Site position determination method and apparatus, electronic device, and storage medium

Abstract

This application discloses a method, apparatus, electronic device, and storage medium for determining station locations. The method includes: generating a first population randomly, where each individual in the population corresponds to a set of stations at different locations; processing the first population using the pole offsetting technique to obtain a second population; and using the second population, determining the locations of different stations based on an distribution estimation algorithm (EDA). The technical solution provided in this application, by processing the randomly generated initial population using the pole offsetting technique, ensures a uniform distribution of stations within the population, guaranteeing the diversity of the population during the iteration process, avoiding getting trapped in local optima, and thus increasing the probability of obtaining the global optimum, i.e., improving the accuracy of station location determination.

Description

Technical Field

[0001] This application relates to the field of evolutionary algorithms, and more particularly to a method, apparatus, electronic device, and storage medium for determining site location. Background Technology

[0002] In recent years, shared resources have begun to appear in people's lives, such as shared bicycles and express delivery locker stations. The emergence of shared resources can improve resource utilization, reduce waste, and make people's lives more convenient. Therefore, how to rationally set up the locations of shared resource stations to maximize coverage and reduce dispatch distances has become an urgent problem to be solved.

[0003] In related technologies, the Estimation of Distribution Algorithm (EDA) has been proposed to determine the location of resource sites. EDA constructs a probability distribution model based on the global distribution information of the population, and iterates through the population using the probability distribution model to select the optimal solution in the population as the location of the resource site. However, this process suffers from low solution accuracy. Summary of the Invention

[0004] To address the related technical issues, embodiments of this application provide a method, apparatus, electronic device, and storage medium for determining site location.

[0005] The technical solution of this application embodiment is implemented as follows:

[0006] This application provides a method for determining a site location, including:

[0007] The first population was generated randomly, with each individual in the population corresponding to a set of stations at different locations;

[0008] The first population is processed using the offsetting pole technique to obtain the second population;

[0009] Using the second type of group, the locations of different sites are determined based on EDA.

[0010] In the above scheme, during the process of determining the location of different sites using the second type of group based on EDA,

[0011] Determine a first parameter, which characterizes the fitness distribution value and mean squared error of the current population;

[0012] Divide the current population into at least two subpopulations and determine a second parameter, which characterizes the fitness distribution value and mean squared error of each subpopulation in the at least two subpopulations;

[0013] A normal distribution model is generated using the first and second parameters;

[0014] Using the normal distribution model, at least two new subpopulations are obtained;

[0015] Using the current population and the new subpopulation, determine the next generation population.

[0016] In the above scheme, dividing the current population into at least two subpopulations includes:

[0017] For each group of sites in the current population, determine the correlation of fitness with other groups of sites;

[0018] Based on the determined relevance, the current population is divided into at least two subpopulations.

[0019] In the above scheme, determining the correlation between the fitness of the site and other groups includes:

[0020] The correlation of fitness between each group of sites and other groups of sites was determined using the Pearson coefficient.

[0021] In the above scheme, generating a normal distribution model using the first parameter and the second parameter includes:

[0022] Determine the learning parameters for each of at least two subpopulations;

[0023] The second parameter is updated based on the first parameter and the corresponding learning parameter;

[0024] A normal distribution model is generated using the updated second parameter.

[0025] In the above scheme, determining the next generation population using the current population and the obtained new subpopulation includes:

[0026] Determine the third parameter, where the first parameter characterizes the fitness of the current population;

[0027] Determine the fourth parameter, whereby the third parameter characterizes the fitness of the new subpopulation;

[0028] Using the third and fourth parameters, the next generation population is determined from the current population and the new subpopulation.

[0029] In the above scheme, determining the first parameter includes:

[0030] For each individual in the current population, determine the location information and fitness of the corresponding individual station, and based on the determined location information and fitness of the station, determine the fitness distribution value of the corresponding individual;

[0031] The root mean square error of the corresponding individual is determined using the fitness distribution value of the corresponding individual and the location information of the corresponding individual's station.

[0032] This application embodiment also provides a site location determination device, including:

[0033] The generation unit is used to generate the first population in a random manner, with each individual in the population corresponding to a set of stations at different locations;

[0034] The processing unit is used to process the first population using the offset pole technique to obtain the second population;

[0035] The determining unit is used to determine the location of different sites based on EDA using the second type of group.

[0036] This application also provides an electronic device, including: a processor and a memory for storing a computer program capable of running on the processor.

[0037] When the processor runs the computer program, it executes the steps of any of the above-described site location determination methods.

[0038] This application also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of any of the above-described site location determination methods.

[0039] The site location determination method, apparatus, electronic device, and storage medium provided in this application embodiment generate a first population in a random manner, with each individual in the population corresponding to a set of sites at different locations; the first population is processed using the pole offsetting technique to obtain a second population; and the locations of different sites are determined using the second population based on EDA. The technical solution provided in this application embodiment uses the pole offsetting technique to process the randomly generated population, ensuring that the sites in the processed population are uniformly distributed in the location space. Thus, when iterating over the processed population using EDA, the diversity of the population during the iteration process can be guaranteed, avoiding getting trapped in local optima and affecting the solution to the global optimum, thereby improving the solution accuracy. Attached Figure Description

[0040] Figure 1 This is a schematic flowchart illustrating the method for determining the site location in an embodiment of this application.

