Fresh food e-commerce-oriented terminal delivery intelligent warehouse division and path optimization method and system

Abstract

The invention relates to the technical field of intelligent warehouse division and path optimization, in particular to a terminal delivery intelligent warehouse division and path optimization method and system for fresh food e-commerce, and the method comprises the steps: obtaining historical order data, real-time order data, weather information and traffic flow data; executing an order space-time distribution prediction algorithm to obtain predicted order space-time distribution; according to the predicted order space-time distribution, executing a dynamic warehouse division optimization algorithm to obtain an optimized warehouse distribution scheme; based on the optimized warehouse allocation scheme, executing a path topology optimization algorithm to obtain an optimized path set; according to the optimized path set, executing a dynamic load balancing algorithm to obtain balanced load distribution; based on the balanced load distribution, executing a global efficiency evaluation algorithm to obtain a system efficiency index; and outputting an optimized warehouse allocation scheme, an optimized path set and an optimized system efficiency index, thereby improving the overall resource utilization rate and the capability of dealing with emergencies.

Description

Technical Field

[0001] The present invention relates to the field of intelligent warehouse distribution and path optimization technology, and more specifically, to an intelligent warehouse distribution and path optimization method and system for terminal distribution of fresh food e-commerce. Background Art

[0002] In recent years, with the booming development of e-commerce, fresh food e-commerce has rapidly emerged as an emerging market segment. However, the special properties of fresh products, such as short shelf life, perishability, and temperature sensitivity, have brought unprecedented challenges to terminal delivery. Traditional delivery methods are unable to cope with the complex scenarios of fresh food e-commerce, and a more intelligent and efficient delivery solution is urgently needed.

[0003] At present, the distribution method commonly used in the industry mainly relies on static warehouse distribution strategies and fixed route planning. This method can play a certain role when the order volume is relatively stable and the distribution environment is relatively simple. However, when faced with the rapidly changing order patterns of fresh food e-commerce and the complex and changeable urban traffic environment, its limitations become particularly obvious.

[0004] In existing research, distribution technology with path optimization as the core has been widely used. For example, Yun chun Cao and Manlin Gong proposed a distribution path optimization method based on genetic algorithm in the paper "Research on the optimization of fresh foodlogistics distribution path under O2O e-commerce model" in 2023 (Cao & Gong, 2023). The study constructed a mathematical model for "last mile" path optimization and verified its effectiveness through MATLAB tools. However, the core of this method is only path optimization, and the impact of the dynamic nature of order distribution on the overall logistics efficiency is not considered, especially in the dynamic adjustment of warehouses and load balancing. In addition, its optimization model has weak adaptability to disturbance conditions (such as order fluctuations and changes in the external environment), and lacks effective evaluation of the global system efficiency.

[0005] Another study that is closer to the present invention was published by Jiajie Li in the paper "Optimization of multi-objective fresh food e-commerce delivery route with fuzzy time window" in 2023 (Li, 2023). This study achieved a trade-off between delivery path and customer satisfaction to a certain extent by constructing a multi-objective path optimization model under a fuzzy time window and using an improved ant colony algorithm to solve it. However, the limitation of this research model is that it focuses too much on the optimization of a single delivery path and ignores the synergy between warehouse layout and load balancing. At the same time, the model fails to dynamically predict the spatiotemporal distribution of orders, and therefore lacks the ability to adapt to complex scenarios.

[0006] In addition, Rayisa Jurat proposed a path optimization method based on genetic algorithm in "Path Optimization of Fresh Food Distribution Based on Genetic Algorithm" in 2023 (Jurat, 2023). This study improved customer satisfaction and enterprise distribution efficiency by constructing a multi-objective optimization model in the fresh food e-commerce distribution scenario in Nanchang. However, the method is limited to a single area in verification and cannot be extended to more complex multi-warehouse and multi-regional distribution networks. At the same time, the study failed to fully integrate the system efficiency evaluation mechanism, such as the lack of dynamic load balancing and global stability optimization, making it difficult to cope with the high volatility and complexity of fresh food e-commerce distribution.

[0007] In addition, existing technologies usually use simple time series analysis methods for order forecasting, such as moving average method or exponential smoothing method. Although these methods are simple to calculate, they are difficult to capture the complex spatiotemporal patterns in order data, resulting in low forecasting accuracy. Especially when encountering promotional activities or emergencies, the forecasting error tends to increase significantly.

[0008] In terms of warehouse optimization, existing technologies mostly use static partitioning methods, such as simple division based on geographical location. This method ignores the dynamic changes in order distribution and cannot adjust warehouse resource allocation in a timely manner, resulting in some warehouses being overloaded while other warehouses have idle resources, greatly affecting the overall distribution efficiency.

[0009] Path optimization is another key link. The classic vehicle routing problem (VRP) solution algorithm is currently widely used. However, these algorithms have high computational complexity when dealing with large-scale, dynamically changing distribution tasks, and it is difficult to obtain a satisfactory solution in a short time. More importantly, they often ignore the topological structure of the path, which may result in an unsmooth distribution path and increase the difficulty of distribution personnel.

[0010] In addition, existing delivery systems generally lack a comprehensive assessment of overall efficiency and stability. Most systems only focus on a single indicator such as delivery time or cost, while ignoring the long-term stability and adaptability of the system. This makes the system prone to confusion when faced with emergencies or market fluctuations, affecting delivery quality. Summary of the invention

[0011] In the face of these problems, the present invention proposes a method and system for intelligent warehouse distribution and path optimization for end-to-end distribution of fresh food e-commerce. The method aims to solve the technical problems of the prior art in terms of low order prediction accuracy, inflexible warehouse distribution strategy, low path optimization efficiency and poor system stability.

[0012] The present invention provides an intelligent warehouse distribution and path optimization method for terminal distribution of fresh food e-commerce, including:

[0013] The acquisition steps include:

[0014] Obtain historical order data, real-time order data, weather information, and traffic flow data;

[0015] Processing steps include:

[0016] Based on the historical order data and the real-time order data, executing an order spatiotemporal distribution prediction algorithm to obtain a predicted order spatiotemporal distribution;

[0017] According to the predicted temporal and spatial distribution of orders, a dynamic warehouse allocation optimization algorithm is executed to obtain an optimized warehouse allocation plan;

[0018] Based on the optimized warehouse allocation plan, executing a path topology optimization algorithm to obtain an optimized path set;

[0019] According to the optimized path set, a dynamic load balancing algorithm is executed to obtain a balanced load distribution;

[0020] Based on the balanced load distribution, a global efficiency evaluation algorithm is executed to obtain a system efficiency index;

[0021] Output steps include:

[0022] Output the optimized warehouse allocation plan, the optimized path set and the system efficiency index.

