An adaptive perimeter defense linkage handling method and system

Abstract

The application discloses a kind of self-adapting perimeter protection linkage handling method and system, specifically related to security linkage alarm control technical field, for solving the problem of multi-target adaptive response linkage difference;The application generates the integrated multi-target perception frame by deploying multiple device equipment to collect multi-source perception data in real time, and carries out time format standardization and space mapping;Extract the dynamic characteristics such as speed, direction, residence time and path of target object, construct feature vector and match with historical behavior template, generate threat level trend;Combined with response device state, response resource scheduling diagram is constructed, response generation value is calculated and optimal strategy sequence is generated, and actions such as voice warning, lighting intervention are executed in linkage;Collect device feedback during execution, judge whether strategy is invalid or abnormal closed loop, if invalid, re-evaluate threat level, dynamically reconstruct response strategy and compress strategy chain length, improve the accuracy and linkage efficiency of perimeter protection.

Description

Technical Field

[0001] The present invention relates to the technical field of security linkage alarm control, and more particularly to an adaptive perimeter protection linkage disposal method and system. Background Art

[0002] Intelligent perimeter protection systems are playing an increasingly critical role in the security of key areas such as substations, oil and gas pipelines, and city boundaries. Existing technologies have extensively integrated multimodal sensors such as infrared radiation, lidar, video surveillance, and geomagnetic sensing. Combined with voice-activated repelling, searchlight warnings, and image tracking, these systems have initially achieved real-time detection and response control of intrusions. With the advancement of artificial intelligence recognition, edge computing, and IoT collaboration, perimeter systems are developing towards multi-target recognition, automatic policy invocation, and closed-loop execution control, gradually increasing their practical value in open perimeter scenarios.

[0003] Deficiencies in existing technologies: Current perimeter protection systems are mostly based on pre-set response rules to trigger linkage operations. That is, when sensors detect an intrusion signal, they automatically execute a fixed combination of disposal actions. However, this rule system exhibits significant response rigidity when faced with sudden, asynchronous intrusions from multiple targets. It is difficult to adaptively adjust strategies based on the dynamic evolution of threat levels, resulting in frequent delays in the disposal of high-priority targets and over-disposal of low-risk targets. This problem arises because the existing system's response logic for intrusion targets only remains at the preliminary classification and identification stage, lacking the ability to comprehensively assess target behavior patterns, path trends, and regional importance. In actual scenarios, when multiple intrusion targets appear simultaneously and their behavioral characteristics differ significantly, such as one quickly running close to a sensitive area while another moves slowly around the periphery, it is impossible to infer their threat ranking based on real-time dynamics, nor is it possible to rationally arrange disposal priorities based on currently available linkage resources. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides an adaptive perimeter protection linkage disposal method and system to solve the problem of poor multi-target adaptive response linkage in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An adaptive perimeter protection linkage disposal method includes the following steps:

[0007] Collect data from infrared radiation devices, laser radars, video surveillance equipment, and microwave sensors deployed within the perimeter protection area in real time, convert format and map spatial coordinates based on a unified time base, and generate a fused multi-target perception frame sequence.

[0008] Extract the dynamic behavior characteristics of each target object from the fused multi-target perception frame sequence, construct the corresponding feature vector, and generate the threat level trend information of the target object according to the preset rules;

[0009] Combine the threat level trend of each target object and the real-time availability of response equipment within the perimeter to establish a response resource scheduling diagram and identify whether there are response conflicts or equipment allocation conflicts between target objects;

[0010] According to the response resource scheduling diagram, a matching response strategy sequence is generated according to the target priority, and the response devices are scheduled to perform the corresponding tasks in sequence;

[0011] During the execution of the response device, the action feedback information of each response device is collected, including whether the task is completed, whether the target object is affected, and the occupancy time of the device. The action feedback information is compared with the strategy sequence to identify whether the current disposal strategy is invalid or interrupted;

[0012] When strategy failure or unclosed-loop response is identified, reassess the threat level of the target object and adjust the response strategy based on the updated priority and response equipment status.

[0013] In a preferred embodiment, format conversion and spatial coordinate mapping are performed based on a unified time reference to generate a fused multi-target perception frame sequence. The specific process is as follows:

[0014] Obtaining raw perception data through infrared radiation devices, laser radars, video surveillance equipment, and microwave sensors;

[0015] Use a unified time base to align timestamps and standardize the format of heterogeneous data structures;

[0016] Based on the preset spatial coordinate reference model, various types of perception data are mapped to the same coordinate system, and repeated targets are fused and de-redundant to generate a structured multi-target perception frame.

[0017] In a preferred embodiment, dynamic behavior features include the target object's moving speed, moving direction change rate, dwell duration, and shortest path length to the key area. Feature vectors perform similarity matching based on historical behavior templates to generate threat level trend information.

[0018] In a preferred embodiment, the feature vector performs similarity matching based on the historical behavior template to generate threat level trend information. The specific process is as follows:

[0019] Constructing the extracted dynamic behavior features into feature vectors;

[0020] Perform similarity matching between the feature vector and the historical behavior template vector, and use the Euclidean distance to calculate the matching value of each historical behavior template;

[0021] The threat level corresponding to the historical behavior template with the smallest distance is selected as the threat level result in the current time window;

[0022] A threat level sequence is formed by sliding the time window, and the threat level trend information of the target object is generated according to the level change trend.

[0023] In a preferred embodiment, a response resource scheduling diagram is established by combining the threat level trend of each target object and the real-time availability of response equipment within the perimeter to identify whether there is a response conflict or equipment allocation conflict between target objects. The specific process is as follows:

[0024] Collect the current idle status, physical location, and expected response delay of each responding device;

[0025] Construct a bipartite graph structure with target objects and device nodes as endpoints, establish connections between each target object and all idle responding devices, and calculate the response cost of each connection edge;

[0026] The response cost is determined based on the spatial distance between the target object and the device, the current task load of the device, and the compatibility between the device capability and the target threat level.

