Intelligent humanistic ancient tree patrol route recommendation method based on particle swarm optimization

Abstract

The invention discloses a humanistic ancient tree patrol route intelligent recommendation method based on particle swarm optimization. The method comprises the following steps: S1, forming a unified ancient tree patrol data set; s2, constructing an improved Markov random field model according to the ancient tree patrol data set; s3, constructing a multi-target and multi-constraint patrol path optimization model by using the ancient tree patrol data set and the improved Markov random field model, and stipulating a feasible patrol sequence and environmental adaptability constraints among ancient trees; s4, based on the ancient tree patrol data set and prior information of the improved Markov random field model, generating an initial patrol path group through a standard particle swarm optimization algorithm; s5, performing iterative optimization on the initial patrol path group by applying a standard particle swarm optimization algorithm to form an updated patrol path group; and S6, selecting the patrol path with the highest fitness from the updated patrol path group according to a preset patrol path fitness evaluation standard. According to the invention, the overall planning of the patrol path is more scientific and reasonable.

Description

Technical Field

[0001] The present invention relates to the technical field of ancient trees, and in particular, to an intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization. Background Art

[0002] With the development of intelligent management technologies, informationization and intelligent means have gradually been introduced into the fields of ecological protection and cultural heritage protection to improve management efficiency and patrol quality. As important natural and cultural heritages, the patrol work of cultural ancient trees is of great significance to ecological environment protection and historical and cultural inheritance. However, in the existing patrol systems, there are still many problems in the patrol route planning and execution, seriously affecting the scientific nature and efficiency of the patrol work.

[0003] Currently, most of the patrol work of cultural ancient trees still adopts the method of manual patrol, that is, patrol personnel conduct inspections on ancient trees along a fixed route according to a preset schedule, and record the health status, environmental risks, disease conditions, and human damage information of ancient trees. Although this method can complete the basic patrol tasks, there are obvious deficiencies in practice: on the one hand, the patrol route is usually formulated based on manual experience and lacks scientific optimization, resulting in too high a patrol frequency for some ancient trees and insufficient patrol coverage for some key ancient trees; on the other hand, the manual patrol route is fixed and cannot be dynamically adjusted according to environmental changes and the health status of ancient trees, making it difficult to achieve precise patrol. In addition, since the patrol task involves multiple factors, it is difficult for manual planning to take into account patrol efficiency, reasonable resource allocation, and the safety of ancient trees, resulting in low patrol work efficiency and serious resource waste.

[0004] In summary, the existing patrol path optimization solutions are difficult to meet the actual patrol needs, and there are problems such as low patrol efficiency, insufficient coverage, and uneven resource allocation. Therefore, there is an urgent need for an intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization to improve the accuracy and intelligence level of the patrol work. Summary of the Invention

[0005] An object of the present invention is to propose an intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization, which makes the overall planning of the patrol route more scientific and reasonable.

[0006] An intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization according to an embodiment of the present invention includes the following steps:

[0007] S1. Install a positioning system for each ancient tree, collect multi-source data within the patrol area of the ancient trees, and perform standardization processing on the collected multi-source data to form a unified ancient tree patrol data set;

[0008] S2. Construct an improved Markov random field model according to the ancient tree patrol data set;

[0009] S3. Use the ancient tree patrol dataset and the improved Markov random field model to construct a multi-objective and multi-constraint patrol path optimization model. The patrol path optimization model aims to optimize the patrol efficiency, optimize the patrol coverage rate, and reduce resource consumption, and stipulates the feasible patrol order between ancient trees and environmental adaptability constraints;

[0010] S4. Based on the prior information of the ancient tree patrol dataset and the improved Markov random field model, generate an initial population of patrol paths through the standard particle swarm optimization algorithm. The initial population of patrol paths satisfies the constraint conditions of the patrol path optimization model;

[0011] S5. Apply the standard particle swarm optimization algorithm to the initial population of patrol paths for iterative optimization. In each iteration process, use the improved Markov random field model to perform feedback correction on the spatial dependence relationship of the patrol paths, and dynamically adjust the patrol paths according to the data constraints in the ancient tree patrol dataset to form an updated population of patrol paths;

[0012] S6. Select the patrol path with the highest fitness from the updated population of patrol paths according to the preset patrol path fitness evaluation criteria, and transmit it to the APP for display.

[0013] Optionally, the S1 includes the following steps:

[0014] S11. Install a positioning system on each ancient tree in the ancient tree patrol area, use the positioning system to obtain the geographical location information of the ancient tree, and store it in the form of three-dimensional coordinates (x i , y i , z i ) to form the ancient tree position information set P;

[0015] S12. Collect the historical and cultural information of the ancient trees, including the cultural value, historical event association, tree age, species, and protection level of each ancient tree, and assign a cultural value score to each ancient tree to form the historical and cultural information set C. The historical and cultural information set contains the cultural value scores of i ancient trees, and the scores are set according to historical documents and expert evaluations;

[0016] S13. Collect the health status data of the ancient trees, including the growth status, disease conditions, withering degree, and soil moisture content of the ancient trees, and establish a health score for each ancient tree to form the health status information set H. The health status information set contains the health scores of i ancient trees;

[0017] S14. Collect environmental risk information of patrol areas, including climate factors, soil conditions, air pollution index and human damage risk, and assign an environmental risk score to each patrol area to form an environmental risk information set R. The environmental risk information set includes the environmental risk score of i ancient trees, and the score is set based on historical meteorological data, air pollution index and surrounding environmental risk assessment;

[0018] S15. Perform data formatting on the ancient tree location information set P, historical and cultural information set C, health status information set H, and environmental risk information set R, remove abnormal data, fill in missing data, and use normalization method to process all scoring data to form a unified ancient tree patrol data set:

[0019] D={(p i ,c i ,h i ,r i )|i=1,2,...,N};

[0020] Where D represents the standardized ancient tree patrol dataset, p i =(x i ,y i ,z i ) is the geographical coordinates of the i-th ancient tree, c i Score the cultural value of the i-th ancient tree, h i Score the health status of the i-th ancient tree, r i is the environmental risk score of the i-th ancient tree, and N is the total number of ancient trees in the patrol area.