[0041] Figure 2 This is a site distribution map corresponding to the first type of group in the embodiments of this application;

[0042] Figure 3 This is a site distribution map corresponding to the second group in the embodiments of this application;

[0043] Figure 4 This is a schematic diagram of the site planning method in an application embodiment of this application;

[0044] Figure 5 This is a schematic diagram of the site location determination device according to an embodiment of this application;

[0045] Figure 6 This is a schematic diagram of the electronic device structure according to an embodiment of this application. Detailed Implementation

[0046] The present application will now be described in further detail with reference to the accompanying drawings and embodiments.

[0047] To determine the location of resources at designated sites, traditional intelligent optimization algorithms can be used. These algorithms, including the waterdrop algorithm and brainstorming algorithm, generate randomized sites (also known as an initial population) and iteratively evolve the population to select the optimal value as the best solution for the resource's site location. However, these algorithms suffer from low efficiency and low accuracy.

[0048] Specifically, firstly, since the initial population is randomly generated, it may lack an evolutionary trend, potentially failing to reach the optimal value even after multiple iterations. Secondly, in each iteration, individuals with high fitness are selected as parents to produce the next generation. This leads to a gradual decrease in individual diversity after multiple iterations, causing the evolutionary trend to gradually cease, resulting in redundant iterations. Thirdly, in the above process, the population typically requires multiple iterations before converging and approaching the optimal solution, resulting in low solution efficiency. Furthermore, during iterative evolution, it is prone to getting trapped in local optima, failing to obtain the globally optimal value.

[0049] To address the aforementioned issues, related technologies have proposed using EDA (Electronic Distribution Analysis) to determine the site locations of resources. When using EDA, after generating an initial population, the global distribution information of the population is acquired, and a probability distribution model is constructed using this information. Then, sampling is performed based on the probability distribution model to obtain the next generation of individuals. Since EDA omits the individual evolution process and uses the global distribution information of the population for iteration, it can shorten the iteration cycle and improve the efficiency of finding the optimal value. However, the above algorithm suffers from low accuracy.

[0050] Specifically, since the initial population is usually randomly generated, there may be uneven distribution of individuals or overfitting of the distribution, making it impossible for the probability distribution model built based on global distribution information to accurately represent the distribution information of individuals in the population. As a result, during the iterative process of using the probability distribution model to solve the population, the diversity of the population will decrease, leading to premature convergence and affecting the accuracy of the solution.

[0051] Based on this, in various embodiments of this application, after generating the initial population randomly, the cofactoring technique is used to process the initial population, and EDA is used to iterate the processed initial population to determine the locations of different stations. This ensures a uniform distribution of individuals in the processed initial population. Iterating the processed initial population based on EDA guarantees the diversity of individuals in the population, avoids premature convergence, and improves the solution accuracy.

[0052] This application provides a site location determination method, applied to electronic devices, such as... Figure 1 As shown, the method includes the following steps:

[0053] Step 101: Generate the first population using a random method, with each individual in the population corresponding to a set of stations at different locations;

[0054] Step 102: Process the initial first group using the offset pole technique to obtain the second group;

[0055] Step 103: Using the second group, determine the location of different sites based on EDA.

[0056] The electronic device may include terminal devices, such as desktop computers, laptops, and other related devices.

[0057] In practical applications, in step 101, the electronic device can generate M individuals in the location space in a random manner according to the preset population size M (which can be understood as the number of individuals in the population and the number of stations corresponding to each individual), thereby obtaining the first population; each individual in the population corresponds to M stations at different locations (each station at a location can be understood as the individual's gene); where M is an integer greater than 1.

[0058] Here, when an electronic device generates a set of stations at different locations corresponding to an individual, it can set corresponding identifiers for the stations at different locations; wherein, the identifiers can reflect the order in which the stations are generated.

[0059] For example, such as Figure 2As shown, within a cube (which can be understood as a spatial location), assuming a preset population size of 10 groups of stations and an individual gene count of 10 stations, the electronic device randomly generates 10 groups of stations at different locations based on the preset population size and individual gene count. Each group contains 10 stations, resulting in a total of 100 stations. Figure 2 As can be seen, since the site locations are generated randomly, their distribution in the location space will be uneven. In this case, if the randomly generated population is iterated based on EDA, it is easy to get stuck in local optima during the iteration process, failing to obtain the global optimum and thus failing to determine the location of different sites.

[0060] In step 102, the electronic device uses the offset pole technique to process the first population so that the site locations corresponding to all individuals in the processed population can be evenly distributed.

[0061] Specifically, for the location of a group of stations corresponding to an individual, the electronic device can determine the conversion factor corresponding to the location of each station in the group, and use the determined conversion factor to convert the location of the corresponding station to obtain the converted location of the station, and thus obtain the location of the group of stations after conversion. Furthermore, by converting the location of the station corresponding to each individual in the population, the locations of the stations corresponding to all individuals after conversion can be obtained, i.e., the second population.

[0062] For example, the electronic device generates the first group G in a random manner. n The first group contains n groups of stations at different locations, and each group contains k stations; the position of each station can be represented as (x, y) (x represents the position of the station along the X-axis in two-dimensional space; y represents the position of the station along the Y-axis in two-dimensional space); a group of stations r can be represented as ((x1, y1), (x2, y2), ..., (x k ,y k )).