[0027] Spatial distance represents the physical cost required for the device to respond, task load represents the load or power consumption level of the device's currently occupied resources, and the adaptation relationship indicates whether the device has the execution capability under the current policy. The spatial distance, task load, and adaptation relationship are weighted according to the preset weight coefficient to obtain the response cost of the connection edge.

[0028] A minimum conflict matching graph is constructed based on the response cost values ​​of all connected edges to identify response conflicts and device allocation conflicts.

[0029] In a preferred embodiment, according to the response resource scheduling diagram, a matching response strategy sequence is generated according to the target priority, and the response devices are scheduled to perform corresponding tasks in sequence. The specific process is as follows:

[0030] The matching response strategy sequence includes voice warning actions, area lighting intervention, image tracking actions, and drone intervention actions. The queues are sorted according to preset priority rules, and the strategy execution timeout window and device redundant alternative paths are set.

[0031] In a preferred embodiment, the action feedback information is compared with the response strategy sequence to identify whether the current handling strategy is invalid or interrupted. The specific process is as follows:

[0032] Compare the action completion status of the responding device with the expected execution identifier of the corresponding action in the policy instruction. If the action is not completed, it is determined that the device response is interrupted;

[0033] Compare the feedback behavior impact information of the response device with the behavior status of the target object. If the target does not produce the expected evacuation or movement response, the response is determined to be invalid.

[0034] Compare the actual occupancy time of the responding device with the time window set by the policy. If it exceeds the upper limit, it is determined to be a handling timeout;

[0035] Based on the comparison results, determine whether the current response strategy is invalid, interrupted or in an abnormal closed-loop state.

[0036] In a preferred embodiment, when it is identified that the strategy fails or the response is not closed-loop, the threat level of the target object is reassessed. The specific process is as follows:

[0037] Based on the new behavioral feature changes of the target object after the response action, the speed, direction and stay behavior indicators are re-extracted;

[0038] Update the threat trend curve of the target object based on the feedback results of the unclosed-loop disposal;

[0039] Restructure response strategies and adjust priorities based on updated threat levels and device status.

[0040] An adaptive perimeter protection linkage disposal system, used to implement the above-mentioned adaptive perimeter protection linkage disposal method, comprising:

[0041] The data acquisition module is used to collect data output by infrared radiation devices, laser radars, video surveillance equipment, and microwave sensors deployed within the perimeter protection area in real time, perform format conversion and spatial coordinate mapping based on a unified time base, and generate a fused multi-target perception frame sequence;

[0042] The response conflict determination module is used to extract the dynamic behavior characteristics of each target object from the fused multi-target perception frame sequence, construct the corresponding feature vector, and generate the threat level trend information of the target object according to the preset rules. The response resource scheduling diagram is established by combining the threat level trend of each target object and the real-time availability status of the response equipment within the perimeter to identify whether there is a response conflict or equipment allocation conflict between the target objects;

[0043] A disposal strategy identification module generates a matched response strategy sequence according to the target priority based on the response resource scheduling graph, and sequentially schedules the response equipment to perform corresponding tasks, collects action feedback information of each response equipment in the execution process of the response equipment, including whether the task is completed, whether the target object is affected, and the occupation time length of the equipment, and compares the action feedback information with the strategy sequence to identify whether the current disposal strategy is invalid or interrupted;

[0044] An adjustment response module re-evaluates the threat level of the target object when the strategy is identified as invalid or the response is not closed-loop, and adjusts the response strategy according to the updated priority and the state of the response equipment.

[0045] The technical effects and advantages of the present application are:

[0046] The present application constructs a multi-source perception system by deploying infrared counter-shooting devices, laser radars, video monitoring equipment and microwave sensors in the perimeter protection area, collects the spatial position, motion trajectory and behavior state information of the target object in real time, and converts the heterogeneous data into a unified time reference and spatial coordinate mapping to generate a structured fusion multi-target perception frame sequence, which effectively improves the accuracy of target identification and the consistency of behavior analysis; On this basis, the dynamic behavior characteristics of the target object, such as moving speed, direction change rate, stay time and shortest path length between the target object and the key area, are extracted, a feature vector is constructed and similarity matching with the historical behavior template is performed to generate threat level trend information that can evolve over time, realizing continuous perception and intelligent judgment of potential intrusion behavior;

[0047] At the same time, a response resource scheduling graph with the target object and the response equipment as the end points is constructed, the spatial position, current load and adaptive capacity of the response equipment are comprehensively considered, the response cost value of the connection edge is calculated and the response conflict is identified, the optimal strategy sequence is generated through graph matching algorithm to realize dynamic linkage and closed-loop execution of multi-level disposal actions such as voice warning, lighting intervention, image tracking and unmanned aerial vehicle response; In the response process, the system collects the device action feedback information in real time and compares it with the expected response state to identify whether the strategy is interrupted, invalid or timed out, if the strategy is invalid, the threat level update and strategy chain reconstruction are automatically triggered, the dynamic compression mechanism is used to preferentially call the device group with the lowest response cost, effectively controlling the response chain complexity, which can continuously monitor multiple targets in complex perimeter scenes and adaptively dispose the strategy, significantly improving the intelligent, collaborative and precision level of perimeter protection, effectively reducing the misjudgment rate and resource waste, and ensuring the continuity and effectiveness of the perimeter safety operation. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 The flowchart of the adaptive perimeter protection linkage disposal method of the present application.

