Remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithm

Abstract

A remote sensing agricultural monitoring energy-saving path planning method based on a discrete algorithm comprises the following steps: firstly, setting an initial path set and upper and lower bounds of variables to initialize a population, and decoding through a specific rule to determine a monitoring point position access sequence; sorting and grouping according to fitness, and simulating carnivorous plant and prey environment; an attraction probability is introduced in the growth stage, carnivorous plants or preys grow according to respective models according to the sizes of the carnivorous plants or preys, and the offspring quality is improved by adopting an improved position updating formula; in the breeding stage, only the optimal carnivorous plant is bred according to a specific model, the breeding process is improved, and different breeding formulas are selected according to random numbers; the neighborhood perturbation strategy is optimized to avoid the too far distance between monitoring points, new and old populations are combined and sorted through recombination operation, and the first nn individuals are selected to maintain the population scale, so that efficient and energy-saving path planning is realized. According to the method, the optimal energy-saving path can be obtained while all farmland areas are covered by the monitoring points, and the solving quality and efficiency of the remote sensing agricultural monitoring energy-saving path planning problem are improved.

Description

Technical Field

[0001] This invention relates to the field of remote sensing agricultural monitoring technology, and in particular to an energy-saving path planning method for remote sensing agricultural monitoring based on discrete algorithms. Background Technology

[0002] In recent years, with the rapid development of remote sensing technology, its application in agriculture, logistics, healthcare, and other fields has been increasing. Especially in agriculture, remote sensing agricultural monitoring has become an emerging method, capable of efficiently and accurately monitoring farmland. However, how to plan an optimal energy-saving path for remote sensing agricultural monitoring to ensure coverage of all farmland areas and efficiently complete monitoring tasks has become a pressing technical problem.

[0003] Traditional path planning methods, such as genetic algorithms, particle swarm optimization, and ant colony optimization, have achieved certain results in some applications, but they still have some limitations when solving the energy-saving path planning problem in remote sensing agriculture. For example, these methods may get trapped in local optima, or they may fail to find solutions that meet practical needs in large-scale farmland areas.

[0004] To address these issues, researchers have begun exploring new optimization algorithms and methods. Among them, the energy-saving path planning method for remote sensing agricultural monitoring based on discrete algorithms has attracted widespread attention. This method combines traditional discrete optimization algorithms with modern artificial intelligence technology, aiming to find an energy-saving path for remote sensing agricultural monitoring that both meets practical needs and has high optimization performance. Summary of the Invention

[0005] The technical problem to be solved by this invention is to address the shortcomings of existing technologies by providing a remote sensing agricultural monitoring energy-saving path planning method based on a discrete algorithm. This method is based on an improved carnivorous plant algorithm, which combines various improvement methods at each stage of the carnivorous plant algorithm. This can improve the global and local search capabilities of the algorithm, ensuring that the monitoring points cover all farmland areas while obtaining the optimal energy-saving path, thus effectively improving the solution quality and efficiency of the remote sensing agricultural monitoring energy-saving path planning problem.

[0006] The technical problem to be solved by this invention is achieved through the following technical solution. This invention is a remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithms. First, an initial path set and upper and lower bounds of variables are set to initialize the population. The access order of monitoring points is determined by decoding according to specific rules. Then, the plants are sorted and grouped according to fitness to simulate the environment of carnivorous plants and prey. During the growth stage, an attraction probability is introduced, and carnivorous plants or prey grow according to their respective models based on their size. An improved position update formula is also used to improve the quality of offspring. During the reproduction stage, only the optimal carnivorous plants reproduce according to a specific model, and the reproduction process is improved and different reproduction formulas are selected based on random numbers. The neighborhood perturbation strategy is optimized to avoid excessive distance between monitoring points. Finally, the old and new populations are merged and sorted through a recombination operation, and the top n individuals are selected to maintain the population size, so as to achieve efficient and energy-saving path planning.

