A multi-source detection sensor layout intelligent optimization method and system

Abstract

This invention discloses an intelligent optimization method and system for multi-source detection sensor layout, belonging to the field of sensor deployment technology. This method overcomes the shortcomings of traditional methods in terms of insufficient 3D modeling accuracy by constructing a high-precision three-dimensional spatial model and explicitly defining sensor detection models and deployment constraints, providing a realistic and reliable physical environment constraint for optimization. Then, a quantity minimization optimization module is introduced to calculate the minimum number and cost required to meet the target coverage for a single sensor model, generating a single-type benchmark scheme. This effectively addresses the challenge of multi-objective trade-offs and provides a benchmark for hybrid deployment. Finally, using the lowest-cost single-type scheme as the initial solution, a strategy of gradually reducing the primary sensor model and supplementing other models is employed, combined with a modified genetic algorithm to search for a hybrid deployment scheme. This overcomes the limitations of a single type and utilizes global search capabilities to avoid getting trapped in local optima. This progressive approach significantly improves computational efficiency and reduces deployment costs while ensuring coverage performance.

Description

Technical Field

[0001] This invention belongs to the field of sensor deployment technology and relates to a method and system for intelligent optimization of multi-source detection sensor layout. Background Technology

[0002] In low-altitude security systems, the continuous monitoring and accurate identification of dynamic targets relies on ground-based distributed information processing and target recognition technologies. This technology requires the effective coordination of multiple types of detection nodes within a complex three-dimensional spatial environment to achieve real-time data acquisition, fusion, and processing.

[0003] Under three-dimensional geographical constraints, the optimal deployment of static multi-source detection sensors faces a multi-objective trade-off between coverage, resource cost, and terrain adaptability. Since the solution space for sensor deployment grows exponentially with the number of nodes, and 3D scene modeling needs to accurately depict terrain undulations, obstacle distribution, and various physical constraints, traditional optimization methods struggle to balance computational efficiency and coverage performance when dealing with large-scale complex scenes.

[0004] Existing sensor deployment methods are usually studied from the perspectives of coverage problems, comprehensive performance evaluation, or multi-objective optimization. However, in the 3D deployment scenario of static multi-source detection sensors, there are still problems such as unclear 3D environment modeling parameters, insufficiently specific line-of-sight occlusion calculation rules, unclear definition of sensor economics, and difficulty in reproducing the mixed search process of multiple models.

[0005] Therefore, there is a need for a method and system for intelligent optimization of static multi-source detection sensor layout that can clearly construct a mesh model in complex three-dimensional space, calculate obstacle occlusion, distinguish between optimization modules that minimize the number of sensors and optimization modules that maximize coverage, and perform multi-model economic search under the premise of satisfying the target coverage constraint. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention provides an intelligent optimization method and system for the layout of multi-source detection sensors. This solves the technical problem that existing technologies, when dealing with the optimization of static multi-source detection sensor layout under three-dimensional geographical constraints, suffer from the inherent NP-hardness of the problem, insufficient accuracy of three-dimensional modeling, difficulty in balancing multiple objectives, and susceptibility to local optima. As a result, it is difficult to balance computational efficiency, coverage performance, and deployment cost.

[0007] This invention is achieved through the following technical solution: A method for intelligent optimization of multi-source detection sensor layout includes the following steps: S1. Obtain the geographic information of the monitoring area, construct a three-dimensional spatial model, and define the sensor detection model and deployment constraints; S2. Based on the three-dimensional spatial model, calculate the cost-effectiveness of various types of sensors; for each sensor model, use the quantity minimization optimization module to calculate the minimum number and corresponding cost required to meet the target coverage under the deployment constraints, and generate a single-type benchmark solution. S3. Select the lowest cost among the single-type benchmark schemes as the initial optimal scheme; based on the sensor detection model, by gradually reducing the number of main models and supplementing other models, use a modified genetic algorithm to search for a hybrid deployment scheme in the three-dimensional space model that meets the coverage requirements and has a lower cost, and output the sensor layout result with the best economic efficiency.

[0008] Optionally, in step S1, constructing the three-dimensional spatial model specifically includes: The monitoring area is divided into cubic grid units according to the preset horizontal and vertical grid spacing, a three-dimensional coordinate system is established, and the grid unit center point coordinates are used for identification. Based on GIS and DEM data, determine whether there are entities within the grid cells. If they exist, mark them as obstacle voxels; otherwise, mark them as free space voxels and construct a three-dimensional KD tree spatial index structure.

[0009] Optionally, in step S1, the definition of the sensor's detection model specifically refers to: Using a Boolean detection model, the sensor's detection range is defined as having a sphere centered at the installation location and a detection radius of [missing information]. The region within the spherical space that satisfies the pitch angle constraint has an inverted cone-shaped blind zone at the center of the upper part and an inverted cone-shaped blind zone at the center of the lower part. To determine the sensor's ability to detect target voxels, the following conditions must be met simultaneously: the distance between the target voxel and the sensor is less than or equal to the detection radius, and the pitch angle of the target voxel relative to the sensor is within the effective pitch angle range. Within, and the line-of-sight path between the sensor and the target voxel does not contain any grid cells marked as obstacles; The semi-cone angle of the inverted cone-shaped blind zone directly above The semi-cone angle of the inverted cone-shaped blind spot directly below They respectively satisfy: ; ; in, For the first The maximum pitch angle of this type of sensor, For the first The minimum pitch angle of this type of sensor; The process of determining whether the line-of-sight path between the sensor and the target voxel contains obstacles specifically includes: Within the sensor detection enclosure, a ray direction is generated according to a preset azimuth angle sampling step size and a preset elevation angle sampling step size; The three-dimensional Bresenham voxel line algorithm is used to obtain the voxel sequence through which the ray passes for each ray; The voxel attributes are determined sequentially along the ray direction. When a voxel is a free space voxel, it is marked as a visible voxel. When an obstacle voxel is hit for the first time, the marking of subsequent voxels for that ray is terminated. After all ray processing is completed, a three-dimensional neighborhood expansion with a preset radius is performed on the visible voxel set, and the obstacle voxels are forcibly cleared to zero.

[0010] Optionally, in step S2, the cost-effectiveness is the ratio of the theoretical three-dimensional coverage volume of the sensor to its unit cost; wherein, the first Theoretical three-dimensional coverage volume of sensor-like devices satisfy: ; No. Cost-effectiveness of similar sensors satisfy:

[0011] in, For the detection radius, and These are the minimum pitch angle and the maximum pitch angle, respectively. For a single cost.