[0063] Given a set of sites r, the electronic device selects one site r from r. i Then, the first formula is used to determine the station r respectively. i Position (x) i ,y i ) in x i The corresponding conversion factor p x and y i The corresponding conversion factor p y The conversion factor is determined using the following formula:

[0064]

[0065] Where, px and p y The value of is the smallest prime number that satisfies formula (1).

[0066] Then, the electronic device uses the conversion formula and the corresponding conversion factor to convert the station r. i Position (x) i ,y i ) in x i and y i Perform the transformations separately to obtain the transformed position (x). i ',y i The following formula can be used for conversion:

[0067]

[0068] Where, p x (max) represents p in r. x The maximum value of p; x (min) characterizes p in r x The minimum value of p; y (max) represents p in r. y The maximum value of p; y (min) characterizes p in r y The minimum value.

[0069] Using formulas (1) and (2), the electronic device can convert the position of all stations in a set of stations, and then convert the position of each set of stations in the population, to obtain a second population in which the station positions of all individuals are evenly distributed.

[0070] Here, if the electronic device uses the above conversion steps to... Figure 2 The randomly generated population is processed, that is, the position of each of the 100 stations is transformed, such as... Figure 3 As shown, it is possible to obtain the uniformly distributed station locations within a unit location space.

[0071] In practical applications, if the global distribution information of the population is directly used to construct the probability distribution model in step 103, the global search capability of the probability distribution model will be strong, while the local search capability will be weak. That is, the global search capability and the local search capability are unbalanced, which affects the accuracy of the solution.

[0072] To balance the global and local search capabilities of EDA, electronic devices can divide the current population into subpopulations during the process of using the second population to determine the location of different sites.

[0073] Therefore, in one embodiment, during the process of determining the locations of different sites using the second group based on EDA,

[0074] Determine a first parameter, which characterizes the fitness distribution value and mean squared error of the current population;

[0075] Divide the current population into at least two subpopulations and determine a second parameter, which characterizes the fitness distribution value and mean squared error of each subpopulation in the at least two subpopulations;

[0076] A normal distribution model is generated using the first and second parameters;

[0077] Using the normal distribution model, at least two new subpopulations are obtained;

[0078] Using the current population and the new subpopulation, determine the next generation population.

[0079] In practical applications, when electronic devices iterate over a population based on EDA, they can determine the first parameter of the current population. This first parameter characterizes the fitness distribution and mean squared error of the current population and reflects its global information. This enables the normal distribution model generated by the electronic device based on the first and second parameters to have global search capabilities.

[0080] In practical applications, electronic devices determine the first parameter of a population based on the site-related information corresponding to each individual within the population.

[0081] Specifically, in one embodiment, determining the first parameter includes:

[0082] For each individual in the current population, determine the location information and fitness of the corresponding individual station, and based on the determined location information and fitness of the station, determine the fitness distribution value of the corresponding individual;

[0083] The root mean square error of the corresponding individual is determined using the fitness distribution value of the corresponding individual and the location information of the corresponding individual's station.

[0084] The fitness characteristic represents the ability to pass on the location information of an individual site to the next generation of individuals.

[0085] In practical applications, for each individual in the population, the electronic device can first determine the location information of a set of stations corresponding to the individual, and then determine the fitness of the individual based on the location information of the set of stations corresponding to the individual.

[0086] For example, the electronic device can determine the order in which the stations are generated based on the station identifiers corresponding to each individual, thereby determining the current station and the next station. It can also determine the location information (x1, y1) of the current station r1 and the location information (x2, y2) of the next station r2. For one dimension x in the location information of the current station, the electronic device can determine the location distance (|x1-x2|) between the current station and the next station, and also determine the location distance (|x1-x2|) corresponding to the stations in this group. max -x min Then, by calculating the ratio of the distance between the current station and the next station to the distance between the corresponding stations in the same group, the fitness of the current station in dimension x is obtained. Correspondingly, the electronic device can also obtain the fitness of the current station in dimension y in the same way.

[0087] Based on the above operation process, the electronic device can calculate the fitness of each station in a set of stations corresponding to a given individual in different dimensions (e.g., X and Y dimensions in two-dimensional space). By processing the fitness of each station in the corresponding dimension, the fitness corresponding to the given individual can be obtained, and the fitness corresponding to the given individual can be presented in a matrix format.

[0088] In practical applications, after determining the site location information *r* and site fitness of the corresponding individual, the electronic device can use the fitness distribution value formula to determine the fitness distribution value of the corresponding individual in each dimension; wherein, the fitness distribution value is determined by the following formula:

[0089]

[0090] Where n represents the number of stations in r; r i Represents the location information of a site in the corresponding dimension; f i Characterize the site's fitness in the corresponding dimension.

[0091] Using formula (3), the electronic device can determine the fitness distribution value of each individual in the population in the corresponding dimension. Then, by processing the fitness distribution value of each individual in the corresponding dimension, the fitness distribution value of the current population can be obtained. The fitness distribution value of the current population can be presented in the form of a matrix.