[0007] The technical problem to be solved by this invention can also be further achieved through the following technical solution: the steps of the remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithms described above are as follows: (1) Encoding: Let the initial set of paths be X 0 =( X 0 1, X 0, 2, ...), determine the upper bound of the variable's values. a and the lower world b Variables in a and b For solving the CPA problem of energy-saving path planning in remote sensing agricultural monitoring, the population is randomly and uniformly selected. During population initialization, let... a =0, b = m Therefore, the i-th individual in the initial population is initialized as follows: m *rand(1, m ), i =1,2,…, nn (1) In the formula, rand (1 ,m () is a random vector between [0,1]; (2) Decoding: Let the population size be nn The number of monitoring points is m, with the k-th individual in the population as the starting point. X k =( x 1, x 2,…, x m Taking (e.g., ) as an example, the specific decoding steps are as follows: Step 1: Randomly generate an integer μ between [1, m], and use μ as the initial remote sensing agricultural monitoring point; Step 2: Delete X k The vector following the μ-th component is X ' k , X ' k , X ' k =( x 1, x 2,…, x μ-1 , x μ+1 , …, x m Based on the distance between μ and the other monitoring points d =( d μ,1 , d μ,2 , …, d μ,μ-1 , d μ,μ+1 , …, d μ,m ), calculate the vector related to the distance between monitoring points and the size of gene loci. Tp , Tp The calculation formula is (2) In the formula, Ä represents vector d and vector d. Multiply elements in the same position; Step 3: Find the vector Tp The minimum value, and the monitoring point corresponding to the minimum value. i As the next monitoring point to be visited; Step 4: Order μ = i Repeat Step 2 and Step 3 to determine the access order of all monitoring points; (3) Grouping: Let the population size be nn Sort the individuals in the population according to their fitness values ​​from smallest to largest, and select the n best individuals as carnivorous plants, and the rest... n One individual is treated as prey. The grouping process simulates the environment of carnivorous plants and their prey. Each individual in the group consists of one carnivorous plant and multiple prey. The best prey is attracted by the best carnivorous plant, the second best prey is attracted by the second best carnivorous plant, and so on. The n1th prey is assigned to the nth carnivorous plant. Let the population before sorting be... X = ( X 1, X 2, ..., X n + n 1), the sorted population X ´=( X ’ 1, X ’ 2,. . . , X ’ n+n ),satisfy F ( X ’ 1)≤ F ( X ’ 2)≤…≤ F ( X ’ n+n ); (4) Growth stage: The carnivorous plant algorithm introduces an attraction probability γ, where γ = 0.8; (5) If the attraction probability γ is greater than the random number l This means that the carnivorous plant has successfully lured its prey. The growth model of the carnivorous plant is as follows: (3) (4) In the formula, Ä represents the multiplication of elements at the same position of two vectors. xp i Refers to the first i The carnivorous plants in the group, p iv Refers to the first i The first in the group v One prey, rand Let be a random vector between [0, 1]. gr The growth rate is usually taken as 2; (6) When the attraction probability γ is less than the random number l, the prey is not attracted by the carnivorous plant and continues to grow. The mathematical model for prey growth is: (5) (6) In the formula, Ä represents the multiplication of elements at the same position of two vectors. p iu , p iv They refer to the i-th group and the i-th u The and the first v There are 1 target prey, and rand is a random vector between [0, 1]. f ( p iv )and f ( p iu) respectively refer to the first i The first in the group v The and the first u The fitness value of each prey; (7) Adaptive attraction probability formula: The formula for calculating the probability of attraction is: (7) In the formula, g is the attraction probability, m is the number of monitoring points, and r is... xp i and p iv The distance between them; xp i and p iv Distance between r The calculation formula is as follows: (8) In the formula, n For the number of carnivorous plants, n 1 represents the number of prey. xp i For the carnivorous plants in the i-th group. p iv For the first i The first in the group v One prey; (8) Adopt the improved position update formula: This formula improves the quality of offspring produced during the growth stage by increasing the attraction term of the optimal carnivorous plants to prey in the population and the attraction term of carnivorous plants to prey in the same group, as detailed below: (9) (10) (11) (12) In the formula, Ä represents the multiplication of elements at the same position of two vectors. xp i Refers to the i-th group of carnivorous plants. xp 1 refers to the best individual in the population. p iv , p iu and p iw They refer to the first i For the v-th, u-th, and w-th prey in a group, rand is a random vector between [0, 2, 1], and rand1 is a random vector between [-1, 1]. grLet m be the growth rate, m be the number of monitoring points, and r be the value in equation (9). xp i and p iv The distance between them, r is in equation (10) p iv and p iu The distance between them, where t is the current algorithm running time. t max This represents the maximum running time of the algorithm. (9) When the attraction probability γ is greater than the random number l, the carnivorous plant grows according to equation (9); otherwise, the prey updates its position according to equation (10). (10) Reproduction stage: Only the top-ranked carnivorous plant, i.e., the best solution in the population, can perform the reproduction operation. This helps reduce the time complexity of the algorithm. The mathematical model for optimal carnivorous plant reproduction is as follows: (13) (14) In the formula, Ä represents the element-wise multiplication of two vectors at the same position, and rand is a random vector between [0, 1]. xp 1 represents the optimal individual in the population, and μ is the reproduction rate, which is usually 1.8. xp i and xp j These refer to the carnivorous plants in the i-th and j-th groups, respectively. f ( xp i )and f ( xp j ) refer to the fitness values ​​of carnivorous plants in group i and group j, respectively; (11) The improved breeding process is as follows: (15) (16) In the formula, Ä represents the element-wise multiplication of two vectors at the same position, and β is an m-dimensional random vector in the range [0.5, 1.8]. xp i and xp j These refer to the carnivorous plants in the i-th and j-th groups, respectively. xp 1 is the optimal carnivorous plant for the population; (12) Improved breeding: A random number d between [0,1] is generated. When d ≥ 0.6, the carnivorous plant reproduces according to formula (15); otherwise, it reproduces according to formula (16). (13) Neighborhood perturbation strategy optimization: To avoid the situation where the distance between non-adjacent monitoring points in the randomly selected 4 monitoring points is too far; (14) Given a certain circuit X=(x1, x2,…, x) in the population m Taking m as an example, where m is the number of monitoring points, the specific steps for neighborhood disturbance are as follows: Step 1: From m Randomly select one monitoring point from the monitoring points. a ,by a With the center as the radius as the radius, r The circle containing 1 is called the neighborhood, denoted as . U ( a , r 1), r The formula for calculating 1 is shown in equation (17): (17) In the formula, Z is the route length of the best individual in the population; Step 2: Determine the neighborhood U ( a , r 1) The number of monitoring points included is u. If u≥2, two monitoring points are randomly selected from the neighborhood. aa 1, aa 2; If u=1, let aa 1= a Then in [1,2,…, a -1, a +1,…, m Randomly select a monitoring point as... aa 2; Step 3: In X Find with aa 1. Adjacent monitoring points bb 1, with aa 2 Adjacent monitoring points bb 2, and bb 1≠ aa 2, bb 2≠ aa 1. If bb 1= bb 2. Repeat Step 1 and Step 2 until... bb 1≠ bb 2; Step 4: Delete edges ( aa 1, bb1) and ( aa 2, bb 2), connecting edge ( bb 1, aa 2) and ( aa 1, bb 2); (15) Reorganization operation: This is used to ensure that the population obtained in each iteration is not inferior to the original population.

