Logistics navigation method and device for catering business

Abstract

The invention relates to the technical field of data analysis, in particular to a logistics navigation method and device for catering businesses, and the method comprises the steps: analyzing the identification information, submitted by logistics personnel, of meal delivery customers, and extracting the attribute features of the meal delivery customers from the identification information; according to the attribute characteristics of the meal delivery customers, positioning node paths associated with the meal delivery customers to generate meal delivery sequence chains in the node paths, allocating weight values to the meal delivery sequence chains, and generating a weighted meal delivery sequence chain list; calculating navigation route time consumption of meal delivery sequence chains in the weighted meal delivery sequence chain list, identifying potential route accident points by using an anomaly detection algorithm, and generating a navigation route report containing a navigation route time consumption result and the route accident points; and drawing a comprehensive navigation route map by using a multi-dimensional visualization algorithm to obtain a navigation route view, and generating a navigation route relieving strategy.

Description

Technical Field

[0001] The present application relates to the field of data analysis technology, and in particular to a logistics navigation method and device for catering business. Background Art

[0002] Developers and businesses need an intelligent analysis method that can parse the delivery customer identification information submitted by users and extract attribute features from it to accurately locate the node paths between delivery customers and generate a weighted dependency chain list containing direct and indirect dependency chains.

[0003] Existing risk assessment solutions for catering business navigation suffer from several significant flaws. First, traditional static analysis and vulnerability scanning tools can only identify known vulnerabilities and are incapable of addressing new attacks and unknown threats. Second, these tools often overlook indirect dependencies within dependency chains, resulting in incomplete and inaccurate risk assessments. Furthermore, most existing solutions calculate risk scores based on predefined rules, lacking flexibility and adaptability, and are unable to dynamically adjust weights and assessment criteria based on actual conditions. Finally, existing visualization tools and risk mitigation measures are relatively simplistic, making it difficult to intuitively display complex dependencies and risk patterns, let alone provide device-based optimization strategies. Summary of the Invention

[0004] To achieve the above objectives, this application provides the following technical solutions:

[0005] According to a first aspect of the present invention, the present invention claims protection for a logistics navigation method for a catering business, comprising:

[0006] Parsing the identification information of the delivery customer submitted by the logistics personnel, and extracting the attribute characteristics of the delivery customer from the identification information;

[0007] According to the attribute characteristics of the delivery customers, the node paths associated with the delivery customers are located to generate a delivery sequence chain in the node path, and weight values ​​are assigned to the delivery sequence chains to generate a weighted delivery sequence chain list, wherein the delivery sequence chain includes a direct delivery sequence chain and an indirect delivery sequence chain;

[0008] A Bayesian network model is used in combination with a predefined navigation route indicator system to calculate the navigation route time of the delivery sequence chain in the weighted delivery sequence chain list, and an anomaly detection algorithm is used to identify potential route accident points to generate a navigation route report including the navigation route time result and the route accident points;

[0009] Based on the navigation route report, a comprehensive navigation route map is drawn using a multidimensional visualization algorithm to obtain a navigation route view, and a genetic algorithm and a simulated annealing algorithm are integrated to explore adjustment suggestions for high navigation route areas of the navigation route view to generate a navigation route mitigation strategy; wherein, the comprehensive navigation route map is grouped and displayed using a hierarchical clustering algorithm to display components with similar navigation route patterns, and the navigation route pattern is obtained by performing a similarity judgment on the food delivery sequence chain, the route accident points, and the navigation route time consumption, and the navigation route mitigation strategy is used to reduce the navigation route time consumption of the food delivery customer.

[0010] Furthermore, the Bayesian network model is combined with a predefined navigation route indicator system to calculate the navigation route time of the delivery sequence chain in the weighted delivery sequence chain list, including:

[0011] According to the Bayesian network model, a low-dimensional vector representation of the weighted delivery sequence chain list is obtained by using a graph embedding algorithm to obtain a dependency vector;

[0012] Based on the dependency vector, a predefined navigation route indicator system is used in combination with a multi-dimensional weight adjustment mechanism to calculate an initial navigation route score, and the initial navigation route score is assigned to the delivery sequence chain, wherein the navigation route indicator system includes the number of vulnerabilities, update frequency, and community activity;

[0013] Using the delivery sequence chain assigned initial navigation route scores, the Bayesian network model is combined with the gradient boosting decision tree algorithm to optimize the navigation route probability distribution of the child nodes that the delivery customer directly or indirectly depends on under the condition of the parent node of the delivery customer, thereby obtaining the optimized conditional probability;

[0014] According to the optimization conditional probability, combined with the ensemble learning method, the overall navigation route score of the food delivery sequence chain is evaluated to obtain the navigation route time result;

[0015] Furthermore, the use of an anomaly detection algorithm to identify potential route accident points to generate a navigation route report containing navigation route time results and the route accident points includes:

[0016] Based on the navigation route time results, an anomaly detection algorithm is used in combination with an adaptive threshold setting strategy to analyze the navigation route time of the delivery sequence chain, identify data points that deviate from the normal range, and obtain route accident points;

[0017] Using natural language processing technology to parse the safety notices and technical documents of the accident point on the route, automatically extracting information directly related to the accident point in the safety notices and technical documents, and generating details of the accident point;

[0018] Summarize and organize all the navigation route time results and the accident point details, and apply time series analysis to predict future navigation route trends to generate comprehensive navigation route data;

[0019] An interactive interface is created using a visualization tool to organize the comprehensive navigation route data, and a navigation route report including the navigation route time result and the accident point archive is generated.