[0023] Preferably, the order spatiotemporal distribution prediction algorithm specifically includes:

[0024] Constructing time and space complexes;

[0025] Computing a persistent homology group based on the time complex and the space complex;

[0026] The persistence graph is used to extract features to obtain the predicted spatiotemporal distribution of the orders.

[0027] Preferably, the dynamic warehouse optimization algorithm specifically includes:

[0028] Define permutation groups;

[0029] Based on the permutation group, define a group action;

[0030] According to the group action, the objective function is optimized to obtain the optimized warehouse allocation plan.

[0031] Preferably, the path topology optimization algorithm specifically includes:

[0032] Construct path space;

[0033] Define homotopy relations;

[0034] Based on the homotopy relationship, the objective function is optimized to obtain the optimized path set.

[0035] Preferably, the dynamic load balancing algorithm specifically includes:

[0036] define integer splitting;

[0037] Construct a split generation function;

[0038] Based on the split generation function, the objective function is optimized to obtain the balanced load distribution.

[0039] Preferably, the global efficiency evaluation algorithm specifically includes:

[0040] Construct system dynamics models;

[0041] Compute the Jacobian matrix;

[0042] Based on the Jacobian matrix, calculating the Lyapunov exponent;

[0043] According to the Lyapunov index, the system efficiency is evaluated to obtain the system efficiency index.

[0044] Preferably, the persistent homology function in the order spatiotemporal distribution prediction algorithm is defined as:

[0045]

[0046] in, is the predicted spatiotemporal distribution of orders, Φ is the persistent homology function, For the time complex, is the spatial complex and α is the persistence parameter.

[0047] Preferably, the group action function in the dynamic warehouse optimization algorithm is defined as:

[0048]

[0049] in, is the optimized warehouse allocation plan, Ψ is the group action function, G is the permutation group, and β is the optimization parameter.

[0050] Preferably, the homotopy optimization function in the path topology optimization algorithm is defined as:

[0051]

[0052] in, is the set of optimized paths, Θ is the homotopy optimization function, H is the homotopy group, and γ is the optimization parameter.

[0053] The terminal distribution intelligent warehouse distribution and route optimization system for fresh food e-commerce that executes the method includes:

[0054] Data acquisition module, used to obtain historical order data, real-time order data, weather information and traffic flow data;

[0055] An order prediction module, used to execute an order spatiotemporal distribution prediction algorithm based on the historical order data and the real-time order data to obtain a predicted order spatiotemporal distribution;

[0056] A warehouse allocation optimization module is used to execute a dynamic warehouse allocation optimization algorithm according to the predicted temporal and spatial distribution of orders to obtain an optimized warehouse allocation plan;

[0057] A path optimization module, used to execute a path topology optimization algorithm based on the optimized warehouse allocation plan to obtain an optimized path set;

[0058] A load balancing module, used to execute a dynamic load balancing algorithm according to the optimized path set to obtain a balanced load distribution;

[0059] An efficiency evaluation module, configured to execute a global efficiency evaluation algorithm based on the balanced load distribution to obtain a system efficiency index;

[0060] The output module is used to output the optimized warehouse allocation plan, the optimized path set and the system efficiency index.

[0061] The beneficial effects of the present invention are mainly reflected in the following aspects:

[0062] The core of this invention is to combine advanced mathematical theories with the actual needs of fresh food e-commerce to build a comprehensive, efficient and stable intelligent distribution system. By introducing the theory of persistent homology for order prediction, this method can more accurately capture the spatiotemporal structural characteristics of order data and significantly improve the prediction accuracy. This not only provides a reliable data basis for subsequent warehouse division and path optimization, but also enables the system to better cope with sudden order fluctuations, such as a surge in orders during holidays or promotional activities.

[0063] In the warehouse optimization link, the present invention innovatively applies the permutation group theory in group theory. This method enables the system to dynamically adjust the warehouse resource configuration according to the predicted order distribution, and realizes the real-time optimization of the warehouse strategy. Compared with the static warehouse method, the dynamic warehouse strategy of the present invention can better balance the load of each warehouse, improve the overall resource utilization, and also enhance the system's ability to deal with emergencies.

[0064] Path optimization is another outstanding innovation of the present invention. By introducing the homotopy theory in algebraic topology, this method can optimize the delivery path while maintaining the topological structure of the path. This not only improves the efficiency of path planning, but also ensures that the generated path is smoother and easier to execute. For time-sensitive tasks such as fresh food delivery, this optimization can significantly reduce the risk of delivery delays and improve customer satisfaction.

[0065] More importantly, the present invention proposes a system efficiency evaluation method based on chaos theory. By calculating the Lyapunov exponent, the system can comprehensively evaluate its stability and efficiency. This method not only considers the short-term distribution efficiency, but also pays attention to the long-term stability of the system, providing an important basis for management decision-making. In practical applications, this comprehensive evaluation mechanism can help the system find the best balance between efficiency and stability, ensuring long-term, continuous and efficient operation.

[0066] The innovative points of each link of the present invention are not isolated, but mutually coordinated and complementary. The high accuracy of order prediction provides reliable input for warehouse optimization, and dynamic warehouse optimization creates favorable conditions for path optimization, and the results of path optimization in turn affect load balancing and system efficiency. This interlocking design ensures that the entire system can operate in an optimal state and fully taps the potential of each innovative point.

[0067] In general, the method provided by the present invention has shown significant advantages in improving delivery efficiency, reducing operating costs, and enhancing system stability. It can not only effectively solve the urgent problems faced by the terminal delivery of fresh food e-commerce, but also provide new ideas for the intelligent and refined development of the entire industry. With the continuous expansion of the fresh food e-commerce market and the continuous improvement of consumers' requirements for delivery services, the method of the present invention is expected to play an increasingly important role in practical applications and promote the development of the entire industry in a more efficient and intelligent direction. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a system logic block diagram of the present invention.

[0069] Figure 2 The figure is a flow chart of the method of the present invention.

[0070] Figure 3 It is the order distribution prediction curve of the present invention.

[0071] Figure 4 This is the path optimization comparison curve of the present invention.

[0072] Figure 5 It is the curve of the system efficiency of the present invention changing with disturbance. DETAILED DESCRIPTION

[0073] Please refer to Figure 1-5 The present invention provides a method and system for intelligent warehouse division and path optimization for terminal delivery of fresh food e-commerce, aiming to solve the problem of intelligent warehouse division and path optimization in terminal delivery of fresh food e-commerce. The specific implementation of the present invention will be described in detail below.

[0074] The method of the present invention includes an acquisition step, a processing step and an output step. In the acquisition step, the system acquires historical order data, real-time order data, weather information and traffic flow data. These data are the basis for subsequent processing and are crucial for accurately predicting order distribution and optimizing delivery routes.

[0075] In the processing steps, the present invention firstly executes an order spatiotemporal distribution prediction algorithm based on historical order data and real-time order data. The algorithm utilizes the persistent homology theory in topology and can effectively capture the spatiotemporal structural characteristics of order data.