[0021] Optionally, S2 includes the following steps:

[0022] S21. Constructing an improved Markov random field model G based on the ancient tree patrol dataset D + :

[0023] G + =(V,E,Ψ + );

[0024] Where V represents the node set in the improved Markov random field model, and each node v in the node set i ∈V corresponds to the i-th ancient tree in the patrol area, E represents the edge set, which indicates the spatial dependency relationship between ancient trees. At the same time, the edge weight is dynamically adjusted in combination with the continuity constraint of the patrol path:

[0025] E + ={e ij ∣v i ,v j ∈V,i≠j,γ ij >τ};

[0026] Among them, γ ij is the weight of the continuity of the patrol path, τ is the threshold of path effectiveness. Only when the path feasibility between two ancient trees meets the set conditions, an edge connection is established. Ψ + is the set of optimized potential energy functions. Combining with the dynamic adaptation mechanism of the patrol strategy, the dynamic optimization of the patrol priority is realized;

[0027] S22. Set the node state set X + in the improved Markov random field model G + , where the state x i of each ancient tree node represents the patrol priority of the ancient tree, and the set of state variables is defined as:

[0028] X + ={x1, x2,..., x N};

[0029] Among them, x i is the patrol priority of the i-th ancient tree;

[0030] S23. Combining the continuity constraint of the patrol path and the dynamic adaptive patrol strategy, optimize the state transition probability distribution of the improved Markov random field model G + so that the patrol path planning adapts to environmental changes. The optimized joint probability is defined as:

[0031]

[0032] Among them, λ ij is the path adjustment factor, which dynamically adjusts the dependence of the patrol path according to the changes in the patrol environment;

[0033] S24. Calculate the optimized potential energy function Ψ + (X + , D):

[0034]

[0035] Among them, φ + (x i , D) is the optimized single-node potential energy function, ψ + (x i , x j , D) is the optimized edge potential energy function, δ + (x i , D) is the patrol path dynamic adjustment function, and η i is the weight factor;

[0036] S25. Calculate the optimized single-node potential energy function φ + (x i , D):

[0037] φ + (x i , D) = w1c i + w2h i - w3r i + ξ i β i ;

[0038] Among them, w1, w2, and w3 are weight parameters, ξ i is the correction factor for the urgency of patrol, and β i is the adjustment factor for the patrol plan;

[0039] S26. Calculate the optimized edge potential function ψ + (x i , x j , D):

[0040]

[0041] Among them, λ ij is the correlation parameter between patrol paths, d ij is the geographical distance between ancient tree i and ancient tree j, σ is the normalization parameter, and α ij is the dynamic environment factor, and Ω ij calculates the impact degree of path emergencies based on real-time patrol data;

[0042] S27. Calculate the dynamic adjustment function δ of the patrol path + (x i , D):

[0043]

[0044] Among them, γ i is the adaptive adjustment parameter of the patrol path, ρ j is the completion degree of the patrol of the jth ancient tree, N(i) is the adjacent patrol point of ancient tree i, and ∈ is the smoothing factor;

[0045] S28. Update the improved Markov random field model G using the adaptive step size optimization method + :

[0046] X * = argmax P(X + | D);

[0047] Among them, P(X + | D) is the optimal state distribution under the condition of the given ancient tree patrol dataset D, and X * is the optimized optimal patrol priority state set;

[0048] S29. Output the improved Markov random field model G after optimization + =(V, E + , Ψ + ).

[0049] Optionally, the S3 includes the following steps:

[0050] S31. Based on the ancient tree patrol dataset D and the improved Markov random field model G + Construct a multi-objective and multi-constraint patrol path optimization model M + :

[0051] M + =(O, C1, Θ);

[0052] Among them, O is the set of patrol optimization objective functions, C1 is the set of patrol path optimization constraints, and Θ is the path optimization decision variable;

[0053] S32. Determine the set of patrol path optimization objective functions O. The set of patrol path optimization objective functions includes the patrol efficiency optimization function O eff , the patrol coverage rate optimization function O cov and the resource consumption minimization function O res :

[0054]

[0055] Among them, T ij is the patrol time from ancient tree i to ancient tree j, x ij is the path decision variable. When the patrol path selects i to j, x ij =1, otherwise 0;

[0056]

[0057] Among them, γ i is the reachability factor of the patrol path. When ancient tree i is covered by the patrol, x i =1, otherwise 0;

[0058]

[0059] Among them, E ij is the energy consumption of the patrol path, and R ij is the resource usage of the patrol path;

[0060] S33. Determine the set of patrol path optimization constraints C1. The set of patrol path optimization constraints includes the patrol feasibility constraint C feas , the patrol order constraint C seq and the environmental adaptability constraint C env :

[0061]

[0062]

[0063]

[0064] Among them, W j is the residence time at the patrol point j, T max is the maximum allowable time, and R max is the maximum available resource quantity;

[0065] S34. Set the path optimization decision variable Θ. The path optimization decision variable includes the patrol path selection variable X, the patrol weight adjustment variable Λ, and the path adaptive adjustment variable Δ;

[0066] S35. According to the patrol path optimization model M + , combined with the improved Markov random field model G + calculate the optimal decision variable Θ of the patrol path * :

[0067]

[0068] Among them, w1, w2, and w3 are weight coefficients that control the optimization priorities of patrol efficiency, coverage rate, and resource consumption;

[0069] S36. Output the optimization result of the patrol path optimization model M + , including the optimal patrol path selection X * , the optimal patrol path adjustment factor Λ * and the optimal path adaptive adjustment variable Δ * .