[0008] The technical problem to be solved by the present invention can also be further realized by the following technical solution: For the remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithm described above, the value range of random number l in the method is [0,1].

[0009] The technical problem to be solved by the present invention can also be further realized through the following technical solution. For the remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithm described above, the recombination operation process of step (15) is as follows: Step 1: Merge the newly generated carnivorous plant and prey populations to obtain a new population; Step 2: Sort the individuals in the new population according to their fitness values ​​from smallest to largest; Step 3: In order to keep the population size of the carnivorous plant algorithm unchanged, select the first n individuals from the new population.

[0010] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention can achieve a balance between global and local search capabilities: The algorithm proposed in this invention combines the advantages of global search and local search, ensuring that in the process of finding the optimal solution, it can both broadly explore the solution space and deeply search for better solutions in a certain region.

[0011] 2. The method of this invention has a high convergence speed: Compared with other algorithms, the algorithm of this invention exhibits a faster convergence speed in most cases. This means that the algorithm can find a better or near-optimal solution in a shorter time.

[0012] 3. The method of this invention has strong adaptability: It proposes a heuristic decoding method based on distance and bit value, which can show good performance in different metaheuristic algorithms.

[0013] 4. This invention has wide applicability: Although this algorithm is mainly aimed at the energy-saving path planning problem of remote sensing agricultural monitoring, the proposed method and ideas can be widely applied to other optimization problems and have high practical value.

