An AI-based intelligent security identification collaboration method and system

Abstract

The present invention discloses an intelligent security identification collaboration method and system based on AI monitoring. S1: Initialize various devices in the security system; S2: Based on the peacock courtship algorithm introducing a memory mechanism, collect and calculate the real-time performance indicators of each device; S3: Establish a multi-level competition and collaboration mechanism based on the memory mechanism; S4: When an abnormal event occurs in a low-level device, transfer the complex event to a high-level device for processing through competition; S5: Dynamically adjust the resource allocation strategy of each device according to the weighted performance indicators. The security system adjusts the resource allocation ratio according to the real-time and historical performance of the device through the weighted performance indicators; S6: Continuously monitor the real-time performance of each device and regularly update its memory score. The present invention realizes the intelligence, self-adaptability and long-term optimization of resource allocation, and improves the response speed, resource utilization rate and overall operation efficiency of the intelligent security system in a complex environment.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent security, and particularly to an intelligent security recognition cooperation method and system based on AI monitoring. Background Art

[0002] In the prior art, intelligent security systems are widely used in various scenarios, including public security, traffic monitoring, and enterprise security. They mainly rely on the collaborative work of cameras, sensors, and central processing systems to detect and respond to abnormal events through real-time monitoring and data processing. However, existing intelligent security systems usually rely on preset rules and fixed resource allocation strategies to allocate device resources and schedule tasks, lacking flexibility and adaptability. In complex monitoring scenarios, the response speed and resource utilization rate of the system are easily affected by fluctuations in device performance, resulting in unstable monitoring effects.

[0003] The prior art usually adopts a single-level resource allocation mechanism, and each device processes events according to fixed priorities. The traditional method cannot dynamically adjust according to different device performances and scenario requirements, and there is often a problem of uneven resource allocation when devices process tasks. Inefficient devices may occupy a large amount of resources, while efficient devices cannot obtain sufficient resource support in a timely manner, resulting in low overall operating efficiency of the system. In addition, most of the existing abnormal event detection technologies are independent of the cooperation between devices, and devices cannot transfer tasks and share resources according to each other's performance and workload, easily leading to resource waste and response delays when dealing with complex events.

[0004] On the other hand, in the case of long-term operation of existing intelligent security systems, the historical performance of devices is not effectively recorded and utilized. The system cannot optimize resource allocation and task scheduling based on the long-term performance of devices, and cannot effectively respond to the dynamic changes in device performance. When some devices show long-term stability or high efficiency, the prior art does not give these devices more resources and task priorities, reducing the adaptability and overall efficiency of the system.

[0005] In summary, the prior art mainly has the following disadvantages: First, resource allocation and task scheduling lack flexibility and are difficult to dynamically adjust according to the real-time and historical performance of devices; second, the collaborative ability between devices is poor, and effective multi-level competition and task transfer cannot be achieved, resulting in low efficiency in dealing with complex events; finally, the system fails to make full use of the historical performance of devices and cannot optimize resource utilization and task allocation strategies in the case of long-term operation. Summary of the Invention

[0006] An object of the present invention is to provide a smart security identification cooperation method and system based on AI monitoring. The present invention realizes the intelligence, self - adaptability and long - term optimization of resource allocation, and improves the response speed, resource utilization rate and overall operation efficiency of the smart security system in complex environments.

[0007] A smart security identification cooperation method based on AI monitoring according to an embodiment of the present invention includes the following steps:

[0008] S1. Initialize various devices in the security system, and set the initial resource allocation strategy and task assignment rules for each device;

[0009] S2. Based on the peacock courtship algorithm introducing a memory mechanism, collect and calculate the real - time performance indicators of each device, and calculate the weighted performance indicators of each device in combination with the memory scores of the devices;

[0010] S3. Establish a multi - level competition and cooperation mechanism based on the memory mechanism. The low - level devices pre - process simple events through edge computing, and compete according to their weighted performance indicators and historical performance. The competition results affect whether the low - level devices are eligible to process complex tasks or whether to transfer tasks to high - level devices. The high - level devices decide whether to accept the tasks transferred by the low - level devices according to the weighted performance indicators and historical performance of the low - level devices, and adjust the priority of complex tasks based on the weighted performance;

[0011] S4. When an abnormal event occurs in a low - level device, transfer the complex event to a high - level device for processing through competition. The high - level device determines the priority of each device according to the weighted performance indicators and memory scores of the low - level devices, and preferentially processes the devices with excellent historical performance and weighted performance;

[0012] S5. Dynamically adjust the resource allocation strategy of each device according to the weighted performance indicators. The security system adjusts the resource allocation ratio through the weighted performance indicators according to the real - time performance and historical performance of the devices, and preferentially allocates more resources to the devices with higher weighted performance in the case of limited resources to optimize the overall operation efficiency of the devices;

[0013] S6. Continuously monitor the real - time performance of each device, regularly update its memory score, and dynamically adjust the resource allocation strategy and task assignment rules of the security system in combination with the weighted performance indicators, so that the devices with excellent long - term performance continuously obtain priority in subsequent task assignments.