[0049] Figure 2 FIG. 1 is a structural schematic diagram of an adaptive perimeter defense linkage handling system according to the present application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0051] Embodiment 1: As shown in the following table, an adaptive perimeter defense linkage handling method includes the following steps: Figure 1

[0052] Real-time collection of data output by infrared transceiver devices, laser radars, video monitoring devices and microwave sensors deployed in the perimeter defense area, format conversion and space coordinate mapping based on a unified time reference, generation of a fused multi-target perception frame sequence;

[0053] Extraction of dynamic behavior features of each target object from the fused multi-target perception frame sequence, construction of a corresponding feature vector, and generation of threat level trend information of the target object according to a preset rule;

[0054] Combination of the threat level trend of each target object and the real-time available state of the response device within the perimeter, establishment of a response resource scheduling diagram, and identification of whether there is a response conflict or device allocation conflict between target objects;

[0055] According to the response resource scheduling diagram, a matching response strategy sequence is generated according to the target priority, and the response device is sequentially scheduled to perform the corresponding task;

[0056] During the execution of the response device, action feedback information of each response device is collected, including whether the task is completed, whether the target object is affected, and the occupation time length of the device, and the action feedback information is compared with the strategy sequence to identify whether the current handling strategy is invalid or interrupted;

[0057] When it is identified that the strategy is invalid or the response is not closed-loop, the threat level of the target object is re-evaluated, and the response strategy is adjusted according to the updated priority and response device state.

[0058] Step 1, real-time collection of data output by infrared transceiver devices, laser radars, video monitoring devices and microwave sensors deployed in the perimeter defense area, format conversion and space coordinate mapping based on a unified time reference, generation of a fused multi-target perception frame sequence, the specific steps including:

[0059] ​In the set perimeter protection area, four types of sensing devices, infrared beam device, laser radar, video monitoring equipment and microwave sensor, are pre-deployed to obtain perimeter target data of different physical characteristic dimensions, wherein the infrared beam device is used to detect the intrusion interruption behavior, the laser radar is used to obtain two-dimensional or three-dimensional space scanning point cloud information, the video monitoring equipment is used to capture continuous image sequence, and the microwave sensor is used to sense the motion direction and speed characteristics of the target. The above sensing devices output raw data at fixed time intervals or in trigger mode. The data output format of each device may be different according to the hardware model or communication protocol, and a unified time reference needs to be configured as the reference standard for data collection and synchronization of all devices. The time reference is usually provided by the system master server in the form of timestamp signal broadcast to the control module of each sensing device. During data collection, all sensing devices are attached with system unified timestamp for subsequent time alignment operation.

[0060] The raw sensing data from the above four types of devices are received and sequentially subjected to timestamp verification and alignment processing. The time alignment operation includes: according to the timestamp attached to the device, the non-synchronous data frames exceeding the tolerance threshold are removed, and interpolation completion or nearest neighbor matching is performed on the slightly time-offset data to ensure that the data of all devices can be processed cooperatively within the same logical time window.

[0061] After time alignment, format standardization processing is performed on each type of sensing data to eliminate the difference in device output format. Specifically, the infrared beam data is parsed into interruption state and time identification pair; the laser radar point cloud is resampled and uniformly encoded into sparse matrix format; the video image sequence is compressed into multi-frame target detection result; and the microwave sensor data is structured into speed vector and target identity label. The above format standardization operation is based on the preset structured data template for subsequent unified analysis by the fusion processing module. Based on the preset spatial coordinate reference model, the standardized data is uniformly mapped to the same three-dimensional coordinate system. The spatial coordinate reference model is based on the building structure diagram or geographic coordinate system of the protection area, combined with the relative orientation parameters of the device installation position, to realize the synchronous projection of data in the spatial dimension. For example, the laser radar and video monitoring data are mapped to the spatial reference plane through device calibration parameters, and the sensing results of the microwave sensor and the infrared beam device are positioned in the spatial range through the physical coverage area.

[0062] After all the perception data have been coordinate-projected, target fusion and redundancy removal are performed. Because different devices may detect the same target at similar times and in similar spatial regions, spatial and temporal overlap thresholds must be set to merge duplicate detections. During the fusion process, the target's spatial position, motion direction, and state attributes are weightedly estimated based on the confidence weights of the target in multiple perception sources to construct a unified multi-target structured perception frame.

[0063] Finally, the output multi-target perception frame sequence is the fusion perception result of all valid targets in a unified coordinate system in each time window.

[0064] According to the confidence weights of the target in multiple perception sources, the spatial position, motion direction and state attributes of the target are weightedly estimated to construct a unified multi-target structured perception frame. The specific process is as follows:

[0065] In a unified coordinate system, a spatial overlap comparison operation is performed on the target lists output by all sensing devices in the current time window to determine whether there are multiple source detection records pointing to the same target. The judgment criteria are: if the Euclidean distance between two or more targets in space is less than the set spatial fusion threshold, and the angle between their motion directions or velocity vectors is less than the set direction consistency threshold, then it is considered a repeated detection item of the same target;

[0066] After identifying a repeated detection target, the data from each sensor device about the target is fused to form a single unified target representation. This fusion process relies on a confidence weight mechanism. Each type of sensor device in the system has a predefined confidence weight, which reflects the device's confidence in identifying a specific target attribute. For example:

[0067] The confidence weight of the target location for video surveillance equipment can be set to 0.6; LiDAR has higher accuracy in spatial positioning, and its confidence weight can be set to 0.8; Microwave sensors are stronger in motion speed recognition, and their confidence weight in the speed dimension is set to 0.7; Infrared counter-radiation devices only provide occlusion events and do not participate in position or speed fusion, but can provide trigger-assisted information on the presence of the target;

[0068] Taking the spatial position of a target as an example, assuming that the video surveillance device detects target A at position P1 and the lidar detects it at position P2, these two position values ​​and their corresponding confidence weights will be weighted and fused. The calculation method is: multiply each position value by its corresponding weight, then add all the results and divide them by the sum of all participating weights to obtain the fused position estimate. In other words, the fused position is equal to the sum of the products of each device's detected position and its corresponding confidence weight, divided by the sum of all weights;