[0012] Optionally, in step S2, the quantity minimization optimization module specifically includes: First, a greedy algorithm is used to progressively select the location that can bring the maximum coverage gain from the candidate points in the deployable area, so as to obtain an initial estimate of the number of sensors that meets the target coverage requirements. Then, starting from the initial sensor number estimate, the coverage maximization optimization module is used for verification, and the sensor number is iteratively reduced until the minimum number of sensors that meet the target coverage requirement is found. The coverage maximization optimization module specifically includes: The internal deployment area is divided into K key regions using the K-means++ clustering algorithm, where K is the number of sensors to be deployed. Candidate deployment points that meet the deployment constraints are selected within each key region. An improved genetic algorithm is used to find the layout scheme with the highest coverage given the sensor type and the number of sensors. If the obtained coverage rate reaches the target coverage rate, the number of sensors continues to decrease; otherwise, it returns to the previous order of magnitude as the minimum number of sensors required to meet the target coverage rate requirement.

[0013] Optionally, the improved genetic algorithm specifically includes: Chromosome coding uses a one-dimensional integer coordinate sequence of region-candidate point, with each gene representing the index of the selected candidate point in the corresponding key region; In genetic operations, a combination of crossover and mutation strategies is used, and an adaptive parameter control mechanism is employed to make the crossover probability decrease with the number of iterations and the mutation probability increase with the number of iterations. When there is no improvement after consecutive preset algebras, a local search optimization mechanism is triggered to conduct in-depth exploration of the current optimal solution.

[0014] Optionally, in step S3, a modified genetic algorithm is used to search for a hybrid deployment scheme, specifically including: Set the decreasing step size of the main model sensor, gradually reduce the number of main models, and calculate the upper limit of the number of auxiliary models that can be replenished based on the remaining cost budget. For each candidate quantity vector, a fixed quantity vector correction genetic algorithm is called to search for a specific deployment location that satisfies the quantity constraint; By employing a performance optimization strategy based on infeasible solution elimination, invalid combinations that do not meet coverage constraints can be quickly eliminated. If a new solution that meets the target coverage and has a lower cost than the current optimal solution is found, the current optimal solution is updated and the cost continues to decrease; otherwise, the search is terminated and the sensor layout result with the best economic efficiency is output.

[0015] Optionally, the performance optimization strategy based on eliminating infeasible solutions includes the following two cases: Scenario 1: Analyze the patterns of verified infeasible solutions. If a solution is infeasible due to an insufficient number of main models, then directly eliminate all solutions with even fewer main models and unchanged or fewer other models. Scenario 2: If the detection radius of the supplementary auxiliary sensor is smaller than that of the main sensor and the pitch angle coverage is not greater than that of the main sensor, and the total number of sensors in the hybrid scheme is not greater than the total number of the verified infeasible schemes, then the scheme will be directly eliminated.

[0016] A static multi-source detection sensor layout intelligent optimization system, used to implement the above-mentioned multi-source detection sensor layout intelligent optimization method, includes: The preprocessing module is used to acquire geographic information of the monitoring area, construct a three-dimensional spatial model, and define the sensor detection model and deployment constraints. The optimization calculation module is used to calculate the cost-effectiveness of various sensors based on the three-dimensional spatial model. For each sensor model, the quantity minimization optimization module is used to calculate the minimum number required to meet the target coverage and the corresponding cost, and generate a single-type benchmark scheme. The output module is used to select the lowest cost among the single-type benchmark schemes as the initial optimal scheme. By gradually reducing the number of main models and supplementing other models, a modified genetic algorithm is used to search for a hybrid deployment scheme that meets the coverage requirements and has a lower cost, and outputs the sensor layout result with the best economic efficiency.

[0017] A computer device includes a memory, a processor, and a computer program stored in the memory, wherein when the processor executes the computer program, it implements the steps of the above-described method.

[0018] Compared with the prior art, the present invention has the following beneficial technical effects: This invention discloses an intelligent optimization method for multi-source detection sensor layout. First, it constructs a high-precision 3D spatial model and clearly defines the sensor detection model and deployment constraints, fundamentally solving the problem of insufficient 3D modeling accuracy in traditional methods and providing realistic and reliable physical environment constraints for subsequent optimization. Second, it introduces a quantity minimization optimization module to calculate the minimum number and cost required to meet the target coverage for a single sensor model, generating a single-type baseline scheme. This effectively addresses the challenge of balancing multiple objectives and provides a reliable cost and performance benchmark for hybrid deployment. Finally, using the lowest-cost single-type scheme as the initial optimal solution, it employs a strategy of gradually reducing the number of primary models and supplementing with other models, combined with a modified genetic algorithm to search for hybrid deployment schemes in the 3D spatial model. This not only overcomes the limitations of single-type deployment but also effectively avoids the problem of getting trapped in local optima through the global search capability of the genetic algorithm. The entire process, from high-precision modeling to single-type benchmark generation and then to multi-source hybrid optimization, progresses step by step, significantly improving computational efficiency and reducing deployment costs while ensuring coverage performance, achieving an effective balance between computational efficiency, coverage performance, and deployment cost. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the intelligent optimization method for static multi-source detection sensor layout in an embodiment of the present invention. Figure 2 This is a schematic diagram of the overall layout optimization process in an embodiment of the present invention; Figure 3 This is a flowchart of the intelligent optimization algorithm for the layout of static multi-source detection sensors in an embodiment of the present invention; Figure 4 This is a flowchart of the quantity minimization optimization module in an embodiment of the present invention; Figure 5 This is a flowchart of the coverage maximization optimization module in an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of the intelligent optimization system for static multi-source detection sensor layout in an embodiment of the present invention. Specific implementation methods To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0024] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0025] The present invention will now be described in further detail with reference to the accompanying drawings.

[0026] Example 1 To address the combinatorial optimization problem under complex three-dimensional spatial constraints, this embodiment proposes an intelligent optimization method for the layout of multi-source detection sensors. It adopts a hierarchical and modular algorithm architecture and supports three application modes: maximizing coverage, minimizing the number of sensors, and optimizing for economic efficiency.

[0027] like Figure 1 As shown in the figure, this embodiment discloses a method for intelligent optimization of multi-source detection sensor layout, including the following steps.

[0028] S1. Layout optimization preprocessing: Obtain the geographic information of the monitoring area, construct a three-dimensional spatial model, and define the sensor detection model and deployment constraints.

[0029] S2. Single-type benchmark scheme generation: Based on the three-dimensional spatial model, calculate the cost-effectiveness of various types of sensors; for each sensor model, use the quantity minimization optimization module to calculate the minimum number and corresponding cost required to meet the target coverage under the deployment constraints, and generate a single-type benchmark scheme.

[0030] S3. Hybrid Deployment Scheme Search: Select the lowest cost among the single-type benchmark schemes as the initial optimal scheme; based on the sensor detection model, by gradually reducing the number of main models and supplementing other models, use a modified genetic algorithm to search for a hybrid deployment scheme in the three-dimensional space model that meets the coverage requirements and has a lower cost, and output the sensor layout result with the best economic efficiency.

[0031] Specifically, the layout optimization preprocessing in step S1 includes map modeling, obstacle modeling, sensor detection model construction, ray projection coverage calculation, and deployment constraint setting.

