm identifier is 0.

3. The comprehensive perception method for environmental parameters of an AIoT-based smart campus according to claim 2, characterized in that, The method for obtaining the historical alarm vector is: Count the number of times the alarm identifier is 1 for each monitoring probe in each time period, denoted as the alarm times; form the historical alarm vector of each monitoring probe with the alarm times of each monitoring probe in multiple time periods.

4. The comprehensive perception method for environmental parameters of an AIoT-based smart campus according to claim 1, wherein The spatial distance is the Euclidean distance between the position coordinates of each monitoring probe and the rest of the monitoring probes.

5. The comprehensive perception method for environmental parameters of an AIoT-based smart campus according to claim 1, characterized in that, The similarity degree is the reciprocal of the Euclidean distance between the historical alarm vectors of each monitoring probe and the rest of the monitoring probes.

6. The comprehensive perception method for environmental parameters of an AIoT-based smart campus according to claim 1, wherein, The alarm correlation degree is the ratio of the similarity degree to the spatial distance.

7. The comprehensive perception method for environmental parameters of an AIoT-based smart campus according to claim 1, characterized in that, The determination of the cache urgency of each monitoring probe at the current moment includes: Take the mean value of the network bandwidth of each monitoring probe at multiple moments before the current moment as the bandwidth eigenvalue of each monitoring probe at the current moment; Take the ratio of the query urgency to the bandwidth eigenvalue as the cache urgency of each monitoring probe at the current moment.

8. The comprehensive perception method for environmental parameters of an AIoT-based smart campus according to claim 1, characterized in that The determination of the video cache length of each monitoring probe at the current moment includes: Normalize the cache urgency of all monitoring probes at the current moment, and use the normalized value as the cache weight of each monitoring probe at the current moment; Allocate a video cache space with a preset duration for each monitoring probe; use the product of the cache weight and the preset duration as the video cache length of each monitoring probe at the current moment.

9. The comprehensive perception method for environmental parameters of an AIoT-based smart campus according to claim 1, characterized in that, The real-time adjustment of the video transmission bitrates of different monitoring probes includes: For the video information corresponding to the monitoring probe whose actual cache length is less than the video cache length, compress and transmit it at a preset minimum transmission bitrate; conversely, compress and transmit it at a preset maximum transmission bitrate.

Images