[0014] In summary, this invention is an improved carnivorous plant algorithm that can effectively solve the energy-saving path planning problem in remote sensing agricultural monitoring. This algorithm combines multiple improvement methods to enhance the algorithm's global and local search capabilities, thereby improving the solution quality and efficiency of the energy-saving path planning problem in remote sensing agricultural monitoring. Attached Figure Description

[0015] Figure 1 This is a diagram illustrating the population grouping process of the present invention; Figure 2 This is a graph showing the trend of attraction probability γ as a function of distance r in the improved attraction probability model of this invention. Figure 3 The diagram shows the distribution of offspring individuals in the search space for the improved position update formula of this invention. Figure 4 This invention generates an excellent individual sequence graph for the neighborhood perturbation strategy. Figure 5 This is a flowchart of the algorithm of the present invention. Detailed Implementation

[0016] 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, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1: A remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithms: (1) Encoding method: Let the initial set of paths be X 0 =( X 0 1, X Before initializing the population (0, 2, ...), the upper and lower bounds of the variable values ​​need to be determined. After the upper and lower bounds of the variable values ​​are determined, the variable can be... a and b Randomly and uniformly selected from within; For CPA (Current Aptitude Test) problems in remote sensing agricultural monitoring energy-saving path planning, during population initialization, one can let... a =0, b = m Therefore, the first in the initial population i Individuals can be initialized to m *rand(1, m ), i =1,2,…, nn (1) In the formula, rand (1 ,m () is a random vector between [0,1]; (2) Decoding method: This decoding method is applicable to all metaheuristic algorithms for real number encoding and has wide applicability; Let the population size be nn The number of monitoring points is m , with the first in the population k individual X k =( x 1, x 2,…, x m Taking (e.g., ) as an example, the specific decoding steps are as follows: Step 1: Randomly generate an integer μ between [1, m], and use μ as the initial monitoring point; Step 2: Delete X k The vector following the μ-th component is X ' k , X ' k =( x 1, x 2,…, x μ-1 , x μ+1 ,…, x m Based on the distance between μ and the other monitoring points d =( d μ,1 , d μ,2 , …, d μ,μ-1 , d μ,μ+1 , …, d μ,m ), calculate the vector related to the distance between monitoring points and the size of gene loci. Tp , Tp The calculation formula is (2) In the formula, Ä represents vector d and vector d. Multiply elements in the same position; Step 3: Find the vector Tp The minimum value, and the monitoring point corresponding to the minimum value. i As the next monitoring point to be visited; Step 4: Order μ =i Repeat Step 2 and Step 3 to determine the access order of all monitoring points; by m Let's take 5 as an example to illustrate; set up μ =2, X k =(1.3575, 4.5155, 3.4863, 0.7845, 0.0637); The distances between the 5 monitoring points are shown in Table 1, and the specific decoding steps are shown in Table 2.

[0018] Table 1. Distances between the 5 monitoring points

[0019] Table 2 Examples of Decoding

[0020] As shown in Table 2, when μ =2, X k When the value is (1.3575, 4.5155, 3.4863, 0.7845, 0.0637), the route obtained according to the distance- and bit-value-based decoding method is 2®5®3®4®1®2.

[0021] From equation (2), we can see that Tp The magnitude of the distance between monitoring points is related to the distance between them. Therefore, in the route obtained after decoding, the probability of adjacent monitoring points being close is relatively high, which is beneficial to improving the quality of the decoded solution and increasing the convergence speed of the algorithm. Tp Size and X k The distance between adjacent monitoring points along the route is not necessarily the minimum, which is related to the size of the median value. This helps to maintain the diversity of the population and reduce the probability of the algorithm getting stuck in a local optimum.

[0022] (3) Grouping strategy: Let the population size be nn Sort the individuals in the population according to their fitness values ​​from smallest to largest, and select the best individuals in the population. n Some individuals are treated as carnivorous plants, while the rest... n One individual was used as prey. n 1> n ,and n 1 can be n (Divisible). The grouping process simulates the environment of carnivorous plants and their prey. Each group consists of one carnivorous plant and multiple prey. The best prey is attracted to the best carnivorous plant, the second best prey is attracted to the second best carnivorous plant, and so on. n One prey was given to the firstn A carnivorous plant.

[0023] For ease of explanation of the grouping process, let the population before sorting be . X = ( X 1, X 2, ..., X n + n 1) The sorted population X ´=( X ’ 1, X ’ 2,. . . , X ’ n+n ), satisfy F ( X ’ 1)≤ F ( X ’ 2)≤…≤ F ( X ’ n+n ), based on population size nn Taking a population of 12 as an example, there are... n =3 carnivorous plants, n 1 = 9 prey, the detailed grouping process is as follows Figure 1 As shown.

[0024] (4) Growth stage: Carnivorous plants grow by attracting, trapping, and digesting prey. Plants lure prey with scents and digest them to obtain the nutrients they need for growth, but prey can sometimes escape or remain unattracted. Therefore, the probability of attraction (CPA) is introduced.