[0014] Optionally, the S1 includes the following sub - steps:

[0015] S11. Initialize the monitoring camera , sensors , edge computing devices and the central processing system , where i, j, k, and l are the indices of each device respectively;

[0016] S12. Set the initial resource allocation strategies for the monitoring cameras, sensors, edge computing devices, and the central processing system:

[0017] ;

[0018] Among them, is the resource allocation ratio for the i-th camera, j-th sensor, k-th edge computing device, and l-th central processing system. , and are the processing capabilities, response speeds, and abnormal event detection accuracies of each device respectively. , , are the weight coefficients of the corresponding devices, used for weighting according to the importance of the devices in different scenarios. The summation term in the denominator represents the overall performance of all devices;

[0019] S13. Set the task assignment rules for the devices. The task assignment rules are based on the task loads and task processing priorities of the devices, and assign tasks to the monitoring cameras, sensors, edge computing devices, and the central processing system according to the task complexity and device performance:

[0020] ;

[0021] Among them, is the task priority of the device. represents the task complexity, which depends on the task load of the current device and the processing capability of the device , is used to control the dynamics of task assignment, so that complex tasks are preferentially assigned to devices with higher performance in the case of limited resources;

[0022] S14. Establish the initial communication channels between the devices to enable the collaborative work among the monitoring cameras, sensors, edge computing devices, and the central processing system, and perform real-time updates and adjustments of the initial resource allocation strategies and task assignment rules through the communication channels.

[0023] Optionally, the S2 includes the following sub-steps:

[0024] S21. Collect the actual performance indicators of each security device, including the processing capability , the response speed and the abnormal event detection accuracy

[0025] S22. Initialize a memory score for each device to record the device's historical performance. , and set the initial memory score to the baseline value preset by the security system. As the security system runs, the processing capacity , response speed and anomaly event detection accuracy of the device are gradually accumulated into the memory score:

[0026] ;

[0027] Among them, is the memory score of the previous iteration, and are weight coefficients;

[0028] S23. Calculate the weighted performance metrics of each device , including the real-time performance of the device and combining its memory score :

[0029] ;

[0030] Among them, , , , are the corresponding weight coefficients, respectively controlling the influence of processing capacity, response speed, anomaly event detection accuracy and memory score on the comprehensive performance evaluation.

[0031] Optionally, the S3 includes the following sub-steps:

[0032] S31. Establish a multi-level competition and cooperation mechanism based on the memory mechanism. The low-level devices include surveillance cameras, sensors and edge computing devices. The low-level devices preprocess simple events through edge computing. The low-level devices participate in the competition according to the weighted performance metrics and the historical performance memory score ;

[0033] S32. During the competition, the low-level devices compete by showing their weighted performance metrics . The low-level devices with higher competition results have the opportunity to obtain the processing authority for complex tasks:

[0034] ;

[0035] Among them, is the competition result of the low-level device;

[0036] S33. According to the competition result Determine whether the low-level device is eligible to handle complex tasks. If the competition result is lower than the threshold set by the security system, transfer the task to the high-level device;

[0037] S34. The high-level device determines whether to accept the task based on the task transferred by the low-level device, the weighted performance index of the low-level device and the memory score and adjusts the task priority of the complex task :

[0038] ;

[0039] wherein, is the task complexity weight coefficient, is a function based on the time difference and is used to control the influence of the device's historical performance on the current task allocation, and is used to introduce the non-linear cumulative effect of the memory mechanism, is the task matching function of the device and is used to measure whether the low-level device is suitable for handling the current task, is the anomaly event detection accuracy of the low-level device and is used to distinguish the ability of the device to handle complex tasks, is the response speed of the low-level device, and adjusts the task priority through the exponential decay function ;

[0040] S35. After the competition ends, the security system updates the resource allocation strategies of the low-level device and the high-level device, so that the device with excellent performance can obtain priority resources and task processing permissions.

[0041] Optionally, the S4 includes the following sub-steps:

[0042] S41. When the low-level device detects an anomaly event, transfer the complex event to the high-level device for processing through competition. The basis of the competition is the weighted performance index and the memory score of the low-level device. The low-level device calculates its competition result through the following formula:

[0043] ;

[0044] wherein, is the time difference and is a function used to reflect the influence of the device's historical performance on the current competition;

[0045] S42. The low-level device transfers the complex event to the high-level device with the highest competition result for processing. The high-level device determines the priority based on the weighted performance index and the memory score of the low-level device, and preferentially processes the events transferred by the low-level devices with excellent historical performance and higher weighted performance indexes;

[0046] S43. Task Priority of Advanced Devices for Calculating Complex Events :

[0047] ;

[0048] Among them, is the task priority adjustment coefficient, which is used to introduce the non - linear effect of the memory score, making the low - level devices with stable performance history have higher priority in event transmission;

[0049] S44. The advanced device determines the order of processing events according to the calculated task priority and preferentially processes the complex events transmitted from the low - level devices with better performance. At the same time, it adjusts the event priority according to the urgency of the complex events;

[0050] S45. After the processing is completed, the advanced device feeds back the processing result to the low - level device and updates the memory score of the low - level device .

[0051] Optionally, the S5 includes the following sub - steps:

[0052] S51. The security system dynamically adjusts the resource allocation strategy of each device according to the weighted performance index of the device. The security system collects the performance data of the device in real - time and calculates the weighted performance index of each device based on the historical performance memory score of the device;

[0053] S52. Determine the resource allocation ratio of each device according to the weighted performance index of the device : :

[0054] ;

[0055] Among them, is the introduced non - linear cumulative effect, which adjusts the growth rate of the device performance, reflects the influence of the device processing ability and detection accuracy on the resource allocation ratio. A higher and will accelerate the growth rate of resource allocation, enabling the device to obtain corresponding resources when processing urgent or complex tasks, is used to standardize the resource allocation ratio of each device so that the sum of the resource allocation ratios of all devices is 1;

[0056] S53. The security system preferentially allocates resources to the devices with higher weighted performance indexes when resources are limited according to the calculated resource allocation ratio ;

[0057] S54. During the resource allocation process, the security system monitors the performance metrics and memory scores of devices in real time. When the real-time performance, response speed, and detection accuracy of a device change, the security system dynamically updates the resource allocation ratio based on the weighted performance metrics.

