A method of material distribution center location selection and path planning based on multi-objectives

Abstract

The present application provides a multi-objective material distribution center site selection and path planning method, which obtains material distribution data; the material distribution data includes geographic information data, traffic condition data, material demand data, meteorological data and socio-economic data; the material distribution data is processed by a missing data processing algorithm based on fuzzy clustering to obtain processed material distribution data; the processed material distribution data is analyzed by a deep learning model to evaluate the material demand characteristics and predict the material demand change trend; a path planning model based on multi-objective optimization is constructed according to the material demand characteristics and material demand change trend; the multi-objective function in the path planning model based on multi-objective optimization is solved based on a microhabitat genetic algorithm optimized by a marine predator algorithm to obtain a globally optimized distribution center site selection and planning path, thereby solving the problem of inaccurate material distribution center site selection and planning path.

Description

Technical Field

[0001] The present application relates to the technical field of material distribution, and in particular to a multi-objective-based material distribution center site selection and path planning method. Background Art

[0002] With the acceleration of urbanization and the continuous development of social economy, the demand for power grid construction and transformation is increasing. Power grid technical transformation projects are important measures to ensure the safe, reliable and stable operation of power systems. However, power grid technical transformation projects are usually widely distributed, with many construction sites and complex environments, which brings great challenges to the distribution of materials. Whether materials can be delivered to the construction site in a timely and accurate manner directly affects the progress and quality of the project. Therefore, how to optimize the material distribution route and improve the distribution efficiency has become a key issue for power grid companies in project management.

[0003] The material distribution path planning methods in related technologies are mostly based on a single goal, such as the shortest path or minimum cost. However, the material distribution of power grid technical transformation projects involves multiple goals and complex constraints, such as delivery time, cost, road conditions, weather conditions, etc. At the same time, with the development of big data and artificial intelligence technology, data-driven decision-making has become a trend. By analyzing historical data and real-time information, it is possible to achieve more accurate path optimization and resource allocation. Summary of the invention

[0004] The present application provides a multi-objective-based material distribution center site selection and path planning method to solve the problem of inaccurate site selection and path planning of material distribution centers.

[0005] The present application provides a multi-objective material distribution center site selection and path planning method, the method comprising:

[0006] Acquire material distribution data; the material distribution data includes geographic information data, traffic condition data, material demand data, meteorological data and socio-economic data;

[0007] Processing the material distribution data using a missing data processing algorithm based on fuzzy clustering to obtain processed material distribution data;

[0008] Analyzing the processed material distribution data using a deep learning model to evaluate material demand characteristics and predict material demand change trends;

[0009] Constructing a path planning model based on multi-objective optimization according to the characteristics of material demand and the changing trend of material demand;

[0010] The microhabitat genetic algorithm optimized based on the marine predator algorithm solves the multi-objective function in the path planning model based on multi-objective optimization to obtain the global optimal distribution center location and planning path.

[0011] The above method comprehensively analyzes the material distribution data, and constructs a path planning model based on multi-objective optimization according to the deep learning model and multi-objective optimization algorithm. It then introduces the marine predator algorithm to enhance the global search capability of the niche genetic algorithm, which can effectively avoid falling into the local optimal solution, and then find a better distribution center site selection and planning path, realize the integrated optimization of distribution center site selection and path planning, and solve the problem of inaccurate site selection and planning path of material distribution centers.

[0012] Optionally, the geographic information data includes terrain feature data and land use data; the traffic condition data includes road network data and traffic flow data; the material demand data includes material types and quantities, data sources and material supply schedules; the meteorological data includes real-time meteorological data and meteorological forecast data; and the socio-economic data includes population distribution data and infrastructure data.

[0013] The geographic information data helps to analyze the feasibility of distribution routes and optimize site selection. Through detailed geographic information data, it can ensure that route planning in the material distribution process is more reasonable and efficient. The traffic condition data helps to avoid traffic congestion areas, select the optimal distribution route, and improve the efficiency and punctuality of material transportation. The material demand data helps to accurately predict and arrange material distribution. Through detailed analysis of material demand, it can ensure that the required materials are delivered to the designated location at the right time to meet project needs. The meteorological data can be used to avoid the risks brought by bad weather in advance, adjust the distribution plan, and ensure that the materials arrive at the destination safely and in time. The socio-economic data can select economically developed and conveniently transported areas as distribution centers to improve overall distribution efficiency and service levels.