[0069] Similarly, the target's speed and direction are weighted and estimated using the same logic. The speed value and direction vector provided by each sensing device are processed according to their confidence weight in speed recognition. For example, if the lidar provides a speed of 1.2 meters per second and the microwave sensor provides a speed of 1.0 meters per second, a weighted average is performed according to their respective confidence weights (e.g., 0.6 and 0.7) to ultimately produce the fused speed value. Furthermore, regarding state attribute fusion, the system establishes a confidence priority based on each device's recognition of the target's activity state (e.g., stationary, moving, occluded, etc.). When multiple devices identify inconsistent states, the state identified by the device with the higher weight is prioritized. If the difference is within the error threshold, the states are merged, for example, "stationary - slow movement" is merged into "slow-speed wandering."

[0070] All target records constitute a multi-target structured perception frame in the current time window, which will serve as input for subsequent threat level assessment and response strategy generation. By weightedly fusing the confidence of repeated targets, it not only retains the complementary advantages of multi-source perception, but also avoids the interference of repeated or conflicting information, ensuring high precision in actual disposal.

[0071] After the fused multi-target perception frame sequence is generated, the dynamic behavior characteristics of the target objects contained in each perception frame are extracted one by one, and the feature vector of the target object is constructed based on this, which is used for subsequent threat level identification and trend judgment;

[0072] Extract the dynamic behavior characteristics of each target object from the fused multi-target perception frame sequence, construct the corresponding feature vector, and generate the threat level trend information of the target object according to the preset rules. The specific process includes:

[0073] Dynamic behavior characteristics mainly include the following four indicators: Movement speed refers to the rate of change of the target object's position within the current time window. The instantaneous speed of the target is calculated based on the difference in the target object's position coordinates in two consecutive perception frames, combined with the time interval. A larger speed value generally indicates a more active target behavior or a risk of crossing.

[0074] The rate of change of movement direction indicates the degree of change in the target object's movement direction in consecutive time frames. By calculating the change in the angle between the target's movement direction vectors in two time periods, the stability of its movement path can be determined. Targets with drastic changes in direction may be in a state of wandering, reconnaissance, or exploration.

[0075] The dwell duration measures the cumulative time a target remains relatively stationary or at a low speed within a certain local area. If a target remains for a long time near a critical area or in a blind spot, it will be considered a potential threat. This indicator is calculated by accumulating the length of time the target's speed is less than a set threshold across multiple consecutive perception frames.

[0076] The shortest path length to the critical area refers to the shortest path from the target's current location to the set critical area (such as entrances and exits, weak points in the perimeter, resource-intensive areas, etc.). Based on the target position in the current perception frame, the spatial map of the deployment area is called to calculate the shortest travel distance between the target and the critical area and dynamically update it. This indicator is used to reflect whether the target is close to a highly sensitive area. The closer the distance, the higher the warning level.

[0077] The above four types of indicators constitute the feature vector of the target object. The feature vector is compared with various typical behavior patterns that have been calibrated in the historical behavior template library to perform similarity matching operations. The matching process compares the difference in each dimension between the current target feature vector and each historical template vector, and calculates the overall difference degree of all templates in turn. The difference degree is calculated as follows: the difference amplitude is calculated for each item in the four feature dimensions, and each difference is multiplied by the corresponding weight factor and then summed up to finally obtain the total difference. The historical template with the smallest difference is selected from all templates as the behavior reference closest to the current target, and The template is assigned a corresponding threat level, which is generally divided into five levels (such as low threat, suspicious, medium threat, high risk, and extremely high risk). Each template has been calibrated in the library. To improve the temporal continuity of judgment and the ability to identify dynamic trends, a sliding time window mechanism is used to collect the threat level results of the target object in several consecutive perception frames to form a time series of the target's threat level. Then, by determining the level change direction of the time series (such as continuous increase, fluctuation, decrease, etc.), the target object's threat level trend information is generated and used as one of the input bases for the policy scheduling module;

[0078] Specifically, the feature vector of the target object is calculated with each template vector in the historical behavior template in turn. The similarity calculation method adopts the Euclidean distance principle. The system calculates the difference between the target object feature vector and the template vector in each dimension, squares the differences in the four dimensions and sums them, and finally takes the square root as the distance value between the target and the current template. All templates execute this step to generate a set of corresponding distance value lists. Then, among all matching results, the historical behavior template with the smallest distance value is selected. It is considered that this template is most similar to the behavioral characteristics of the current target object, and the threat level bound to this template is used as the threat identification result of the target object in the current time window, and is recorded as the threat level result of the object in the current time window. In order to obtain a more timely judgment result, the threat level results of the target object in several consecutive time windows are continuously collected to form a threat level time series. This time series is generally maintained in the form of a sliding time window, such as the level results of 5 frames in the past 10 seconds.

[0079] Analyze the changing trends of the threat level time series. If the threat level in the sequence shows an upward trend (e.g., the level value gradually increases or frequently jumps to a higher level), the target threat level is considered to be on the rise. If the level remains stable, it is considered to be a neutral trend. If the level gradually decreases or remains at a low level, it is considered to be a safe trend.

[0080] In actual deployment, this threat level trend information can be combined with the spatial relative relationship and behavioral synchronization between multiple targets to further determine whether it is a group intrusion or coordinated behavior, thereby enhancing the perimeter protection system's response capability to complex threats.

[0081] To efficiently handle multiple targets and optimally allocate resources, it is necessary to construct a response resource scheduling diagram based on the acquired threat level trend information of the target objects and the current operating status of each response device deployed in the perimeter area. In addition, it is necessary to identify possible response conflicts and equipment allocation conflicts during the scheduling process.