[0058] S55. By regularly evaluating the weighted performance metrics of devices and the historical performance memory scores , the security system optimizes the resource allocation strategy, enabling devices with excellent long-term performance to continuously receive priority resource allocation, while also adjusting the resource allocation of inefficient devices in real time.

[0059] Optionally, the task complexity is calculated as the comprehensive performance score of a device based on its weighted performance metrics and historical performance memory scores , and the comprehensive performance score is used to determine whether a device performs excellently:

[0060] ;

[0061] Among them, and are weight coefficients that respectively adjust the influence of the weighted performance metrics and historical performance memory scores of the device, represents the comprehensive performance score of the device over a period of time;

[0062] Based on the task complexity and the comprehensive performance score of the device , the security system introduces segmented thresholds:

[0063] When , the device is determined to perform excellently and is given priority to receive the most complex tasks and resources;

[0064] When , the device receives medium-complexity tasks and resources;

[0065] When , the device receives low-complexity tasks and resources;

[0066] is the threshold for the i-th complexity level.

[0067] A smart security recognition collaboration system based on AI monitoring includes the following modules:

[0068] A device initialization module, used to initialize various devices in the security system, including surveillance cameras, sensors, edge computing devices, and central processing systems, and set the initial resource allocation strategy and task dispatch rules for each device.

[0069] A performance monitoring module, which is used to collect the performance metrics of the device in real time, including the processing capacity, response speed, and anomaly event detection accuracy of the device, and record the historical performance memory score of the device. The performance monitoring module combines real-time data and historical performance data for comprehensive calculation to generate weighted performance metrics;

[0070] A task scheduling module, based on the weighted performance metrics and historical performance of the device, uses the peacock courtship algorithm for multi-level competition to determine whether a low-level device is eligible to process complex tasks, and transfers complex tasks to a high-level device for processing according to the competition result. Low-level devices include surveillance cameras, sensors, and edge computing devices, and high-level devices include a central processing system;

[0071] A resource allocation module, which is used to dynamically adjust the resource allocation strategy of each device according to the weighted performance metrics and historical performance memory score of the device. The resource allocation module preferentially allocates more resources to devices with excellent performance when resources are limited according to the performance excellence degree of the device;

[0072] An anomaly event handling module. When a low-level device detects an anomaly event, it transfers the complex event to a high-level device through competition. The high-level device adjusts the task priority according to the weighted performance metrics and memory score of the low-level device and processes the anomaly event;

[0073] A memory update module, which is used to continuously monitor and update the historical performance memory score of the device, and dynamically adjust it according to the real-time performance of the device. The memory update module optimizes the resource allocation and task dispatching strategies of the system by recording the task completion situation of the device.

[0074] The beneficial effects of the present invention are as follows:

[0075] (1) The present invention introduces a memory mechanism, enabling the system to accumulate memory scores based on the historical performance of the device. During the task processing of the device, not only its real-time performance is considered, but also the long-term stability and historical efficiency of the device are comprehensively evaluated. Through the combination of weighted performance metrics and memory scores, the system can preferentially allocate more resources to devices with excellent historical performance, ensuring that the system performs task scheduling more stably and efficiently during long-term operation. Compared with the traditional resource allocation method relying on fixed rules, the adaptive mechanism of the present invention significantly improves the rationality of resource allocation and optimizes the overall operation efficiency of the device.

[0076] (2) The present invention realizes effective cooperation between low-level devices and high-level devices through a multi-level competition mechanism. The low-level devices participate in the competition according to their weighted performance indicators to decide whether to process complex events or transfer them to high-level devices for processing. At the same time, the high-level devices decide whether to accept tasks and adjust task priorities according to the weighted performance of the low-level devices, effectively avoiding resource waste, ensuring that the system can respond quickly and schedule reasonably when facing complex events, improving the processing efficiency of the system for abnormal events. In a multi-task scenario, the present invention optimizes the task allocation between devices through competition and cooperation, reducing the response delay caused by unreasonable task allocation.

[0077] (3) The performance monitoring and memory update module of the present invention can continuously monitor the real-time and historical performance of devices, and by dynamically adjusting the resource allocation ratio and task priorities, it ensures that the system can adaptively adjust according to the performance changes of devices. In the case of device performance fluctuations or task complexity changes, the system dynamically adjusts the resource allocation strategy and task scheduling by real-time updating the weighted performance indicators and memory scores, enabling the system to maintain efficient operation under limited resources. Compared with the prior art, the dynamic adjustment mechanism of the present invention significantly enhances the flexibility of the system and its ability to cope with complex environments, ensuring the stability and security of the system during long-term operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] The drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, but do not constitute a limitation to the present invention. In the drawings:

[0079] Figure 1 is a flowchart of a method and system for intelligent security identification collaboration based on AI monitoring proposed by the present invention;

[0080] Figure 2 is a flowchart of resource allocation and task scheduling based on the peacock courtship algorithm in a method and system for intelligent security identification collaboration based on AI monitoring proposed by the present invention;

[0081] Figure 3 is a schematic diagram of task allocation of the multi-level competition and cooperation mechanism in a method and system for intelligent security identification collaboration based on AI monitoring proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0082] Now, the present invention will be further described in detail with reference to the drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0083] Refer to Figures 1-3 , a method for intelligent security identification collaboration based on AI monitoring, includes the following steps:

[0084] S1. Initialize various devices in the security system, and set the initial resource allocation strategy and task assignment rules for each device;

[0085] S2. Based on the peacock courtship algorithm with a memory mechanism, collect and calculate the real-time performance indicators of each device, and calculate the weighted performance indicators of each device in combination with the memory scores of the devices;

[0086] S3. Establish a multi-level competition and cooperation mechanism based on the memory mechanism. The low-level devices perform preprocessing on simple events through edge computing and compete according to their weighted performance indicators and historical performances. The competition results affect whether the low-level devices are eligible to process complex tasks or whether to transfer the tasks to high-level devices. The high-level devices decide whether to accept the tasks transferred by the low-level devices based on the weighted performance indicators and historical performances of the low-level devices, and adjust the priorities of complex tasks based on the weighted performance;

[0087] S4. When an abnormal event occurs in a low-level device, transfer the complex event to a high-level device for processing through competition. The high-level device determines the priorities of each device according to the weighted performance indicators and memory scores of the low-level devices, and preferentially processes the devices with excellent historical performances and weighted performances;

[0088] S5. Dynamically adjust the resource allocation strategy of each device according to the weighted performance indicators. The security system adjusts the resource allocation ratio through the weighted performance indicators according to the real-time and historical performances of the devices, and preferentially allocates more resources to the devices with higher weighted performances in the case of limited resources to optimize the overall operation efficiency of the devices;

[0089] S6. Continuously monitor the real-time performances of each device, regularly update their memory scores, and dynamically adjust the resource allocation strategy and task assignment rules of the security system in combination with the weighted performance indicators, so that the devices with excellent long-term performances can continuously obtain priority in subsequent task assignments.

[0090] In this embodiment, S1 includes the following sub-steps:

[0091] S11. Initialize the monitoring cameras , sensors , edge computing devices and the central processing system , where i, j, k, l are the indexes of each device respectively;

[0092] S12. Set the initial resource allocation strategies for the monitoring cameras, sensors, edge computing devices and the central processing system:

[0093] ;

[0094] Among them, The resource allocation ratio of the i-th camera, the j-th sensor, the k-th edge computing device, and the l-th central processing system, , and They are the processing capacity, response speed and abnormal event detection accuracy of each device. , , is the weight coefficient of the corresponding device, which is used to weight the device according to its importance in different scenarios. The cumulative term in the denominator represents the overall performance of all devices.

[0095] S13. Set the task assignment rules for the device. The task assignment rules are based on the task load and task processing priority of the device. The monitoring cameras, sensors, edge computing devices and central processing systems are assigned tasks according to the task complexity and device performance:

[0096] ;

[0097] in, is the task priority of the device, Represents the complexity of the task, which depends on the task load of the current device and the processing power of the device , It is used to control the dynamics of task dispatching, so that complex tasks can be dispatched to devices with higher performance first when resources are limited;

[0098] S14, establish an initial communication channel between devices to enable collaborative work between surveillance cameras, sensors, edge computing devices and the central processing system, and perform real-time updates and adjustments to initial resource allocation strategies and task dispatching rules through the communication channel.

[0099] In this implementation, S2 includes the following sub-steps:

[0100] S21. Collect the actual performance indicators of each security device, including processing capacity , Response speed and abnormal event detection accuracy ;

[0101] S22. Initialize a memory score for each device to record the device's historical performance The initial memory score is set to the preset benchmark value of the security system. As the security system runs longer, the processing capacity of the device , Response speed and abnormal event detection accuracy Historical performance is gradually accumulated into memory scores:

[0102] ;

[0103] Among them, is the memory score of the previous iteration, and are weight coefficients;

[0104] S23. Calculate the weighted performance indicators of each device , including the real-time performance of the device and combining its memory score :

[0105] ;

[0106] Among them, , , , are the corresponding weight coefficients, respectively controlling the influence of processing capacity, response speed, abnormal event detection accuracy and memory score on the comprehensive performance evaluation.

[0107] In this embodiment, S3 includes the following sub-steps:

[0108] S31. Establish a multi-level competition and cooperation mechanism based on the memory mechanism. The low-level devices include surveillance cameras, sensors and edge computing devices. The low-level devices preprocess simple events through edge computing. The low-level devices participate in the competition according to the weighted performance indicators and the historical performance memory score ;

[0109] S32. During the competition, the low-level devices compete by showing their weighted performance indicators . The low-level devices with higher competition results have the opportunity to obtain the processing authority for complex tasks:

[0110] ;

[0111] Among them, is the competition result of the low-level device;

[0112] S33. Determine whether the low-level device is eligible to process complex tasks according to the competition result . If the competition result is lower than the threshold set by the security system, the task will be transferred to the high-level device;

[0113] S34. The high-level device decides whether to accept the task according to the task transferred by the low-level device and the weighted performance indicator of the low-level device and the memory score , and adjusts the task priority of the complex task :

[0114] ;

[0115] Among them, is the task complexity weight coefficient, is a function based on the time difference and is used to control the impact of the historical performance of the device on the current task allocation, and is used to introduce the non-linear cumulative effect of the memory mechanism, is the task matching function of the device and is used to measure whether a low-level device is suitable for processing the current task, is the detection accuracy of abnormal events of the low-level device and is used to distinguish the ability of the device to handle complex tasks, is the response speed of the low-level device, and the task priority is adjusted through the exponential decay function ;

[0116] S35. After the competition ends, the security system updates the resource allocation strategies of the low-level devices and high-level devices, so that the devices with excellent performance can obtain priority resources and task processing permissions.