[0020] Furthermore, the delivery sequence chain for assigning initial navigation route scores is used to optimize the navigation route probability distribution of the child nodes that the delivery customer directly or indirectly depends on under the given parent node condition of the delivery customer through the Bayesian network model combined with the gradient boosting decision tree algorithm, thereby obtaining the optimized conditional probability, including:

[0021] Collecting and organizing the low-dimensional vector representation of the food delivery sequence chain and the initial navigation route score to construct a training dataset;

[0022] Initializing and processing the training data set based on the Bayesian network model to obtain a gradient boosting decision tree model, wherein the gradient boosting decision tree model is used to assist in estimating conditional probabilities in the optimized Bayesian network model;

[0023] Using the training data set, jointly training the Bayesian network model and the gradient boosting decision tree model, optimizing the estimation of the conditional probability by iteratively calculating the loss function and adjusting the parameters by backpropagation, and obtaining a joint training result;

[0024] According to the joint training results, the conditional probability distribution of each node in the Bayesian network model is refined, and the conditional probability values ​​of the child nodes that the delivery customer directly or indirectly depends on are updated under the condition of the parent node of a given delivery customer to obtain the optimized conditional probability.

[0025] Furthermore, based on the navigation route time result, the anomaly detection algorithm is combined with an adaptive threshold setting strategy to analyze the navigation route time of the delivery sequence chain, identify data points that deviate from the normal range, and obtain route accident points, including:

[0026] Selecting a suitable anomaly detection algorithm based on the navigation route time result, wherein the anomaly detection algorithm includes an isolation forest and a local outlier factor, and the anomaly detection algorithm can effectively identify abnormal patterns in the navigation route time data to generate an anomaly detection algorithm;

[0027] Based on the historical data and domain knowledge of the food delivery sequence chain, an initial threshold is set for the anomaly detection algorithm to preliminarily screen outliers and obtain a preliminarily set threshold;

[0028] Applying an adaptive threshold setting strategy and automatically updating the initially set threshold in combination with monitoring device performance indicators to improve accuracy and generate an optimized threshold, wherein the monitoring device performance indicators include accuracy and recall;

[0029] The anomaly detection algorithm is combined with the optimized threshold to analyze the navigation route time results, identify data points that deviate from the normal range, and obtain route accident points.

[0030] Furthermore, the method of drawing a comprehensive navigation route map based on the navigation route report using a multi-dimensional visualization algorithm to obtain a navigation route view includes:

[0031] According to the navigation route report, data of the food delivery sequence chain and navigation route information are collected and organized so that the data format is consistent and ready for visualization processing, thereby obtaining a navigation route data set;

[0032] Using a multi-dimensional visualization algorithm, the navigation route data set is mapped, and the navigation route time and accident point location of the food delivery sequence chain are mapped into a multi-dimensional space to generate a visual representation;

[0033] Based on the visual representation, a hierarchical clustering algorithm is applied to group components with similar navigation route patterns and perform similarity judgment and classify them into different navigation route pattern groups to obtain grouped navigation route patterns;

[0034] In combination with geographic information or network topology, adjusting the visual layout of the navigation route modes displayed in groups, optimizing the visual effect, and generating a comprehensive navigation route map;

[0035] Highlighting high navigation route areas of the integrated navigation route map through color coding and icon styles helps logistics personnel quickly identify key navigation route points, and utilizing interactive tools to enhance logistics personnel experience and obtain a navigation route view.

[0036] Furthermore, the node paths associated with the delivery customers are located based on the attribute characteristics of the delivery customers to generate a delivery sequence chain in the node path, and a weight value is assigned to the delivery sequence chain to generate a weighted delivery sequence chain list, including:

[0037] Utilizing the parsed delivery customer identification information submitted by the logistics personnel, extracting attribute characteristics of the delivery customer;

[0038] Based on the attribute characteristics of the delivery customers, the delivery sequence between the delivery customers is modeled through a graph database, the node paths associated with the delivery customers are identified and located, and a delivery sequence network is generated;

[0039] Based on the meal delivery sequence network, the meal delivery sequence chain is refined and processed using static code analysis tools and manager log data to obtain the specific source and version information of the meal delivery sequence, and the information is organized into a meal delivery sequence chain record;

[0040] Classifying the meal delivery sequence chain records, distinguishing between direct meal delivery sequence chains and indirect meal delivery sequence chains, assigning corresponding weight values ​​to each type of meal delivery sequence chain, and generating a preliminary weighted meal delivery sequence chain;

[0041] Based on the comprehensive navigation route data, an expert device or a machine learning model is introduced to review and adjust the preliminary weighted meal delivery sequence chain, optimize the weight value, obtain the optimized weighted meal delivery sequence chain, and summarize and organize all the optimized meal delivery sequence chains to form a weighted meal delivery sequence chain list.