[0082] Combine the threat level trends of each target object and the real-time availability of response equipment within the perimeter to establish a response resource scheduling diagram and identify whether there are response conflicts or equipment allocation conflicts between target objects. The specific process includes:

[0083] Collect the current available status information of each response device deployed in the perimeter area. The available status information includes whether the device is in an idle state, the physical location information of the device (which can be expressed in spatial coordinates), and the expected response delay required to get from the current device location to the target object location. The expected response delay can be calculated by combining the device's response action preparation time with the length of its moving path. For example, for devices with mobile capabilities, such as drones and controllable lighting devices, their moving speed and path obstacles need to be considered; for fixed devices such as speakers or cameras, the response delay mainly considers the startup time or switching time. With each target object and each idle response device as the endpoint, a bipartite graph structure is constructed, that is, the target object set and the response device set are the two endpoint sets of the graph, and a connection relationship between the two types of nodes is established. For each target object, a connection edge is established with all available response devices, and the response cost value is calculated for each connection edge;

[0084] The response cost is determined by three components:

[0085] The spatial distance factor is used to measure the physical distance between the responding device and the target object. It is obtained by calculating the path length between the responding device's current position and the target object's current position based on the spatial coordinates. If the path needs to take obstacles into account, the moving distance after path planning is used; if it is a line-of-sight device, the three-dimensional straight-line distance can be used.

[0086] The task load factor represents the device's current resource usage and reflects its ability to handle new tasks. It is obtained by collecting the device's resource usage within the current time window, including operating power consumption, current occupancy time, and task processing queue length. After normalization, a load factor between 0 and 1 is obtained. A larger value indicates a heavier device load.

[0087] The adaptability factor indicates whether the responding device is capable of responding to the target's current threat level. It is obtained by comparing the degree of match between the actions supported by the device and the actions currently required by the target. For example, if the target is assessed as high risk (requiring warning, tracking, and intervention), and the device only has voice warning capabilities, the factor is set to a low value. If the device can perform all required actions, it is set to a high value. The rules can be set as follows: full match is set to 1, partial match is set to 0.5, and no match is set to 0.

[0088] In order to form a unified evaluation standard for the above three factors, a preset weight coefficient needs to be assigned to each factor. The weight coefficient is set according to the usage scenario or policy formulation rules. For example, in the handling of high-risk targets, the weight of the adaptation relationship factor can be set to the highest, followed by response delay, and then equipment load. When dealing with high-risk targets, priority should be given to whether the equipment has the corresponding handling capabilities, so the weight of the adaptation relationship factor is set to the maximum value; in resource constraints or task peak periods, the current load of the equipment must be taken into account, and the weight of the task load factor is moderate; if the scenario response has high timeliness requirements, the weight of the spatial distance factor is moderate or high. Multiply the three factors by the corresponding weight coefficients and sum them to obtain the response cost value of the current connection edge;

[0089] For example, assume the spatial path distance between T1 and D1 is 60 meters, and the maximum scenario response radius is 100 meters. Therefore, their normalized spatial distance factor is 0.6 (indicating a distance close to the maximum value). Assume D1 already has two tasks in its current task queue and a resource load rate of 80%. After normalization, its task load factor is 0.8. Assume D1 supports all types of actions required by the current target object, so its adaptation factor is 1. Substituting these three factors into the weighted response cost framework, the specific steps are: multiplying the spatial distance factor of 0.6 by its weight of 0.3, the result is 0.18; multiplying the task load factor of 0.8 by its weight of 0.2, the result is 0.16; and multiplying the adaptation factor of 1 by its weight of 0.5, the result is 0.5. Summing these three results, the response cost value for the link T1–D1 is 0.84. The lower the value, the more suitable the device is for responding to the target in the current situation. If there are multiple candidate devices, the one with the lowest value is selected as the priority response device.

[0090] After calculating the cost values ​​of all connected edges, a response cost graph between the target object and the responding device is obtained. Based on this, the minimum conflict matching algorithm is executed to construct a minimum conflict matching graph. This graph is used to identify the following two types of conflicts:

[0091] Response conflict refers to the situation where multiple target objects need to call the same response device in the same time window, and their expected response time periods overlap; device allocation conflict refers to the situation where the same device cannot meet the needs of multiple target objects at the same time due to task adaptation or delay exceeding the limit.

[0092] By identifying and recording the above conflicts, we can ensure reasonable resource allocation and avoid policy execution failures caused by resource conflicts.

[0093] Once the response resource scheduling diagram is constructed, the response strategy generation and equipment scheduling tasks will be executed based on the response resource scheduling diagram, combined with the threat level trend information of each target object and the schedulable status of the response equipment within the perimeter;

[0094] Based on the response resource scheduling graph, a matching response strategy sequence is generated according to the target priority, and the response devices are scheduled to perform the corresponding tasks in sequence. The specific process includes:

[0095] Based on the connection relationship between target objects and available response devices in the response resource scheduling graph, as well as the response cost value of each connection edge, the priority response device set for each target object is determined. In the case where multiple target objects share the same response device, the device is preferentially allocated to the target object with a higher threat level trend and a stronger response timeliness requirement to avoid resource conflicts and scheduling delays.

[0096] For each target object to be handled, the priority ranking is determined based on its threat level trend information. The basis for determining the priority is as follows:

[0097] If the threat level trend of a target object continues to rise within the current time window, it will be ranked before all static or declining trend objects;

[0098] If multiple objects have the same threat level, the object closer to the critical area will be dealt with first;

[0099] If there is still a conflict, the decision is made based on the order of the timestamps of the sensed frames, with the first one discovered taking precedence;

[0100] The generated target object priority list is used for queue allocation in subsequent response policies.