[0076] Next, according to the predicted temporal and spatial distribution of orders, the present invention executes a dynamic warehouse optimization algorithm to obtain an optimized warehouse allocation plan. This step utilizes the permutation group theory in group theory, which can achieve dynamic adjustment of warehouse resources while ensuring distribution efficiency.

[0077] After obtaining the optimized warehouse allocation plan, the present invention executes the path topology optimization algorithm to obtain the optimized path set. This algorithm cleverly uses the homology theory in algebraic topology to achieve path optimization while maintaining the path topology structure.

[0078] Preferably, in one embodiment of the present invention, the path topology optimization algorithm also takes into account traffic flow data and weather information. For example, under severe weather conditions, the algorithm will automatically adjust the path to avoid sections of road that are prone to water accumulation or ice formation. At the same time, the algorithm will dynamically adjust the delivery path based on real-time traffic flow data to bypass congested sections. This dynamic adjustment mechanism greatly improves the efficiency and safety of delivery.

[0079] Next, the present invention executes a dynamic load balancing algorithm based on the optimized path set to obtain a balanced load distribution. This step utilizes the integer splitting theory in number theory, which can achieve a balanced distribution of the delivery vehicle load while ensuring delivery efficiency. In practical applications, the load balancing algorithm will take into account the load capacity of each delivery vehicle, the current load status, and the expected delivery time. For example, for fresh fruit products, the algorithm will give priority to vehicles with better temperature control performance and minimize the number of stops to ensure the freshness of the product.

[0080] Finally, the present invention executes a global efficiency evaluation algorithm based on the balanced load distribution to obtain a system efficiency index. This algorithm draws on the Lyapunov exponent in chaos theory and can comprehensively evaluate the overall efficiency and stability of the system. The system efficiency index not only reflects the current distribution efficiency, but also predicts the stability of the system in the future. For example, if the system efficiency index is lower than the preset threshold (usually set to 0.8), the system will automatically trigger an alarm, prompting managers to adjust the distribution strategy or increase distribution resources.

[0081] In the output step, the present invention outputs the optimized warehouse allocation plan, the optimized path set and the system efficiency index. These output results provide a comprehensive optimization plan for the terminal delivery of fresh food e-commerce, effectively improving the delivery efficiency and reducing the operating cost.

[0082] The order spatiotemporal distribution prediction algorithm of the present invention specifically includes the steps of constructing a time complex and a space complex, calculating a persistent homology group, and extracting features using a persistent graph. This topology-based prediction method can effectively capture the spatiotemporal structural characteristics of order data and has stronger prediction capabilities than traditional time series analysis methods.

[0083] When constructing the time complex, the system discretizes the time axis into a series of time units, such as 15 minutes as a time unit. The space complex divides the delivery area into several grids, each grid represents a space unit. The granularity of this time and space discretization needs to be determined according to the specific delivery needs. For example, for cities with a large delivery range, larger space units may be required; while for periods of high delivery frequency, smaller time units may be required.

[0084] When calculating the persistent homology group, the system uses the homology theory in algebraic topology. Specifically, the system calculates the persistent homology groups of different dimensions (0, 1, 2, etc.). The 0-dimensional persistent homology group reflects the clustering of order points, the 1-dimensional persistent homology group reflects the ring structure of the order distribution, and the 2-dimensional persistent homology group reflects the hollow structure of the order distribution. These topological features of different dimensions provide rich information for subsequent predictions.

[0085] Finally, the system uses the persistence graph to extract features and obtain the predicted spatiotemporal distribution of orders. The persistence graph is a tool for visualizing the results of persistent homology, which records the "birth" and "death" time of topological features. By analyzing the persistence graph, the system can identify stable structures and noise in the order distribution, thereby making more accurate predictions.

[0086] The dynamic warehouse optimization algorithm of the present invention specifically includes the steps of defining a permutation group, defining a group action, optimizing an objective function, etc. This group theory-based optimization method can achieve dynamic adjustment of warehouse resources while ensuring distribution efficiency.

[0087] When defining the permutation group, the system considers each possible warehouse allocation as an element in the permutation group. For example, if there are n warehouses and m distribution points, then the size of the permutation group will be n to the power of m. Although this number can be very large, through clever group theory techniques, the system can efficiently handle this optimization problem.

[0088] When defining group actions, the system defines the effects of permutation groups on warehouse allocation plans. Specifically, each permutation corresponds to a way to reallocate orders to different warehouses. This definition of group actions allows the system to systematically explore all possible allocation plans.

[0089] Finally, the system optimizes the objective function to obtain the optimal warehouse allocation plan. The objective function usually includes multiple indicators, such as total delivery distance, warehouse load balance, order fulfillment rate, etc. The system will assign different weights to these indicators according to specific business needs. For example, for some fresh products with extremely high timeliness requirements, more attention may be paid to the order fulfillment rate; while for some products with a long shelf life, more attention may be paid to the optimization of the total delivery distance.

[0090] Through the above detailed description, it can be seen that the intelligent warehouse distribution and path optimization method for terminal distribution of fresh food e-commerce provided by the present invention is highly innovative and practical. It can not only accurately predict the order distribution, but also dynamically optimize the warehouse allocation and distribution path, realizing the intelligence and efficiency of the terminal distribution of fresh food e-commerce. This method is particularly suitable for dealing with the special challenges faced by fresh food e-commerce, such as large order fluctuations, high timeliness requirements, and perishable products, and provides strong support for improving the operational efficiency and customer satisfaction of fresh food e-commerce. Next, the present invention will elaborate on the specific implementation steps of the path topology optimization algorithm.

[0091] The algorithm includes key steps such as constructing path space, defining homotopy relations and optimizing objective function.

[0092] When constructing the path space, the method of the present invention regards each possible delivery path as a point in the path space. Specifically, the path space X is defined as follows:

[0093]

[0094] Among them, p represents a path from the warehouse to the order delivery point, W is the optimized warehouse set, is the set of order delivery points. This definition allows the system to fully consider all possible delivery paths.

[0095] In defining homotopy relations, this method adopts the classical definition in algebraic topology. Two paths p0 and p1 are considered homotopic if and only if there exists a continuous transformation that transforms p0 into p1. Mathematically, this can be expressed as:

[0096]

[0097] Among them, F is called a homotopy mapping. The definition of this homotopy relationship allows the system to maintain the topological structure of the path during the optimization process, thus avoiding some unreasonable path deformations, such as crossing buildings or crossing rivers.

[0098] Preferably, in one embodiment of the present invention, the system also considers the smoothness of the path. By introducing curvature constraints, it is possible to ensure that the generated path is smoother and reduce sharp turns, thereby improving the safety and comfort of delivery. For example, a curvature constraint condition can be added to the homotopy mapping F:

[0099]

[0100] Among them, κ represents the curvature, K max is the preset maximum curvature threshold. In practical applications, K max The value is usually between 0.1 and 0.5, and the specific value needs to be determined based on the performance of the delivery vehicle and road conditions.