[0042] According to a second aspect of the present invention, the present invention claims protection for a logistics navigation device for a catering business, comprising:

[0043] A parsing module, configured to parse the identification information of the delivery customer submitted by the logistics personnel and extract attribute characteristics of the delivery customer from the identification information;

[0044] a positioning module, configured to locate, based on the attribute characteristics of the delivery customers, the node paths associated with the delivery customers, to generate a delivery sequence chain in the node path, assign weight values ​​to the delivery sequence chains, and generate a weighted delivery sequence chain list, wherein the delivery sequence chain includes a direct delivery sequence chain and an indirect delivery sequence chain;

[0045] a calculation module, configured to calculate the navigation route duration of the delivery sequence chains in the weighted delivery sequence chain list using a Bayesian network model in combination with a predefined navigation route indicator system, and identify potential route accident points using an anomaly detection algorithm to generate a navigation route report including the navigation route duration results and the route accident points;

[0046] A drawing module is used to draw a comprehensive navigation route map based on the navigation route report using a multidimensional visualization algorithm to obtain a navigation route view, and to integrate a genetic algorithm and a simulated annealing algorithm to explore adjustment suggestions for high navigation route areas of the navigation route view to generate a navigation route mitigation strategy; wherein the comprehensive navigation route map is grouped and displayed using a hierarchical clustering algorithm to display components with similar navigation route patterns, the navigation route patterns are obtained by similarity judgment of the delivery sequence chain, the route accident points and the navigation route time, and the navigation route mitigation strategy is used to reduce the navigation route time of the delivery customer.

[0047] The present application relates to the field of data analysis technology, and in particular to a logistics navigation method and device for catering business, which parses identification information of delivery customers submitted by logistics personnel and extracts attribute characteristics of the delivery customers from the identification information; locates node paths associated with the delivery customers based on the attribute characteristics of the delivery customers to generate a delivery sequence chain in the node path, assigns weight values ​​to the delivery sequence chains, and generates a weighted delivery sequence chain list; calculates the navigation route time of the delivery sequence chains in the weighted delivery sequence chain list, and uses an anomaly detection algorithm to identify potential route accident points, and generates a navigation route report including navigation route time results and route accident points; uses a multidimensional visualization algorithm to draw a comprehensive navigation route map, obtains a navigation route view, and generates a navigation route mitigation strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A workflow diagram of a logistics navigation method for catering business claimed in an embodiment of the present application;

[0049] Figure 2 This is a structural module diagram of a logistics navigation for catering business, which is requested to be protected by an embodiment of the present application. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0051] The terms "first", "second" and "third" in this application are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, a feature defined as "first", "second" and "third" may explicitly or implicitly include at least one of such features. In the description of this application, "multiple" means at least two, for example, two, three, etc., unless otherwise clearly and specifically defined. All directional indications in the embodiments of this application (such as up, down, left, right, front, back...) are only used to explain the relative positional relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but may optionally also include steps or units that are not listed, or may optionally also include other steps or units inherent to these processes, methods, products or equipment.

[0052] References to "embodiments" herein mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0053] Figure 1 The present application provides a flow chart of a logistics navigation method for catering business, such as Figure 1 As shown, the method includes:

[0054] 101. Parse the identification information of the delivery customer submitted by the logistics personnel, and extract attribute characteristics of the delivery customer from the identification information;

[0055] In this solution, the device first receives the delivery customer identification information submitted by logistics personnel and reads and processes it using a parsing tool (such as a JSON or XML parser). Next, it uses natural language processing (NLP) techniques and pattern matching algorithms to extract attribute features related to safety and delivery order. These features serve as the basis for subsequent navigation route evaluation and delivery sequence analysis.

[0056] 102. Locate node paths associated with the delivery customers based on their attribute characteristics to generate delivery sequence chains in the node paths, assign weight values ​​to the delivery sequence chains, and generate a weighted delivery sequence chain list, wherein the delivery sequence chains include direct delivery sequence chains and indirect delivery sequence chains.

[0057] The node path refers to the delivery sequence link between delivery customers, including direct dependencies (such as A depends on B) and indirect dependencies (such as A depends on B, and B depends on C).

[0058] A delivery sequence chain is a specific representation of these paths, with weights reflecting the importance of each delivery sequence or the degree of navigation route. By constructing a list of weighted delivery sequence chains, we can more accurately assess the potential impact of each delivery sequence on the overall device.

[0059] Based on the extracted attribute features, the device constructs a graph structure to represent the delivery order between customers. For each pair of components with a delivery order, the weight between them is calculated, taking into account factors such as the depth and frequency of dependencies, as well as historical vulnerability records. Ultimately, a weighted delivery order chain list containing all delivery orders and their weights is generated, providing detailed dependency information for subsequent navigation route evaluation.

[0060] 103. Using a Bayesian network model in combination with a predefined navigation route indicator system, calculate the navigation route time of the delivery sequence chain in the weighted delivery sequence chain list, and use an anomaly detection algorithm to identify potential route accident points, so as to generate a navigation route report including the navigation route time result and the route accident points;

[0061] The Bayesian network model is a probabilistic graphical model that can capture the conditional delivery order between variables and is suitable for complex navigation route evaluation scenarios.

[0062] The navigation route indicator system includes multiple dimensions, such as the number of vulnerabilities, update frequency, community activity, etc., which are used to quantify the navigation route level of each delivery sequence chain.

[0063] Anomaly detection algorithms identify data points that significantly deviate from normal behavior patterns, thereby identifying potential route incidents. Navigation route reports summarize route duration and incident details, helping decision-makers fully understand and address navigation routes.