[0014] Optionally, the step of processing the material distribution data using a missing data processing algorithm based on fuzzy clustering to obtain processed material distribution data includes:

[0015] Preprocessing the material distribution data to obtain material distribution data with outliers removed;

[0016] Initialize missing values ​​by using the mean or median method to facilitate clustering;

[0017] Iteratively update the cluster centers and membership matrix by minimizing the weighted square error objective function;

[0018] Use the membership matrix to calculate new cluster centers;

[0019] Obtain missing values ​​based on new cluster centers and membership matrices;

[0020] Generate processed material distribution data based on the missing values ​​and the material distribution data with outliers removed.

[0021] Outliers are removed through preprocessing to ensure the accuracy and reliability of material distribution data. The mean or median method is used to preliminarily fill in missing values, providing a complete data set for cluster analysis, which helps to avoid analysis bias caused by missing data and ensure the representativeness and accuracy of clustering results. The clustering center and membership matrix are iteratively updated by minimizing the weighted square error objective function to continuously optimize the clustering effect. This dynamic adjustment process helps to find more accurate clustering divisions and improve the model's ability to capture data structure. The new cluster center is calculated using the membership matrix to further refine the clustering results. This method can more accurately reflect the intrinsic distribution of data and provide a more reasonable basis for subsequent path planning. Missing values ​​are obtained based on the new cluster center and membership matrix, and the processed material distribution data is generated in combination with the data without outliers. This process not only fills the data gaps, but also improves the integrity and consistency of the data, providing high-quality input for deep learning models and path planning models.

[0022] Optionally, the updating formula of the membership matrix is:

[0023]

[0024] Among them, m is the fuzzy factor, which is 2; n is the number of data points; c is the number of clusters; is the position of the ith data point in the rth iteration; x k is the feature vector of the kth data point; is the position of the jth cluster center in the rth iteration.

[0025] Optionally, the update formula of the cluster center is:

[0026]

[0027] Where n is the number of data points; m is the fuzzy factor; is the membership of the i-th data point to the k-th data point after the r+1th iteration; x j is the eigenvector of the jth cluster center.

[0028] Optionally, the step of using a deep learning model to analyze the processed material distribution data to evaluate material demand characteristics and predict material demand change trends includes:

[0029] Extracting time series features and spatial features from the processed material distribution data through the LSTM model to predict material demand and material distribution;

[0030] Determining the priority of material demand in the processed material distribution data and identifying key materials through a multi-task learning model and a critical path method;

[0031] The improved particle swarm algorithm is used to cluster and analyze the material demand, material distribution and priority of material demand to evaluate the characteristics of material demand and predict the changing trend of material demand.

[0032] The LSTM model can accurately capture the multi-dimensional factors that affect material demand and improve the accuracy and timeliness of the forecast results. The multi-task learning model and the critical path method can simultaneously process multiple related tasks, dynamically adjust the priority of tasks, and use the improved particle swarm algorithm to cluster the material demand, distribution and priority to optimize the distribution path, improve the ability to solve the global optimal solution, and enhance the algorithm's ability to jump out of the local optimal.

[0033] Optionally, the step of constructing a path planning model based on multi-objective optimization according to the material demand characteristics and material demand change trends includes:

[0034] According to the characteristics of the material demand and the trend of material demand changes, the objective function and constraint conditions are set; the objective function includes minimizing the total transportation cost, minimizing the total transportation time and maximizing the path reliability; the constraint conditions include vehicle capacity restrictions, time window constraints and road condition restrictions;

[0035] Determine the model input according to the material demand characteristics; the model input includes the geographical location of the construction site, material demand, vehicle information, road network data, real-time road condition data and weather forecast;

[0036] A path planning model based on multi-objective optimization is constructed according to the objective function, constraints and model input.

[0037] Setting objective functions and constraints based on the characteristics of material demand and the changing trend of material demand can comprehensively consider multiple factors, flexibly respond to real-time changing traffic conditions and emergency needs, and ensure that the distribution plan has a high degree of adaptability and flexibility, thereby meeting the dynamic needs of power grid technical transformation projects. Based on the geographical location of the construction site, material demand, vehicle information, road network data, real-time road condition data and weather forecast model input, the accuracy and feasibility of the planning can be ensured. The multi-objective optimization model not only considers logistics costs and time, but also pays attention to the reliability of the path. By comprehensively evaluating multiple factors, the optimal balance point is found, thereby improving the overall distribution efficiency and reducing operational risks.