[0025] (5) If the attraction probability γ ( γ =0.8) is greater than random number l This means that the carnivorous plant has successfully lured its prey. The growth model of a carnivorous plant is as follows: (3) (4) In the formula, Ä represents the multiplication of elements at the same position of two vectors. xp i Refers to the first i The carnivorous plants in the group, p iv Refers to the first i The first in the group v One prey, rand Let be a random vector between [0, 1]. gr The growth rate is usually taken as 2.

[0026] (6) When the probability of attraction γ Less than random number l (0£ lAt £1), the prey is no longer attracted to carnivorous plants and continues to grow. The mathematical model for prey growth is: (5) (6) In the formula, Ä represents the multiplication of elements at the same position of two vectors. p iu , p iv They refer to the first i The first in the group u The and the first v One prey, rand Let be a random vector between [0, 1]. f ( p iv )and f ( p iu ) respectively refer to the first i The first in the group v The and the first u The fitness value of each prey.

[0027] (7) Adaptive attraction probability formula: An adaptive formula for calculating the attraction probability based on distance and the number of monitoring points is adopted. This formula effectively improves the algorithm's global and local search capabilities. The formula for calculating the attraction probability is as follows: (7) In the formula, g To attract probability, m To determine the number of monitoring points, r for xp i and p iv The distance between them.

[0028] xp i and p iv Distance between r The calculation formula is as follows: (8) In the formula, n For the number of carnivorous plants, n 1 represents the number of prey. xp i For the first i The carnivorous plants in the group, p iv For the first i The first in the group v One prey.

[0029] (8) An improved position update formula is adopted; This formula improves the quality of offspring produced during the growth stage by increasing the attraction term of the optimal carnivorous plants to prey in the population and the attraction term of carnivorous plants to prey in the same group, as detailed below. (9) (10) (11) (12) In the formula, Ä represents the multiplication of elements at the same position of two vectors. xp i Refers to the first i Group of carnivorous plants, xp 1 refers to the best individual in the population. p iv , p iu and p iw They refer to the first i Group 1 v The, the u The and the first w One prey, rand Let be a random vector between [0, 2, 1]. ​ 1 is a random vector in the interval [-1, 1]. gr For growth rate, m To determine the number of monitoring points, in equation (9) r yes ​ i and p iv The distance between them is given in equation (10). r yes p iv and p iu The distance between, t This represents the current algorithm runtime. t max This represents the maximum running time of the algorithm.

[0030] (9) When the probability of attraction ​ ( ​ =0.8) is greater than random number l (0£ l When £1), carnivorous plants grow according to formula (9); otherwise, prey updates its position according to formula (10). (10) Reproduction stage: Only the top-ranked carnivorous plant, i.e., the best solution in the population, can undergo reproduction, which helps reduce the algorithm's time complexity. The mathematical model for optimal carnivorous plant reproduction is as follows: (13) (14) In the formula, Ä represents the multiplication of elements at the same position of two vectors. ​ Let be a random vector between [0, 1]. ​ 1 represents the optimal individual in the population. ​ The reproduction rate is typically 1.8. ​ i and ​ j They refer to the first i Group and No. j The carnivorous plants in the group, f ( ​ i )and f ( ​ j ) respectively refer to the first i Groups and j Fitness values ​​of carnivorous plants in the first group; (11) The improved breeding process is as follows: (15) (16) In the formula, Ä represents the multiplication of elements at the same position of two vectors. β for m A random vector in the dimension [0.5, 1.8]. ​ i and ​ j They refer to the first i Group and No. j Group of carnivorous plants, ​ 1 is the optimal carnivorous plant in the population.

[0031] (12) Reproduction strategy: Generate a random number between [0, 1]. d ,when d When the price is 0.6, carnivorous plants are propagated according to formula (15); otherwise, they are propagated according to formula (16).

[0032] (13) Neighborhood perturbation strategy optimization: This can increase the probability of producing outstanding individuals and avoid the phenomenon that the distance between non-adjacent monitoring points in the four randomly selected monitoring points is too far. (14) Using a certain circuit in the population X=( x 1, x 2,…, x m For example, m To determine the number of monitoring points, the specific steps for neighborhood disturbance are as follows: Step 1: From m Randomly select one monitoring point from the monitoring points. a ,by a With the center as the radius as the radius, r The circle containing 1 is called the neighborhood, denoted as . U ( a , r 1). r The formula for calculating 1 is shown in equation (17): (17) In the formula, Z This represents the route length of the optimal individual in the population.