[0064] Using a Bayesian network model and a predefined navigation route metric system, the device calculates the navigation route duration for each delivery chain in a weighted delivery chain list. It then applies an anomaly detection algorithm (such as an isolation forest or autoencoder) to identify delivery chains with anomalous behavior and identify route failure points. Finally, it generates a detailed navigation route report, including route duration results and a file of failure points, for decision makers to reference.

[0065] 104. Based on the navigation route report, a comprehensive navigation route map is drawn using a multi-dimensional visualization algorithm to obtain a navigation route view, and a genetic algorithm and a simulated annealing algorithm are integrated to explore adjustment suggestions for high navigation route areas of the navigation route view to generate a navigation route mitigation strategy.

[0066] A multi-dimensional visualization algorithm transforms complex delivery sequences and navigation time into an intuitive graphical interface, helping logistics personnel quickly understand navigation route distribution. A comprehensive navigation route map uses a hierarchical clustering algorithm to group components with similar navigation route patterns, facilitating the identification of high-risk navigation route areas. Genetic algorithms and simulated annealing algorithms are used to optimize adjustment recommendations, explore effective ways to reduce navigation route time, and generate specific navigation route mitigation strategies.

[0067] Based on the data from the navigation route reports, the device uses a multidimensional visualization algorithm to create an interactive navigation route map, visually displaying the navigation route time and accident points of the delivery sequence chain. A hierarchical clustering algorithm then classifies the navigation route patterns and displays them in groups. Next, combining a genetic algorithm and simulated annealing algorithms, the device explores optimal adjustments to high-risk navigation route areas and proposes specific mitigation measures, such as replacing dependent libraries or strengthening security audits. Ultimately, a detailed navigation route mitigation strategy is generated to guide decision makers in taking action.

[0068] Through the implementation of steps 101 to 104, the method realizes the intelligent analysis of the entire process from parsing the identification information of the delivery customer to generating a detailed navigation route mitigation strategy. First, by accurately parsing the identification information and extracting attribute features, the accuracy of the data for subsequent analysis is ensured. Secondly, a weighted delivery sequence chain list is constructed to comprehensively capture the direct and indirect delivery sequences, providing a detailed basis for navigation route evaluation. Thirdly, by utilizing the Bayesian network model and anomaly detection algorithm, the device can accurately identify potential route accident points and generate reliable navigation route reports. Finally, through multi-dimensional visualization and optimization algorithms, the device not only intuitively displays the distribution of navigation routes, but also proposes effective mitigation strategies, significantly improving the safety of software routes and the navigation route management capabilities. This series of steps complement each other to form a complete navigation route management closed loop, providing strong support for software development, distribution and deployment.

[0069] In order to improve the accuracy and reliability of the software route navigation route evaluation and further optimize the navigation route prediction of the delivery sequence chain, in some embodiments, the Bayesian network model is used in combination with a predefined navigation route indicator system in step 103 to calculate the navigation route time of the delivery sequence chain in the weighted delivery sequence chain list, including:

[0070] According to the Bayesian network model, the weighted delivery order chain list is represented by a low-dimensional vector using a graph embedding algorithm to obtain a dependency vector; based on the dependency vector, an initial navigation route score is calculated using a pre-defined navigation route indicator system in combination with a multi-dimensional weight adjustment mechanism, and the initial navigation route score is assigned to the delivery order chain, wherein the navigation route indicator system includes the number of vulnerabilities, update frequency, and community activity; using the delivery order chain assigned the initial navigation route score, the Bayesian network model is combined with a gradient boosting decision tree algorithm to optimize the navigation route probability distribution of the child nodes that the delivery customer directly or indirectly depends on under the condition of the parent node of the given delivery customer to obtain an optimized conditional probability; based on the optimized conditional probability, combined with an ensemble learning method, the overall navigation route score of the delivery order chain is evaluated to obtain a navigation route time result;

[0071] In this embodiment, a graph embedding algorithm is used to convert the weighted delivery order chain list into a low-dimensional vector representation, namely a dependency vector. The dependency vector captures the structural characteristics and semantic information of the delivery order chain, facilitating subsequent navigation route evaluation calculations.

[0072] The multi-dimensional weight adjustment mechanism dynamically adjusts the importance of each delivery order based on different navigation route indicators (such as the number of vulnerabilities, update frequency, and community activity), ensuring the flexibility and adaptability of the evaluation results;

[0073] Ensemble learning methods further enhance the robustness and reliability of the overall navigation route evaluation by combining the results of multiple models.

[0074] In the embodiment of the present application, first, a graph embedding algorithm is used to convert the weighted delivery order chain list into a low-dimensional vector representation to obtain a dependency vector;

[0075] Then, based on these dependency vectors and a pre-defined navigation route indicator system (including the number of vulnerabilities, update frequency, and community activity), a multi-dimensional weight adjustment mechanism is used to calculate the initial navigation route score and assign this score to each delivery sequence chain;

[0076] Next, we use the Bayesian network model combined with the gradient boosting decision tree algorithm to optimize the navigation route probability distribution of the child nodes that the customer directly or indirectly depends on, given the parent node of the delivery customer, and obtain the optimized conditional probability.

[0077] Finally, based on the optimized conditional probability and combined with the ensemble learning method, the overall navigation route score of the delivery sequence chain is comprehensively evaluated to obtain the final navigation route time result.