[0038] Optionally, the steps of solving the multi-objective function in the path planning model based on multi-objective optimization by using a niche genetic algorithm optimized by a marine predator algorithm to obtain a globally optimized distribution center location and planning path include:

[0039] Initialize a population, where each individual in the population represents a candidate solution;

[0040] Dividing the population into niches based on a distance metric or a sharing function, each niche representing a local region of the solution space;

[0041] fusing the marine predator algorithm with the niche to optimize the niche genetic algorithm;

[0042] In each iteration, selection, crossover and mutation operations are performed according to the fitness function to update the population;

[0043] The fitness function is used to evaluate each individual in the updated population to obtain the global optimal distribution center location and planning path.

[0044] By introducing the marine predator algorithm, the global search capability of the niche genetic algorithm can be enhanced, which can effectively avoid falling into the local optimal solution and obtain the global optimal distribution center location and planning path.

[0045] Optionally, the step of integrating the marine predator algorithm with the niche to optimize the niche genetic algorithm further includes:

[0046] The Levy flight mechanism of the marine predator algorithm is introduced within or between niches to enhance the global search capability of the niche genetic algorithm.

[0047] Microhabitat maintenance through dynamic predation behavior of marine predator algorithms to optimize individual distribution;

[0048] The Brownian motion strategy of the marine predator algorithm is introduced in each niche to fine-tune the local optimal solution to improve the accuracy of local search within the niche.

[0049] The adaptive adjustment mechanism of the marine predator algorithm is used to dynamically adjust the key parameters of the microhabitat; the key parameters include the microhabitat size, mutation rate, and crossover rate.

[0050] It can be seen from the above technical scheme that the present application provides a multi-objective material distribution center site selection and path planning method, by acquiring material distribution data; the material distribution data includes geographic information data, traffic condition data, material demand data, meteorological data and socio-economic data; the material distribution data is processed by a missing data processing algorithm based on fuzzy clustering to obtain processed material distribution data; the processed material distribution data is analyzed using a deep learning model to evaluate the material demand characteristics and predict the material demand change trend; a path planning model based on multi-objective optimization is constructed according to the material demand characteristics and material demand change trend; the multi-objective function in the path planning model based on multi-objective optimization is solved based on a microhabitat genetic algorithm optimized by a marine predator algorithm to obtain a globally optimized distribution center site selection and planning path, so as to solve the problem of inaccurate material distribution center site selection and planning path. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0052] Figure 1 A flowchart of a multi-objective material distribution center site selection and path planning method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0053] The embodiments will be described in detail below, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation methods described in the following embodiments do not represent all implementation methods consistent with the present application. They are only examples of systems and methods consistent with some aspects of the present application.

[0054] To solve the problem of inaccurate site selection and planning path of material distribution centers, see Figure 1 The present application embodiment provides a multi-objective material distribution center site selection and path planning method, the method comprising:

[0055] S100: Acquire material distribution data.

[0056] Material distribution data includes geographic information data, traffic conditions data, material demand data, meteorological data and socio-economic data.

[0057] In some embodiments, geographic information data includes terrain feature data and land use data; traffic condition data includes road network data and traffic flow data; material demand data includes material types and quantities, data sources, and material supply schedules; meteorological data includes real-time meteorological data and meteorological forecast data; and socio-economic data includes population distribution data and infrastructure data.

[0058] It should be understood that terrain feature data refers to terrain feature data around the construction site (such as altitude, slope, and terrain complexity), which can be obtained through geographic information systems (GIS) or remote sensing technology. Land use data includes land use (such as cultivated land, construction land, and forests) and infrastructure distribution (such as roads, water sources, and power supply facilities). Road network data includes road types (such as highways, main roads, and secondary roads), road length, traffic capacity, and road conditions (such as flatness, width, and traffic flow). Traffic flow data refers to real-time traffic flow information and historical traffic data obtained through a traffic monitoring system, which are used to analyze the possibility and time period of traffic congestion. Material types and quantities include the types, specifications, quantities, and priorities of materials required for each construction site. Data sources include project plans, material scheduling plans, and historical demand data. The material supply schedule is a table that collects and formulates a time schedule for material supply based on project progress requirements to ensure that materials arrive on time at key construction nodes. Real-time meteorological data refers to real-time meteorological information in the construction area, such as temperature, rainfall, wind speed, and humidity, provided by meteorological sensors or meteorological stations. Weather forecast data is used to analyze weather trends over a period of time in the future and predict weather conditions that may affect delivery, such as rainstorms and typhoons. Population distribution data is used to understand the socioeconomic conditions of the construction area and its surroundings, including population density and types of economic activities, to assist in site selection decisions. Infrastructure data includes power grid layout, communication networks, and energy supply.