[0101] For each target object, the corresponding response action type is matched according to its threat level and behavioral characteristics, and a specific policy sequence is generated based on the current response device status. Policy action types include but are not limited to:

[0102] Voice warning action, for calling fixed broadcast equipment or directional speaker to play warning voice to target direction, for psychological intervention or preliminary deterrence to low to medium level target; regional lighting intervention, for controlling perimeter lighting equipment to quickly light up corresponding area, enhancing target exposure, while cooperating with video monitoring equipment to improve picture clarity; UAV intervention action, for sending low-altitude UAV equipped with voice broadcast and highlight light to the area above the target, for close-range tracking, warning or even blocking flight.

[0103] Set the following key parameters for each policy sequence:

[0104] Set the policy execution time window, that is, set the maximum tolerance execution time of each response action, for example, the voice warning action is completed within 5 seconds, and the image tracking holding time is not more than 15 seconds, and if the time exceeds, it is considered as a failure;

[0105] Set the device redundancy replacement path, that is, configure at least one standby device for each response action, the standby device is selected according to the suboptimal response value and has the same functional capability, if the main device cannot participate in execution due to task conflict or failure, switch to the redundant device

[0106] According to the above queue and policy sequence, send control instructions to the corresponding response device in turn, the instruction content includes: action type (such as warning, lighting, tracking, etc.), target object ID and coordinate, expected start execution time, policy number and time window, redundant device list, after dispatching, the system continuously monitors the state change of each response device, records its execution feedback.

[0107] In the process of response device execution, collect the action feedback information of each response device, including whether the task is completed, whether the target object is affected, and the occupation time of the device, and compare the action feedback information with the policy sequence to identify whether the current disposal policy is invalid or interrupted, the specific process includes:

[0108] In the process of response device executing the linkage task, the action feedback information generated is collected and discriminant analyzed in real time, the action feedback information includes three types of content: action completion state of response device, feedback behavior influence information and actual occupation time of device, and is compared with the preset task indicators in the response policy sequence respectively, to identify whether the current disposal policy appears invalid, interrupted or abnormal closed loop situation;

[0109] For the comparison of the action completion status, according to the feedback mark of each responding device after executing the task, confirm whether it has completed the linkage action specified in the strategy instruction, such as whether the voice playback is finished, whether the lighting is triggered, whether the image tracking is started, or whether the drone has reached the specified location. If the device does not return the completion status mark within the set time window, or the feedback information shows that the action execution is interrupted, the current response of the device is determined to be a response interruption and marked as an unclosed loop state; secondly, for the comparison of the feedback behavior impact information, the behavioral changes of the target object after the response action are synchronously obtained, including key behaviors such as position movement, direction deflection, and change in residence time. As an indicator, it is compared with the behavioral effect expected to be triggered by the response strategy. For example, if the voice expulsion strategy is implemented, the target object should show an evacuation trend; if the image locking strategy is implemented, the target should take evasive actions; if the target object does not produce the behavioral response expected by the strategy, then the corresponding strategy segment of the responding device is judged to be invalid response; thirdly, the actual device occupation time is compared, and the duration of the entire process from the start of task scheduling to the release of the device is recorded and compared with the time window upper limit set in the response strategy. If the occupation time exceeds the strategy time threshold and the expected behavioral response is not formed, then the task is deemed to have timed out and not closed;

[0110] After completing the comparison of the above three dimensions, based on the comparison results, determine whether the current response strategy segment has failure (i.e. invalid feedback), interruption (i.e. incomplete task) or timeout (i.e. closed-loop abnormality). If any of the conditions is met, the strategy reconstruction mechanism is triggered and the next stage of response strategy evaluation and scheduling adjustment process is entered. Through the above process, while ensuring the continuity of strategy execution, the judgment transparency of the linkage response process and the timeliness of strategy adjustment are enhanced, thereby achieving high reliability and high closed-loop rate response control of perimeter abnormal events.

[0111] When a strategy failure or an incomplete response is identified, the threat level of the target object is reassessed and the response strategy is adjusted based on the updated priority and the status of the response equipment. The specific process includes:

[0112] When the current response strategy is identified as failing, interrupted, or not closed-loop, the threat level of the corresponding target object must be immediately reassessed to ensure the accuracy and pertinence of the subsequent response strategy. Specifically, the continuous behavioral trajectory data of the target object after executing the response action is used as input to re-extract dynamic behavioral characteristic indicators, including but not limited to: updated movement speed value, frequency of change in movement direction, duration of current stop position, and shortest path length to key areas (such as fences, passages, or restricted areas);

[0113] The speed value is obtained by sampling the target position change rate at adjacent time points; the direction change frequency refers to the number of times the target's movement direction changes per unit time, reflecting the instability of the target's behavior; the dwell time refers to the continuous length of time the target stays at a fixed position, which is used to determine whether it has behavioral tendencies such as wandering and snooping; the shortest path length is obtained by calculating the Euclidean shortest path distance between the current target position and the preset key area in the coordinate space, which is used to assess the risk of approaching sensitive areas; after obtaining new behavioral features, the corresponding updated feature vector is constructed and matched with the original historical behavior template again for similarity, and behavioral sequence analysis is performed within the sliding time window to update the threat level label of the target object. On this basis, the current threat level trend curve is generated. The threat level trend curve is a level sequence arranged in chronological order, which is used to reflect the dynamic change trend of the target threat intensity over time;

[0114] Based on the updated threat level trend results, combined with the current response device's idle state, response load, physical location, and task occupancy records, the response strategy reconstruction process is triggered. To improve the efficiency of strategy reconstruction, a dynamic compression mechanism is introduced to optimize the resource call structure of the strategy chain. The core logic of this mechanism is as follows:

[0115] First, a set of response devices that are currently idle or can be quickly released is identified as a resource pool that can be scheduled during policy reconfiguration. Then, the response cost of each device relative to the target object is calculated within this resource pool. This value takes into account the spatial distance between the target and the device (the length of the physical response path), the device's current load (the remaining resource ratio), and the match between the device's capabilities and the target threat level. The lower the response cost, the more suitable the device is for quickly executing the new response task.