[0032] In one specific implementation, the three-dimensional spatial model is configured according to the horizontal grid spacing. , and vertical grid spacing Discretize to form a voxel set Each voxel has its center point coordinates. As a unique identifier, the horizontal coordinate range of the monitoring area is... The vertical coordinate range is ,in This is the lowest ground elevation in the region. This is the preset maximum monitoring height.

[0033] If a voxel is located inside a terrain or building entity, it is marked as an obstacle voxel; otherwise, it is marked as a free space voxel.

[0034] In this embodiment, the horizontal scale of the monitoring area is 30 km. 30km, with a horizontal grid spacing of 300m, a vertical grid spacing of 10m, and 100 horizontal grid cells. The vertical grid number is 100, and the target height ranges from 30m to 1000m. The above values ​​are only used to illustrate the specific implementation process of this embodiment and are not intended to limit the invention.

[0035] In this embodiment, the latitude of the center of the monitoring area is taken. Calculate the conversion factor between latitude and longitude and the actual distance: ; ; in, This indicates the number of kilometers per degree of longitude. This represents the number of kilometers per degree of latitude. The number of grid cells for each axis is calculated as follows: ; ; ; The coordinate vectors for each axis are generated as follows: ; ; ; The obstacle modeling specifically involves: obtaining the location of each horizontal grid based on GIS and DEM data using bilinear interpolation. ground elevation at the location For each cell in the 3D mesh Perform the following attribute identification: ; Grid cells marked as obstacles are not included in coverage statistics; those marked as free space and located within the preset target height range are excluded. The grid cells within the target area are used as the target voxels to be covered. After marking, the center coordinates of all obstacle grid cells are used to construct a 3D KD tree spatial index structure.

[0036] The total number of free space voxels within the target height range. : ; in This is an indicator function; it takes the value 1 if the condition within the parentheses is met, and 0 otherwise.

[0037] Extract the voxel mesh index coordinates of all objects marked as obstacles, and construct a 3D KD tree spatial index: ; KD-trees employ a standard recursive median partitioning algorithm, alternating between selections... , , The nodes are constructed using three dimensions, allowing for the creation of any query point. The nearest neighbor search is completed in time complexity, where The total number of obstacle elements.

[0038] In step S1, the sensor's detection model is defined as follows: a Boolean detection model is adopted, and the sensor's detection range is set as the area within a spherical space centered on the installation position that satisfies the pitch angle constraint. The central area directly above and the central area directly below each have an inverted cone-shaped blind zone. The sensor's ability to detect a target point is determined by simultaneously satisfying the following conditions: the distance between the target point and the sensor is less than or equal to the detection radius; the relative pitch angle of the target point is within the effective detection range; and the line-of-sight path between the sensor and the target point does not contain any grid cells marked as obstacles.

[0039] Specifically, for the first A type of sensor, whose detection radius is... The effective detection range of the pitch angle is The distance between the target voxel and the sensor is The relative pitch angle is The detectable conditions are: ; ; Furthermore, the line-of-sight path between the sensor and the target voxel was not blocked by any obstacle voxel. When At that time, the target voxel is located in the blind zone directly above; when At that time, the target voxel was located in the blind zone directly below.

[0040] The semi-cone angle of the inverted cone-shaped blind zone directly above The semi-cone angle of the inverted cone-shaped blind spot directly below Each by the first The pitch angle range of this type of sensor is determined to satisfy: ; ; in, A negative value indicates the angle at which the sensor can detect downwards.

[0041] In this embodiment, the sensor's detection radius is 6km, and the pitch angle range is... to ,thus: The semi-cone angle of the inverted cone-shaped blind spot directly above is: ; The semi-cone angle of the inverted cone-shaped blind spot directly below is: ; That is, directly above the sensor The area within the cone angle is the upper detection blind zone, directly below the sensor. The area within the cone angle range constitutes a blind spot for detection below. In other embodiments, the sensor detection radius and pitch angle range can be configured according to the sensor type.

[0042] For installation location The Sensors that determine whether they can cover the target voxel The following three conditions must be met simultaneously: Condition 1: Distance constraint. The three-dimensional Euclidean distance between the sensor and the target voxel. Not exceeding the detection radius: ; in, , , (Unit: meters) Distance The unit is kilometers.

[0043] Condition 2: Pitch angle constraint. The relative pitch angle between the sensor and the target voxel. Within the effective detection range: ; ; Condition 3: Unobstructed line of sight. The line of sight between the sensor and the target voxel does not contain any mesh cells marked as obstacles. Line of sight occlusion detection is implemented using the 3D Bresenham line algorithm.

[0044] The deployment constraints defined in step S1 are as follows: the sensor can only be deployed in an internal area at a certain distance from the area boundary, and must be installed within the deployable area on the top of the building.

[0045] In one specific implementation, line-of-sight occlusion detection includes: sampling within the sensor detection bounding box according to a preset azimuth angle step size. and preset pitch angle sampling step size Generate the ray direction; use the 3D Bresenham voxel line algorithm to obtain the voxel sequence that the ray passes through for each ray; judge the voxel attributes sequentially along the ray direction; when the voxel is a free space voxel, mark it as a visible voxel; when the obstacle voxel is hit for the first time, terminate the subsequent voxel marking of the ray.

[0046] Taking the sensor position as the origin, in the azimuth angle Within the range of preset azimuth step size Uniform sampling at pitch angle Within the range of preset pitch angle steps Uniform sampling, generating Rays: ; For the azimuth angle and pitch angle The given ray direction has a unit direction vector as follows: ; The endpoint of the ray, after coordinate conversion, is: ; Convert the sensor position and ray endpoint to grid index coordinates: ; ; For the 3D Bresenham voxel line algorithm, let the starting grid index be... The endpoint grid index is Calculate the absolute difference of the three axes and the step direction: ; ; Select , , The axis corresponding to the largest value is used as the drive axis. Taking the drive shaft as an example (i.e.) and Initialize error parameters: ; ; Proceed step by step along the drive axis, updating coordinates at each step: ; like ,but: ; like ,but: ; ; ; Will the current Add to the waypoint list. Repeat until... .

[0047] Similarly, when When the shaft is a drive shaft (i.e.) and ),initialization: ; along Updated after each axis advance step: ; like ; like ; ; when When the shaft is a drive shaft (i.e.) and ),initialization: ; along Updated after each axis advance step: ; like ; like ; ; Iterate through each point in the path point list Query its obstacle marker attributes: Traversing the voxel sequence along the ray path, the occlusion determination condition is: ; That is, if any voxel in the path sequence is marked as an obstacle, the ray is determined to be blocked.

[0048] After occlusion occurs, all subsequent voxels of the ray starting from the occlusion point are not marked as visible.

[0049] After all ray processing is complete, a three-dimensional neighborhood dilation with a preset radius is performed on the visible voxel set to compensate for voxel gaps between discrete rays. Let the dilation radius be... For three-dimensional visibility arrays The expansion operation specifically involves: for each voxels, and their All valid index voxels within the neighborhood are set to 1:

[0050]

[0051] The expansion assignment is performed only if the above conditions are met.