[0059] Geographic information data helps to analyze the feasibility of distribution routes and optimize site selection. Detailed geographic information data can ensure that route planning during material distribution is more reasonable and efficient. Traffic condition data helps to avoid traffic congestion areas, select the optimal distribution route, and improve the efficiency and punctuality of material transportation. Material demand data helps to accurately predict and arrange material distribution. Through detailed analysis of material demand, it can ensure that the required materials are delivered to the designated location at the right time to meet project needs. Meteorological data can be used to avoid the risks brought by bad weather in advance, adjust distribution plans, and ensure that materials arrive at the destination safely and in a timely manner. Socioeconomic data can select economically developed and conveniently transported areas as distribution centers to improve overall distribution efficiency and service levels.

[0060] S200: Processing the material distribution data using a missing data processing algorithm based on fuzzy clustering to obtain processed material distribution data.

[0061] It should be understood that the missing data processing algorithm based on fuzzy clustering is an effective filling method, which estimates and fills missing values ​​through fuzzy clustering of data. The core idea of ​​this algorithm is to use the flexible classification ability of fuzzy clustering to assign each data point to multiple clusters to capture the intrinsic structure of the data and estimate the missing values ​​on this basis.

[0062] In some embodiments, the step of processing the material distribution data using the missing data processing algorithm based on fuzzy clustering to obtain processed material distribution data includes:

[0063] The material distribution data is preprocessed to obtain material distribution data with outliers removed.

[0064] Initialize missing values ​​by using the mean or median method to facilitate clustering.

[0065] The cluster centers and membership matrices are iteratively updated by minimizing the weighted squared error objective function.

[0066] It should be understood that the membership matrix represents the membership of each data point in each cluster. By updating the membership matrix, the algorithm can dynamically adjust the cluster assignment of data points to make the estimated missing values ​​more accurate. The cluster center is the center point of the cluster, which is updated in each iteration and affects the next step of missing value estimation. The update formula of the cluster center is based on the membership matrix and the current complete data.

[0067] The membership matrix is ​​used to calculate new cluster centers.

[0068] Obtain missing values ​​based on new cluster centers and membership matrices.

[0069] It should be understood that in each iteration, the above process is repeated until the convergence condition is met. The missing values ​​are obtained based on the weighted average of the membership of the data points and the centers of each cluster. The cluster with higher membership has a greater impact on the missing values. This weighted estimation method makes full use of the flexible classification characteristics of fuzzy clustering, so that the missing values ​​filled can reflect the inherent structure of the data.

[0070] Generate processed material distribution data based on the material distribution data with missing values ​​and outliers removed.

[0071] Specifically, calculate (update) the membership matrix:

[0072]

[0073] Among them, m is the fuzzy factor, which is 2; n is the number of data points; c is the number of clusters; is the position of the ith data point in the rth iteration; x k is the feature vector of the kth data point; is the position of the jth cluster center in the rth iteration.

[0074] The membership matrix must meet the following conditions:

[0075]

[0076] Among them, 1<c≤n.

[0077] Update cluster centers:

[0078]

[0079] Where n is the number of data points; m is the fuzzy factor; is the membership of the i-th data point to the k-th data point after the r+1th iteration; x j is the eigenvector of the jth cluster center.

[0080] If ||v (r+1) -v (r) ||<ε holds or the maximum number of iterations is reached, the algorithm stops, where the convergence threshold ε>0, otherwise the filling value of the missing data is calculated:

[0081]

[0082] Let r=r+1, and recalculate (update) the membership matrix.