[0078] In order to solve the problem of difficulty in identifying and evaluating potential accident points in a route and further improve the timeliness and accuracy of navigation route warnings, in some embodiments, the use of an anomaly detection algorithm in step 103 to identify potential accident points on the route to generate a navigation route report including navigation route time results and the accident points on the route may include:

[0079] Based on the navigation route time-consuming results, an anomaly detection algorithm is used in combination with an adaptive threshold setting strategy to analyze the navigation route time-consuming of the food delivery sequence chain, identify data points that deviate from the normal range, and obtain route accident points; natural language processing technology is used to parse the safety announcements and technical documents of the route accident points, and information directly related to the route accident points in the safety announcements and technical documents is automatically extracted to generate accident point details; all the navigation route time-consuming results and the accident point details are summarized and organized, and time series analysis is applied to predict future navigation route trends to generate comprehensive navigation route data; visualization tools are used to create an interactive interface to organize the comprehensive navigation route data, and a navigation route report containing the navigation route time-consuming results and the accident point archive is generated.

[0080] In this embodiment, an anomaly detection algorithm is a technique used to identify unusual patterns in a data set, discovering data points that deviate from normal behavior. In this solution, an adaptive threshold setting strategy is combined to dynamically adjust the threshold based on the characteristics of different delivery sequence chains, ensuring flexibility and accuracy in anomaly detection.

[0081] Natural language processing (NLP) technology is used to parse safety announcements and technical documents related to route incidents, automatically extracting key information such as vulnerability descriptions and remediation suggestions, and generating detailed incident profiles.

[0082] Time series analysis predicts future trends by analyzing historical data, helping decision makers identify potential navigation routes in advance;

[0083] The interactive visualization tool converts complex integrated navigation route data into an intuitive graphical interface, making it easier for logistics personnel to understand and operate.

[0084] In an embodiment of the present application, first, based on the calculated navigation route time results, the device uses an anomaly detection algorithm combined with an adaptive threshold setting strategy to analyze the navigation route time of the food delivery sequence chain, identify data points that significantly deviate from the normal range, and determine them as route accident points.

[0085] Next, natural language processing technology is used to parse the safety announcements and technical documents related to these accident points, automatically extract directly relevant information, and generate detailed accident point details.

[0086] Then, the device summarizes and organizes the time-consuming results of all navigation routes and the details of the accident points, and applies time series analysis to predict future navigation route trends to form comprehensive navigation route data.

[0087] Finally, an easy-to-understand interface is created using interactive visualization tools to display comprehensive navigation route data and generate a navigation route report containing navigation route time results and accident point archives for decision makers' reference.

[0088] In order to solve the problem of inaccurate probability distribution of navigation routes of child nodes in the delivery sequence chain and further improve the accuracy and reliability of navigation route evaluation, in some embodiments, the delivery sequence chain using the assigned initial navigation route score is used to optimize the navigation route probability distribution of child nodes that the delivery customer directly or indirectly depends on under the condition of the parent node of the delivery customer through the Bayesian network model combined with the gradient boosting decision tree algorithm to obtain the optimized conditional probability, including:

[0089] The low-dimensional vector representation of the food delivery sequence chain and the initial navigation route score are collected and organized to construct a training data set; based on the Bayesian network model, the training data set is initialized and processed to obtain a gradient boosting decision tree model, wherein the gradient boosting decision tree model is used to assist in estimating the conditional probability in the optimized Bayesian network model; using the training data set, the Bayesian network model and the gradient boosting decision tree model are jointly trained, and the estimation of the conditional probability is optimized by iteratively calculating the loss function and adjusting the parameters by backpropagation to obtain a joint training result; based on the joint training result, the conditional probability distribution of each node in the Bayesian network model is refined, and the conditional probability values ​​of the child nodes that the delivery customer directly or indirectly depends on are updated under the condition of the parent node of the given delivery customer to obtain the optimized conditional probability.

[0090] In this embodiment, the low-dimensional vector representation is achieved by converting the food delivery sequence chain into vectors in a low-dimensional space through a graph embedding algorithm. These vectors not only capture the structural characteristics of the food delivery sequence, but also contain semantic information.

[0091] The training dataset consists of these low-dimensional vectors and their corresponding initial navigation route scores, which are used to train the machine learning model;

[0092] The gradient boosted decision tree (GBDT) model is an integrated learning method that can gradually optimize the prediction results and is particularly suitable for dealing with complex nonlinear relationships;

[0093] The loss function is used to measure the gap between the model prediction and the actual value, and the model parameters are adjusted through backpropagation to minimize the gap;

[0094] Conditional probability distribution refers to the probability distribution of the navigation route of the child node under the given parent node condition, reflecting the mutual influence between the nodes in the delivery sequence chain.

[0095] In the embodiment of the present application, first, the device collects and organizes the low-dimensional vector representation of the food delivery sequence chain and its initial navigation route score to construct a training data set.

[0096] Next, the training data set is initialized and processed based on the Bayesian network model to obtain a gradient boosting decision tree model, which is used to assist in optimizing the estimation of conditional probabilities in the Bayesian network model.

[0097] Then, the training dataset is used to jointly train the Bayesian network model and the gradient boosting decision tree model. By iteratively calculating the loss function and adjusting the parameters through backpropagation, the estimation of the conditional probability is optimized, and finally the joint training result is obtained.

[0098] Finally, based on the joint training results, the conditional probability distribution of each node in the Bayesian network model is refined, and the conditional probability values ​​of the child nodes that it directly or indirectly depends on are updated under the condition of the parent node of the given delivery customer to obtain the optimized conditional probability distribution.