[0052] ; After the expansion is complete, the obstacle object pixels are forcibly cleared to zero to ensure that the cover set does not contain obstacle object pixels: ; That is, for all voxels marked as obstacles, their visibility is forcibly set to zero.

[0053] In this embodiment, the azimuth sampling step size Pitch angle sampling step size All The three-dimensional neighborhood expansion radius is 1 voxel. The elevation angle range of the first type of sensor... For example, the number of azimuth angle samples is Number of pitch angle samples: The total number of rays is: strip; In this embodiment, the physical basis for taking the expansion radius as 1 is: within the sensor detection radius... At this location, the distance between adjacent rays is approximately This value is on the same order of magnitude as the actual size of a single horizontal grid (300m), so an expansion operation with a radius of 1 can effectively compensate for the ray gap.

[0054] In one specific implementation, the core operation of multi-sensor joint coverage calculation is to merge multiple sorted coverage voxel index arrays and remove duplicates. The coverage voxel set of each sensor is represented as a sorted one-dimensional linearized index array, and the linearized index is calculated as follows: ; Two ascending covered index arrays and The algorithm for merging and deduplicating (union operation) is as follows: Initialize three pointers , , Output array Pre-allocated length .

[0055] The loop condition is and : like ; like ; like ; After the loop ends, or The remaining elements are appended to The end. The final. That is, the number of unique voxels in the joint cover. The time complexity of this algorithm is O(n log n). .

[0056] Sorted array difference operation (used for cover gain calculation in greedy algorithms) )as follows: initialization , , .

[0057] like (Skip repetitions); like ; like ; After the loop ends, The remaining elements are appended to end. That is, the number of elements in the difference set.

[0058] In one specific implementation, the number minimization optimization module in step S2 specifically includes: firstly, using a greedy algorithm to select the location that can bring the maximum coverage gain from the candidate points in the deployable area, and obtaining an initial sensor number estimate that meets the coverage requirements; then, starting from the estimated value, using the coverage maximization optimization module to verify it, and iteratively decreasing the number until the minimum number of sensors that meets the coverage requirements is found.

[0059] The detailed execution flow of the greedy algorithm is as follows: Step a: Initialize the covered voxel set (Represented by a sorted array) Set of deployed sensor locations Iteration counting .

[0060] Step b: For all candidate deployment locations ( ), calculate their respective coverage gains: ; Select the position with the highest gain: ; Step c: Update , , .

[0061] Step d: Calculate the current coverage ,like Then output Otherwise, return to step b.

[0062] The verification process using the coverage maximization optimization module in step S2 specifically includes: dividing the internal deployment area into regions using the K-means++ clustering algorithm. For each key region, candidate deployment points are selected. An improved genetic algorithm is used to find the layout scheme with the highest coverage rate given the number of sensors. If the coverage rate is met, the number of sensors is reduced; otherwise, the algorithm returns to the previous order of magnitude as the optimal solution.

[0063] It's important to note that the number minimization optimization module and the coverage maximization optimization module are not the same. The number minimization optimization module searches for the minimum number of sensors required to achieve the target coverage; the coverage maximization optimization module searches for the deployment location with the highest coverage given a given sensor type and number. The number minimization optimization module uses the coverage maximization optimization module as a verification tool, repeatedly calling the coverage maximization optimization module to determine whether different numbers of sensors meet the target coverage.

[0064] In one specific implementation, target coverage The coverage maximization optimization module is 0.85; for a given number of... At that time, the internal deployment area can be divided into One key region; each key region is divided according to a second-level equal division parameter. Divided into The process involves dividing the region into sub-regions and selecting candidate building tops that meet the deployment constraints from each sub-region. , and The values ​​shown in this embodiment are specific to this example. In other embodiments, the values ​​can be set according to the number of sensors and the size of the deployment area.

[0065] The initial center selection process for K-means++ clustering is as follows: Let the set of candidate deployment points be... .

[0066] Step 1: From One of the candidate locations is randomly selected as the first cluster center. .

[0067] Step 2: For each candidate position ( ), calculate the shortest distance from it to the set of selected cluster centers: ; proportional to probability Select the next cluster center: ; Step 3: Repeat step 2 until all are selected. Cluster centers.

[0068] Step 4: Perform standard K-means iteration—assign each candidate position to the nearest cluster center, then update the cluster center to the mean coordinates of the position to which each cluster belongs, repeating until the cluster center no longer changes or the maximum number of iterations is reached.

[0069] In this embodiment, when At that time, perform two-level division for each cluster region: using the division parameter Each cluster region is divided into Each sub-region is used to select one candidate point for the tallest building, thus expanding the number of candidate locations for the genetic algorithm search to [number]. This increases the diversity of population initialization.

[0070] The improved genetic algorithm specifically includes the following elements.

[0071] Chromosome encoding: Employing a one-dimensional integer coordinate sequence of region-candidate points, the chromosome is composed of... Individual genetic composition: ; in, Indicates the first The index of the selected candidate points in the key regions.

[0072] Fitness function: fitness value based on coverage: ; in, Indicated by chromosomes The deployment plan obtained through decoding For the first The set of visible voxels corresponding to the selected candidate points in each key region.

[0073] Adaptive parameter adjustment: crossover probability and mutation probability With iterative algebra Dynamic adjustment: ; ; in, For the maximum number of generations, and These represent the base value and the magnitude of change of the crossover probability, respectively. and These represent the base value and the magnitude of change of the mutation probability, respectively.

[0074] In this embodiment, the genetic algorithm parameters of the coverage maximization optimization module are set as follows: population size is 40, and the maximum number of generations is... Basic crossover probability Crossover probability change magnitude Basic mutation probability Variation in mutation probability .

[0075] Under the above parameters, the crossover probability and mutation probability change with the number of iterations as follows: when (Initial generation): , High crossover and low mutation rate, which is beneficial for global exploration.

[0076] when (Mid-term) period: , Crossover and mutation tend to reach equilibrium.

[0077] when (At the end of the dynasty): , This reduces crossover and increases variation, which is beneficial for escaping local optima.

[0078] Selection Operation: Combining tournament selection and elite retention strategies, a predetermined number of individuals are randomly selected from the population each time, and the individuals with the highest fitness are chosen as parents. The individuals with the best fitness are retained in each generation and directly enter the next generation.

[0079] Crossover operation: Genetic operations employ a combination of single-point crossover, multi-point crossover, and uniform crossover strategies. Taking single-point crossover as an example, crossover points are randomly selected. ( ), father generation and The offspring are: ; Mutation operation: Randomly select a gene location on a chromosome. Replace its value with the index of another candidate point within the critical region: ; Local search optimization: When the genetic algorithm fails to improve after a preset number of generations, a local search optimization mechanism is automatically triggered to explore the current optimal solution in depth. In this embodiment, when the coverage maximization optimization module is executed independently, a local search is triggered after 10 consecutive generations without improvement; when the number minimization module calls the coverage maximization module for decreasing verification, a local search is triggered after 15 consecutive generations without improvement.