[0083] Outliers are removed through preprocessing to ensure the accuracy and reliability of material distribution data. The mean or median method is used to preliminarily fill in missing values, providing a complete data set for cluster analysis, which helps to avoid analysis bias caused by missing data and ensure the representativeness and accuracy of clustering results. The clustering center and membership matrix are iteratively updated by minimizing the weighted square error objective function to continuously optimize the clustering effect. This dynamic adjustment process helps to find more accurate clustering divisions and improve the model's ability to capture data structure. The new cluster center is calculated using the membership matrix to further refine the clustering results. This method can more accurately reflect the intrinsic distribution of data and provide a more reasonable basis for subsequent path planning. Missing values ​​are obtained based on the new cluster center and membership matrix, and the processed material distribution data is generated in combination with the data without outliers. This process not only fills the data gaps, but also improves the integrity and consistency of the data, providing high-quality input for deep learning models and path planning models.

[0084] S300: Use deep learning models to analyze processed material distribution data to evaluate the characteristics of material demand and predict the changing trend of material demand.

[0085] In some embodiments, the steps of analyzing the processed material distribution data using a deep learning model to evaluate the characteristics of material demand and predict the trend of material demand changes include:

[0086] The LSTM model is used to extract the time series and spatial features of the processed material distribution data to predict material demand and material distribution.

[0087] It should be understood that the LSTM (Long Short-Term Memory) model is a special deep learning model suitable for processing and predicting time series data. LSTM can effectively capture long-term dependencies and prevent the gradient vanishing problem through its gating mechanism (input gate, forget gate, output gate). This feature makes LSTM particularly suitable for capturing the law of material demand changes over time, such as peak and periodic fluctuations in material demand. When training the LSTM model, the historical material demand data is first input, and then the future demand forecast results are gradually output through the LSTM network. This process enables the LSTM model to gradually learn and capture the time-varying pattern of material demand by adjusting the weights and biases of the network.

[0088] The input data of the LSTM model include construction progress, meteorological conditions, and weather conditions; the construction progress is used to record the schedule and actual progress of the construction project. Meteorological conditions include temperature, precipitation, and wind speed, which affect the construction environment and material demand. Meteorological conditions include road capacity and congestion, which affect the time of material distribution. The LSTM model can automatically extract various features and combine historical data to predict future material demand and its distribution. It can accurately capture the multidimensional factors that affect material demand and improve the accuracy and timeliness of the prediction results.

[0089] The multi-task learning model and critical path method are used to determine the priority of material demand in the processed material distribution data and identify key materials.

[0090] It should be understood that the multi-task learning model can handle multiple related tasks at the same time, improving the generalization ability of the model. In the material demand priority analysis, the multi-task learning model determines the priority of each type of material by sharing the network structure and jointly analyzing factors such as material type, construction stage and urgency. The input of the multi-task learning model includes material type, construction stage, and material urgency; the output is the priority ranking of each type of material, and the key materials are identified for priority distribution. Ensure that key materials are supplied in place first to avoid construction delays due to material shortages. The critical path method (CPM) is used to identify key tasks that affect the project schedule. In the material demand analysis, CPM can identify material requirements that are critical to the construction schedule and prioritize these materials. Based on the project's construction plan, all construction activities are listed, and the dependencies and time requirements between activities are analyzed. Through CPM, the material demand path that has the greatest impact on the construction schedule is identified.

[0091] The improved particle swarm algorithm is used to cluster analyze the material demand, material distribution and priority of material demand to evaluate the characteristics of material demand and predict the changing trend of material demand.

[0092] It should be understood that cluster analysis is used to identify the spatial distribution characteristics of material demand and optimize the location and coverage of distribution centers. Through cluster analysis, we can identify areas where material demand is concentrated, ensure that distribution centers can effectively cover these areas, and optimize distribution routes and resource allocation. The specific steps of cluster analysis of material demand, material distribution, and material demand priority using the improved particle swarm algorithm are as follows:

[0093] The following initial parameters are set: particle swarm size, sample feature dimension, number of clusters, maximum number of iterations, position boundary, and maximum speed.

[0094] Chaotic numbers are generated through chaotic mapping, and then mapped to position intervals. At the same time, the reverse position of the particle is constructed through reverse learning, and the fitness of the particle and its reverse position are compared, so that the position corresponding to the better fitness is taken as the initial position of the particle.

[0095] The search starts from the initial position of the particle and the particle is updated through normal iteration, otherwise the particle position is mutated through Cauchy and boundary control is performed.

[0096] The denoised index data set is divided according to the nearest neighbor rule, and the fitness of the particles is recalculated.

[0097] Compare and update the optimal positions and optimal fitness values ​​of individual particles and groups.

[0098] If the iteration stopping condition is met, the denoised index data is divided according to the nearest neighbor rule with the optimal solution as the clustering center to obtain clusters, thus completing the clustering of different demand characteristics.