[0101] When optimizing the objective function, the method of the present invention comprehensively considers multiple factors. The main optimization goal is to minimize the path length, while also considering factors such as traffic conditions and time window constraints. The optimization objective function can be expressed as:

[0102]

[0103] Among them, π1(X) represents the basic group of the path space, ||p′(t)|| represents the length of the path, C(p(t)) represents the congestion cost on the path, T(p(t)) represents the time window default cost, and w1, w2, and w3 are the corresponding weight coefficients. In practical applications, these weight coefficients need to be adjusted according to specific business needs. For example, for some fresh products with extremely high timeliness requirements, the weight of w3 may be increased; and during peak traffic hours, the weight of w2 may be increased.

[0104] Integer partition is an important concept in number theory, which is used to decompose an integer into the sum of several positive integers. In the present invention, the integer partition theory is used to solve the distribution problem of delivery tasks. Specifically, the system will allocate all orders to different delivery vehicles to ensure that the load of each vehicle is balanced.

[0105] Next, the present invention will describe in detail the specific implementation steps of the dynamic load balancing algorithm. The algorithm includes key steps such as defining integer splitting, constructing splitting generation function and optimizing objective function.

[0106] When defining integer splitting, this method transforms the distribution problem of delivery tasks into an integer splitting problem. Specifically, if there are n orders to be delivered and the load of each delivery vehicle is m, then the integer splitting set I can be defined as:

[0107]

[0108] Where k represents the number of delivery vehicles, a i represents the number of orders assigned to the i-th vehicle. When constructing the split generation function, the method of the present invention adopts the classical techniques in number theory. The split generation function P(x) is defined as follows:

[0109]

[0110] The coefficients of the expansion of this generating function just give the number of different integer splits, providing a theoretical basis for subsequent optimization.

[0111] Preferably, in one embodiment of the present invention, the system also considers the time window constraint of the order. By introducing the time dimension, the split generation function can be expanded to:

[0112]

[0113] Where t represents the time variable and the index i represents the hour in the 24-hour system. This extension allows the system to consider both the number of orders and the time distribution when performing load balancing. When optimizing the objective function, this method comprehensively considers load balancing and delivery efficiency. The optimization objective function can be expressed as:

[0114]

[0115] Where l(p) represents the load assigned to path p, represents the average load, T(p,l(p)) represents the time required to complete path p under a given load, and λ is the balance factor. In practical applications, the value of λ is usually between 0.1 and 1. A smaller λ value will pay more attention to the balance of the load, while a larger λ value will pay more attention to the efficiency of distribution.

[0116] Through integer splitting theory, the system can reasonably allocate orders to different delivery vehicles to ensure that each vehicle has a balanced load. This helps to avoid overloading or idling some vehicles and improve overall delivery efficiency. The system also considers the time window constraints of the order to ensure that each order can be delivered within the specified time. For example, for fresh fruit products, the system will give priority to vehicles with better temperature control performance and minimize the number of stopovers to ensure the freshness of the products. By introducing the balancing factor λ, the system can find the best balance between load balancing and delivery efficiency. For example, in periods when the order volume fluctuates greatly, a larger λ value (such as 1) can be selected to prioritize delivery efficiency; while in periods when the order volume is relatively stable, a smaller λ value (such as 0.1) can be selected to prioritize load balancing.

[0117] The system will predict the results based on the temporal and spatial distribution of orders in the previous step As the basic input for load balancing. This ensures that the load balancing solution can reflect the latest order requirements. The system obtains the current status of each delivery vehicle (such as location, load, etc.) in real time so that the actual availability of the vehicle can be considered when allocating orders. For example, for fresh fruit products, the system will give priority to vehicles with better temperature control performance to ensure the freshness of the products. The system will also consider the time window constraints of each order to ensure that each order can be delivered within the specified time. For example, for orders with high timeliness requirements, the system will give priority to warehouses that are closer for delivery to ensure on-time delivery.

[0118] Fresh fruit products have high requirements for delivery time and temperature. Through the dynamic load balancing algorithm, the system can give priority to vehicles with better temperature control performance and minimize the number of stops to ensure the freshness of the products. The order volume of fresh food e-commerce fluctuates greatly, especially during promotional activities. Through the dynamic load balancing algorithm, the system can flexibly adjust the load of the delivery vehicle according to the temporal and spatial distribution of orders, ensuring that the delivery task can be completed efficiently even when the order volume fluctuates greatly.

[0119] The global efficiency evaluation algorithm of the present invention includes key steps such as constructing a system dynamics model, calculating the Jacobian matrix, calculating the Lyapunov index and evaluating the system efficiency.

[0120] When constructing the system dynamics model, this method regards the entire distribution system as a dynamic system. The system state vector x includes information such as the position, speed, and load of each distribution vehicle. The system dynamics equation can be expressed as:

[0121]

[0122] Wherein, f is a nonlinear function that describes the change of the system state over time. When calculating the Jacobian matrix, the method of the present invention linearizes the system dynamics equation. The Jacobian matrix J is defined as:

[0123]

[0124] This matrix describes the impact of small changes in the system state on the system dynamics and is key to assessing the stability of the system.

[0125] Preferably, in one embodiment of the present invention, the system also takes into account the influence of external disturbances. By introducing random disturbance terms, the system dynamics equation can be expanded to:

[0126]

[0127] Among them, σ represents the disturbance intensity and ξ(t) is Gaussian white noise. This extension enables the system to evaluate the stability in the face of random factors such as traffic accidents and weather changes.

[0128] When calculating the Lyapunov exponent, this method uses the following definition:

[0129]

[0130] Among them, δx i Represents a small disturbance along the i-th characteristic direction. The Lyapunov exponent describes the sensitivity of the system to small changes in initial conditions and is an important indicator for evaluating whether the system is in a chaotic state.

[0131] Lyapunov Exponent is an important concept in chaos theory, which is used to describe the stability and sensitivity of the system. In the present invention, Lyapunov Exponent is used to evaluate the stability of the entire distribution system. Specifically, the system calculates multiple Lyapunov exponents λ i , and sum them up to get the system efficiency index ε.

[0132] Finally, when evaluating the system efficiency, the method of the present invention comprehensively considers all Lyapunov exponents. The system efficiency index ε is defined as:

[0133]

[0134] Generally speaking, if ε>0, the system is in a chaotic state, indicating that there may be unstable factors in the distribution system; if ε<0, the system is stable, and the distribution process is controllable and efficient. In practical applications, a threshold ε can be set threshold (usually between -0.1 and 0.1), when ε>ε threshold The system optimization process is triggered when

[0135] By evaluating the overall efficiency and stability of the system, the system can provide optimization decision support for managers. For example, if the stability of the system is poor, managers can take measures such as increasing the number of warehouses and adjusting delivery routes to improve the overall efficiency of the system.