[0116] When reconstructing a policy chain, the device group with the lowest response cost is prioritized as the starting node of the new policy path, and a compressed policy chain is constructed based on the linkage capabilities between devices (for example, voice devices and lighting devices can be executed in parallel). The compression mechanism sets a control parameter: the maximum length of the policy chain, which is equal to the length of the original policy chain. This prevents the newly added policy chain from being too long, causing response delays or resource dispersion. If a closed-loop path cannot be constructed within this limit, the system activates the device substitution mechanism, automatically selecting alternative devices with similar response effects but slightly higher costs to construct a redundant response path;

[0117] Through the above process, the target threat level can be re-evaluated and the linkage strategy can be reconstructed as soon as the identification strategy fails. It can also ensure that the reconstructed strategy has the characteristics of rapid response, controllable resource usage, and closed-loop linkage action, thereby improving the intelligent response capability and actual combat reliability of the entire perimeter protection system.

[0118] It should be noted that the thresholds involved in the embodiment can be determined according to specific scenarios and requirements.

[0119] The present invention builds a multi-source perception system by deploying infrared radiation devices, laser radars, video surveillance equipment, and microwave sensors in perimeter protection areas. This system collects the spatial position, motion trajectory, and behavioral status information of target objects in real time, and performs format conversion and spatial coordinate mapping on heterogeneous data using a unified time base to generate a structured fusion multi-target perception frame sequence, effectively improving the accuracy of target recognition and the consistency of behavioral analysis. On this basis, the present invention extracts dynamic behavioral characteristics such as the target object's movement speed, direction change rate, residence time, and the shortest path length between the target object and the key area, constructs feature vectors, and performs similarity matching with historical behavior templates to generate threat level trend information that can evolve over time, thereby achieving continuous perception and intelligent judgment of potential intrusion behaviors.

[0120] At the same time, a response resource scheduling graph with target objects and response devices as endpoints is constructed, comprehensively considering the spatial position, current load and adaptability of the response devices, calculating the response cost of the connection edge and identifying response conflicts, and generating the optimal strategy sequence through the graph matching algorithm to realize the dynamic linkage and closed-loop execution of multi-level disposal actions such as voice warnings, lighting intervention, image tracking and drone response; during the response process, the system collects device action feedback information in real time and compares it with the expected response status, identifies whether the strategy has interruption, invalidity or timeout problems, and automatically triggers the threat level update and strategy chain reconstruction if the strategy fails. A dynamic compression mechanism is used to prioritize the call of the device group with the lowest response cost, effectively controlling the complexity of the response chain, and can continuously monitor and adaptively handle multiple targets in complex perimeter scenarios, significantly improving the intelligence, coordination and precision of perimeter protection, effectively reducing the misjudgment rate and resource waste, and ensuring the continuity and effectiveness of perimeter safety operation.

[0121] Example 2: An adaptive perimeter protection linkage disposal system, such as Figure 2 As shown, specifically including:

[0122] The data acquisition module is used to collect data output by infrared radiation devices, laser radars, video surveillance equipment, and microwave sensors deployed within the perimeter protection area in real time, perform format conversion and spatial coordinate mapping based on a unified time base, and generate a fused multi-target perception frame sequence;

[0123] The response conflict determination module is used to extract the dynamic behavior characteristics of each target object from the fused multi-target perception frame sequence, construct the corresponding feature vector, and generate the threat level trend information of the target object according to the preset rules. The response resource scheduling diagram is established by combining the threat level trend of each target object and the real-time availability status of the response equipment within the perimeter to identify whether there is a response conflict or equipment allocation conflict between the target objects;

[0124] The disposal strategy identification module generates a matching response strategy sequence based on the target priority according to the response resource scheduling diagram, and sequentially schedules the response devices to perform the corresponding tasks. During the execution of the response devices, the module collects the action feedback information of each response device, including whether the task is completed, whether it has an impact on the target object, and the device's occupancy time. The module then compares the action feedback information with the strategy sequence to identify whether the current disposal strategy has failed or been interrupted.

[0125] Adjust the response module. When it identifies that the strategy has failed or the response is not closed-loop, it re-evaluates the threat level of the target object and adjusts the response strategy based on the updated priority and response device status.

[0126] The above formulas are all dimensionless and calculated numerically. Specific dimension removal can be achieved by various means such as standardization, which will not be elaborated here. The formula is a formula obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0127] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, ATA hard drives, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state ATA hard drive.