[0099] Otherwise, continue to iterate and update the particles.

[0100] The LSTM model can accurately capture the multi-dimensional factors that affect material demand and improve the accuracy and timeliness of the forecast results. The multi-task learning model and the critical path method can simultaneously process multiple related tasks, dynamically adjust the priority of tasks, and use the improved particle swarm algorithm to cluster the material demand, distribution and priority to optimize the distribution path, improve the ability to solve the global optimal solution, and enhance the algorithm's ability to jump out of the local optimal.

[0101] S400: Construct a path planning model based on multi-objective optimization according to the characteristics of material demand and the changing trend of material demand.

[0102] In some embodiments, the steps of constructing a path planning model based on multi-objective optimization according to the characteristics of material demand and the trend of material demand changes include:

[0103] The objective function and constraints are set according to the characteristics of material demand and the changing trend of material demand; the objective function includes minimizing the total transportation cost, minimizing the total transportation time and maximizing the path reliability; the constraints include vehicle capacity limitation, time window constraint and road condition limitation.

[0104] It should be understood that minimizing the total transportation cost refers to minimizing the total transportation cost of the distribution vehicles, including fuel costs, vehicle usage costs and driver fees. Minimizing the total transportation time refers to minimizing the total transportation time of materials from the distribution center to each construction site to ensure that the materials can arrive in time. Maximizing the reliability of the path is to select a more reliable path based on real-time road conditions and weather conditions to reduce the risk in the distribution process. Vehicle capacity restrictions refer to the restrictions on the load and volume of each distribution vehicle. Time window constraints mean that the material receiving time of each construction site has a certain window period, and the distribution vehicle must arrive within the specified time. Road condition restrictions are the paths selected based on road capacity and prohibited regulations to ensure that the path is feasible and unobstructed.

[0105] The model input is determined based on the characteristics of material demand; the model input includes the geographical location of the construction site, material demand, vehicle information, road network data, real-time traffic data and weather forecast.

[0106] A path planning model based on multi-objective optimization is constructed according to the objective function, constraints and model input.

[0107] Setting objective functions and constraints based on the characteristics of material demand and the changing trend of material demand can comprehensively consider multiple factors, flexibly respond to real-time changing traffic conditions and emergency needs, and ensure that the distribution plan has a high degree of adaptability and flexibility, thereby meeting the dynamic needs of power grid technical transformation projects. Based on model inputs such as the geographical location of the construction site, material demand, vehicle information, road network data, real-time traffic data, and weather forecasts, the accuracy and feasibility of the planning can be ensured. The multi-objective optimization model not only considers logistics costs and time, but also pays attention to the reliability of the path. By comprehensively evaluating multiple factors, the optimal balance point is found, thereby improving overall distribution efficiency and reducing operational risks.

[0108] S500: A niche genetic algorithm optimized based on the marine predator algorithm solves the multi-objective function in the path planning model based on multi-objective optimization to obtain the global optimal distribution center location and planning path.

[0109] It should be understood that the Niche Genetic Algorithm (NGA) introduces the concept of "niche" to imitate the mechanism of survival competition of biological populations in nature under different ecological environments to maintain population diversity and avoid premature convergence. The Marine Predator Algorithm (MPA) simulates the natural law of survival of the fittest in the ocean, where various marine organisms constantly switch their identities as predators and prey, and switch foraging strategies according to different situations, thereby optimizing the process.

[0110] In some embodiments, the steps of solving the multi-objective function in the path planning model based on multi-objective optimization by using a niche genetic algorithm optimized by a marine predator algorithm to obtain a globally optimized distribution center location and planning path include:

[0111] Initialize the population, and each individual in the population represents a candidate solution.

[0112] The population is divided into niches based on a distance metric or a sharing function, and each niche represents a local area of ​​the solution space.

[0113] Fusion of marine predator algorithm with niche to optimize niche genetic algorithm.

[0114] In each iteration, selection, crossover, and mutation operations are performed according to the fitness function to update the population.

[0115] The fitness function is used to evaluate each individual in the updated population to obtain the global optimal distribution center location and planning path.

[0116] By introducing the marine predator algorithm, the global search capability of the niche genetic algorithm can be enhanced, which can effectively avoid falling into the local optimal solution and obtain the global optimal distribution center location and planning path.