[0092] In practical applications, for a set of stations corresponding to a specific individual, electronic devices can also use the mean squared error formula to determine the mean squared error corresponding to that individual; the mean squared error is determined using the following formula:

[0093]

[0094] Where, ε iCharacterizes the fitness distribution of a site in the corresponding dimension.

[0095] Using formula (4), the electronic device can determine the mean squared error of each individual in the population in the corresponding dimension, and by processing the mean squared error of each individual in the corresponding dimension, the mean squared error of the current population can be obtained. The mean squared error of the current population is presented in the form of a matrix.

[0096] Here, after determining the first parameter of the current population, the electronic device can divide the current population according to the relationship between individuals in the current population, so that the electronic device can perform corresponding operations based on the characteristics of the divided subpopulations.

[0097] Specifically, in one embodiment, dividing the current population into at least two subpopulations includes:

[0098] For each group of sites in the current population, determine the correlation of fitness with other groups of sites;

[0099] Based on the determined relevance, the current population is divided into at least two subpopulations.

[0100] In practical applications, electronic devices can divide the current population based on the correlation between individuals in the population, that is, the correlation between each group of sites in the population, and generate at least two sub-populations.

[0101] Here, electronic devices can use the Pearson coefficient (also known as the Pearson correlation coefficient) to calculate the correlation between each group of sites in the population.

[0102] Specifically, in one embodiment, determining the relevance to the fitness of other groups of sites includes:

[0103] The correlation of fitness between each group of sites and other groups of sites was determined using the Pearson coefficient.

[0104] The correlation of fitness among each group of sites can be determined using the following formula:

[0105]

[0106] Among them, f i Characterizes the fitness of each site in a set of sites r; f represents the average fitness of each site in a set of sites r; j Characterizes the fitness of each site in other group sites e; The average fitness of each site in the other group of sites e.

[0107] In practical applications, when electronic devices use the above formula (5) to determine the correlation between the fitness of each group of sites and other groups of sites, the obtained correlation values ​​will be between [-1, 1]. If the absolute value of the correlation is larger, it means that the correlation between the two groups of sites is stronger and they can be classified into a subpopulation; if the absolute value of the correlation is smaller, it means that the correlation between the two groups of sites is weaker and they cannot be classified into a subpopulation.

[0108] Here, after dividing at least two subpopulations based on the correlation of fitness between each group of sites, the electronic device can use formulas (4) and (5) to determine the fitness distribution value and mean square error of each subpopulation in the corresponding dimension, i.e., the second parameter.

[0109] Then, when the electronic device generates a normal distribution model based on the first and second parameters, it can perform different operations according to the site location corresponding to each of the at least two subpopulations.

[0110] Specifically, in one embodiment, generating a normal distribution model using the first parameter and the second parameter includes:

[0111] Determine the learning parameters for each of at least two subpopulations;

[0112] The second parameter is updated based on the first parameter and the corresponding learning parameter;

[0113] A normal distribution model is generated using the updated second parameter.

[0114] In practical applications, since the distribution of site locations differs within each subpopulation, the electronic device can determine the distribution pattern of site locations for each subpopulation based on its corresponding fitness. Then, based on the fitness of each subpopulation, the electronic device can set different learning parameters λ for each subpopulation.

[0115] For example, for a subpopulation with high fitness, the electronic device can set a larger value of λ to improve the convergence rate of the subpopulation during iteration, thereby achieving a local search of the subpopulation; for a subpopulation with low fitness, the electronic device can set a smaller value of λ to reduce the convergence rate of the subpopulation during iteration, thereby achieving a global search of the subpopulation.

[0116] Then, for each subpopulation, the electronic device can update the second parameter corresponding to the subpopulation based on the first parameter corresponding to the current population and the set learning parameters, so that the updated second parameter can simultaneously indicate the global information of the current population and the corresponding subpopulation.

[0117] Specifically, the electronic device can update the fitness distribution value of each dimension in the second parameter using the fitness distribution value update formula; it can be updated using formula (6):

[0118] ε=λε1+(1-λ)ε old (6)

[0119] Where, ε old ε1 represents the fitness distribution value of the corresponding dimension in the first parameter; λ represents the fitness distribution value of the corresponding dimension in the second parameter; λ represents the learning parameter of the current subpopulation; ε represents the fitness distribution value of the corresponding dimension in the updated second parameter.

[0120] In addition, electronic devices can also use the mean squared error update formula to update the mean squared error of each dimension in the second parameter; it can be updated using formula (7):

[0121]

[0122] Where, σ old σ1 represents the mean squared error of the corresponding dimension in the first parameter; σ1 represents the mean squared error of the corresponding dimension in the second parameter; σ represents the mean squared error of the corresponding dimension in the updated second parameter.

[0123] In other words, through formulas (6) and (7), the electronic device can update the fitness distribution value and mean square error of each dimension in the second parameter.

[0124] Then, the electronic device can generate a normal distribution model based on the updated second parameter. Since the updated second parameter can simultaneously indicate global information of the current population and subpopulations, the generated normal distribution model has both global and local search capabilities. This allows the dominant individuals in the population to be preserved during the sampling of subpopulations using the generated normal distribution model to generate at least two new subpopulations, avoiding premature convergence that could affect the solution to the global optimum, thus improving the solution accuracy.