[0070] Optionally, the S4 includes the following steps:

[0071] S41. Initialize the particle swarm set according to the improved Markov random field model G + and the patrol path optimization model M + :

[0072] S = {s1, s2,..., s M};

[0073] Among them, M is the total number of particles, and each particle s j represents a patrol path plan, and the state vector is defined as:

[0074] sj = (s1j, s2j,..., s N j);

[0075] Among them, s ij represents the patrol path plan s jThe access status of the patrol point i, s ij = 1 indicates that the patrol path includes this patrol point, s ij = 0 indicates that it does not include;

[0076] Based on the improved Markov random field model G + , adopt a weighted random initialization strategy to generate the initial particle swarm:

[0077]

[0078] Among them, is the patrol priority of the patrol point i;

[0079] According to the path constraint conditions in the patrol path optimization model M + , eliminate the particles that do not meet the constraints, and use the random insertion strategy to complete the path, so that each particle meets all constraints after initialization:

[0080]

[0081] Among them, is the path adjustment operator, so that the initial solution of the particle swarm meets the constraint conditions of the patrol path optimization model M + ;

[0082] S42. Set the velocity vector of the particle:

[0083] v j (t) = (v 1j (t), v 2j (t),..., v Nj (t));

[0084] Among them, v ij (t) represents the search step size of the particle s j at the patrol point i, and the initial velocity is set to:

[0085] v ij (0) = η·(g best - s j (0));

[0086] Among them, η is the initialization weight coefficient, and g best is the global optimal solution;

[0087] S43. Calculate the fitness function of the particle. The fitness function combines the patrol efficiency optimization function O + in the patrol path optimization model M eff , the patrol coverage rate optimization function O cov and the resource consumption minimization function O res , and is defined as follows:

[0088] F(sj ) = w4O eff (s j ) + w5O cov (s j ) + w6O res (s j );

[0089] Among them, w4, w5, and w6 are the weight coefficients for the optimization of the patrol path;

[0090] S44. Update the particle position according to the standard particle swarm optimization algorithm:

[0091]

[0092] s j (t + 1) = s j (t) + v j (t + 1);

[0093] Among them, χ is the contraction factor, c1 and c2 are the learning factors, r1 and r2 are random numbers between [0, 1], is the historical optimal solution of particle s j , and g best is the global optimal solution;

[0094] If the updated position vector s j (t + 1) does not satisfy the constraint condition C of the patrol path optimization model M + , then adopt the path correction strategy:

[0095]

[0096] Adjust the path according to the optimization constraints;

[0097] S45. Adopt a random perturbation mechanism to enhance the search ability, and randomly perturb some particles with probability P in each iteration: mut For part of the particles:

[0098] s j (t + 2) = s j (t + 1) + μ j r mut ;

[0099] Among them, μ j is the perturbation amplitude, and r mut is a random number between [-1, 1];

[0100] S46. Set the termination condition of the standard particle swarm optimization algorithm. When the number of iterations reaches the maximum value T max or the fitness function F(s jWhen the change of () is less than the set threshold ∈, terminate the optimization and output the initial patrol path population S * :

[0101]

[0102] Among them, is the optimized patrol path plan, which meets the constraints C1 of the patrol path optimization model M + of the constraint condition C1.

[0103] Optionally, the S6 includes the following steps:

[0104] S61. According to the patrol path optimization result S * , construct a patrol path evaluation index matrix Q. The patrol path evaluation index matrix includes key patrol factors such as patrol time, patrol coverage rate, ancient tree safety, and resource consumption. In the patrol path evaluation index matrix, q ij represents the score of the patrol path plan under the i-th evaluation index;

[0105] S62. Use the adaptive hierarchical weighted sorting algorithm to calculate the final score of the patrol path The sorting rule is as follows:

[0106]

[0107] Among them, α i is the index weight, which is dynamically adjusted according to the patrol task requirements, represents the dynamic environment score of the patrol path , β is the environmental adaptability coefficient, and K represents the total number of dynamic environmental factors;

[0108] S63. According to the calculated patrol path score select the patrol path plan with the highest score from the patrol path population S * and according to the visualization requirements of the patrol path, convert the optimal patrol path plan into visualization path data, and through the wireless communication module, transmit the optimal patrol path plan to the APP side for display. The APP side renders the optimal patrol path in real time according to the patrol requirements and provides path navigation and task progress tracking functions.

[0109] The beneficial effects of the present invention are:

[0110] ​(1) The present invention uses an improved Markov random field to model the spatial dependence relationship in the patrol path, enabling the optimization of the patrol path not only based on single-point optimization but also considering the geographical proximity between patrol points, the similarity of environmental risks, and the risk of ancient tree disease transmission. By constructing an improved Markov random field to define node states and edge weights, the spatial dependence structure of the patrol path is optimized, making the overall planning of the patrol path more scientific and reasonable.

[0111] (2) The present invention uses a standard particle swarm optimization algorithm. By introducing a contraction factor and a path correction strategy, it ensures that the patrol path optimization process converges stably and can meet the patrol constraint conditions. By dynamically adjusting the search step size and the global optimal solution search strategy, the convergence efficiency and the quality of the optimal solution of the patrol path optimization are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0112] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. In the drawings:

[0113] Figure 1 is a flowchart of an intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization proposed by the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0114] Now, the present invention will be further described in detail with reference to the drawings. These drawings are all simplified schematic diagrams, only showing the basic structure of the present invention in a schematic way, so they only show the components related to the present invention.