[0117] In some embodiments, the step of integrating the marine predator algorithm with the niche to optimize the niche genetic algorithm further comprises:

[0118] The Levy flight mechanism of the marine predator algorithm is introduced within or between niches to enhance the global search capability of the niche genetic algorithm.

[0119] It should be understood that the Levy flight mechanism is an optimization algorithm based on random walks. Its core idea is to generate random steps and directions through Levy distribution, so as to achieve a wide range of exploration during the search process. The Levy flight mechanism can be used to enhance the global search capability of NGA. By introducing the Levy flight mechanism within or between small habitats, individuals can make large-scale jumps and explore a wider solution space, thereby discovering new local optimal points and avoiding falling into local optimal solutions.

[0120] Niche maintenance through dynamic predation behavior of a marine predator algorithm to optimize individual distribution.

[0121] It should be understood that in NGA, the maintenance of microhabitats usually relies on a certain distance metric or sharing function. The dynamic predation behavior of MPA (such as multi-stage search strategy in different water layers) can be used to optimize the individual distribution and diversity of microhabitats. In the case of sparse or over-concentrated habitats, the predation behavior of MPA can guide the redistribution of individuals to achieve a more uniform coverage of the solution space.

[0122] The Brownian motion strategy of the marine predator algorithm is introduced in each niche to fine-tune the local optimal solution to improve the accuracy of local search within the niche.

[0123] It should be understood that the Brownian motion of MPA is suitable for fine-tuning local solutions. By introducing the Brownian motion mechanism in each niche, individuals are able to conduct small-scale searches and adjustments, thereby refining the local optimal solution. This helps to improve the optimization accuracy of NGA in each niche.

[0124] The adaptive adjustment mechanism of the marine predator algorithm is used to dynamically adjust the key parameters of the microhabitat; the key parameters include microhabitat size, mutation rate, and crossover rate.

[0125] It should be understood that MPA has the characteristics of adaptively adjusting the predation strategy. Specifically, the size of habitats, sharing parameters, and mutation rate in NGA can be dynamically adjusted to make the algorithm better adapt to the search environment at different stages. For example, the number of habitats can be increased in the early stage and reduced in the later stage to concentrate development.

[0126] The above method comprehensively analyzes the material distribution data, and constructs a path planning model based on multi-objective optimization according to the deep learning model and multi-objective optimization algorithm. It then introduces the marine predator algorithm to enhance the global search capability of the niche genetic algorithm, which can effectively avoid falling into the local optimal solution, and then find a better distribution center site selection and planning path, realize the integrated optimization of distribution center site selection and path planning, and solve the problem of inaccurate site selection and planning path of material distribution centers.

[0127] It can be seen from the above technical solutions that the embodiment of the present application provides a multi-objective material distribution center site selection and path planning method, by acquiring material distribution data; the material distribution data includes geographic information data, traffic condition data, material demand data, meteorological data and socio-economic data; the material distribution data is processed by a missing data processing algorithm based on fuzzy clustering to obtain processed material distribution data; the processed material distribution data is analyzed using a deep learning model to evaluate the material demand characteristics and predict the material demand change trend; a path planning model based on multi-objective optimization is constructed according to the material demand characteristics and material demand change trend; the multi-objective function in the path planning model based on multi-objective optimization is solved based on a microhabitat genetic algorithm optimized by a marine predator algorithm to obtain a globally optimized distribution center site selection and planning path, thereby solving the problem of inaccurate material distribution center site selection and planning path.

[0128] Similar parts between the embodiments provided in this application can be referenced to each other. The specific implementation methods provided above are only a few examples under the general concept of this application and do not constitute a limitation on the protection scope of this application. For those skilled in the art, any other implementation methods expanded based on the scheme of this application without creative work belong to the protection scope of this application.

Claims

1. A multi-objective material distribution center site selection and path planning method, characterized in that: The method comprises: Acquire material distribution data; the material distribution data includes geographic information data, traffic condition data, material demand data, meteorological data and socio-economic data; Processing the material distribution data using a missing data processing algorithm based on fuzzy clustering to obtain processed material distribution data; Analyzing the processed material distribution data using a deep learning model to evaluate material demand characteristics and predict material demand change trends; Constructing a path planning model based on multi-objective optimization according to the characteristics of material demand and the changing trend of material demand; The microhabitat genetic algorithm optimized based on the marine predator algorithm solves the multi-objective function in the path planning model based on multi-objective optimization to obtain the global optimal distribution center location and planning path.