[0128] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0129] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0130] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0131] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0132] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0133] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. An adaptive perimeter protection linkage processing method, characterized in that: The steps include: Collect data from infrared radiation devices, laser radars, video surveillance equipment, and microwave sensors deployed within the perimeter protection area in real time, convert format and map spatial coordinates based on a unified time base, and generate a fused multi-target perception frame sequence. Extract the dynamic behavior characteristics of each target object from the fused multi-target perception frame sequence, construct the corresponding feature vector, and generate the threat level trend information of the target object according to the preset rules; Combine the threat level trend of each target object and the real-time availability of response equipment within the perimeter to establish a response resource scheduling diagram and identify whether there are response conflicts or equipment allocation conflicts between target objects; According to the response resource scheduling diagram, a matching response strategy sequence is generated according to the target priority, and the response devices are scheduled to perform the corresponding tasks in sequence; During the execution of the response device, the action feedback information of each response device is collected, including whether the task is completed, whether the target object is affected, and the occupancy time of the device. The action feedback information is compared with the strategy sequence to identify whether the current disposal strategy is invalid or interrupted; When identifying strategy failure or unclosed-loop response, reassess the threat level of the target object and adjust the response strategy based on the updated priority and response equipment status; The feature vector performs similarity matching based on historical behavior templates to generate threat level trend information. The specific process is as follows: Constructing the extracted dynamic behavior features into feature vectors; Perform similarity matching between the feature vector and the historical behavior template vector, and use the Euclidean distance to calculate the matching value of each historical behavior template; The threat level corresponding to the historical behavior template with the smallest distance is selected as the threat level result in the current time window; A threat level sequence is formed by sliding a time window, and threat level trend information of the target object is generated based on the level change trend; Combine the threat level trend of each target object and the real-time availability of response equipment within the perimeter to establish a response resource scheduling diagram to identify whether there are response conflicts or equipment allocation conflicts between target objects. The specific process is as follows: Collect the current idle status, physical location, and expected response delay of each responding device; Construct a bipartite graph structure with target objects and device nodes as endpoints, establish connections between each target object and all idle responding devices, and calculate the response cost of each connection edge; The response cost is determined based on the spatial distance between the target object and the device, the current task load of the device, and the compatibility between the device capability and the target threat level. Spatial distance represents the physical cost required for the device to respond, task load represents the load or power consumption level of the device's currently occupied resources, and the adaptation relationship indicates whether the device has the execution capability under the current policy. The spatial distance, task load, and adaptation relationship are weighted according to the preset weight coefficient to obtain the response cost of the connection edge. A minimum conflict matching graph is constructed based on the response cost values ​​of all connected edges to identify response conflicts and device allocation conflicts.

2. The adaptive perimeter protection linkage treatment method according to claim 1, characterized in that: Based on a unified time base, format conversion and spatial coordinate mapping are performed to generate a fused multi-target perception frame sequence. The specific process is as follows: Obtaining raw perception data through infrared radiation devices, laser radars, video surveillance equipment, and microwave sensors; Use a unified time base to align timestamps and standardize the format of heterogeneous data structures; Based on the preset spatial coordinate reference model, various types of perception data are mapped to the same coordinate system, and repeated targets are fused and de-redundant to generate structured perception frames of various targets.

3. The adaptive perimeter protection linkage treatment method according to claim 2, characterized in that: Dynamic behavior features include the target object's movement speed, rate of change of movement direction, duration of stay, and shortest path length to key areas. Feature vectors perform similarity matching based on historical behavior templates to generate threat level trend information.

4. The adaptive perimeter protection linkage treatment method according to claim 3, characterized in that: According to the response resource scheduling diagram, a matching response strategy sequence is generated according to the target priority, and the response devices are scheduled to perform the corresponding tasks in sequence. The specific process is as follows: The matching response strategy sequence includes voice warning actions, area lighting intervention, image tracking actions, and drone intervention actions. The queues are sorted according to preset priority rules, and the strategy execution timeout window and device redundant alternative paths are set.

5. The adaptive perimeter protection linkage treatment method according to claim 4, characterized in that: The action feedback information is compared with the response strategy sequence to identify whether the current disposal strategy is invalid or interrupted. The specific process is as follows: Compare the action completion status of the responding device with the expected execution identifier of the corresponding action in the policy instruction. If the action is not completed, it is determined that the device response is interrupted; Compare the feedback behavior impact information of the response device with the behavior status of the target object. If the target does not produce the expected evacuation or movement response, the response is determined to be invalid. Compare the actual occupancy time of the responding device with the time window set by the policy. If it exceeds the upper limit, it is determined to be a handling timeout; Based on the comparison results, determine whether the current response strategy is invalid, interrupted or in an abnormal closed-loop state.

6. The adaptive perimeter protection linkage treatment method according to claim 5, characterized in that: When a strategy failure or an unclosed-loop response is identified, the threat level of the target object is reassessed. The specific process is as follows: Based on the new behavioral feature changes of the target object after the response action, the speed, direction and stay behavior indicators are re-extracted; Update the threat trend curve of the target object based on the feedback results of the unclosed-loop disposal; Restructure response strategies and adjust priorities based on updated threat levels and device status.

7. The adaptive perimeter protection linkage treatment method according to claim 6, characterized in that: A dynamic compression mechanism is introduced when reconstructing the response strategy, giving priority to calling the device group with the lowest response cost among the remaining available resources, and controlling the length of the readjusted strategy chain to not exceed the set upper limit of the initial strategy chain.

8. An adaptive perimeter protection linkage processing system, used to implement an adaptive perimeter protection linkage processing method according to any one of claims 1 to 7, characterized in that: include: The data acquisition module is used to collect data output by infrared radiation devices, laser radars, video surveillance equipment, and microwave sensors deployed within the perimeter protection area in real time, perform format conversion and spatial coordinate mapping based on a unified time base, and generate a fused multi-target perception frame sequence; The response conflict determination module is used to extract the dynamic behavior characteristics of each target object from the fused multi-target perception frame sequence, construct the corresponding feature vector, and generate the threat level trend information of the target object according to the preset rules. The response resource scheduling diagram is established by combining the threat level trend of each target object and the real-time availability status of the response equipment within the perimeter to identify whether there is a response conflict or equipment allocation conflict between the target objects; The disposal strategy identification module generates a matching response strategy sequence based on the target priority according to the response resource scheduling diagram, and sequentially schedules the response devices to perform the corresponding tasks. During the execution of the response devices, the module collects the action feedback information of each response device, including whether the task is completed, whether it has an impact on the target object, and the device's occupancy time. The module then compares the action feedback information with the strategy sequence to identify whether the current disposal strategy has failed or been interrupted. Adjust the response module. When it identifies that the strategy has failed or the response is not closed-loop, it re-evaluates the threat level of the target object and adjusts the response strategy based on the updated priority and response device status.

Images