[0125] In practical applications, the electronic device can also generate a normal distribution model for each subpopulation based on the relevant parameters corresponding to each subpopulation in the updated second parameter, specifically including the updated fitness distribution value and mean squared error. The function expression corresponding to the normal distribution model is as follows:

[0126]

[0127] Where x represents the location information of a corresponding individual in the subpopulation at a station in the corresponding dimension; f(x) represents the location information of a corresponding individual in the candidate subpopulation at a station in the corresponding dimension.

[0128] Then, the electronic device can use the normal distribution model corresponding to the corresponding subpopulation to generate candidate subpopulations and sample the candidate subpopulations to obtain new subpopulations with the same size as the corresponding subpopulations. Since the normal distribution model corresponding to each subpopulation is generated based on the site distribution characteristics of the subpopulation and the current population, it can sample the subpopulations more specifically to retain the dominant individuals in the subpopulations.

[0129] For example, the electronic device can obtain the site location information of each individual in the candidate subpopulation in the corresponding dimension using the above formula (8). In order to select excellent individuals from the candidate subpopulation as individuals in the new subpopulation, the electronic device will pre-set the fitness threshold of the site, calculate the fitness of the site corresponding to each individual in the candidate subpopulation, and compare the fitness of the site with the pre-set fitness threshold of the site. The site location information with fitness greater than the fitness threshold corresponding to the site location information is selected from the candidate subpopulation, that is, the site location information of the dominant individual, and is used as the site location information of the individual in the new subpopulation. If the size of the new subpopulation is smaller than the size of the subpopulation after sampling the dominant individuals in the subpopulation, the electronic device can generate a new subpopulation with the same size as the subpopulation by repeated sampling. In addition, the electronic device can also generate new individuals based on the distribution pattern of the dominant individuals in the subpopulation to generate a new subpopulation with the same size as the subpopulation. This application embodiment does not limit this.

[0130] In practical applications, after obtaining at least two new subpopulations, electronic devices can select dominant individuals from the current population and the new subpopulations as the next generation population to achieve population updates and iterations, and ultimately determine the locations of different sites.

[0131] Specifically, in one embodiment, determining the next generation population using the current population and the obtained new subpopulation includes:

[0132] Determine the third parameter, where the first parameter characterizes the fitness of the current population;

[0133] Determine the fourth parameter, whereby the third parameter characterizes the fitness of the new subpopulation;

[0134] Using the third and fourth parameters, the next generation population is determined from the current population and the new subpopulation.

[0135] In practical applications, the electronic device can determine the site fitness of each individual in the current population, and also the site fitness of each individual in the new subpopulation. Since site fitness characterizes the ability to transmit the site location information of an individual to the next generation, the electronic device can sort the fitness values ​​of individuals in the current and new subpopulations from highest to lowest, and select the M individuals with the highest fitness values ​​as the next generation population; the population size of the next generation population is the same as the population size of the current population.

[0136] In practical applications, the electronic device can preset a maximum number of iterations (expressed as gen_max); the maximum number of iterations is an integer greater than or equal to 1. Then, by repeatedly executing the above steps, the current population can be updated and iterated, generating the next generation population and accumulating iterations until the population reaches the maximum number of iterations, at which point the iteration operation can be stopped. At this point, the electronic device can select a dominant individual from the last generation population and use the set of site locations corresponding to that individual as the final site locations.

[0137] For example, an electronic device can calculate the fitness of each individual in the last generation of the population and use the set of site locations corresponding to the individual with the highest fitness value as the final site location.

[0138] The site location determination method provided in this application involves generating a first population randomly, where each individual corresponds to a set of sites at different locations. The first population is then processed using the offsetting pole technique to obtain a second population. Using this second population, and based on EDA, the locations of different sites are determined. The technical solution provided in this application uses the offsetting pole technique to process the randomly generated population, ensuring that the sites in the processed population are evenly distributed in the location space. Thus, when iterating through the processed population using EDA, dominant individuals in the population are always retained, ensuring population diversity and avoiding premature convergence that could affect the solution to the global optimum, thereby improving the solution accuracy.

[0139] The present application will be further described in detail below with reference to application examples.

[0140] This application proposes a site planning method based on the EDA evolutionary algorithm, applicable to electronic devices. Compared with traditional site planning methods, the method of this application can effectively balance local and global searches in the planning process, and can effectively reduce redundant iterations and improve the diversity of site distribution.

[0141] like Figure 4As shown, the process by which electronic devices determine site locations using the above site planning method includes the following steps:

[0142] Step 401: The electronic device initializes the population using the redundant pole technique;

[0143] Here, before step 401, the electronic device generates M groups of stations at arbitrary locations (i.e., M individuals) in a randomized manner using a preset population size M; where each group of stations at different locations corresponds to one individual, and each station at a location corresponds to one gene in the individual.

[0144] In practical applications, a set of M individuals can also be called a population (i.e., the first population).

[0145] Since the station locations are generated in a randomized manner, they are usually unevenly distributed in the station location distribution space. Therefore, the electronic device uses the offset pole technique to initialize the population, that is, to transform a set of station locations corresponding to each individual in the population to obtain a transformed population (i.e., the second population). In the transformed population, the station locations corresponding to individuals are evenly distributed in the station location distribution space.