[0099] In order to solve the problem of difficulty in accurately identifying abnormal navigation route points in the food delivery sequence chain and further improve the accuracy and timeliness of route navigation route management, in some embodiments, based on the navigation route time results, an anomaly detection algorithm is used in combination with an adaptive threshold setting strategy to analyze the navigation route time of the food delivery sequence chain, identify data points that deviate from the normal range, and obtain route accident points, including:

[0100] According to the navigation route time consumption result, a suitable anomaly detection algorithm is selected, wherein the anomaly detection algorithm includes an isolation forest and a local outlier factor, and the anomaly detection algorithm can effectively identify abnormal patterns in the navigation route time consumption data to generate an anomaly detection algorithm; based on the historical data and domain knowledge of the food delivery sequence chain, an initial threshold is set for the anomaly detection algorithm for preliminary screening of abnormal points to obtain a preliminarily set threshold; an adaptive threshold setting strategy is applied, and the preliminarily set threshold is automatically updated in combination with the performance indicators of the monitoring device to improve the accuracy and generate an optimized threshold, wherein the performance indicators of the monitoring device include accuracy and recall rate; the anomaly detection algorithm is used in combination with the optimized threshold to analyze the navigation route time consumption result, identify data points that deviate from the normal range, and obtain route accident points.

[0101] The initial threshold is a preliminary screening criterion set based on historical data and domain knowledge of the delivery sequence chain to distinguish normal and abnormal data points;

[0102] The adaptive threshold setting strategy automatically adjusts the threshold according to the performance indicators of the monitoring device (such as accuracy and recall rate) to improve the accuracy of anomaly detection.

[0103] In an embodiment of the present application, first, the device selects a suitable anomaly detection algorithm (such as an isolation forest or a local outlier factor) based on the navigation route time consumption result. These algorithms can effectively identify abnormal patterns in the navigation route time consumption data.

[0104] Next, based on the historical data and domain knowledge of the food delivery sequence chain, an initial threshold is set for the anomaly detection algorithm to preliminarily screen outliers.

[0105] Then, an adaptive threshold setting strategy is applied, and the initially set threshold is automatically updated in combination with the performance indicators of the monitoring device (such as precision and recall) to optimize the detection effect.

[0106] Finally, the optimized threshold is used in combination with the selected anomaly detection algorithm to analyze the navigation route time results, identify data points that deviate from the normal range, and ultimately determine the route accident points.

[0107] In order to solve the problem that the delivery sequence chain and navigation route information are difficult to intuitively display and understand, and to further improve the visualization effect and decision-making efficiency of navigation route management, in some embodiments, step 104 uses a multi-dimensional visualization algorithm to draw a comprehensive navigation route map based on the navigation route report to obtain a navigation route view, including:

[0108] According to the navigation route report, data of the meal delivery sequence chain and navigation route information are collected and organized to make the data format consistent and ready for visualization processing, thereby obtaining a navigation route data set; using a multidimensional visualization algorithm, the navigation route data set is mapped, and the navigation route time and accident point location of the meal delivery sequence chain are mapped into a multidimensional space to generate a visualization representation; based on the visualization representation, a hierarchical clustering algorithm is applied to group and display components with similar navigation route patterns, and similarity is judged and classified into different navigation route pattern groups to obtain grouped navigation route patterns; in combination with geographic information or network topology, the visualization layout of the grouped navigation route patterns is adjusted to optimize the visual effect and generate a comprehensive navigation route map; high navigation route areas of the comprehensive navigation route map are highlighted through color coding and icon styles to help logistics personnel quickly identify key navigation route points, and interactive tools are used to enhance the logistics personnel experience to obtain a navigation route view.

[0109] In this embodiment, the navigation route data set includes data of the meal delivery sequence chain and navigation route information collected and organized from the navigation route report, and these data are formatted to ensure consistency and visualization preparation;

[0110] Multidimensional visualization algorithms are used to map complex navigation route times and accident point locations into multidimensional space to generate intuitive visualization representations.

[0111] The hierarchical clustering algorithm groups and displays components with similar navigation route patterns based on similarity judgment, and classifies them into different navigation route pattern groups;

[0112] Geographic information or network topology structure is combined with this information to adjust the visual layout, optimize the visual effect, and make the navigation route map more intuitive;

[0113] Color coding and icon styles are used to highlight high-traffic route areas, helping logisticians quickly identify key route points. Interactive tools enhance the logistician experience, allowing logisticians to dynamically explore and analyze the route view.

[0114] In the embodiment of the present application, first, the device collects and organizes data on the meal delivery sequence chain and navigation route information based on the navigation route report, so that the format is consistent and ready for visualization processing, thereby forming a navigation route data set.

[0115] Then, a multidimensional visualization algorithm is used to map this data set, mapping information such as the navigation route time and accident point location of the food delivery sequence chain into a multidimensional space to generate a visual representation.

[0116] Next, based on the visual representation, a hierarchical clustering algorithm is applied to group and display components with similar navigation route patterns, and similarity is judged to classify them into different navigation route pattern groups.

[0117] Subsequently, the visual layout of the navigation route patterns displayed in groups is adjusted in combination with the geographic information or network topology, the visual effect is optimized, and a comprehensive navigation route map is generated.