[0033] Step 2: Determine the neighborhood U ( a , r 1) Number of monitoring points included u ,like u ≥2, randomly select 2 monitoring points in the neighborhood. ​ 1, ​ 2; if u =1, let ​ 1= a Then in [1,2,…, a -1, a +1,…, m Randomly select a monitoring point as... ​ 2; Step 3: In X Find with ​ 1. Adjacent monitoring points ​ 1, with ​ 2 Adjacent monitoring points ​ 2, and ​ 1≠ ​ 2, ​ 2≠ ​ 1. If ​ 1= ​ 2. Repeat Step 1 and Step 2 until... ​ 1≠ ​ 2; Step 4: Delete edges ( ​ 1, ​ 1) and ( ​ 2, ​ 2), connecting edge ( ​1, ​ 2) and ( ​ 1, ​ 2); (15) The following examples illustrate the energy-saving path planning for remote sensing agricultural monitoring with and without geometrically intersecting individuals in the population: 1) Randomly select an individual with geometrically intersecting edges from the population. X =(1, 2, 10, 9, 8, 7, 6, 5, 4, 3), let's assume... X Three monitoring points were randomly selected, and a circle was formed with point 3 as the center. r If the monitoring points within the neighborhood with a radius of 1 are 2 and 3, then... ​ 1=2, ​ 2=3. According to Step 3, ​ 1 = 10 ​ 2=11. Delete edge ( ​ 1, ​ 1) = (2, 10), ( ​ 2, ​ 2) = (3, 11), connect the edges ( ​ 1, ​ 2) = (10, 11) and ( ​ 1, ​ 2) = (2, 3). Individual X And the route obtained after neighborhood perturbation, d (2,10)+ d (3,11) =96, d (2,3)+ d (10,11) =41, the path length after reconnection is better than the path length before the perturbation, and the individual generated by the neighborhood perturbation is better than the individual. X ’ .

[0034] 2) Randomly select an individual from the population that does not contain a geometric cross edge. X ’ =(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11), let's say from... X ’ Six monitoring points were randomly selected, and a circle was formed with point 6 as the center. r The monitoring points within the neighborhood of radius 5, 6, 7, and 8 are 6, 7, and 8. Two monitoring points are randomly selected from this neighborhood, 6 and 7. ​ 1 = 6 ​ 2=7. According to Step 3, ​ 1 = 5 ​ 2=8. Delete edges (​ 1, ​ 1) = (6, 5), ( ​ 2, ​ 2) = (7, 8), connect the edges ( ​ 1, ​ 2) = (5, 7) and ( ​ 1, ​ 2) = (6, 8). Individual X And the route obtained after neighborhood perturbation, d (5,7)+ d (6,8)=46, d (5,6)+ d (7,8)=61, the path length after reconnection is better than the path length before the perturbation, and the individual generated by the neighborhood perturbation is better than the individual. X ’ .

[0035] (16) Reorganization operation: The recombination operation ensures that the population obtained in each iteration is no worse than the original population.

[0036] Step 1: Merge the newly generated carnivorous plant and prey populations to obtain a new population.

[0037] Step 2: Sort the individuals in the new population according to their fitness values ​​from smallest to largest.

[0038] Step 3: To maintain the population size of the carnivorous plant algorithm, select the top seed from the new population. ​ Individual.

[0039] This invention proposes an improved carnivorous plant algorithm for solving energy-saving paths in remote sensing agricultural monitoring. The algorithm mainly includes heuristic decoding methods, improved location update formulas, adaptive change probabilities, neighborhood perturbation, and other operations.

[0040] A heuristic decoding method that comprehensively considers distance and position value, unlike sorting and rounding methods, is heuristic and can therefore improve the quality of the decoded solution, and has wide applicability. Furthermore, inspired by the correlation between the fragrance concentration emitted by carnivorous plants and the distance between prey and the plant, an adaptive formula for calculating the attraction probability based on distance and the number of monitoring points is proposed. It strikes a good balance between the ability to explore and develop algorithms.

[0041] Secondly, inspired by the growth and reproduction patterns of carnivorous plants in nature, the reproduction strategy of standard CPA carnivorous plants has been improved. In this improvement, each carnivorous plant is allowed to reproduce, and the reproduction process of the optimal carnivorous plant in the population is different from that of other carnivorous plants. This can effectively reduce the probability of the algorithm getting stuck in local optima.

[0042] Finally, a neighborhood perturbation is introduced into the improved carnivorous plant algorithm to further enhance the probability of generating superior individuals. The algorithm's performance is validated using standard computational examples. The results show that, compared with algorithms proposed in current technology, the improved carnivorous plant algorithm proposed in this invention has the best solution accuracy, stability, and convergence speed.