[0117] In this embodiment, S4 includes the following sub-steps:

[0118] S41. When a low-level device detects an abnormal event, it transfers the complex event to a high-level device for processing through competition. The basis for the competition is the weighted performance index and memory score of the low-level device. The low-level device calculates its competition result through the following formula:

[0119] ;

[0120] Among them, is a function of the time difference and is used to reflect the impact of the historical performance of the device on the current competition;

[0121] S42. The low-level device transfers the complex event to the high-level device with the highest competition result for processing. The high-level device determines the priority according to the weighted performance index and memory score of the low-level device, and preferentially processes the events transferred by the low-level devices with excellent historical performance and higher weighted performance indexes;

[0122] S43. The high-level device calculates the task priority of the complex event :

[0123] ;

[0124] Among them, is the task priority adjustment coefficient, and is used to introduce the non-linear effect of the memory score, so that the low-level devices with stable historical performance have more priority in event transfer;

[0125] S44. The high-level device adjusts the task priority Determine the order of processing events, giving priority to processing complex events transmitted from lower-level devices with better performance, and adjusting the event priority according to the urgency of the complex events;

[0126] S45. After the processing is completed, the high-level device feeds back the processing result to the low-level device and updates the memory score of the low-level device .

[0127] In this embodiment, S5 includes the following sub-steps:

[0128] S51. The security system dynamically adjusts the resource allocation strategy of each device according to the weighted performance index of the device. The security system collects the performance data of the device in real time and calculates the weighted performance index of each device based on the historical performance memory score of the device;

[0129] S52. According to the weighted performance index of the device Determine the resource allocation ratio of each device :

[0130] ;

[0131] Among them, is the introduced non-linear cumulative effect, which adjusts the growth rate of the device performance, reflects the influence of the device processing ability and detection accuracy on the resource allocation ratio. A higher and will accelerate the resource allocation growth rate, enabling the device to obtain corresponding resources when processing urgent or complex tasks, is used to standardize the resource allocation ratio of each device, so that the sum of the resource allocation ratios of all devices is 1;

[0132] S53. The security system allocates resources preferentially to devices with higher weighted performance indexes in the case of limited resources according to the calculated resource allocation ratio ;

[0133] S54. During the resource allocation process, the security system monitors the performance index and memory score of the device in real time. When the real-time performance, response speed and detection accuracy of the device change, the security system dynamically updates the resource allocation ratio according to the weighted performance index;

[0134] S55. By regularly evaluating the weighted performance index of the device and the historical performance memory score , the security system optimizes the resource allocation strategy, enabling devices with excellent long-term performance to continuously obtain preferential resource allocation, and at the same time, adjusting the resource allocation of inefficient devices in real time.

[0135] In this embodiment, the task complexity is to calculate the comprehensive performance score of the device according to the weighted performance index and the historical performance memory score of the device. The comprehensive performance score is used to determine whether the device performs excellently:

[0136] ;

[0137] Among them, and are weight coefficients, which respectively adjust the influence of the weighted performance index and the historical performance memory score of the device. represents the comprehensive performance score of the device over a period of time;

[0138] According to the task complexity and the comprehensive performance score of the device , the security system introduces segmented thresholds:

[0139] When , the device is determined to perform excellently and is given priority to obtain the tasks and resources with the highest complexity;

[0140] When , the device obtains tasks and resources with medium complexity;

[0141] When , the device obtains tasks and resources with low complexity;

[0142] is the threshold for the i-th complexity level.

[0143] An intelligent security recognition collaboration system based on AI monitoring includes the following modules:

[0144] The device initialization module is used to initialize various devices in the security system, including surveillance cameras, sensors, edge computing devices, and the central processing system, and set the initial resource allocation strategy and task dispatching rules for each device;

[0145] The performance monitoring module is used to collect the performance indicators of the device in real time, including the processing capacity, response speed, and abnormal event detection accuracy of the device, and record the historical performance memory score of the device. The performance monitoring module combines real-time data and historical performance data for comprehensive calculation to generate a weighted performance index;

[0146] The task scheduling module, based on the weighted performance index and historical performance of the device, uses the peacock courtship algorithm for multi-level competition to determine whether a low-level device is eligible to process complex tasks, and transfers complex tasks to a high-level device for processing according to the competition results. Low-level devices include surveillance cameras, sensors, and edge computing devices, and high-level devices include the central processing system;

[0147] A resource allocation module, which is used to dynamically adjust the resource allocation strategies of each device according to the weighted performance indicators and historical performance memory scores of the devices. The resource allocation module preferentially allocates more resources to the devices with excellent performance in the case of limited resources according to the excellent degree of the device performance;

[0148] An abnormal event handling module. When a low-level device detects an abnormal event, it competes to transfer the complex event to a high-level device. The high-level device adjusts the task priority according to the weighted performance indicators and memory scores of the low-level device and processes the abnormal event;

[0149] A memory update module, which is used to continuously monitor and update the historical performance memory scores of the devices and dynamically adjust according to the real-time performance of the devices. The memory update module optimizes the resource allocation and task dispatching strategies of the system by recording the task completion situations of the devices.

[0150] Example 1: To verify the actual application effect of the present invention in the intelligent security identification collaboration system, the experimenter selected two publicly available datasets, the Stanford Drone Dataset (SDD) and the DukeMTMC dataset, which are respectively used for the high-traffic monitoring environment and the complex abnormal behavior detection scenario. These two datasets respectively provide different types of monitoring scenarios to test the performance improvement of the present invention in multi-level device collaboration, resource allocation, and abnormal event detection.