[0136] The system obtains the location, speed, load and other information of each delivery vehicle in real time to construct the system state vector x. These data are used to describe the current state of the system. The system also considers the impact of external disturbances, such as traffic accidents and weather changes. By introducing random disturbance terms, the system can evaluate its stability in the face of these random factors. The system combines historical data to evaluate the long-term stability and efficiency of the system. For example, by analyzing the trend of the Lyapunov exponent over a period of time, the system can predict the system performance over a period of time in the future.

[0137] During holidays, the number of orders surges, which may cause the system to be overloaded. Through the global efficiency evaluation algorithm, the system can promptly detect potential unstable factors and take corresponding optimization measures to ensure that delivery tasks can be completed efficiently even during peak periods. When encountering emergencies (such as traffic accidents, bad weather, etc.), the system can use the global efficiency evaluation algorithm to evaluate the stability of the system and adjust the delivery strategy in time to avoid delivery delays or customer complaints caused by emergencies.

[0138] The persistent homology function in the order spatiotemporal distribution prediction algorithm of the present invention is defined as:

[0139]

[0140] The specific implementation of this function involves complex algebraic topological calculations. In practical applications, discretization methods can be used to approximate the calculation of persistent homology. The specific steps are as follows: 1. Substitute the time complex T and the space complex Discretize into a finite simplicial complex. 2. Construct the boundary operator matrix Among them C k 3. Calculate the persistent homology group 4. Use the persistence parameter α to filter the topological features with short lifecycles. 5. Convert the filtered persistent homology information into the order spatiotemporal distribution prediction O.

[0141] Persistent Homology is an important tool in algebraic topology, which is used to capture the topological structure characteristics of data. In this paper, it is used to analyze the spatiotemporal distribution of order data. Specifically, the time complex T and the space complex Respectively represent the order distribution in time and space, while the persistence parameter α is used to filter out short-term fluctuations and retain the long-term stable structure. The order distribution of fresh food e-commerce often has complex spatiotemporal patterns, especially in special circumstances such as holidays and weather changes. Persistent coherence can effectively capture these patterns and help the system more accurately predict future order distribution.

[0142] The system collects order data from the past period of time, including the time, location, quantity, etc. of the order. These data are used to construct the time complex T and the space complex The system also obtains current order data in real time to dynamically adjust the forecast model. This helps to cope with unexpected changes in demand, such as sudden promotions or weather changes. Weather conditions (such as rain, snow, etc.) will affect the distribution and delivery efficiency of orders. The system will adjust the forecast model based on weather forecast data to ensure reasonable delivery arrangements even under adverse weather conditions. Traffic conditions (such as congestion, accidents, etc.) will also affect the delivery path of orders. The system will obtain traffic flow data in real time to optimize delivery routes and avoid delays caused by traffic problems.

[0143] By adjusting the value of α, the system can find a balance between short-term fluctuations and long-term trends. For example, in daily delivery, a smaller α value (such as 0.1) can be used to capture the order fluctuations of the day; while in monthly or quarterly forecasts, a larger α value (such as 0.5) can be used to highlight long-term trends. . Reduce the impact of noise: Persistent coherence can automatically filter out short-term noise and anomalies to avoid the impact of these factors on the forecast results. This is especially important for fresh food e-commerce, because the order volume may be affected by a variety of factors such as weather and promotional activities, resulting in large fluctuations in the short term.

[0144] In this process, the persistence parameter α plays a key role. A smaller α value will retain more local features and is suitable for short-term predictions; a larger α value will highlight long-term stable structures and is suitable for long-term trend analysis. In practical applications, the α value can be dynamically adjusted according to the time scale of the prediction. For example, a smaller α value (such as 0.1) may be used for same-day delivery predictions, while a larger α value (such as 0.5) may be used for monthly trend analysis.

[0145] During holidays, the order volume of fresh food e-commerce usually increases significantly. Through the persistent homology algorithm, the system can predict the temporal and spatial distribution of orders in advance, reasonably arrange warehouse resources and delivery vehicles, and ensure that delivery tasks can be completed efficiently even during peak periods. In severe weather conditions (such as rain, snow, etc.), the order distribution may change significantly. Through the persistent homology algorithm, the system can identify the affected areas and automatically adjust the delivery route to avoid sections of the road that are prone to waterlogging or icing, ensuring the safety and punctuality of delivery.

[0146] Through the above detailed description, it can be seen that the terminal distribution intelligent warehouse division and path optimization method for fresh food e-commerce provided by the present invention has significant innovations in terms of path optimization, load balancing and system efficiency evaluation. These innovations not only improve the efficiency and stability of the distribution system, but also provide strong technical support for fresh food e-commerce to cope with the complex and changing market environment. Especially when dealing with large-scale, high-timeliness fresh food delivery tasks, the advantages of this method are more obvious. By comprehensively using cutting-edge mathematical tools such as topology, group theory, number theory and chaos theory, the present invention provides a comprehensive and in-depth solution to the terminal distribution problem of fresh food e-commerce.

[0147] Permutation Group is an important concept in group theory, which is used to describe the arrangement relationship between objects. In this invention, the permutation group G represents all possible warehouse allocation schemes. The group action function Ψ transforms the predicted order spatiotemporal distribution Combined with the permutation group G, the optimal warehouse allocation plan is generated

[0148] Next, the present invention will elaborate on the group action function in the dynamic warehouse optimization algorithm.

[0149] The function is defined as:

[0150]

[0151] In this function, Ψ represents the group action function, which predicts the spatiotemporal distribution of orders The permutation group G and the optimization parameter β are used as input, and the optimized warehouse allocation plan W is output. The core idea of ​​this function is to use the permutation group theory in group theory to realize the dynamic adjustment of warehouse resources.

[0152] Specifically, the permutation group G can be represented as the set of all possible warehouse allocations. Each group element g∈G corresponds to a way to reallocate orders to different warehouses. The group action function Ψ defines how these permutations act on the current allocation. Mathematically, this can be expressed as:

[0153] Ψ:G×W→W,

[0154] Ψ(g,w)=g(w),

[0155] in, Represents a specific warehouse allocation plan. Preferably, in one embodiment of the present invention, the group action function Ψ also takes into account the capacity constraints of the warehouse and the timeliness requirements of the order. For example, a capacity constraint function C(w) and a timeliness scoring function T(w) can be introduced, and the modified group action function can be expressed as:

[0156]

[0157] Here, g(w) represents the set of all possible allocations obtained after applying the permutation g. The optimization parameter β plays a key regulatory role in this process. It controls the "temperature" of the optimization process, that is, the probability that the system accepts a suboptimal solution. Specifically, in each iteration, the system accepts a new allocation with probability P = exp(-ΔE / β, where ΔE is the energy change of the new solution relative to the current solution.