[0115] Refer to Figure 1 , an intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization, includes the following steps:

[0116] S1. Install a positioning system for each ancient tree, collect multi-source data within the patrol area of the ancient trees, and perform standardization processing on the collected multi-source data to form a unified ancient tree patrol data set;

[0117] S2. Construct an improved Markov random field model based on the ancient tree patrol data set;

[0118] S3. Use the ancient tree patrol data set and the improved Markov random field model to construct a multi-objective and multi-constraint patrol path optimization model. The patrol path optimization model aims to optimize the patrol efficiency, optimize the patrol coverage rate, and reduce resource consumption, and stipulates the feasible patrol order between ancient trees and environmental adaptability constraints;

[0119] S4. Based on the prior information of the ancient tree patrol data set and the improved Markov random field model, generate an initial patrol path population through a standard particle swarm optimization algorithm, and the initial patrol path population meets the constraint conditions of the patrol path optimization model;

[0120] S5. Apply the standard particle swarm optimization algorithm to iteratively optimize the initial patrol path group. In each iteration, use the improved Markov random field model to feedback and correct the spatial dependency of the patrol path, and dynamically adjust the patrol path according to the data constraints in the ancient tree patrol dataset to form an updated patrol path group.

[0121] S6. According to the preset patrol path fitness evaluation standard, the patrol path with the highest fitness is selected from the updated patrol path group, and transmitted to the APP for display.

[0122] In this implementation, S1 includes the following steps:

[0123] S11. Install a positioning system on each ancient tree in the ancient tree patrol area, use the positioning system to obtain the geographical location information of the ancient tree, and use the three-dimensional coordinates (x i ,y i ,z i ) to form an ancient tree location information set P;

[0124] S12. Collect the historical and cultural information of ancient trees, including the cultural value, historical event association, age, species and protection level of each ancient tree, and assign a cultural value score to each ancient tree to form a historical and cultural information set C. The historical and cultural information set includes the cultural value scores of i ancient trees, and the scores are set according to historical documents and expert evaluations;

[0125] S13. Collect health data of ancient trees, including growth status, disease status, wilting degree and soil moisture content of ancient trees, and establish a health score for each ancient tree to form a health information set H, which includes the health scores of i ancient trees;

[0126] S14. Collect environmental risk information of patrol areas, including climate factors, soil conditions, air pollution index and human damage risk, and assign an environmental risk score to each patrol area to form an environmental risk information set R. The environmental risk information set includes the environmental risk score of i ancient trees, and the score is set based on historical meteorological data, air pollution index and surrounding environmental risk assessment;

[0127] S15. Perform data formatting on the ancient tree location information set P, historical and cultural information set C, health status information set H, and environmental risk information set R, remove abnormal data, fill in missing data, and use normalization method to process all scoring data to form a unified ancient tree patrol data set:

[0128] D={(p i ,c i ,h i ,ri ) | i = 1, 2, ..., N};

[0129] Among them, D represents the standardized ancient tree patrol dataset, and p i = (x i , y i , z i ) is the geographical coordinate of the i-th ancient tree, c i is the cultural value score of the i-th ancient tree, h i is the health status score of the i-th ancient tree, r i is the environmental risk score of the i-th ancient tree, and N is the total number of ancient trees in the patrol area.

[0130] In this embodiment, S2 includes the following steps:

[0131] S21. Construct an improved Markov random field model G based on the ancient tree patrol dataset D + :

[0132] G + = (V, E, Ψ + );

[0133] Among them, V represents the node set in the improved Markov random field model, and each node v i ∈V corresponds to the i-th ancient tree in the patrol area, E represents the edge set, representing the spatial dependence relationship between ancient trees, and dynamically adjusts the edge weights in combination with the continuity constraint of the patrol path:

[0134] E + = {e ij | v i , v j ∈V, i ≠ j, γ ij > τ};

[0135] Among them, γ ij is the patrol path continuity weight, τ is the path effectiveness threshold, and only when the path feasibility between two ancient trees meets the set conditions, an edge connection is established. Ψ + is the optimized potential function set, and realizes the dynamic optimization of the patrol priority in combination with the dynamic adaptation mechanism of the patrol strategy;

[0136] S22. Set the node state set X + in the improved Markov random field model G + , where the state x i of each ancient tree node represents the patrol priority of the ancient tree, and define the state variable set:

[0137] X + = {x1, x2, ..., x N};

[0138] where x i is the patrol priority of the i-th ancient tree;

[0139] S23. Combine the patrol path continuity constraint and the dynamic adaptive patrol strategy to optimize and improve the state transition probability distribution of the Markov random field model G + so that the patrol path planning adapts to environmental changes. The optimized joint probability is defined as:

[0140]

[0141] where λ ij is the path adjustment factor, which dynamically adjusts the dependence of the patrol path according to changes in the patrol environment;

[0142] S24. Calculate the optimized potential energy function Ψ + (X + , D):

[0143]

[0144] where φ + (x i , D) is the optimized single-node potential energy function, ψ + (x i , x j , D) is the optimized edge potential energy function, δ + (x i , D) is the patrol path dynamic adjustment function, and η i is the weight factor;

[0145] S25. Calculate the optimized single-node potential energy function φ + (x i , D):

[0146] φ + (x i , D) = w1c i + w2h i - w3r i + ξ i β i ;

[0147] where w1, w2, w3 are weight parameters, ξ i is the patrol emergency degree correction factor, and β i is the patrol plan adjustment factor;

[0148] S26. Calculate the optimized edge potential energy function ψ + (x i , x j , D):

[0149]

[0150] Among them, λ ij is the correlation parameter between patrol paths, d ij is the geographical distance between ancient tree i and ancient tree j, σ is the normalization parameter, and α ij is the dynamic environment factor, Ω ij calculates the influence degree of path emergencies based on real-time patrol data;

[0151] S27. Calculate the dynamic adjustment function δ + (x i , D):

[0152]

[0153] Among them, γ i is the adaptive adjustment parameter of the patrol path, ρ j is the completion degree of the patrol of the j-th ancient tree, N(i) is the neighboring patrol point of ancient tree i, and ∈ is the smoothing factor;

[0154] S28. Update and improve the Markov random field model G using the adaptive step size optimization method + :

[0155] X * = argmax P(X + |D);

[0156] Among them, P(X + |D) is the optimal state distribution under the condition of the given ancient tree patrol data set D, and X * is the optimized optimal patrol priority state set;

[0157] S29. Output the optimized improved Markov random field model G + = (V, E + , Ψ + ).