Claims

1. A remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithms, characterized in that: First, an initial path set and upper and lower bounds for variables are set to initialize the population. The access order of monitoring points is determined by decoding according to specific rules. Then, the plants are sorted and grouped according to fitness to simulate the environment of carnivorous plants and prey. During the growth stage, an attraction probability is introduced, and carnivorous plants or prey grow according to their respective models based on their size. An improved position update formula is also used to improve the quality of offspring. During the reproduction stage, only the best carnivorous plants reproduce according to a specific model. The reproduction process is improved and different reproduction formulas are selected based on random numbers. The neighborhood perturbation strategy is optimized to avoid the monitoring points being too far apart. Finally, the old and new populations are merged and sorted through a recombination operation. The top n individuals are selected to maintain the population size, so as to achieve efficient and energy-saving path planning.

2. The remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithms according to claim 1, characterized in that: The steps of this method are as follows: (1) Encoding: Let the initial set of paths be X 0 =( X 0 1, X 0, 2, ...), determine the upper bound of the variable's values. a and the lower world b Variables in a and b For solving the CPA problem of energy-saving path planning in remote sensing agricultural monitoring, the population is randomly and uniformly selected. During population initialization, let... a =0, b = m Therefore, the i-th individual in the initial population is initialized as follows: m *rand(1, m ), i =1,2,…, nn (1) In the formula, rand (1 ,m () is a random vector between [0,1]; (2) Decoding: Let the population size be nn The number of monitoring points is m, with the k-th individual in the population as the starting point. X k =( x 1, x 2,…, x m Taking (e.g., ) as an example, the specific decoding steps are as follows: Step 1: Randomly generate an integer μ between [1, m], and use μ as the initial remote sensing agricultural monitoring point; Step 2: Delete X k The vector following the μ-th component is X ' k , X ' k , X ' k =( x 1, x 2,…, x μ-1 , x μ+1 , …, x m Based on the distance between μ and the other monitoring points d =( d μ,1 , d μ,2 , …, d μ,μ-1 , d μ,μ+1 , …, d μ,m ), calculate the vector related to the distance between monitoring points and the size of gene loci. Tp , Tp The calculation formula is （2） In the formula, Ä represents vector d and vector d. Multiply elements in the same position; Step 3: Find the vector Tp The minimum value, and the monitoring point corresponding to the minimum value. i As the next monitoring point to be visited; Step 4: Order μ = i Repeat Step 2 and Step 3 to determine the access order of all monitoring points; (3) Grouping: Let the population size be nn Sort the individuals in the population according to their fitness values ​​from smallest to largest, and select the n best individuals as carnivorous plants, and the rest... n One individual is treated as prey. The grouping process simulates the environment of carnivorous plants and their prey. Each individual in the group consists of one carnivorous plant and multiple prey. The best prey is attracted by the best carnivorous plant, the second best prey is attracted by the second best carnivorous plant, and so on. The n1th prey is assigned to the nth carnivorous plant. Let the population before sorting be... X = ( X 1, X 2, ..., X n + n 1), the sorted population X ´=( X ’ 1, X ’ 2,. . . , X ' n+n ),satisfy F ( X ’ 1)≤ F ( X ’ 2)≤…≤ F ( X ' n+n ); (4) Growth stage: The carnivorous plant algorithm introduces an attraction probability γ, where γ = 0.8; (5) If the attraction probability γ is greater than the random number l This means that the carnivorous plant has successfully lured its prey. The growth model of the carnivorous plant is as follows: (3) (4) In the formula, Ä represents the multiplication of elements at the same position of two vectors. xp i Refers to the first i The carnivorous plants in the group, p iv Refers to the first i The first in the group v One prey, rand Let be a random vector between [0, 1]. gr The growth rate is usually taken as 2; (6) When the attraction probability γ is less than the random number l, the prey is not attracted by the carnivorous plant and continues to grow. The mathematical model for prey growth is: (5) (6) In the formula, Ä represents the multiplication of elements at the same position of two vectors. p iu , p iv They refer to the i-th group and the i-th u The and the first v There are 1 target prey, and rand is a random vector between [0, 1]. f ( p iv )and f ( p iu ) respectively refer to the first i The first in the group v The and the first u The fitness value of each prey; (7) Adaptive attraction probability formula: The formula for calculating the probability of attraction is: （7） In the formula, g is the attraction probability, m is the number of monitoring points, and r is... xp i and p iv The distance between them; xp i and p iv Distance between r The calculation formula is as follows: （8） In the formula, n For the number of carnivorous plants, n 1 represents the number of prey. xp i For the first i The carnivorous plants in the group, p iv For the first i The first in the group v One prey; (8) Adopt the improved position update formula: This formula improves the quality of offspring produced during the growth stage by increasing the attraction term of the optimal carnivorous plants to prey in the population and the attraction term of carnivorous plants to prey in the same group, as detailed below: （9） （10） (11) (12) In the formula, Ä represents the multiplication of elements at the same position of two vectors. xp i Refers to the first i Group of carnivorous plants, xp 1 refers to the best individual in the population. p iv , p iu and p iw Let refer to the v-th, u-th, and w-th prey in the i-th group, respectively. `rand` is a random vector between [0, 2, 1], and `rand1` is a random vector between [-1, 1]. gr Let m be the growth rate, m be the number of monitoring points, and r be the value in equation (9). xp i and p iv The distance between them, r is in equation (10) p iv and p iu The distance between them, where t is the current algorithm running time. t max This represents the maximum running time of the algorithm. (9) When the attraction probability γ is greater than the random number l, the carnivorous plant grows according to equation (9); otherwise, the prey updates its position according to equation (10). (10) Reproduction stage: Only the top-ranked carnivorous plant, i.e., the best solution in the population, can perform the reproduction operation. This helps reduce the time complexity of the algorithm. The mathematical model for optimal carnivorous plant reproduction is as follows: (13) (14) In the formula, Ä represents the element-wise multiplication of two vectors at the same position, and rand is a random vector between [0, 1]. xp 1 represents the optimal individual in the population, and μ is the reproduction rate, which is usually 1.