[0158] A larger β value increases the system's ability to explore new solutions, while a smaller β value makes the system more inclined to local optimal solutions.

[0159] In practical applications, the value of β is usually between 0.01 and 0.1. For example, during periods of drastic fluctuations in order volume (such as holidays), a larger β value (such as 0.1) may be selected to increase the adaptability of the system; while during periods of relatively stable order volume, a smaller β value (such as 0.01) may be selected to improve the accuracy of optimization. The data acquisition module 1 plays an important role in this process. It is not only responsible for obtaining real-time order data, but also needs to collect inventory information of each warehouse, the status of delivery vehicles, etc. These data provide the necessary input for the group action function Ψ to ensure that the optimization results can reflect the actual operating conditions.

[0160] The order volume of fresh food e-commerce fluctuates greatly, especially during holidays or promotional activities. Through the permutation group theory, the system can dynamically adjust warehouse resources according to the real-time order distribution to ensure the load balance of each warehouse and avoid overloading or idleness of some warehouses. The system will select the most suitable warehouse for delivery according to the temporal and spatial distribution of orders. For example, for orders that are close in distance, the same warehouse can be selected for centralized delivery to reduce the empty driving rate of delivery vehicles and reduce operating costs. By adjusting the optimization parameter β, the system can flexibly respond to changes in demand in different business scenarios. For example, during periods of large fluctuations in order volume (such as holidays), a larger β value (such as 0.1) can be selected to increase the adaptability of the system; while during periods of relatively stable order volume, a smaller β value (such as 0.01) can be selected to improve the accuracy of optimization.

[0161] The system will predict the results based on the temporal and spatial distribution of orders in the previous step As the basic input for dynamic warehouse optimization. This ensures that the warehouse allocation plan can reflect the latest order demand. The system obtains inventory information of each warehouse in real time, including inventory, storage capacity, etc. This data is used to constrain the warehouse allocation plan to ensure that the load of each warehouse does not exceed its capacity limit. The system also obtains the current status of each delivery vehicle (such as location, load, etc.) so that the actual availability of the vehicle can be considered when allocating warehouses. For example, for fresh fruit products, the system will give priority to vehicles with better temperature control performance to ensure the freshness of the products. During holidays, the order volume surges, which may cause some warehouses to be overloaded. Through the dynamic warehouse optimization algorithm, the system can dynamically adjust warehouse resources according to the real-time order distribution to ensure that the load of each warehouse is balanced and avoid delivery delays caused by warehouse overload. The order volume of fresh food e-commerce fluctuates greatly, especially during promotional activities. Through the dynamic warehouse optimization algorithm, the system can flexibly adjust warehouse resources according to the temporal and spatial distribution of orders to ensure that the delivery task can be completed efficiently even when the order volume fluctuates greatly.

[0162] Next, the present invention will describe in detail the homotopy optimization function in the path topology optimization algorithm. The function is defined as:

[0163]

[0164] In this function, Θ represents the homology optimization function, which converts the optimized warehouse allocation plan The homotopy group H and the optimization parameter γ are used as input, and the optimized path set is output. The core idea of ​​this function is to use the homotopy theory in algebraic topology to optimize the delivery path while maintaining the topological structure of the path.

[0165] Homotopy is an important concept in algebraic topology, which is used to describe the continuous deformation relationship between paths. In the present invention, the homotopy group H represents all topologically equivalent path categories. The homotopy optimization function Θ optimizes the distribution path while maintaining the path topology structure to ensure that the length and cost of the path are minimized.

[0166] Specifically, the homotopy group H can be represented as the set of all topologically equivalent path categories. Each group element h∈H corresponds to a class of topologically equivalent paths. The homotopy optimization function Θ defines how to optimize the path while maintaining the topological structure. Mathematically, this can be expressed as:

[0167]

[0168] Where L(p) represents the length or cost of path p.

[0169] Preferably, in one embodiment of the present invention, the homology optimization function Θ also takes into account the smoothness and traffic conditions of the path. For example, a smoothness scoring function S(p) and a traffic condition scoring function T can be introduced. (p) ,The modified homology optimization function can be expressed as:

[0170]

[0171] Among them, w1, w2 and w3 are weight coefficients used to balance the three factors of path length, smoothness and traffic conditions.

[0172] Homotopy theory ensures that the optimized path will not pass through inaccessible areas such as buildings and rivers, avoiding the generation of unreasonable paths. This is especially important for fresh food e-commerce, because delivery vehicles need to comply with the actual road network and cannot cross obstacles at will. Through homotopy optimization, the system can minimize the length and cost of the path while maintaining the path topology. This helps to reduce the driving distance of delivery vehicles, reduce fuel consumption and carbon emissions, and improve delivery efficiency.

[0173] The system will allocate warehouses according to the warehouse allocation plan in the previous step. As the basic input for route optimization. This ensures that the optimized route can start from the correct warehouse and reach the designated delivery point. The system will obtain traffic flow data in real time and dynamically adjust the delivery route. For example, during peak traffic hours, the system will give priority to relatively unobstructed routes to avoid delivery delays caused by traffic congestion. The system will also optimize the delivery route based on weather forecast data. For example, in severe weather conditions, the system will automatically adjust the route to avoid sections of the road that are prone to waterlogging or ice formation, ensuring the safety and punctuality of delivery.

[0174] The system can also introduce a smoothness scoring function S(p) and a traffic condition scoring function T(p) to further optimize the smoothness and traffic adaptability of the path. For example, during peak traffic hours, the system will give priority to relatively unobstructed paths to avoid delivery delays caused by traffic congestion.

[0175] In the city center, traffic conditions are complex, roads are narrow, and traffic jams are prone to occur. Through the homology optimization algorithm, the system can select the most unobstructed path while maintaining the path topology structure to avoid delivery delays caused by traffic jams. In the suburbs, the roads are relatively simple, but the distances are longer. Through the homology optimization algorithm, the system can minimize the driving distance of delivery vehicles, reduce fuel consumption and carbon emissions, and improve delivery efficiency while ensuring path safety.

[0176] The optimization parameter γ plays a fine-tuning role in this process. It controls the maximum degree of deformation allowed during the optimization process. Specifically, in each iteration, the system only considers candidate paths that do not differ from the current path by more than γ. Larger values ​​of γ increase the flexibility of the optimization but may increase the computational complexity; smaller values ​​of γ limit the search space but can speed up the optimization.

[0177] In practical applications, the value of γ is usually between 0.001 and 0.01. For example, in the central area of ​​the city with complex traffic conditions, a larger γ value (such as 0.01) may be selected to increase the flexibility of path selection; while in the suburbs with relatively simple traffic conditions, a smaller γ value (such as 0.001) may be selected to speed up the optimization.

[0178] The path optimization module 4 plays a core role in this process. It not only needs to implement the homology optimization function θ, but also needs to process the real-time traffic data from the traffic data collection submodule 41 and the weather forecast information from the weather information acquisition submodule 42. This additional information enables the path optimization to better adapt to the actual distribution environment.