[0146] Step 402: The electronic device calculates the fitness distribution and mean squared error of the current population.

[0147] For the first iteration, the electronic device can use formulas (3) and (4) to calculate the fitness distribution value and mean square error of each individual in the current population; the fitness distribution value and mean square error represent the distribution pattern of the site location corresponding to the individual.

[0148] The electronic device can calculate the fitness distribution value and mean squared error of the site location corresponding to the individual in different dimensions, and obtain the fitness distribution value and mean squared error of the individual based on the fitness distribution value and mean squared error of each dimension.

[0149] Step 403: The electronic device is divided into q subpopulations based on the Pearson correlation coefficient;

[0150] The value of q is an integer greater than or equal to 2; the value of q can be set as needed, for example, to 3.

[0151] In practical applications, electronic devices can determine the fitness of an individual based on the location of that individual site in the current population and the distribution of each site location.

[0152] For example, the electronic device uses the ratio of the distance between the current station and the next station to the coverage area of ​​the corresponding individual station as the fitness of the current station, and obtains the fitness of the corresponding individual by summing the fitness of all stations in the corresponding individual's station.

[0153] Then, the correlation between individuals is calculated using formula (5), and individuals with high correlation are assigned to a subpopulation, thus obtaining q subpopulations.

[0154] Step 404: The electronic device assigns an evolution rate (i.e., a learning rate) to the subpopulation;

[0155] In practical applications, electronic devices assign a corresponding evolution rate λ to each of the q subpopulations based on its fitness. For subpopulations with high fitness, a larger λ value is assigned to achieve local search; for subpopulations with low fitness, a smaller λ value is assigned to achieve global search.

[0156] Step 405: The electronic device calculates the fitness distribution and mean squared error of the subpopulation;

[0157] For each of the q subpopulations, the electronic device can use formulas (3) and (4) to calculate the fitness distribution value and mean square error corresponding to each dimension of the subpopulation.

[0158] In order to correlate the site distribution patterns of the subpopulation with the site distribution patterns of the current population, and to retain the dominant individuals in the population, the electronic device will use formulas (6) and (7) to update the fitness distribution value and mean square error of each subpopulation, and obtain the updated fitness distribution value and mean square error for each dimension.

[0159] Step 406: The electronic device generates a normal distribution model;

[0160] Here, for each of the q subpopulations, the electronic device uses the updated fitness distribution value and mean squared error corresponding to each dimension to generate a normal distribution model for the corresponding subpopulation in order to execute step 407.

[0161] Step 407: The electronic device generates a new subpopulation of the same size;

[0162] Here, the electronic device uses the normal distribution model corresponding to the corresponding subpopulation in the q subpopulations to sample the corresponding subpopulations and generate a new subpopulation with the same population size as the corresponding subpopulation; the population size of each new subpopulation generated is the same as the population size of the current population, containing M individuals.

[0163] For example, based on the normal distribution model of the corresponding subpopulation, the electronic device can determine the site location distribution characteristics of the corresponding subpopulation; according to the site location distribution characteristics of the corresponding subpopulation, it can determine the distribution of dominant individuals in the corresponding subpopulation, and sample the dominant individuals in the corresponding subpopulation as individuals in the corresponding new subpopulation. Alternatively, based on the site location distribution characteristics of the corresponding subpopulation, the electronic device can also generate new individuals through randomization, and these new individuals can be included in the corresponding new subpopulation.

[0164] In other words, for each of the q subpopulations, the electronic device can use the normal distribution model corresponding to the subpopulation to generate a new subpopulation, thus obtaining q new subpopulations.

[0165] Step 408: The electronic device merges the populations and selects the best individuals for the next generation.

[0166] In practical applications, the electronic device merges the generated q new subpopulations with the current population and determines the fitness of each individual in the merged population. Then, by arranging the individuals with the highest fitness in descending order, the M individuals with the highest fitness, representing M different site locations, are selected as the next generation population. This completes one iteration of the population process.

[0167] Accordingly, the electronic devices will accumulate and record the number of iterations of the population.

[0168] Step 409: The electronic device determines whether the termination condition is met;

[0169] The termination condition can be set to the maximum number of iterations; the value of the maximum number of iterations can be set as needed, for example, to 20.

[0170] If the electronic device determines that the cumulative number of iterations has reached the maximum number of iterations, then proceed to step 410; otherwise, proceed to step 402.

[0171] Step 410: The electronic device determines the optimal solution.

[0172] Here, when the population reaches its maximum iteration count, the electronic device calculates the fitness of each individual in the population for the last generation. Then, the locations of a set of different stations corresponding to the individual with the highest fitness are used as the final station distribution result, and the current operation ends.

[0173] In this application embodiment, the initial population is processed using the offsetting pole technique to generate uniformly distributed stations within the station location space. This ensures that the initial population maintains diversity during iteration, preventing premature entrapment in local optima and thus improving the accuracy of finding the global optimum.

[0174] Secondly, during population iteration, the current population is divided into multiple subpopulations using the Pearson coefficient. Different learning rates are set based on the site distribution characteristics (including fitness distribution values ​​and mean squared error) of each subpopulation. Then, a corresponding normal distribution model is generated for each subpopulation based on the current population's distribution characteristics and the subpopulation's learning rate. This normal distribution model is then used to generate new subpopulations, thus determining the next generation population. In this process, by setting different learning rates for different subpopulations and generating corresponding normal distribution models, the local and global searches during iteration can be balanced, thereby improving the accuracy of finding the global optimum.