8. xp i and xp j They refer to the first i Carnivorous plants in group j and group j f ( xp i )and f ( xp j ) refer to the fitness values ​​of carnivorous plants in group i and group j, respectively; (11) The improved breeding process is as follows: (15) (16) In the formula, Ä represents the element-wise multiplication of two vectors at the same position, and β is an m-dimensional random vector in the range [0.5, 1.8]. xp i and xp j These refer to the carnivorous plants in groups i and j, respectively. xp 1 is the optimal carnivorous plant for the population; (12) Improved breeding: A random number d between [0,1] is generated. When d ≥ 0.6, the carnivorous plant reproduces according to formula (15); otherwise, it reproduces according to formula (16). (13) Neighborhood perturbation strategy optimization: To avoid the situation where the distance between non-adjacent monitoring points in the randomly selected 4 monitoring points is too far; (14) Given a certain circuit X=(x1, x2,…, x) in the population m Taking m as an example, where m is the number of monitoring points, the specific steps for neighborhood disturbance are as follows: Step 1: From m Randomly select one monitoring point from the monitoring points. a ,by a With the center as the radius as the radius, r The circle containing 1 is called the neighborhood, denoted as . U ( a , r 1), r The formula for calculating 1 is shown in equation (17): (17) In the formula, Z is the route length of the best individual in the population; Step 2: Determine the neighborhood U ( a , r 1) The number of monitoring points included is u. If u≥2, two monitoring points are randomly selected from the neighborhood. aa 1, aa 2; If u=1, let aa 1= a Then in [1,2,…, a -1, a +1,…, m Randomly select a monitoring point as... aa 2; Step 3: In X Find with aa 1. Adjacent monitoring points bb 1, with aa 2 Adjacent monitoring points bb 2, and bb 1≠ aa 2, bb 2≠ aa 1. If bb 1= bb 2. Repeat Step 1 and Step 2 until... bb 1≠ bb 2; Step 4: Delete edges ( aa 1, bb 1) and ( aa 2, bb 2), connecting edge ( bb 1, aa 2) and ( aa 1, bb 2); (15) Reorganization operation: This is used to ensure that the population obtained in each iteration is not inferior to the original population.

3. The remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithms according to claim 2, characterized in that: In this method, the random number l takes values ​​in the range [0, 1].

4. The remote sensing agricultural monitoring energy-saving path planning method based on discrete algorithms according to claim 2, characterized in that: The recombination process in step (15) is as follows: Step 1: Merge the newly generated carnivorous plant and prey populations to obtain a new population; Step 2: Sort the individuals in the new population according to their fitness values ​​from smallest to largest; Step 3: In order to keep the population size of the carnivorous plant algorithm unchanged, select the first n individuals from the new population.

Images