[0080] The complete iterative process of the genetic algorithm is as follows: Step a: Population initialization, generating a preset number of chromosomes.

[0081] Step b: Fitness assessment, calculating coverage for each chromosome.

[0082] Step c: Elite retention. The population is sorted in descending order of fitness, and a number of the best individuals are retained to directly enter the next generation.

[0083] Step d: Select an operation to generate parent pairings.

[0084] Step e: Crossover operation, with probability Perform crossover.

[0085] Step f: Mutation operation, with probability Perform mutation.

[0086] Step g: Population update, determine if local search is triggered.

[0087] Step h: Termination judgment. If the maximum number of iterations is reached, terminate; otherwise, return to step b.

[0088] Minimum quantity iterative decreasing process: the quantity obtained by the greedy algorithm Starting with the coverage maximization optimization module, the optimal coverage is obtained. .like Then let Then re-execute the coverage maximization optimization module; repeat the above decreasing process until the optimal coverage is achieved in a certain iteration. Then take the previous number as the minimum number of sensors: ; The corresponding minimum deployment cost is: ; In a more preferred embodiment, step S3, which uses a modified genetic algorithm to search for a hybrid deployment scheme, specifically includes: setting a step size for reducing the number of primary model sensors, calculating the number of other model sensors that need to be added; generating a hybrid deployment scheme that includes multiple model combinations, and using a performance optimization strategy based on eliminating infeasible schemes to quickly eliminate invalid combinations that do not meet coverage constraints.

[0089] Specifically, let the sensor model in the lowest-cost single-type benchmark scheme be the primary model, and the initial quantity of the primary model be... The cost per unit is The main model decreases by a step size of 1. , No. The number of main models after the wheel reduction is: ; After determining the quantity of the main model, if the cost of the current optimal solution is denoted as... The remaining cost budget is: ; like If the decreasing process ends, the process will terminate.

[0090] For any auxiliary model The upper limit of its replenishment quantity satisfies: ; in, auxiliary models The individual cost.

[0091] For each candidate quantity vector The algorithm uses a fixed-quantity vector correction genetic algorithm to search for specific deployment locations. After the genetic operation, the fixed-quantity vector correction genetic algorithm counts the number of each sensor model. When the number of a certain model exceeds the candidate quantity vector, the corresponding gene is deleted or replaced. When the number of a certain model is insufficient, the corresponding model gene is added to the candidate points that meet the deployment constraints, so that the corrected chromosome meets the candidate quantity vector again.

[0092] The performance optimization strategy based on eliminating infeasible solutions specifically includes two cases: Scenario 1: Infeasible solutions with decreasing numbers of primary models are eliminated. Analyzing the patterns of verified infeasible solutions, if a solution is infeasible due to an insufficient number of primary models, then all solutions with even fewer primary models and unchanged or fewer other models are directly eliminated. The rationale is that if the number of primary models further decreases while the number of secondary models does not increase, the overall coverage capability can only decrease, making it impossible to meet the target coverage rate.

[0093] Scenario 2: Auxiliary models are eliminated due to insufficient performance. Let the detection radius of the primary model be... The pitch angle coverage area is Auxiliary model The detection radius is The pitch angle coverage area is If the following conditions are met: and ; That is, the detection radius of the auxiliary sensor is no greater than that of the primary sensor, and the elevation angle coverage is no greater than that of the primary sensor. Meanwhile, the total number of sensors in the candidate hybrid scheme... If the number of sensors is not greater than the total number of sensors in verified infeasible solutions, the solution is deemed unable to meet the coverage constraint and is directly eliminated. The criterion is that coverage cannot be improved by replacing better-performing sensors with worse or equivalent sensors without increasing the total number of sensors.

[0094] Furthermore, when a candidate solution fails to meet the target coverage constraint after verification by the modified genetic algorithm, based on the above two rules, all related solutions derived from that solution that meet the elimination conditions are simultaneously recorded as infeasible solutions. When these solutions are encountered in subsequent searches, they are skipped directly without needing to call the genetic algorithm for verification again, thereby improving the efficiency of hybrid solution search.

[0095] In step S2, the first Theoretical three-dimensional coverage volume of sensor-like devices Calculate using the following formula: ; in, For the first Detection radius of the sensor and These are the minimum and maximum pitch angles for this type of sensor, respectively.

[0096] No. Cost-effectiveness of similar sensors Defined as the ratio of theoretical 3D coverage volume to unit cost: ; in, For the first The cost per unit of this type of sensor. This cost-effectiveness ratio represents the theoretical three-dimensional coverage capability per unit cost, not the two-dimensional area coverage capability.

[0097] For any deployment scheme Its three-dimensional coverage Defined as: ; in, Deployment plan The sensor set in the middle, Indicates sensor The set of visible free space voxels, This represents the set of free-space voxels within the target altitude range. Deployment plan. Total cost is defined as The total number of sensors is defined as .

[0098] This embodiment adopts an economic optimization rule under hard coverage constraints, that is, it first requires... In order to meet the target coverage rate Choose from the set of solutions The option with the smallest total cost; if multiple options have the same total cost, then choose the one with the smallest total cost. The smaller option; if the total cost and quantity are the same, then choose the smaller option. A more advanced option.

[0099] To characterize the weighted relationship between coverage, cost, and quantity, the above constraint priority rule can be expressed as an evaluation function with the inadequate coverage penalty term as the highest priority: ; in, and These are the normalized cost and quantity items, respectively. , , , To evaluate the weights, and If the coverage is greater than the sum of the other weighted items, then the solution with insufficient coverage will be excluded first.

[0100] Example 2 To further explain the solution of the present invention, this embodiment provides a more systematic description of the overall implementation process and key sub-algorithms of the intelligent optimization method for static multi-source detection sensor layout.

[0101] This invention discloses an intelligent optimization method for the layout of multi-source detection sensors, the core process of which is as follows: Figure 2 As shown, it mainly includes two parts: layout optimization preprocessing and intelligent optimization algorithm.

[0102] Step 1: Layout Optimization Preprocessing. First, this invention models the deployment environment through layout optimization preprocessing, transforming the complex physical space problem into a mathematical optimization problem.

[0103] Map and sensor modeling: 3D grid discretization, specifically, dividing the monitoring area into cubic grid cells with a certain resolution, establishing a 3D coordinate system, and uniquely identifying each grid cell through the coordinates of its center point.

[0104] Furthermore, the monitoring area is discretized in three dimensions, specifically by dividing the monitoring area into several non-overlapping cubic grid units according to a preset spatial resolution; establishing a unified three-dimensional rectangular coordinate system within the monitoring area; and using the spatial coordinates of the geometric center point of each cubic grid unit as the unique identifier of the grid unit to characterize the position information of the grid unit in physical space, thereby converting the continuous physical space into a discrete mathematical space model.