[0151] The Stanford Drone Dataset (SDD) contains video data of large outdoor areas monitored from different angles, covering dynamic scenarios of crowd behavior and vehicle activities. The experimenter selected a subset of the data monitoring the shopping center area for the experiment. The data collection time was the peak period from 9 am to 9 pm every day, covering the whole year of 2020. The DukeMTMC dataset is mainly used for pedestrian trajectory analysis and abnormal behavior detection, and contains indoor and outdoor monitoring data obtained from multiple cameras to detect abnormal behaviors in the crowd. The experimenter divided the two datasets into a training set and a test set according to a ratio of 8:2 to train and test the effect of the peacock courtship algorithm proposed by the present invention in intelligent security identification collaboration.

[0152] In this experiment, a pre-trained YOLOv5 model was used to extract the object detection features in the video, including the position information and categories of pedestrians, vehicles, and other dynamic objects. For these multi-modal monitoring video data, the experimenter further used ResNet50 to extract visual features and performed weighted calculation in combination with the historical performance memory scores of each camera to adjust the resource allocation strategy of each camera. The historical performance of each device is based on its detection accuracy, response time, and processing ability in different scenarios. The historical data uses the performance in the past 30 days as the weight input to the memory module.

[0153] During the resource allocation and task scheduling process, the experimenter compared the traditional fixed resource allocation method with the dynamic resource allocation strategy of the present invention. In the traditional method, the resource allocation for each camera and sensor is based on a preset fixed ratio, while in the present invention, the peacock courtship algorithm is used to dynamically adjust resources, and the resource allocation priority is determined according to the weighted performance indicators and historical performance of each device. In the experiment, the peacock courtship algorithm not only considers the real-time performance of the device, but also combines the performance of the device in past events, enabling resources to be more effectively allocated to devices that perform well in complex environments.

[0154] During the experiment, the experimenter used the Adam optimizer for training, with a learning rate of 0.0001, trained for 100 epochs, and the mini-batch was set to 32. The experimenter adopted accuracy, response time, resource utilization rate, and the average false alarm rate of event processing as the main evaluation indicators. To better evaluate the application effect of the present invention in complex scenarios, in the StanfordDrone Dataset (SDD), the experimenter introduced indicators such as the accuracy of pedestrian flow monitoring, the detection accuracy of abnormal events, and the resource scheduling delay. These indicators can better reflect the performance of the system under the monitoring pressure during peak hours.

[0155] Table 1 Comparison between the present invention and traditional methods on the SDD dataset

[0156]

[0157] The experimental results on the SDD dataset show that the peacock courtship algorithm of the present invention has obvious advantages in processing pedestrian flow monitoring and abnormal behavior detection. Compared with the traditional method, the response time is shortened by 61.9%, the accuracy of pedestrian flow monitoring is increased by about 14%, and the accuracy of abnormal event detection is also significantly increased by 17.6%. In addition, through the multi-level device cooperation mechanism and the resource adaptive allocation strategy, the resource utilization rate is increased from 65% of the traditional method to 88.5%, effectively reducing the waste of system resources. At the same time, the false alarm rate is reduced by 8.1 percentage points, indicating that the present invention has higher robustness in dealing with complex scenarios.

[0158] Table 2 Comparison between the present invention and traditional methods on the DukeMTMC dataset

[0159]

[0160] In the experimental results of the DukeMTMC dataset, the accuracy of the present invention in detecting abnormal behaviors reached 91.2%, which is nearly 17% higher than that of traditional methods. In addition, the response time of the system was shortened from 10.1 seconds of traditional methods to 4.7 seconds, significantly improving the reaction speed of the system to abnormal behaviors. At the same time, the resource allocation optimization rate of the present invention is significantly higher than that of traditional methods, and the resource utilization rate has increased by 22.9 percentage points. Especially when dealing with complex abnormal events, the average processing duration has been shortened by 54.2%, providing a more efficient solution for real-time security monitoring.

[0161] It can be seen from the results of Example 1 that the present invention significantly optimizes the resource allocation and task scheduling mechanisms in complex scenarios, not only improving the processing efficiency of the security system, but also reducing the false alarm rate and resource waste. In the experiment, the intelligent scheduling and competition cooperation mechanism of the present invention significantly improves the utilization rate of devices by combining weighted performance indicators and historical performance scores, ensuring that complex events can still be efficiently handled under limited resources, fully verifying the feasibility and effectiveness of the present invention in practical applications.

[0162] The present invention introduces a memory mechanism, enabling the system to accumulate memory scores based on the historical performance of devices. During the task processing of devices, it not only considers their real-time performance, but also comprehensively evaluates the long-term stability and historical efficiency of devices. By combining weighted performance indicators and memory scores, the system can preferentially allocate more resources to devices with excellent historical performance, ensuring that the system performs task scheduling more stably and efficiently during long-term operation. Compared with the traditional resource allocation method relying on fixed rules, the adaptive mechanism of the present invention significantly improves the rationality of resource allocation and optimizes the overall operation efficiency of devices.

[0163] The present invention realizes effective cooperation between low-level devices and high-level devices through a multi-level competition mechanism. Low-level devices participate in the competition according to their weighted performance indicators to decide whether to process complex events or transfer them to high-level devices for processing. At the same time, high-level devices decide whether to accept tasks and adjust task priorities according to the weighted performance of low-level devices, effectively avoiding resource waste and ensuring that the system can respond quickly and schedule reasonably when facing complex events, improving the processing efficiency of the system for abnormal events. In multi-task scenarios, the present invention optimizes the task allocation between devices through competition and cooperation, reducing the response delay caused by unreasonable task allocation.