[0158] In this embodiment, S3 includes the following steps:

[0159] S31. Construct a multi-objective and multi-constraint patrol path optimization model M based on the ancient tree patrol data set D and the improved Markov random field model G + : + M

[0160] = (O, C1, Θ); + Among them, O is the set of patrol optimization objective functions, C1 is the set of patrol path optimization constraints, and Θ is the path optimization decision variable;

[0161] ​

[0162] S32. Determine the optimized objective function set O for the patrol path. The optimized objective function set for the patrol path includes the optimized function O for patrol efficiency eff , the optimized function O for patrol coverage cov and the function O for minimizing resource consumption res :

[0163]

[0164] where T ij is the patrol time from ancient tree i to ancient tree j, and x ij is the path decision variable. When the patrol path selects from i to j, x ij = 1; otherwise it is 0.

[0165]

[0166] where γ i is the reachability factor of the patrol path. When ancient tree i is covered by the patrol, x i = 1; otherwise it is 0.

[0167]

[0168] where E ij is the energy consumption of the patrol path, and R ij is the resource usage of the patrol path.

[0169] S33. Determine the optimized constraint set C1 for the patrol path. The optimized constraint set for the patrol path includes the patrol feasibility constraint C feas , the patrol sequence constraint C seq and the environmental adaptability constraint C env :

[0170]

[0171]

[0172]

[0173] where W j is the residence time at patrol point j, T max is the maximum allowable time, and R max is the maximum available resource quantity.

[0174] S34. Set the path optimization decision variable Θ. The path optimization decision variable includes the patrol path selection variable X, the patrol weight adjustment variable Λ, and the path adaptive adjustment variable Δ;

[0175] S35. According to the optimized model M for the patrol path + , combined with the improved Markov random field model G+ Calculate the optimal decision variable Θ of the patrol path * :

[0176]

[0177] where w1, w2, and w3 are weight coefficients that control the optimization priorities of patrol efficiency, coverage rate, and resource consumption;

[0178] S36. Output the optimized result of the patrol path optimization model M + including the optimal patrol path selection X * the optimal patrol path adjustment factor Λ * and the optimal path adaptive adjustment variable Δ * .

[0179] In this embodiment, S4 includes the following steps:

[0180] S41. Initialize the particle swarm set according to the improved Markov random field model G + and the patrol path optimization model M + :

[0181] S = {s1, s2,..., s M};

[0182] where M is the total number of particles, and each particle s j represents a patrol path plan, and the state vector is defined as:

[0183] sj = (s1j, s2j,..., s N j);

[0184] where s ij represents the access status of the patrol path plan s j at the patrol point i, s ij = 1 indicates that the patrol path includes this patrol point, and s ij = 0 indicates that it does not include;

[0185] Based on the improved Markov random field model G + , use the weighted random initialization strategy to generate the initial particle swarm:

[0186]

[0187] where, is the patrol priority of the patrol point i;

[0188] According to the path constraint conditions in the patrol path optimization model M + , eliminate the particles that do not meet the constraints and use the random insertion strategy to complete the path so that each particle meets all constraints after initialization:

[0189]

[0190] Among them, is a path adjustment operator, which makes the initial solution of the particle swarm satisfy the constraint conditions of the patrol path optimization model M + ;

[0191] S42. Set the velocity vector of the particle:

[0192] v j (t) = (v 1j (t), v 2j (t),..., v Nj (t));

[0193] Among them, v ij (t) represents the search step size of particle s j at the patrol point i, and the initial velocity is set as:

[0194] v ij (0) = η·(g best - s j (0));

[0195] Among them, η is the initialization weight coefficient, and g best is the global optimal solution;

[0196] S43. Calculate the fitness function of the particle. The fitness function combines the patrol efficiency optimization function O + in the patrol path optimization model M eff , the patrol coverage rate optimization function O cov and the resource consumption minimization function O res , and is defined as follows:

[0197] F(s j ) = w4O eff (s j ) + w5O cov (s j ) + w6O res (s j );

[0198] Among them, w4, w5, w6 are the weight coefficients of the patrol path optimization;

[0199] S44. Update the particle position according to the standard particle swarm optimization algorithm:

[0200]

[0201] s j (t + 1) = s j (t) + v j (t + 1);

[0202] Among them, χ is the contraction factor, c1 and c2 are learning factors, and r1 and r2 are random numbers between [0, 1]. is the historical optimal solution of particle s j , and g best is the global optimal solution;

[0203] If the updated position vector s j (t + 1) does not satisfy the constraint condition C of the patrol path optimization model M + , then adopt the path correction strategy:

[0204]

[0205] Adjust the path according to the optimization constraints;

[0206] S45. Adopt a random perturbation mechanism to enhance the search ability, and randomly perturb some particles with probability P mut in each iteration:

[0207] s j (t + 2) = s j (t + 1) + μ j r mut ;

[0208] Among them, μ j is the perturbation amplitude, and r mut is a random number between [-1, 1];

[0209] S46. Set the termination condition of the standard particle swarm optimization algorithm. When the number of iterations reaches the maximum value T max or the change in the fitness function F(s j ) is less than the set threshold ∈, terminate the optimization and output the initial patrol path population S * :

[0210]

[0211] Among them, is the optimized patrol path plan, which satisfies the constraint condition C1 of the patrol path optimization model M + .