[0179] Finally, the present invention will elaborate on the overall architecture of the terminal distribution intelligent warehouse distribution and path optimization system for fresh food e-commerce. The system includes a data acquisition module 1, an order prediction module 2, a warehouse distribution optimization module 3, a path optimization module 4, a load balancing module 5, an efficiency evaluation module 6 and an output module 7.

[0180] The data acquisition module 1 is responsible for acquiring historical order data, real-time order data, weather information and traffic flow data.

[0181] The order prediction module 2 executes an order spatiotemporal distribution prediction algorithm based on the historical order data and real-time order data obtained from the data acquisition module 1.

[0182] The warehouse optimization module 3 executes a dynamic warehouse optimization algorithm according to the predicted order temporal and spatial distribution output by the order prediction module 2.

[0183] The path optimization module 4 executes the path topology optimization algorithm based on the optimized warehouse allocation plan output by the warehouse distribution optimization module 3.

[0184] The load balancing module 5 executes a dynamic load balancing algorithm according to the optimized path set output by the path optimization module 4 .

[0185] The efficiency evaluation module 6 executes a global efficiency evaluation algorithm based on the balanced load distribution output by the load balancing module 5 .

[0186] The output module 7 is responsible for outputting the optimized warehouse allocation plan, the optimized path set and the system efficiency index to the user or other systems.

[0187] Preferably, in one embodiment of the present invention, a feedback adjustment module 8 may be added. This module is responsible for collecting feedback information during the actual delivery process and dynamically adjusting the parameters of each module based on the information. For example, it may adjust the weight coefficient in the path optimization module 4 based on the difference between the actual delivery time and the predicted time, or adjust the persistence parameter in the order prediction module 2 based on the difference between the actual order volume and the predicted order volume.

[0188] Through the above detailed description, it can be seen that the terminal distribution intelligent warehouse distribution and path optimization system for fresh food e-commerce provided by the present invention is a highly integrated and fully functional intelligent system. It can not only accurately predict order distribution, optimize warehouse allocation and distribution paths, but also dynamically balance loads and evaluate the overall efficiency of the system. This system is particularly suitable for dealing with the complex and changeable distribution environment faced by fresh food e-commerce, and can significantly improve distribution efficiency, reduce operating costs, and improve customer satisfaction. Through the comprehensive use of a variety of advanced mathematical theories and algorithms, this system provides a comprehensive and in-depth solution to the terminal distribution problem of fresh food e-commerce, which has important theoretical value and practical application prospects.

[0189] In order to verify the superiority of the intelligent warehouse distribution and path optimization method and system for fresh food e-commerce terminal delivery proposed in this invention, we designed a set of comparative experiments. These experiments are aimed at simulating real fresh food e-commerce delivery scenarios and comparing them with existing delivery methods.

[0190] Example 1 adopts the complete method proposed by the present invention, including order prediction based on persistent homology, dynamic warehouse optimization based on group theory, path optimization based on homology theory, and system efficiency evaluation based on Lyapunov index. Comparative Example 1 adopts the traditional time series prediction method (ARIMA) and static warehouse strategy, combined with the classic vehicle routing problem (VRP) solution algorithm. Comparative Example 2 adopts a machine learning method (random forest) for order prediction, combined with a greedy algorithm for warehouse and path optimization.

[0191] We selected a fresh food e-commerce platform in a medium-sized city as the test object. The platform handles about 10,000 orders per day, has 5 distribution centers and 100 delivery vehicles. The test period is 30 days, covering normal working days, weekends, and a small promotion event. The main test indicators include order prediction accuracy, average delivery time, delivery success rate, vehicle utilization rate, and system stability.

[0192] The following are the testing standards and methods for various indicators:

[0193] 1. Order forecast accuracy: The mean absolute percentage error (MAPE) is used to evaluate forecast accuracy. The difference between the predicted order volume and the actual order volume is calculated after the close of each day.

[0194] 2. Average delivery time: The average time from order allocation to delivery to the customer. Calculated using log data from the delivery system.

[0195] 3. Delivery success rate: The proportion of orders successfully delivered within the promised time. Calculated based on customer feedback and system records.

[0196] 4. Vehicle utilization rate: The average load rate of delivery vehicles. It is calculated by the ratio of the actual load of each vehicle to its maximum load capacity.

[0197] 5. System stability: The method based on Lyapunov index proposed in the present invention is used for evaluation. The Lyapunov index of the system is calculated once a day and the average value for 30 days is taken.

[0198] The test results are shown in the following table:

[0199] index Example 1 Comparative Example 1 Comparative Example 2 Order Forecast Accuracy (MAPE) 5.2％ 12.7％ 8.9％ Average delivery time (minutes) 47 68 59 Delivery success rate 98.5％ 92.3％ 95.1％ Vehicle Utilization 78.6％ 65.2％ 71.4％ System stability (Lyapunov index) -0.042 0.103 0.025

[0200] It can be clearly seen from the test results that the method of the present invention (Example 1) is significantly superior to the traditional method (Comparative Example 1) and the simple machine learning method (Comparative Example 2) in all key indicators.

[0201] The significant improvement in order prediction accuracy (MAPE is only 5.2%) proves that the prediction method based on persistent homology can better capture the spatiotemporal structural characteristics of order data. This high-precision prediction lays a solid foundation for subsequent warehouse division and path optimization.

[0202] The significant reduction in average delivery time (only 47 minutes) and the improvement in delivery success rate (up to 98.5%) fully demonstrate the superiority of dynamic warehouse optimization based on group theory and path optimization based on homology theory. The combination of these two technologies enables the system to respond more flexibly to changes in order distribution and achieve efficient path planning while maintaining the path topology structure.

[0203] The increase in vehicle utilization (reaching 78.6%) not only reflects the optimization of system resource allocation, but also means lower operating costs and higher environmental friendliness. This result is mainly due to the dynamic load balancing algorithm proposed in the present invention, which can achieve near-optimal load distribution while considering multiple constraints.

[0204] Finally, the significant improvement in the system stability index (Lyapunov index is -0.042) proves that the method of the present invention can effectively suppress the chaotic behavior of the system, so that the distribution system can still maintain stable and efficient operation in the face of complex and changing external environments. This is especially important for fresh food e-commerce, because the distribution of fresh products often faces more uncertainty and time pressure.

[0205] It is worth noting that the method of the present invention showed stronger adaptability in the small-scale promotional activities encountered during the test. When the order volume suddenly increased, the indicators of Example 1 only fluctuated slightly, while Comparative Examples 1 and 2 showed obvious performance degradation. This further proves the superiority of the method of the present invention in dealing with complex and changeable fresh food e-commerce delivery scenarios.