[0164] The performance monitoring and memory update module of the present invention can continuously monitor the real-time and historical performance of the device, and by dynamically adjusting the resource allocation ratio and task priority, it ensures that the system can adaptively adjust according to the performance changes of the device. In the case of fluctuations in device performance or changes in task complexity, the system dynamically adjusts the resource allocation strategy and task scheduling by real-time updating the weighted performance metrics and memory scores, enabling the system to maintain efficient operation even under limited resources. Compared with the prior art, the dynamic adjustment mechanism of the present invention significantly enhances the flexibility of the system and its ability to handle complex environments, ensuring the stability and security of the system during long-term operation.

[0165] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, shall be covered by the protection scope of the present invention.

Claims

1. A smart security identification and collaboration method based on AI monitoring, characterized in that: The steps include: S1. Initialize various devices in the security system and set the initial resource allocation strategy and task assignment rules for each device; S2. Based on the peacock courtship algorithm with the memory mechanism introduced, the real-time performance indicators of each device are collected and calculated, and the weighted performance indicators of each device are calculated in combination with the memory score of the device; The S2 comprises the following sub-steps: S21. Collect the actual performance indicators of each security device, including the processing capacity P′ proc (x i ), response speed R′ speed (y j ) and abnormal event detection accuracy D′ acc (z k ); S22. Initialize a memory score M for each device to record the historical performance of the device. score (i, j, k), the initial memory score is set to the preset benchmark value of the security system. As the security system continues to run, the processing capacity of the device P′ proc (x i ), response speed R′ speed (y j ) and abnormal event detection accuracy D′ acc (z k ) Historical performance is gradually accumulated into memory score: M score (i,j,k)=α·M score_prev (i,j,k)+β·(P′ proc (x i )+R′ speed (y j )+D′ acc (z k )); Among them, M score_prev (i, j, k) is the memory score of the previous iteration, α and β are weight coefficients; S23, calculate the weighted performance index P′ of each device total (i,j,k), including the real-time performance of the device and its memory score M score (i,j,k): P′ total (i,j,k)=w1·P′ proc (x i )+w2·R′ speed (y j )+w3·D′ acc (z k )+w4·M score (i,j,k); Among them, w1, w2, w3, and w4 are corresponding weight coefficients, which respectively control the impact of processing capacity, response speed, abnormal event detection accuracy, and memory score on the comprehensive performance evaluation; S3. Establish a multi-level competition and cooperation mechanism based on the memory mechanism. The low-level devices pre-process simple events through edge computing and compete according to their weighted performance indicators and historical performance. The competition results affect whether the low-level devices are qualified to handle complex tasks or whether to pass the tasks to high-level devices. The high-level devices decide whether to accept the tasks passed by the low-level devices according to the weighted performance indicators and historical performance of the low-level devices, and adjust the priority of the complex tasks based on the weighted performance. The S3 comprises the following sub-steps: S31. Establish a multi-level competitive cooperation mechanism based on the memory mechanism. The low-level devices include surveillance cameras, sensors and edge computing devices. The low-level devices pre-process simple events through edge computing. The low-level devices perform weighted performance indicators P′ total (i,j,k) and historical performance memory score M score (i,j,k) participate in the competition; S32, during the competition, the lower-level devices demonstrate their weighted performance index P′ total (i, j, k) compete, and the low-level device with the higher competition result has the opportunity to obtain the processing rights of complex tasks: Among them, C result (i, j, k) is the competition result of the low-level device; S33. According to the competition results C result (i, j, k) Determine whether the low-level device is qualified to handle the complex task. If the competition result is lower than the threshold set by the security system, the task will be passed to the high-level device; S34, the high-level device is based on the task delivered by the low-level device and the weighted performance index P′ of the low-level device total (i,j,k) and the memory score M score (i,j,k) decides whether to accept the task and adjusts the task priority Q of the complex task priority_complex (l): Among them, γ is the task complexity weight coefficient, f(ΔT) is a function based on the time difference ΔT, which is used to control the impact of the historical performance of the device on the current task allocation, log(1+M score (i, j, k)·f(ΔT)) is used to introduce the nonlinear cumulative effect of the memory mechanism. h(i, j, k) is the task matching function of the device, which is used to measure whether the low-level device is suitable for processing the current task. D′ acc (z k ) is the abnormal event detection accuracy of low-level devices, which is used to distinguish the ability of devices to handle complex tasks, R′ speed (y j ) is the response speed of the low-level device, which decays through an exponential function Adjust task priorities; S35. After the competition is over, the security system updates the resource allocation strategy for low-level and high-level devices, so that devices with outstanding performance can obtain priority resources and task processing permissions; S4. When an abnormal event occurs in a low-level device, the complex event is transmitted to the high-level device for processing through competition. The high-level device determines the priority of each device according to the weighted performance index and memory score of the low-level device, and gives priority to the devices with excellent historical performance and weighted performance; S5. Dynamically adjust the resource allocation strategy of each device based on weighted performance indicators. The security system adjusts the resource allocation ratio based on the real-time and historical performance of the device through weighted performance indicators. When resources are limited, more resources are allocated to devices with higher weighted performance first, thereby optimizing the overall operating efficiency of the device. S6. Continuously monitor the real-time performance of each device, regularly update its memory score, and dynamically adjust the resource allocation strategy and task assignment rules of the security system based on weighted performance indicators, so that devices with long-term excellent performance continue to receive priority in subsequent task assignments.