[0118] Finally, a detailed navigation route view is obtained by highlighting high navigation route areas of the comprehensive navigation route map through color coding and icon styles, and utilizing interactive tools to enhance the logistics personnel experience.

[0119] In order to solve the problem of difficulty in accurately identifying and quantifying the delivery order between delivery customers and further improve the accuracy and reliability of delivery order chain management, in some embodiments, the node paths associated with the delivery customers are located according to the attribute characteristics of the delivery customers in step 102 to generate a delivery order chain in the node path, and weight values ​​are assigned to the delivery order chains to generate a weighted delivery order chain list, including:

[0120] Utilize the parsed delivery customer identification information submitted by the logistics personnel to extract the attribute characteristics of the delivery customer; based on the attribute characteristics of the delivery customer, model the delivery sequence between the delivery customers through a graph database, identify and locate the node paths associated with the delivery customers, and generate a delivery sequence network; based on the delivery sequence network, utilize static code analysis tools and manager log data to refine the delivery sequence chain, obtain the specific source and version information of the delivery sequence, and organize it into delivery sequence chain records; classify the delivery sequence chain records, distinguish between direct delivery sequence chains and indirect delivery sequence chains, assign corresponding weight values ​​to each type of delivery sequence chain, and generate a preliminary weighted delivery sequence chain; introduce an expert device or a machine learning model based on the comprehensive navigation route data, review and adjust the preliminary weighted delivery sequence chain, optimize the weight value, and obtain an optimized weighted delivery sequence chain; summarize and organize all the optimized delivery sequence chains to form a weighted delivery sequence chain list.

[0121] In this embodiment, attribute features refer to specific characteristics extracted from the delivery customer identification information submitted by logistics personnel, such as function description, dependency library list, and open source license type. These features are used to model the delivery sequence between delivery customers.

[0122] A graph database is a data management device specifically designed to store and query complex network structures. In this solution, it is used to model the delivery sequence network between delivery customers, enabling efficient identification and location of node paths.

[0123] Static code analysis tools are used to automatically analyze source code, detect potential problems, and provide detailed delivery sequence information. The manager's log data records the actual behavior of delivery customers in operation, helping to refine the delivery sequence chain.

[0124] The delivery order chain record includes the specific source and version information of each delivery order, ensuring the transparency and traceability of the delivery order;

[0125] Expert devices or machine learning models are used to review and adjust the preliminary weighted delivery sequence chain, optimizing the weight values ​​by introducing domain knowledge or automated learning mechanisms.

[0126] In an embodiment of the present application, first, the device uses the parsed delivery customer identification information submitted by the logistics personnel to extract its attribute characteristics.

[0127] Then, based on these attribute features, the delivery sequence between delivery customers is modeled through a graph database, the associated node paths between them are identified and located, and a delivery sequence network is generated.

[0128] Next, based on the meal delivery sequence network, the device uses static code analysis tools and manager log data to refine the meal delivery sequence chain, obtain the specific source and version information of the meal delivery sequence, and organize it into a meal delivery sequence chain record.

[0129] Subsequently, these records are classified to distinguish between direct meal delivery sequence chains and indirect meal delivery sequence chains, and a corresponding weight value is assigned to each type of meal delivery sequence chain to generate a preliminary weighted meal delivery sequence chain.

[0130] Finally, based on the comprehensive navigation route data, an expert device or machine learning model is introduced to review and adjust the preliminary weighted meal delivery sequence chain, optimize the weight value, and obtain the optimized weighted meal delivery sequence chain. All optimized meal delivery sequence chains are summarized and organized to form the final weighted meal delivery sequence chain list.

[0131] This application considers that in order to accurately evaluate the level of navigation routes in the delivery sequence chain, a comprehensive navigation route evaluation formula system has been developed; this system is based on the low-dimensional vector representation of the delivery sequence chain (standard dependency vector v), combined with multi-dimensional navigation route indicators, and by introducing time decay factors and nonlinear transformations, it quantifies the attributes of each navigation route and generates a quantitative navigation route characteristic r quant,i To further improve the accuracy, the weights are dynamically adjusted according to the importance of different navigation route indicators, and adaptive learning rate and Bayesian optimization method are introduced to generate navigation route weights w that reflect the influence of different navigation route factors. adj Finally, the adjusted weights are used to perform weighted calculation on the quantitative navigation route characteristics, and exponential decay and nonlinear terms are introduced to comprehensively evaluate the overall navigation route level of the delivery sequence chain to obtain the initial navigation route score S initial , so a new alternative is proposed, which includes:

[0132] The initial navigation route score is calculated based on the dependency vector using a predefined navigation route indicator system in combination with a multi-dimensional weight adjustment mechanism, including:

[0133] The data in the delivery sequence chain is converted into a low-dimensional vector representation using a graph embedding algorithm, and a regularization term is introduced to prevent overfitting to generate a dependency vector v. The dependency vector v is obtained by calculating the following formula:

[0134]

[0135] Among them, v is the dependency vector of the delivery sequence chain, G is the graph data corresponding to the delivery sequence chain, t j is the time since the last update, is the mean of the time interval, σ t is the standard deviation of the time interval, W (1),b (1) are the weight matrix and bias vector of the first layer, σ is the activation function, l is the number of network layers, Θ represents the set of all parameters, θ i is the i-th parameter, W (l) is the weight matrix of the lth layer in the neural network, b (l) is the bias vector of the lth layer in the neural network, σ t is the standard deviation of the time interval;

[0136] The following is a detailed explanation of each parameter:

[0137] V: Low-dimensional vector representation of the delivery sequence chain. A graph embedding algorithm is used to convert the complex graph data in the delivery sequence chain into a low-dimensional vector, capturing the key features of the delivery sequence chain.