[0175] In addition, by using the fitness distribution value and mean square error of the population to generate a normal distribution model, the fitness value of the population can be effectively linked to the fitness value of each group of stations, so that the dominant individuals in the population are retained in the iteration process, effectively reducing redundant iterations.

[0176] To implement the solutions in the embodiments of this application, the embodiments of this application also provide a site location determination device, such as... Figure 5 As shown, the device includes:

[0177] The generation unit 501 is used to generate the first population in a random manner, where each individual in the population corresponds to a set of stations at different locations;

[0178] Processing unit 502 is used to process the first population using the offset pole technique to obtain the second population;

[0179] The determination unit 503 is used to determine the location of different sites based on EDA using the second group.

[0180] In one embodiment, the determining unit 503 is configured to:

[0181] Determine a first parameter, which characterizes the fitness distribution value and mean squared error of the current population;

[0182] Divide the current population into at least two subpopulations and determine a second parameter, which characterizes the fitness distribution value and mean squared error of each subpopulation in the at least two subpopulations;

[0183] A normal distribution model is generated using the first and second parameters;

[0184] Using the normal distribution model, at least two new subpopulations are obtained;

[0185] Using the current population and the new subpopulation, determine the next generation population.

[0186] In one embodiment, the determining unit 503 is configured to:

[0187] For each group of sites in the current population, determine the correlation of fitness with other groups of sites;

[0188] Based on the determined relevance, the current population is divided into at least two subpopulations.

[0189] In one embodiment, the determining unit 503 is configured to:

[0190] The correlation of fitness between each group of sites and other groups of sites was determined using the Pearson coefficient.

[0191] In one embodiment, the determining unit 503 is configured to:

[0192] Determine the learning parameters for each of at least two subpopulations;

[0193] The second parameter is updated based on the first parameter and the corresponding learning parameter;

[0194] A normal distribution model is generated using the updated second parameter.

[0195] In one embodiment, the determining unit 503 is configured to:

[0196] Determine the third parameter, where the first parameter characterizes the fitness of the current population;

[0197] Determine the fourth parameter, whereby the third parameter characterizes the fitness of the new subpopulation;

[0198] Using the third and fourth parameters, the next generation population is determined from the current population and the new subpopulation.

[0199] In one embodiment, the determining unit 503 is configured to:

[0200] For each individual in the current population, determine the location information and fitness of the corresponding individual station, and based on the determined location information and fitness of the station, determine the fitness distribution value of the corresponding individual;

[0201] The root mean square error of the corresponding individual is determined using the fitness distribution value of the corresponding individual and the location information of the corresponding individual's station.

[0202] In practical applications, the generation unit 501, the processing unit 502, and the determination unit 503 can be implemented by the processor in the site location determination device.

[0203] It should be noted that the site location determination device provided in the above embodiments is only illustrated by the division of the above-described program units when determining the site location. In practical applications, the above processing can be assigned to different program units as needed, that is, the internal structure of the device can be divided into different program units to complete all or part of the processing described above. In addition, the site location determination device and the site location determination method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0204] Based on the hardware implementation of the above program modules, and in order to implement the site location determination method of the embodiments of this application, the embodiments of this application also provide an electronic device, such as... Figure 6 As shown, the electronic device 600 includes:

[0205] The communication interface 601 enables interaction with other devices;

[0206] The processor 602 is connected to the communication interface 601 to enable interaction with other devices and to execute the methods provided by one or more of the above-mentioned technical solutions when running a computer program;

[0207] The computer program is stored in memory 603.

[0208] Specifically, the processor 602 is used for:

[0209] The first population was generated randomly, with each individual in the population corresponding to a set of stations at different locations;

[0210] The first population is processed using the offsetting pole technique to obtain the second population;

[0211] Using the second type of group, the locations of different sites are determined based on EDA.

[0212] In one embodiment, the processor 602 is configured to:

[0213] Determine a first parameter, which characterizes the fitness distribution value and mean squared error of the current population;

[0214] Divide the current population into at least two subpopulations and determine a second parameter, which characterizes the fitness distribution value and mean squared error of each subpopulation in the at least two subpopulations;

[0215] A normal distribution model is generated using the first and second parameters;

[0216] Using the normal distribution model, at least two new subpopulations are obtained;

[0217] Using the current population and the new subpopulation, determine the next generation population.

[0218] In one embodiment, the processor 602 is configured to:

[0219] For each group of sites in the current population, determine the correlation of fitness with other groups of sites;

[0220] Based on the determined relevance, the current population is divided into at least two subpopulations.

[0221] In one embodiment, the processor 602 is configured to:

[0222] The correlation of fitness between each group of sites and other groups of sites was determined using the Pearson coefficient.

[0223] In one embodiment, the processor 602 is configured to:

[0224] Determine the learning parameters for each of at least two subpopulations;

[0225] The second parameter is updated based on the first parameter and the corresponding learning parameter;

[0226] A normal distribution model is generated using the updated second parameter.