2. The multi-objective material distribution center site selection and path planning method according to claim 1 is characterized in that: The geographic information data includes terrain feature data and land use data; The traffic condition data includes road network data and traffic flow data; The material demand data includes material types and quantities, data sources, and material supply schedule; The meteorological data includes real-time meteorological data and meteorological forecast data; The socio-economic data include population distribution data and infrastructure data.

3. The multi-objective material distribution center site selection and path planning method according to claim 1 is characterized in that: The step of processing the material distribution data using a missing data processing algorithm based on fuzzy clustering to obtain processed material distribution data comprises: Preprocessing the material distribution data to obtain material distribution data with outliers removed; Initialize missing values ​​by using the mean or median method to facilitate clustering; Iteratively update the cluster centers and membership matrix by minimizing the weighted square error objective function; Use the membership matrix to calculate new cluster centers; Obtain missing values ​​based on new cluster centers and membership matrices; Generate processed material distribution data based on the missing values ​​and the material distribution data with outliers removed.

4. The multi-objective material distribution center site selection and path planning method according to claim 3 is characterized in that: The updating formula of the membership matrix is: Among them, m is the fuzzy factor, which is 2; n is the number of data points; c is the number of clusters; is the position of the ith data point in the rth iteration; x k is the feature vector of the kth data point; is the position of the jth cluster center in the rth iteration.

5. The multi-objective material distribution center site selection and path planning method according to claim 4 is characterized in that: The update formula of the cluster center is: Where n is the number of data points; m is the fuzzy factor; is the membership of the i-th data point to the k-th data point after the r+1th iteration; x j is the eigenvector of the jth cluster center.

6. The multi-objective material distribution center site selection and path planning method according to claim 1 is characterized in that: The steps of analyzing the processed material distribution data using a deep learning model to evaluate the characteristics of material demand and predict the trend of material demand changes include: Extracting time series features and spatial features from the processed material distribution data through the LSTM model to predict material demand and material distribution; Determining the priority of material demand in the processed material distribution data and identifying key materials through a multi-task learning model and a critical path method; The improved particle swarm algorithm is used to cluster and analyze the material demand, material distribution and priority of material demand to evaluate the characteristics of material demand and predict the changing trend of material demand.

7. The multi-objective material distribution center site selection and path planning method according to claim 1 is characterized in that: The steps of constructing a path planning model based on multi-objective optimization according to the material demand characteristics and material demand change trends include: According to the characteristics of the material demand and the changing trend of the material demand, the objective function and the constraint conditions are set; the objective function includes minimizing the total transportation cost, minimizing the total transportation time and maximizing the path reliability; the constraint conditions include vehicle capacity restrictions, time window constraints and road condition restrictions; Determine the model input according to the material demand characteristics; the model input includes the geographical location of the construction site, material demand, vehicle information, road network data, real-time road condition data and weather forecast; A path planning model based on multi-objective optimization is constructed according to the objective function, constraints and model input.

8. The multi-objective material distribution center site selection and path planning method according to claim 1 is characterized in that: The steps of solving the multi-objective function in the path planning model based on multi-objective optimization by using a niche genetic algorithm optimized by a marine predator algorithm to obtain a global optimal distribution center location and planning path include: Initialize a population, where each individual in the population represents a candidate solution; Dividing the population into niches based on a distance metric or a sharing function, each niche representing a local region of the solution space; fusing the marine predator algorithm with the niche to optimize the niche genetic algorithm; In each iteration, selection, crossover and mutation operations are performed according to the fitness function to update the population; The fitness function is used to evaluate each individual in the updated population to obtain the global optimal distribution center location and planning path.

9. The multi-objective material distribution center site selection and path planning method according to claim 8 is characterized in that: The step of integrating the marine predator algorithm with the niche to optimize the niche genetic algorithm also includes: The Levy flight mechanism of the marine predator algorithm is introduced within or between niches to enhance the global search capability of the niche genetic algorithm. Microhabitat maintenance through dynamic predation behavior of marine predator algorithms to optimize individual distribution; The Brownian motion strategy of the marine predator algorithm is introduced in each niche to fine-tune the local optimal solution to improve the accuracy of local search within the niche. The adaptive adjustment mechanism of the marine predator algorithm is used to dynamically adjust the key parameters of the microhabitat; the key parameters include the microhabitat size, mutation rate, and crossover rate.

Images