[0105] In one specific embodiment of the present invention, preprocessing for layout optimization is first performed, with the core step of map modeling being the digital transformation of continuous physical space. The system receives geographic range data of the monitoring area and sets the spatial resolution according to actual optimization requirements. The system divides the entire three-dimensional monitoring area according to this resolution, generating fine-grained cubic grid cells. To establish a mathematical coordinate system, the system sets an origin within this area and constructs a three-dimensional coordinate system. To uniquely locate and index each grid cell, the present invention uses the coordinates of its center point as the unique identifier of that cell. This discretization process transforms the originally continuous and complex physical spatial environment into a mathematical model composed of discrete point sets, providing a data foundation for calculating sensor coverage, determining obstacles, and iterating optimization algorithms.

[0106] Terrain and obstacle modeling: Using multi-source data such as GIS and DEM, obstacle attributes are assigned to each grid cell. If buildings, mountains, etc., exist within the grid, they are marked as obstacles with an attribute value of 1; otherwise, they are considered free space with an attribute value of -1. At the same time, a three-dimensional KD tree spatial index structure is constructed to accelerate obstacle lookup.

[0107] Furthermore, the terrain and obstacle modeling steps specifically include: determining the attributes of the cubic grid cells based on geographic information system data and digital elevation model data; if the cubic grid cell contains entities, such as buildings, mountains, or ground entities, it is marked as an obstacle and assigned a first attribute value; if the cubic grid cell does not contain any entities, it is marked as free space and assigned a second attribute value; after completing the attribute marking of all cubic grid cells, a three-dimensional KD tree spatial index structure is constructed for quickly querying the distribution of obstacles on the target path during subsequent line-of-sight determination.

[0108] In a specific embodiment of this invention, terrain and obstacle modeling is a crucial step in assigning physical features to digital maps. The system uses high-precision geographic information system (GIS) data and digital elevation model (DEM) data as its basic data sources. The system traverses each generated cubic grid cell and analyzes whether buildings, mountains, or other ground entities exist within the cell through spatial location matching. If an entity is detected, the grid cell is determined to be a physically impenetrable area, and its obstacle attribute is marked as 1; conversely, if the grid cell is an open area, it is marked as free space -1.

[0109] To improve the computational efficiency of subsequent line-of-sight occlusion detection and avoid excessive computation due to traversing all grids, this invention further constructs a three-dimensional KD-tree spatial index structure based on the marked grid data. This three-dimensional KD-tree spatial index structure can organize disordered grid data according to spatial location, enabling the system to quickly retrieve relevant grids on the path when determining whether there is occlusion between the sensor and the target point, providing a computational foundation for optimization in large-scale scenes.

[0110] Sensor detection model construction: A Boolean detection model is adopted. The effective detection range of the sensor is the region within a spherical space that satisfies the pitch angle constraint, with an inverted cone-shaped blind zone directly above and below. The sensor's detection capability of the target point is determined by distance, pitch angle, and line of sight: the distance must be less than or equal to the detection radius, the pitch angle must be within the sensor's allowed effective pitch angle range, and the line of sight path between the sensor and the target point must not be blocked by any obstacles.

[0111] Specifically, the sensor detection model construction steps include: constructing a spherical physical detection space with the sensor installation location as the center; defining an inverted cone-shaped invalid detection area as the upper blind zone in the central region directly above the spherical space; and defining an inverted cone-shaped invalid detection area as the lower blind zone in the central region directly below the spherical space. The semi-cone angle of the upper blind zone is determined by... Calculate the semi-cone angle of the lower blind zone according to Calculation. The effective detection area is an annular strip of space located between the upper and lower blind zone cones within a spherical space.

[0112] Step 2: Implementation of the intelligent optimization algorithm for the static multi-source detection sensor layout. The core of this invention is an economy-oriented hybrid layout optimization algorithm, the overall process of which is as follows: Figure 3 As shown, the algorithm achieves cost optimization under coverage constraints through a three-stage architecture.

[0113] Phase 1: Cost-Effectiveness Screening and Single-Type Benchmark Solution Generation. First, the three-dimensional cost-effectiveness indicators for various types of sensors are calculated. The cost-effectiveness ratio mentioned is not the ratio of two-dimensional area to unit price, but rather the ratio of theoretical three-dimensional coverage volume to unit cost, satisfying... A higher cost-performance ratio indicates a more significant cost advantage per unit area for the sensor in terms of theoretical three-dimensional coverage volume.

[0114] Secondly, for each sensor model, the quantity minimization optimization module is invoked to calculate the minimum coverage required to meet the target coverage. Under the premise of using only this model of sensor, the minimum number required and the corresponding total cost.

[0115] This embodiment provides a specific description of a decreasing verification process: Assume the greedy algorithm obtains an initial estimated number of a certain type of sensor. Target coverage The decreasing verification process is as follows: First verification: The genetic algorithm outputs the optimal coverage. , through, continue to decrease; Second verification: The genetic algorithm outputs the optimal coverage. , through, continue to decrease; Third verification: The genetic algorithm outputs the optimal coverage. , through, continue to decrease; 4th verification: The genetic algorithm outputs the optimal coverage. It was not approved.

[0116] Determine the minimum number of this type of sensor. .

[0117] The values ​​above are for illustrative purposes only; actual values ​​depend on specific terrain data and sensor parameters.

[0118] Phase Two: Hybrid Scheme Search Based on Modified Genetic Algorithm. The lowest-cost scheme among the single-type schemes is selected as the primary scheme, and this single-type scheme is used as the initial optimal scheme.

[0119] During the iterative optimization phase, the number of primary sensor models is gradually reduced, and other sensor models are attempted to supplement them. For each newly generated hybrid deployment scheme, a modified genetic algorithm is invoked to verify whether it can achieve the target coverage, and a sensor number correction mechanism is designed to ensure that the generated individuals meet the number constraints.

[0120] In this process, the present invention proposes a performance optimization strategy based on the elimination of infeasible solutions, which includes two elimination rules. The first rule addresses infeasibility caused by a decrease in the number of primary models: if a solution is infeasible due to insufficient primary models, then all solutions with fewer primary models and no increase in the number of secondary models are eliminated. The second rule addresses infeasibility due to insufficient performance of secondary models: if the supplementary secondary model sensors are not superior to the primary models in terms of both detection radius and elevation angle coverage, and the total number of sensors in the hybrid solution does not increase, then the solution is directly determined to be infeasible and eliminated. Through the synergistic effect of the two rules, invalid candidate solution combinations are quickly eliminated, improving search efficiency. All feasible solutions that meet the coverage requirements are sorted by cost, and the layout solution with the lowest cost is finally output.

[0121] This embodiment provides a specific example of a hybrid scheme search process: There are two types of sensors, the primary model (the...) Unit cost (category) 10,000 yuan, initial minimum quantity Benchmark cost 10,000 yuan; auxiliary model (number) Unit cost (category) 10,000 yuan. Decreasing step size. .