[0212] In this embodiment, S6 includes the following steps:

[0213] S61. According to the patrol path optimization result S * , construct a patrol path evaluation index matrix Q. The patrol path evaluation index matrix includes key patrol factors such as patrol time, patrol coverage rate, ancient tree safety, and resource consumption. In the patrol path evaluation index matrix, q ijRepresents the patrol path plan The score under the i-th evaluation indicator;

[0214] S62. Adopting adaptive hierarchical weighted sorting algorithm to calculate the final score of patrol path The sorting rules are as follows:

[0215]

[0216] Among them, α i The indicator weight is dynamically adjusted according to the patrol task requirements. Indicates patrol path The dynamic environmental score, β is the environmental adaptability coefficient, and K represents the total number of dynamic environmental factors;

[0217] S63. Based on the calculated patrol path score From the patrol path group S * Select the patrol path plan with the highest score And according to the needs of patrol path visualization, the optimal patrol path plan It is converted into visual path data, and the optimal patrol path plan is transmitted to the APP for display through the wireless communication module. The APP renders the optimal patrol path in real time according to patrol needs, and provides path navigation and task progress tracking functions.

[0218] Embodiment 1:

[0219] At 8 a.m. on May 10, 2024, patrol captain A of the Forestry Bureau of City A opened the "Ancient Tree Patrol Intelligent Management System" APP, and the system automatically pushed an "Emergency Patrol Route Update Notification", indicating that due to the sudden thunderstorm last night, some ancient trees in Area B may have been damaged, and the patrol route for the day needed to be adjusted to prioritize the inspection of the disaster situation.

[0220] A checked the notification details, and the system automatically generated a new patrol route recommendation, focusing on covering the ancient trees numbered #208, #215, #219, and #225 in Zone B. Due to abnormal soil moisture in the area and a high probability of being affected by lightning strikes, the recommended route was 3.4 kilometers shorter than the original patrol route, while covering more high-risk ancient trees. The patrol time was expected to be reduced by 23 minutes. After A confirmed the adjustment plan, the system automatically pushed the patrol task to patrolmen B and C, and provided patrol point details, recommended patrol sequence, and expected completion time information.

[0221] At 9:15, Person B and Person C arrived at Area B. The APP showed their current locations and the progress of the patrol route. The patrol route guided them to go to Ancient Tree #208 first. This ancient tree is an ancient camphor tree with an age of over 320 years. Due to its historical and cultural value, it is listed as a first-class protected ancient tree. After arrival, the patrol officer scanned the identity code of the ancient tree through the APP, and the system automatically retrieved the historical health data, the latest soil humidity monitoring value, and the pest and disease records of this ancient tree.

[0222] The patrol officer inspected and found that there were obvious cracks in the main trunk of Ancient Tree #208, some branches were broken, and the bark was damaged. It was initially judged that the tree damage was caused by a thunderstorm. The APP prompted the patrol officer to record the damage situation and upload photos. At the same time, the system automatically synchronized the data to the ancient tree database and recommended that the Forestry Bureau arrange professional tree care personnel to conduct emergency treatment on it. After the patrol officer selected "Mark as a key patrol object", the system automatically adjusted the patrol plan for the next three months and increased the patrol frequency of this ancient tree to once a week.

[0223] At 10:05, the patrol officer went to Ancient Tree #215. This ancient tree is located in a relatively secluded forest area with dense surrounding vegetation. The APP prompted that there was less human activity in this area, but the soil humidity detection data showed that the water content in this area had increased sharply in the past 24 hours, and there might be a risk of tree toppling. After the patrol officer arrived, it was found that Ancient Tree #215 had indeed tilted, the soil at the root was loose, and the crack on the left side of the trunk extended to the main trunk. According to the system prompt, the patrol officer measured the tilting angle of the tree body and uploaded photos. After the system integrated the data, it pushed a "Tree Toppling Warning" to the Forestry Bureau of City A, suggesting to immediately arrange reinforcement or relocation treatment.

[0224] At 12:30, the patrol task was basically completed. The APP showed that 95% of the daily patrol goals had been achieved. Compared with the original patrol route, the optimized patrol plan reduced the patrol distance by 4.7 kilometers, covered 3 additional damaged ancient trees, and shortened the patrol time by 42 minutes. The system summarized the patrol data of this time and recorded the optimization suggestions into the patrol model to improve the accuracy of future path recommendations.

[0225] The following table shows the comparison of the patrol route optimization data between the present invention and the prior art:

[0226]

[0227] Through this implementation, the feasibility and effectiveness of the present invention are proved, which are mainly reflected in the following points:

[0228] 1. Dynamically adjust the patrol route to improve the patrol response ability: Compared with the traditional fixed patrol plan, the present invention adjusts the patrol route in real time, accurately locks high-risk areas, reduces ineffective patrols, and improves resource utilization.

[0229] 2. Intelligent patrol task allocation to optimize patrol efficiency: The system automatically assigns tasks to ensure that patrolmen can efficiently cover key ancient trees, reducing the patrol time by 19.6% and increasing the patrol coverage rate by 3%.

[0230] 3. Data-driven decision-making to enhance the scientific nature of patrol: By using Markov random field to model the patrol space relationship and combining with standard particle swarm optimization to dynamically adjust the patrol path, the patrol efficiency is improved, the resource consumption is reduced by 13%, providing a more scientific management means for ancient tree protection.

[0231] The present invention uses an improved Markov random field to model the spatial dependence relationship in the patrol path, so that the optimization of the patrol path is not only based on single-point optimization, but also takes into account the geographical proximity between patrol points, the similarity of environmental risks and the risk of ancient tree disease transmission. By constructing an improved Markov random field to define the node state and edge weight, the spatial dependence structure of the patrol path is optimized, making the overall planning of the patrol path more scientific and reasonable.

[0232] The present invention uses a standard particle swarm optimization algorithm. By introducing a contraction factor and a path correction strategy, it ensures that the patrol path optimization process converges stably and can meet the patrol constraint conditions. By dynamically adjusting the search step size and the global optimal solution search strategy, the convergence efficiency and the quality of the optimal solution of the patrol path optimization are improved.