[0206] In summary, the intelligent warehouse distribution and path optimization method and system for terminal distribution of fresh food e-commerce proposed in the present invention have shown significant advantages in improving distribution efficiency, reducing operating costs, and enhancing system stability. This comprehensive optimization method based on advanced mathematical theory can not only cope with the challenges currently faced by fresh food e-commerce, but also provide a scalable solution for more complex distribution scenarios in the future. With the continuous development of the fresh food e-commerce market and the continuous improvement of consumers' requirements for distribution services, the method of the present invention is expected to play an increasingly important role in practical applications.

[0207] Figure 4For order distribution prediction, the blue solid line represents the order distribution predicted by the method of the present invention. The orange dotted line represents the actual order distribution. The predicted curve is highly consistent with the actual curve, especially in both order-intensive and sparse areas, the peak and trend of the real distribution can be captured. The present invention uses the theory of topological persistent homology to capture the multidimensional characteristics of order distribution (dynamic changes in time and space and local aggregation), significantly improving the robustness and accuracy of the prediction. Traditional statistical models often ignore boundary fluctuations (such as a sharp increase in peak areas) in predictions, while the present invention can model global and local spatiotemporal characteristics through persistent homology groups. Accurately predict order distribution and provide a reliable data basis for warehouse distribution and path optimization. Dynamically respond to order changes to adapt to the high volatility and diversity needs of fresh food e-commerce.

[0208] Figure 5 For path optimization comparison, the blue solid line represents the total length of the path before optimization. The orange dotted line represents the total length of the path after optimization. The total length of the path after optimization is shortened by an average of about 15.2%, especially when the path is complex and the intersection is frequent, the optimization effect is more obvious. The present invention constructs a topological optimization model of the path space through the homology theory of algebraic topology, which can effectively avoid path redundancy (such as intersections and loops). Traditional path optimization algorithms (such as genetic algorithms) cannot ensure the global optimality of the path and are prone to fall into local optimality, while the topological optimization model of the present invention can globally optimize the path structure. Shorten the delivery path and reduce vehicle fuel consumption and time cost. The globality of path optimization ensures the timeliness and stability of delivery.

[0209] Figure 4 For path optimization comparison, the blue solid line represents the total length of the path before optimization. The orange dotted line represents the total length of the path after optimization. The total length of the path after optimization is shortened by an average of about 15.2%, especially when the path is complex and the intersection is frequent, the optimization effect is more obvious. The present invention constructs a topological optimization model of the path space through the homology theory of algebraic topology, which can effectively avoid path redundancy (such as intersections and loops). Traditional path optimization algorithms (such as genetic algorithms) cannot ensure the global optimality of the path and are prone to fall into local optimality, while the topological optimization model of the present invention can globally optimize the path structure. Shorten the delivery path and reduce vehicle fuel consumption and time cost. The globality of path optimization ensures the timeliness and stability of delivery.

[0210] These effects provide theoretical support and practical value for the terminal delivery of fresh food e-commerce, and at the same time highlight the core competitiveness of the present invention in logistics optimization.

[0211] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modification, replacement, and improvement made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. The intelligent warehouse distribution and route optimization method for terminal distribution of fresh food e-commerce is characterized by: include: The acquisition steps include: Obtain historical order data, real-time order data, weather information, and traffic flow data; Processing steps include: Based on the historical order data and the real-time order data, executing an order spatiotemporal distribution prediction algorithm to obtain a predicted order spatiotemporal distribution; According to the predicted temporal and spatial distribution of orders, a dynamic warehouse allocation optimization algorithm is executed to obtain an optimized warehouse allocation plan; Based on the optimized warehouse allocation plan, executing a path topology optimization algorithm to obtain an optimized path set; According to the optimized path set, a dynamic load balancing algorithm is executed to obtain a balanced load distribution; Based on the balanced load distribution, a global efficiency evaluation algorithm is executed to obtain a system efficiency index; Output steps include: Output the optimized warehouse allocation plan, the optimized path set and the system efficiency index.

2. The method according to claim 1, characterized in that The order spatiotemporal distribution prediction algorithm specifically includes: Constructing time and space complexes; Computing a persistent homology group based on the time complex and the space complex; The persistence graph is used to extract features to obtain the predicted spatiotemporal distribution of the orders.

3. The method according to claim 1, characterized in that The dynamic warehouse optimization algorithm specifically includes: Define permutation groups; Based on the permutation group, define a group action; According to the group action, the objective function is optimized to obtain the optimized warehouse allocation plan.

4. The method according to claim 1, characterized in that: The path topology optimization algorithm specifically includes: Construct path space; Define homotopy relations; Based on the homotopy relationship, the objective function is optimized to obtain the optimized path set.

5. The method according to claim 1, characterized in that The dynamic load balancing algorithm specifically includes: define integer splitting; Construct a split generation function; Based on the split generation function, the objective function is optimized to obtain the balanced load distribution.

6. The method according to claim 1, characterized in that The global efficiency evaluation algorithm specifically includes: Construct system dynamics models; Compute the Jacobian matrix; Based on the Jacobian matrix, calculating the Lyapunov exponent; According to the Lyapunov index, the system efficiency is evaluated to obtain the system efficiency index.

7. The method according to claim 1, characterized in that The persistent homology function in the order spatiotemporal distribution prediction algorithm is defined as: in, is the predicted spatiotemporal distribution of orders, Φ is the persistent homology function, For the time complex, is the spatial complex and α is the persistence parameter.

8. The method according to claim 1, characterized in that The group action function in the dynamic warehouse optimization algorithm is defined as: in, is the optimized warehouse allocation plan, Ψ is the group action function, G is the permutation group, and β is the optimization parameter.

9. The method according to claim 1, characterized in that: The homotopy optimization function in the path topology optimization algorithm is defined as: in, is the set of optimized paths, Θ is the homotopy optimization function, H is the homotopy group, and γ is the optimization parameter.

10. An intelligent warehouse distribution and route optimization system for terminal distribution of fresh food e-commerce that implements the method described in any one of claims 1 to 9, characterized in that: include: Data acquisition module, used to obtain historical order data, real-time order data, weather information and traffic flow data; An order prediction module, used to execute an order spatiotemporal distribution prediction algorithm based on the historical order data and the real-time order data to obtain a predicted order spatiotemporal distribution; A warehouse allocation optimization module is used to execute a dynamic warehouse allocation optimization algorithm according to the predicted temporal and spatial distribution of orders to obtain an optimized warehouse allocation plan; A path optimization module, used to execute a path topology optimization algorithm based on the optimized warehouse allocation plan to obtain an optimized path set; A load balancing module, used to execute a dynamic load balancing algorithm according to the optimized path set to obtain a balanced load distribution; An efficiency evaluation module, configured to execute a global efficiency evaluation algorithm based on the balanced load distribution to obtain a system efficiency index; The output module is used to output the optimized warehouse allocation plan, the optimized path set and the system efficiency index.

Images