[0122] First round of decreasing: Remaining budget: Ten thousand yuan; Auxiliary models available in the following quantities: The budget is insufficient to supplement the auxiliary models, so this round will be skipped.

[0123] Second round of decrease: Remaining budget: Ten thousand yuan; Auxiliary models available in the following quantities: Hybrid scheme candidates: 12 units Class +1 unit Class, Total Cost: Ten thousand yuan; If the cost reduction condition is met, execute the modified genetic algorithm to verify the coverage. If the verification result... Then update Ten thousand yuan.

[0124] Third round of decrease: With new Based on a budget of 10,000 yuan, the remaining budget is: Ten thousand yuan; Auxiliary models available in the following quantities: Total cost of the hybrid solution: Ten thousand yuan; Continue executing the modified genetic algorithm verification.

[0125] Repeat the above decreasing process until the termination condition is met: no better solution is found after a preset number of consecutive decreasing rounds (i.e., the optimal coverage output by the corrected genetic algorithm is achieved). ), or the number of main models decreases to 1.

[0126] The values ​​above are for illustrative purposes only; actual values ​​depend on specific sensor parameters and terrain data.

[0127] The module is explained in detail below.

[0128] I. Minimize quantity optimization module. For example... Figure 4 As shown, the optimization module for minimizing the number of items adopts a greedy-genetic hybrid optimization strategy.

[0129] Greedy estimation: First, the deployable area is gridded, and the tallest building in each small grid is selected as a candidate point. Then, a greedy algorithm is used to select the candidate point that brings the maximum marginal coverage gain each time, quickly obtaining an estimate of the number of sensors required to meet the coverage requirements. And the initial layout scheme. The formal description of the choice made by the greedy algorithm in each round is: Initialize the covered voxel set Gradually select candidate locations with the greatest coverage gain to add to the deployment plan: ; renew Repeat until the current coverage reaches the target coverage: ; in, The set of all free space voxels within the target height range.

[0130] Iterative verification: Starting with the module verification method, maximize call coverage. Can each sensor achieve the target coverage? If so, reduce the number of sensors to make... And verify again; if not, return the previous feasible number as the minimum number of sensors to meet the coverage requirement.

[0131] II. Coverage maximization optimization module. For example... Figure 5 As shown, this coverage maximization optimization module is used to find the layout scheme with the highest coverage given the sensor type and number, and it is the core verification tool of the number minimization optimization module. It adopts a clustering-genetic hybrid optimization strategy.

[0132] Key point pre-setting technology: To solve the problem of large-scale search space, the K-means++ clustering algorithm is first used to divide the internal deployment area into regions equal to the number of sensors. There are three key regions. Then, each key region is further divided into two equal parts, i.e., divided into... The global search problem is decomposed into three sub-regions, and the tallest building in each sub-region is selected as a candidate deployment point. An independent local choice problem.

[0133] Improved genetic algorithm optimizations include: chromosome encoding using a one-dimensional integer coordinate sequence encoding of region-candidate points, and chromosomes being composed of... Each gene is composed of a gene, and the value of each gene represents the index of the selected candidate point in its corresponding key region. The genetic operations combine tournament selection and elite retention strategies, and the crossover operation mixes single-point crossover, two-point crossover, multi-point combination crossover, uniform crossover and differential candidate point replacement strategies. An adaptive parameter regulation mechanism is adopted to balance global exploration and local development.

[0134] Local search optimization: When the genetic algorithm fails to improve after a set number of generations, a local search-based optimization mechanism is automatically triggered to explore the current optimal solution in depth and help the algorithm escape local optima.

[0135] Through the combined efforts of the modular design and intelligent optimization algorithms described above, this invention can efficiently solve the layout optimization problem of static multi-source detection sensors in complex three-dimensional environments.

[0136] This invention proposes an intelligent optimization scheme for sensor layout in low-altitude security scenarios. Its core lies in constructing a hierarchical and modular sensor layout optimization architecture. This architecture first performs preliminary planning of sensor deployment using a clustering-genetic hybrid optimization method based on key point pre-setting. Then, a greedy-genetic hybrid strategy is introduced in the quantity minimization optimization module to achieve efficient resource allocation. To further improve search efficiency and solution quality, this invention employs an improved genetic algorithm incorporating local search, combined with a performance optimization strategy based on infeasible solution elimination, effectively filtering out invalid solutions. Finally, in the optimal solution search stage, a refined search is performed using a modified genetic algorithm, thereby obtaining a more economical sensor layout scheme under complex three-dimensional geographical constraints.

[0137] Example 3 like Figure 6 As shown, the present invention also includes a static multi-source detection sensor layout intelligent optimization system, comprising a preprocessing module, an optimization calculation module, and an output module.

[0138] The preprocessing module is used to acquire geographic information of the monitoring area, construct a three-dimensional spatial model, and define the sensor detection model and deployment constraints.

[0139] The optimization calculation module is used to calculate the cost-effectiveness of various types of sensors based on the three-dimensional spatial model; for each sensor model, the quantity minimization optimization module is used to calculate the minimum number and corresponding cost required to meet the target coverage under the deployment constraints, and generate a single-type benchmark scheme.

[0140] The output module is used to select the lowest cost among the single-type benchmark schemes as the initial optimal scheme; based on the sensor detection model, by gradually reducing the number of main models and supplementing other models, a modified genetic algorithm is used to search for a hybrid deployment scheme in the three-dimensional space model that meets the coverage requirements and has a lower cost, and outputs the sensor layout result with the best economic efficiency.

[0141] Furthermore, one embodiment of the present invention provides a terminal device comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the various method embodiments described above. Alternatively, when the processor executes the computer program, it implements the functions of each module or unit in the various device embodiments described above.

[0142] The computer program can be divided into one or more modules or units, which are stored in the memory and executed by the processor to complete the present invention.

[0143] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0144] The processor may be a central processing unit, or it may be other general-purpose processors, digital signal processors, application-specific integrated circuits, field-programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0145] The memory can be used to store the computer program and / or module. The processor implements various functions of the terminal device by running or executing the computer program and / or module stored in the memory and by calling the data in the memory.

[0146] If the modules or units integrated into the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above.

[0147] The computer program includes computer program code, which may be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium may include any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory, random access memory, electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0148] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for intelligent optimization of multi-source detection sensor layout, characterized in that, Includes the following steps: S1. Obtain the geographic information of the monitoring area, construct a three-dimensional spatial model, and define the sensor detection model and deployment constraints; S2. Based on the three-dimensional spatial model, calculate the cost-effectiveness of various types of sensors; for each sensor model, use the quantity minimization optimization module to calculate the minimum number and corresponding cost required to meet the target coverage under the deployment constraints, and generate a single-type benchmark solution. S3. Select the lowest cost among the single-type benchmark schemes as the initial optimal scheme; based on the sensor detection model, by gradually reducing the number of main models and supplementing other models, use a modified genetic algorithm to search for a hybrid deployment scheme in the three-dimensional space model that meets the coverage requirements and has a lower cost, and output the sensor layout result with the best economic efficiency.