[0233] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. An intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization, characterized in that, The steps include: S1. Install a positioning system for each ancient tree, collect multi-source data in the ancient tree patrol area, and standardize the collected multi-source data to form a unified ancient tree patrol data set; S2. Constructing an improved Markov random field model based on the ancient tree patrol data set; S3. Using the ancient tree patrol dataset and the improved Markov random field model, a multi-objective and multi-constrained patrol path optimization model was constructed. The patrol path optimization model aims to optimize patrol efficiency, optimize patrol coverage, and reduce resource consumption, and stipulates the feasible patrol sequence and environmental adaptability constraints between ancient trees. S4. Based on the ancient tree patrol data set and the prior information of the improved Markov random field model, the initial patrol path group is generated by the standard particle swarm optimization algorithm, and the initial patrol path group meets the constraints of the patrol path optimization model; S5. Apply the standard particle swarm optimization algorithm to iteratively optimize the initial patrol path group. In each iteration, use the improved Markov random field model to feedback and correct the spatial dependency of the patrol path, and dynamically adjust the patrol path according to the data constraints in the ancient tree patrol dataset to form an updated patrol path group. S6. According to the preset patrol path fitness evaluation standard, the patrol path with the highest fitness is selected from the updated patrol path group, and transmitted to the APP for display.

2. The intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization according to claim 1, wherein The S1 comprises the following steps: S11. Install a positioning system on each ancient tree within the patrol area of ancient trees, use the positioning system to obtain the geographical location information of the ancient trees, and store it in the form of three-dimensional coordinates (x i , y i , z i ), forming a set P of ancient tree location information; S12. Collect the historical and cultural information of ancient trees, including the cultural value, historical event association, age, species and protection level of each ancient tree, and assign a cultural value score to each ancient tree to form a historical and cultural information set C. The historical and cultural information set includes the cultural value scores of i ancient trees, and the scores are set according to historical documents and expert evaluations; S13. Collect health data of ancient trees, including growth status, disease status, wilting degree and soil moisture content of ancient trees, and establish a health score for each ancient tree to form a health information set H, which includes the health scores of i ancient trees; S14. Collect environmental risk information of patrol areas, including climate factors, soil conditions, air pollution index and human damage risk, and assign an environmental risk score to each patrol area to form an environmental risk information set R. The environmental risk information set includes the environmental risk score of i ancient trees, and the score is set based on historical meteorological data, air pollution index and surrounding environmental risk assessment; S15. Perform data formatting on the ancient tree location information set P, historical and cultural information set C, health status information set H, and environmental risk information set R, remove abnormal data, fill in missing data, and use normalization method to process all scoring data to form a unified ancient tree patrol data set: D = {(p i , c i , h i , r i ) | i = 1, 2,..., N}; Among them, D represents the standardized ancient tree patrol dataset, p i =(x i , y i , z i ) is the geographical coordinate of the i-th ancient tree, c i is the cultural value score of the i-th ancient tree, h i is the health status score of the i-th ancient tree, r i is the environmental risk score of the i-th ancient tree, and N is the total number of ancient trees in the patrol area.

3. The intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization according to claim 1, characterized in that, The S2 comprises the following steps: S21. Construct an improved Markov random field model G based on the ancient tree patrol dataset D + : G + = (V, E, Ψ + ); Among them, V represents the set of nodes in the improved Markov random field model, and each node v in the node set i ∈ V corresponds to the i-th ancient tree in the patrol area. E represents the edge set, which represents the spatial dependence relationship between ancient trees. At the same time, the edge weights are dynamically adjusted in combination with the continuity constraint of the patrol path: E + = {e ij | v i , v j ∈ V, i ≠ j, γ ij > τ}; Among them, γ ij is the weight of the continuity of the patrol path, τ is the threshold of path effectiveness. An edge connection is established only when the path feasibility between two ancient trees meets the set conditions. Ψ + is the set of optimized potential energy functions. Combining the dynamic adaptation mechanism of the patrol strategy, the dynamic optimization of the patrol priority is realized; S22. Set the improved Markov random field model G + The node state set X in + , where the state x of each ancient tree node i represents the patrol priority of the ancient tree, and define the set of state variables: X + = {x1, x2,..., x N}; where x i is the patrol priority of the i-th ancient tree; S23. Combine the patrol path continuity constraint with the dynamic adaptive patrol strategy to optimize and improve the Markov random field model G + 's state transition probability distribution, so that the patrol path planning adapts to environmental changes. The optimized joint probability is defined as: Among them, λ ij is a path adjustment factor that dynamically adjusts the dependency relationship of the patrol path according to changes in the patrol environment; S24. Calculate the optimized potential energy function Ψ + (X + , D): Among them, φ + (x i , D) is the optimized single-node potential energy function, ψ + (x i , x j , D) is the optimized edge potential energy function, δ + (x i , D) is the scouting path dynamic adjustment function, η i is the weight factor; S25. Calculate the optimized single-node potential energy function φ + (x i , D): φ + (x i , D) = w1c i + w2h i - w3r i + ξ i β i ; Among them, w1, w2, w3 are weight parameters, and ξ i is the correction factor for the urgency of patrol, and β i is the adjustment factor for the patrol plan; S26. Calculate the optimized edge potential function ψ + (x i , x j , D): Among them, λ ij is the correlation parameter between patrol paths, d ij is the geographical distance between ancient tree i and ancient tree j, σ is the normalization parameter, α ij is the dynamic environment factor, Ω ij calculates the impact degree of path emergencies based on real-time patrol data; S27. Calculate the dynamic adjustment function δ of the patrol path + (x i , D): where γ i is the adaptive adjustment parameter of the patrol path, ρ j is the patrol completion degree of the j-th ancient tree, N(i) is the neighboring patrol point of ancient tree i, and ∈ is the smoothing factor; S28. Update and improve the Markov random field model G using an adaptive step size optimization method + : X * = argmax P(X + | D); Among them, P(X + |D) is the optimal state distribution under the condition of the given ancient tree patrol dataset D, and X * is the optimized optimal patrol priority state set; S29. Output the optimized improved Markov random field model G + =(V, E + , Ψ + ).