[0227] In one embodiment, the processor 602 is configured to:

[0228] Determine the third parameter, where the first parameter characterizes the fitness of the current population;

[0229] Determine the fourth parameter, whereby the third parameter characterizes the fitness of the new subpopulation;

[0230] Using the third and fourth parameters, the next generation population is determined from the current population and the new subpopulation.

[0231] In one embodiment, the processor 602 is configured to:

[0232] For each individual in the current population, determine the location information and fitness of the corresponding individual station, and based on the determined location information and fitness of the station, determine the fitness distribution value of the corresponding individual;

[0233] The root mean square error of the corresponding individual is determined using the fitness distribution value of the corresponding individual and the location information of the corresponding individual's station.

[0234] It should be noted that the specific processing procedure of the processor 602 can be understood by referring to the above method.

[0235] Of course, in practical applications, the various components in electronic device 600 are coupled together through bus system 604. It can be understood that bus system 604 is used to realize the connection and communication between these components. In addition to a data bus, bus system 604 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, in... Figure 6 The general designated all buses as Bus System 604.

[0236] The memory 603 in this embodiment is used to store various types of data to support the operation of the electronic device 600. Examples of such data include any computer program used to operate on the electronic device 600.

[0237] The methods disclosed in the embodiments of this application can be applied to the processor 602, or implemented by the processor 602. The processor 602 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in the processor 602 or by instructions in the form of software. The processor 602 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 602 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly reflected as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, which is located in the memory 603. The processor 602 reads the information in the memory 603 and combines its hardware to complete the steps of the aforementioned method.

[0238] In an exemplary embodiment, the electronic device 600 may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components to perform the aforementioned method.

[0239] In an exemplary embodiment, this application also provides a storage medium, namely a computer storage medium, specifically a computer-readable storage medium, such as a memory 603 storing a computer program, which can be executed by the processor 602 of the electronic device 600 to complete the steps described in the aforementioned site location determination method. The computer-readable storage medium may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory may be a disk storage device or a magnetic tape storage device.

[0240] It should be noted that terms such as "first" and "second" are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0241] Furthermore, the technical solutions described in the embodiments of this application can be combined arbitrarily without conflict.

[0242] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for determining a site location, characterized in that, include: The first population was generated randomly, with each individual in the population corresponding to a set of stations at different locations; The first population is processed using the offset pole technique to obtain the second population. The offset pole technique is used to transform the location of the station corresponding to each individual in the first population. Using the second type of population, and based on the distribution estimation algorithm (EDA), the locations of different stations are determined; in the process of using the second type of population and EDA to determine the locations of different stations, Determine a first parameter, which characterizes the fitness distribution value and mean squared error of the current population; For each group of sites in the current population, determine the correlation of fitness with other groups of sites; Based on the determined relevance, the current population is divided into at least two subpopulations, and a second parameter is determined, which characterizes the fitness distribution value and mean square error of each subpopulation in the at least two subpopulations. Determine the learning parameters for each of at least two subpopulations; The second parameter is updated based on the first parameter and the corresponding learning parameter; A normal distribution model is generated using the updated second parameter; Using the normal distribution model, at least two new subpopulations are obtained; Determine a third parameter, which characterizes the fitness of the current population; Determine a fourth parameter, which characterizes the fitness of the new subpopulation; Using the third and fourth parameters, the next generation population is determined from the current population and the new subpopulation.

2. The method according to claim 1, characterized in that, The determination of the correlation between the fitness of the site and other groups includes: The correlation of fitness between each group of sites and other groups of sites was determined using the Pearson coefficient.

3. The method according to claim 1, characterized in that, Determining the first parameter includes: For each individual in the current population, determine the location information and fitness of the corresponding individual station, and based on the determined location information and fitness of the station, determine the fitness distribution value of the corresponding individual; The root mean square error of the corresponding individual is determined by using the fitness distribution value of the corresponding individual and the location information of the corresponding individual's station.

4. A site location determination device, characterized in that, include, The generation unit is used to generate the first population in a random manner, with each individual in the population corresponding to a set of stations at different locations; The processing unit is used to process the first population using the offset pole technique to obtain the second population, wherein the offset pole technique is used to transform the position of the station corresponding to each individual in the first population. The determining unit is used to determine the locations of different sites based on EDA using the second type of group; wherein, The determining unit is configured to: determine a first parameter, which characterizes the fitness distribution value and mean squared error of the current population; determine the correlation of fitness with other groups of sites for each group in the current population; divide the current population into at least two subpopulations based on the determined correlation, and determine a second parameter, which characterizes the fitness distribution value and mean squared error of each subpopulation in the at least two subpopulations; determine a learning parameter for each subpopulation in the at least two subpopulations; update the second parameter based on the first parameter and the corresponding learning parameter; generate a normal distribution model using the updated second parameter; obtain at least two new subpopulations using the normal distribution model; determine a third parameter, which characterizes the fitness of the current population; determine a fourth parameter, which characterizes the fitness of the new subpopulations; and determine the next generation population from the current population and the new subpopulations using the third parameter and the fourth parameter.

5. An electronic device, characterized in that, include: The processor and the memory used to store computer programs that can run on the processor. When the processor is used to run the computer program, it performs the steps of the method according to any one of claims 1 to 3.

6. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

Images