2. The intelligent optimization method for multi-source detection sensor layout according to claim 1, characterized in that, In step S1, constructing the three-dimensional spatial model specifically includes: The monitoring area is divided into cubic grid units according to the preset horizontal and vertical grid spacing, a three-dimensional coordinate system is established, and the grid unit center point coordinates are used for identification. Based on GIS and DEM data, determine whether there are entities within the grid cells. If they exist, mark them as obstacle voxels; otherwise, mark them as free space voxels and construct a three-dimensional KD tree spatial index structure.

3. The intelligent optimization method for multi-source detection sensor layout according to claim 1, characterized in that, In step S1, the detection model of the sensor is specifically defined as follows: Using a Boolean detection model, the sensor's detection range is defined as having a sphere centered at the installation location and a detection radius of [missing information]. The region within the spherical space that satisfies the pitch angle constraint has an inverted cone-shaped blind zone at the center of the upper part and an inverted cone-shaped blind zone at the center of the lower part. To determine the sensor's ability to detect target voxels, the following conditions must be met simultaneously: the distance between the target voxel and the sensor is less than or equal to the detection radius, and the pitch angle of the target voxel relative to the sensor is within the effective pitch angle range. Within, and the line-of-sight path between the sensor and the target voxel does not contain any grid cells marked as obstacles; The semi-cone angle of the inverted cone-shaped blind zone directly above The semi-cone angle of the inverted cone-shaped blind spot directly below They respectively satisfy: ； ； in, For the first The maximum pitch angle of this type of sensor, For the first The minimum pitch angle of this type of sensor; The process of determining whether the line-of-sight path between the sensor and the target voxel contains obstacles specifically includes: Within the sensor detection enclosure, a ray direction is generated according to a preset azimuth angle sampling step size and a preset elevation angle sampling step size; The three-dimensional Bresenham voxel line algorithm is used to obtain the voxel sequence through which the ray passes for each ray; The voxel attributes are determined sequentially along the ray direction. When a voxel is a free space voxel, it is marked as a visible voxel. When an obstacle voxel is hit for the first time, the marking of subsequent voxels for that ray is terminated. After all ray processing is completed, a three-dimensional neighborhood expansion with a preset radius is performed on the visible voxel set, and the obstacle voxels are forcibly cleared to zero.

4. The intelligent optimization method for multi-source detection sensor layout according to claim 1, characterized in that, In step S2, the cost-effectiveness is the ratio of the theoretical three-dimensional coverage volume of the sensor to its individual cost; wherein, the first Theoretical three-dimensional coverage volume of sensor-like devices satisfy: ； No. Cost-effectiveness of similar sensors satisfy: in, For the detection radius, and These are the minimum pitch angle and the maximum pitch angle, respectively. For a single cost.

5. The intelligent optimization method for multi-source detection sensor layout according to claim 1, characterized in that, In step S2, the quantity minimization optimization module specifically includes: First, a greedy algorithm is used to progressively select the location that can bring the maximum coverage gain from the candidate points in the deployable area, so as to obtain an initial estimate of the number of sensors that meets the target coverage requirements. Then, starting from the initial sensor number estimate, the coverage maximization optimization module is used for verification, and the sensor number is iteratively reduced until the minimum number of sensors that meet the target coverage requirement is found. The coverage maximization optimization module specifically includes: The internal deployment area is divided into K key regions using the K-means++ clustering algorithm, where K is the number of sensors to be deployed. Candidate deployment points that meet the deployment constraints are selected within each key region. An improved genetic algorithm is used to find the layout scheme with the highest coverage given the sensor type and the number of sensors. If the obtained coverage rate reaches the target coverage rate, the number of sensors continues to decrease; otherwise, it returns to the previous order of magnitude as the minimum number of sensors required to meet the target coverage rate requirement.

6. The intelligent optimization method for multi-source detection sensor layout according to claim 5, characterized in that, The improved genetic algorithm specifically includes: Chromosome coding uses a one-dimensional integer coordinate sequence of region-candidate point, with each gene representing the index of the selected candidate point in the corresponding key region; In genetic operations, a combination of crossover and mutation strategies is used, and an adaptive parameter control mechanism is employed to make the crossover probability decrease with the number of iterations and the mutation probability increase with the number of iterations. When there is no improvement after consecutive preset algebras, a local search optimization mechanism is triggered to conduct in-depth exploration of the current optimal solution.

7. The intelligent optimization method for multi-source detection sensor layout according to claim 1, characterized in that, In step S3, a modified genetic algorithm is used to search for a hybrid deployment scheme, specifically including: Set the decreasing step size of the main model sensor, gradually reduce the number of main models, and calculate the upper limit of the number of auxiliary models that can be replenished based on the remaining cost budget. For each candidate quantity vector, a fixed quantity vector correction genetic algorithm is called to search for a specific deployment location that satisfies the quantity constraint; By employing a performance optimization strategy based on infeasible solution elimination, invalid combinations that do not meet coverage constraints can be quickly eliminated. If a new solution that meets the target coverage and has a lower cost than the current optimal solution is found, the current optimal solution is updated and the cost continues to decrease; otherwise, the search is terminated and the sensor layout result with the best economic efficiency is output.

8. The intelligent optimization method for multi-source detection sensor layout according to claim 7, characterized in that, The performance optimization strategy based on eliminating infeasible solutions includes the following two cases: Scenario 1: Analyze the patterns of verified infeasible solutions. If a solution is infeasible due to an insufficient number of main models, then directly eliminate all solutions with even fewer main models and unchanged or fewer other models. Scenario 2: If the detection radius of the supplementary auxiliary sensor is smaller than that of the main sensor and the pitch angle coverage is not greater than that of the main sensor, and the total number of sensors in the hybrid scheme is not greater than the total number of the verified infeasible schemes, then the scheme will be directly eliminated.

9. A static multi-source detection sensor layout intelligent optimization system, characterized in that, A method for intelligent optimization of multi-source detection sensor layout as described in any one of claims 1 to 8 includes: The preprocessing module is used to acquire geographic information of the monitoring area, construct a three-dimensional spatial model, and define the sensor detection model and deployment constraints. The optimization calculation module is used to calculate the cost-effectiveness of various sensors based on the three-dimensional spatial model. For each sensor model, the quantity minimization optimization module is used to calculate the minimum number required to meet the target coverage and the corresponding cost, and generate a single-type benchmark scheme. The output module is used to select the lowest cost among the single-type benchmark schemes as the initial optimal scheme. By gradually reducing the number of main models and supplementing other models, a modified genetic algorithm is used to search for a hybrid deployment scheme that meets the coverage requirements and has a lower cost, and outputs the sensor layout result with the best economic efficiency.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 8.

Images