2. According to claim 1, a smart security identification and coordination method based on AI monitoring is characterized in that: The S1 comprises the following sub-steps: S11, Initialize surveillance camera C i , sensor S j , Edge computing device E k and central processing system P l , where i, j, k, and l are the indexes of each device respectively; S12. Set the initial resource allocation strategy for surveillance cameras, sensors, edge computing devices, and central processing systems: Among them, R alloc (i,j,k,l) ​​is the resource allocation ratio of the i-th camera, j-th sensor, k-th edge computing device and l-th central processing system, P proc (x i ), R speed (y j ) and D acc (z k ) are the processing capability, response speed and abnormal event detection accuracy of each device, respectively, i ,w j ,w k is the weight coefficient of the corresponding device, which is used to weight the device according to its importance in different scenarios. The cumulative term in the denominator represents the overall performance of all devices. S13. Set the task assignment rules for the device. The task assignment rules are based on the task load and task processing priority of the device. The monitoring cameras, sensors, edge computing devices and central processing systems are assigned tasks according to the task complexity and device performance: Among them, Q priority (i,j,k,l) ​​is the task priority of the device, T comp (L task ,P proc (x i )) represents the task complexity, which depends on the task load L of the current device task and the processing power of the device P proc (x i ), It is used to control the dynamics of task dispatching, so that complex tasks can be dispatched to devices with higher performance first when resources are limited; S14, establish an initial communication channel between devices to enable collaborative work between surveillance cameras, sensors, edge computing devices and the central processing system, and perform real-time updates and adjustments to initial resource allocation strategies and task dispatching rules through the communication channel.

3. According to claim 1, the intelligent security identification and coordination method based on AI monitoring is characterized in that: The S4 comprises the following sub-steps: S41. When a low-level device detects an abnormal event, it transmits the complex event to the high-level device for processing through competition. The basis of competition is the weighted performance index and memory score of the low-level device. The low-level device calculates its competition result through the following formula: Where f(ΔT) is a function of the time difference ΔT, which is used to reflect the impact of the device’s historical performance on the current competition; S42: The low-level device transmits the complex event to the high-level device with the highest competition result for processing. The high-level device determines the priority according to the weighted performance index and memory score of the low-level device, and gives priority to processing the events transmitted by the low-level device with excellent historical performance and higher weighted performance index; S43, Task priority of advanced equipment calculation complex events Q priority_event (l): Q priority_event (l)=γ·(P′ total (i,j,k)·(1+log(1+M score (i,j,k)))); Among them, γ is the task priority adjustment coefficient, log(1+M score (i,j,k)) is used to introduce the nonlinear effect of memory score, so that low-level devices with stable historical performance have higher priority in event delivery; S44, advanced equipment based on the calculated task priority Q priority_event (l) Determine the order in which events are processed, give priority to complex events transmitted from lower-level devices with better performance, and adjust event priorities based on the urgency of complex events; S45. After the processing is completed, the high-level device feeds back the processing results to the low-level device and updates the memory score M′ of the low-level device. score (i,j,k).

4. According to the AI ​​monitoring-based intelligent security identification and coordination method of claim 1, it is characterized in that: The S5 comprises the following sub-steps: S51. The security system dynamically adjusts the resource allocation strategy of each device according to the weighted performance index of the device. The security system collects the performance data of the device in real time and the historical performance memory score of the device to calculate the weighted performance index of each device. S52, according to the weighted performance index P′ of the equipment total (i,j,k) determines the resource allocation ratio R of each device alloc (i,j,k): in, To introduce nonlinear cumulative effects, adjust the growth rate of equipment performance, Reflects the impact of equipment processing capacity and detection accuracy on resource allocation ratio. Higher P′ proc (x i ) and D′ acc (z k ) will accelerate the growth rate of resource allocation, so that devices can obtain corresponding resources when handling urgent or complex tasks. Used to standardize the resource allocation ratio of each device so that the total resource allocation ratio of all devices is 1; S53, the security system calculates the resource allocation ratio R alloc (i, j, k) When resources are limited, resources are allocated preferentially to devices with higher weighted performance indicators; S54. During the resource allocation process, the security system monitors the performance indicators and memory scores of the equipment in real time. When the real-time performance, response speed and detection accuracy of the equipment change, the security system dynamically updates the resource allocation ratio based on the weighted performance indicators; S55, by regularly evaluating the weighted performance index P′ of the equipment total (i,j,k) and historical performance memory score M′ score (i, j, k), the security system optimizes the resource allocation strategy so that the equipment with long-term excellent performance can continue to obtain priority resource allocation, and at the same time adjust the resource allocation of inefficient equipment in real time.

5. According to the AI ​​monitoring-based intelligent security identification collaborative method of claim 1, it is characterized in that: The task complexity is calculated based on the weighted performance index and historical performance memory score of the device. performance (i,j,k), the comprehensive performance score is used to determine whether the device performs well: S performance (i,j,k)=λ1·P′ total (i,j,k)+λ2·M′ score (i,j,k); Among them, λ1 and λ2 are weight coefficients, P′ total (i, j, k) is the weighted performance index for periodic evaluation of equipment, M′ score (i, j, k) are the memory scores of low-level devices, which respectively adjust the weighted performance index and historical performance memory score of the device. S performance (i, j, k) represents the overall performance score of the device over a period of time; Score based on task complexity and equipment performance performance (i,j,k), the security system introduces segmentation thresholds: When S performance (i,j,k)≥T excellent1 When the device is judged to have excellent performance, it will be given priority to obtain the most complex tasks and resources; When T excellent2 ≤S performance (i,j,k) <T excellent1 When , the device obtains tasks and resources of medium complexity; When S performance (i,j,k) <T excellent2 When, the device obtains low-complexity tasks and resources; T excellenti is the threshold of the i-th complexity level.

Images