4. An intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization according to claim 1, characterized in that, The S3 comprises the following steps: S31. Based on the ancient tree patrol dataset D and the improved Markov random field model G + Construct a multi-objective and multi-constraint patrol path optimization model M + : M + =(O, C1, Θ); Among them, O is the patrol optimization objective function set, C1 is the patrol path optimization constraint set, and Θ is the path optimization decision variable; S32. Determine the optimized objective function set O for the patrol path. The optimized objective function set for the patrol path includes the optimized function O for patrol efficiency eff , the optimized function O for patrol coverage rate cov , and the function O for minimizing resource consumption res : Among them, T ij is the patrol time from ancient tree i to ancient tree j, and x ij is the path decision variable. When the patrol path selects from i to j, x ij = 1, otherwise it is 0; Among them, γ i is the reachability factor of the patrol path. When the ancient tree i is covered by the patrol, x i = 1, otherwise it is 0; Among them, E ij is the energy consumption of the patrol path, and R ij is the resource usage of the patrol path; S33. Determine the optimized patrol path constraint set C1, and the optimized patrol path constraint set includes the patrol feasibility constraint C feas , the patrol sequence constraint C seq , and the environmental adaptability constraint C env : Among them, W j is the residence time at the patrol point j, T max is the maximum allowable time, R max is the maximum available resource quantity; S34. Set the path optimization decision variable Θ, the path optimization decision variable includes the patrol path selection variable X, the patrol weight adjustment variable Λ and the path adaptive adjustment variable Δ; S35. According to the patrol path optimization model M + , combined with the improved Markov random field model G + calculate the optimal decision variable Θ of the patrol path * : Among them, w1, w2, and w3 are weight coefficients that control the optimization priorities of patrol efficiency, coverage rate, and resource consumption; S36. Output the optimization result of the patrol path optimization model M + , including the selection X of the optimal patrol path * , the adjustment factor Λ of the optimal patrol path * and the adaptive adjustment variable Δ of the optimal path * .

5. The intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization according to claim 1, characterized in that The said S4 includes the following steps: S41. According to the improved Markov random field model G + and the patrol path optimization model M + Initialize the particle swarm set: S = {s1, s2,..., s M}; where M is the total number of particles, and each particle s j represents a scouting path plan, and the state vector is defined as: sj = (s1j, s2j,..., s N j); Among them, s ij represents the patrol path plan s j at the access status of the patrol point i, s ij = 1 indicates that the patrol path includes this patrol point, s ij = 0 indicates not included; Based on the improved Markov random field model G + , a weighted random initialization strategy is adopted to generate the initial particle swarm: where x i + is the patrol priority of patrol point i; According to the path constraint conditions in the patrol path optimization model M + exclude the particles that do not meet the constraints, and use the random insertion strategy to complete the path, so that each particle satisfies all constraints after initialization: Among them, is a path adjustment operator, which enables the initial solution of the particle swarm to satisfy the constraint conditions of the patrol path optimization model M + . S42. Set the velocity vector of the particle: v j (t) = (v 1j (t), v 2j (t),..., v Nj (t)); Among them, v ij (t) represents the search step size of particle s j at the patrol point i, and the initial velocity is set as: v ij (0) = η·(g best -s j (0)); Among them, η is the initial weight coefficient, and g best is the global optimal solution; S43. Calculate the fitness function of the particles. The fitness function combines the scouting path optimization model M + with the scouting efficiency optimization function O eff , the scouting coverage rate optimization function O cov and the resource consumption minimization function O res , which are defined as follows: F(s j ) = w4O eff (s j ) + w5O cov (s j ) + w6O res (s j ); Among them, w4, w5, and w6 are weight coefficients for optimizing the patrol path; S44. Update the particle position according to the standard particle swarm optimization algorithm: s j (t + 1) = s j (t) + v j (t + 1); Among them, χ is the contraction factor, c1 and c2 are the learning factors, and r1 and r2 are random numbers between [0, 1]. is the historical optimal solution of particle s j and g best is the global optimal solution. If the updated position vector s j (t + 1) does not satisfy the constraints C of the patrol path optimization model M + , then the path correction strategy is adopted: Adjust the path according to the optimization constraints; S45. Enhance the search ability by adopting a random perturbation mechanism, and with probability P in each iteration mut perform random perturbation on some particles: s j (t + 2) = s j (t + 1) + μ j r mut ; where μ j is the perturbation amplitude, and r mut is a random number between [-1, 1]; S46. Set the termination condition of the standard particle swarm optimization algorithm. When the number of iterations reaches the maximum value T max or the change in the fitness function F(s j ) is less than the set threshold ∈, terminate the optimization and output the initial scout path population S * : Among them, is the optimized patrol path plan that satisfies the constraint condition C1 of the patrol path optimization model M + .

6. The intelligent recommendation method for the patrol route of cultural ancient trees based on particle swarm optimization according to claim 1, wherein, The said S6 includes the following steps: S61. According to the optimized result S of the patrol route * , construct a patrol route evaluation index matrix Q. The patrol route evaluation index matrix includes key factors of patrol time, patrol coverage rate, ancient tree safety, and resource consumption. In the patrol route evaluation index matrix, q ij represents the patrol route plan score under the i-th evaluation index; S62. Calculate the final score of the patrol path using the adaptive hierarchical weighted sorting algorithm The sorting rules are as follows: Among them, α i is the index weight, which is dynamically adjusted according to the requirements of the patrol task, represents the dynamic environment score of the patrol path , β is the environmental adaptability coefficient, and K represents the total number of dynamic environmental factors; According to the calculated patrol path score Select the patrol path plan with the highest score from the patrol path group S * According to the visualization requirements of the patrol path, convert the optimal patrol path plan into visualization path data, and transmit the optimal patrol path plan to the APP side for display through the wireless communication module. The APP side renders the optimal patrol path in real time according to the patrol requirements, and provides functions such as path navigation and task progress tracking.​

Images