[0138] G: Graph data (adjacency matrix or edge list) corresponding to the delivery sequence chain; extracted from the actual delivery sequence network, describing the connection relationship between nodes;

[0139] W (i) ,b (i) : The weight matrix and bias vector of the i-th layer of the neural network; optimized through a training process (such as gradient descent) to capture nonlinear patterns in graph data;

[0140] σ: Activation function (such as ReLU, Sigmoid, etc.), used to introduce nonlinear characteristics; nonlinear activation functions can enable the model to better capture complex relationships in the data;

[0141] l: The number of layers in the neural network; multi-layer structures can more deeply learn and abstract features in graph data;

[0142] t j : Time since the last update; records the most recent update time of each dependency;

[0143] The mean of the time intervals is calculated by counting the update time intervals of all dependencies and calculating their average value.

[0144] σ t : Standard deviation of time intervals. Count the update time intervals of all dependencies and calculate their standard deviation.

[0145] Θ: The set of all parameters, including weight matrices and bias vectors, contains all parameters optimized during model training.

[0146] θ i : The i-th parameter, obtained through the model training process, specifically refers to the elements in each weight matrix and bias vector.

[0147] The following is an introduction to the design reasons of each sub-item:

[0148] Deep neural network σ(W (l) ·σ(…W (1) G+b (1) )…+b (l) ): Use a multi-layer neural network to encode the graph data of the delivery sequence chain, and capture the complex delivery sequence and features through layer-by-layer transformation and nonlinear activation function. The weight matrix W of each layer (i) and the bias vector b (i) Through training optimization, we ensure that the model can effectively learn and express key information in graph data.

[0149] Time decay factor: A time decay factor in the form of Gaussian distribution is introduced so that the newer delivery order has a greater impact on the final result, while the influence of the older delivery order gradually weakens, which reflects the time sensitivity and ensures that the model can dynamically adapt to changes in the delivery order.

[0150] Regularization term By squaring and summing all parameters, an L2 regularization term is introduced to prevent the model from overfitting. The regularization term helps maintain the generalization ability of the model and ensures that it performs well on unseen data.

[0151] Adding a time decay factor and regularization term to the neural network output comprehensively considers both the static characteristics of the delivery sequence (encoded by the neural network) and temporal variations (via the time decay factor), while also incorporating regularization to prevent overfitting. This combination enables the model to fully capture the spatiotemporal characteristics of the delivery sequence while maintaining good generalization performance.

[0152] Here's a specific example:

[0153] Assume there is a simple food delivery sequence graph data G, which contains 5 nodes and 7 edges. Use a two-layer neural network (l = 2), the activation function is ReLU (σ(x) = max(0, x)), and introduce a regularization term to prevent overfitting.

[0154] The specific parameters are as follows:

[0155] W (1) and b (1) : Layer 1 weight matrix and bias vector;

[0156] W (2) and b (2) : Layer 2 weight matrix and bias vector;

[0157] t j =30 days, Day, σt =5 days;

[0158] Θ contains all parameters, assuming |Θ| = 10, each parameter θ i =0.1;

[0159] Substituting into the formula we get:

[0160]

[0161] Assume W (1) ,b (1) ,W (2) ,b (2) Known and trained, the specific value of v can be calculated:

[0162] v=[0.8,0.6,0.4,0.2,0.1]

[0163] Conclusion Explanation:

[0164] The dependency vector v = [0.8, 0.6, 0.4, 0.2, 0.1] provides a compact representation of the delivery sequence chain, capturing the importance of each node and the order in which they are delivered to each other. Lower values ​​indicate that the node has a relatively small navigation path, while higher values ​​indicate a larger potential navigation path. This result provides a solid foundation for subsequent navigation path evaluation, helping decision makers better understand and address potential navigation paths in the delivery sequence chain, particularly in areas such as route management and network security.

[0165] According to the dependency vector v, combined with the navigation route index system, the navigation route attributes of the delivery sequence chain are quantified, and the time decay factor and nonlinear transformation are introduced to obtain the quantified navigation route characteristics; wherein, the quantified navigation route characteristics r quant,i Obtained by calculating the following formula:

[0166]

[0167] Among them, r quant,i is the quantitative navigation route characteristic, n is the number of navigation route indicators, w j is the weight of the j-th navigation route indicator, φ is a nonlinear transformation function used to increase the expression ability, v·r j represents the vector dot product, r j is the eigenvector or characterizing vector, ||v||2||r j ||2 are v and r respectively j The L2 norm of

[0168] The following is a detailed explanation of each parameter:

[0169] r quant,i: The quantitative navigation route characteristics of the i-th delivery sequence chain. It integrates the influence of multiple navigation route indicators and quantifies the navigation route attributes of the delivery sequence chain by introducing a time decay factor and nonlinear transformation. This value reflects the overall navigation route level of a specific delivery sequence chain in the i-th navigation route dimension and is an important basis for subsequent navigation route evaluation and decision-making.

[0170] n: the number of navigation route indicators; each navigation route indicator R j Each of these metrics measures the navigational path of a specific aspect of the delivery sequence chain, such as the number of vulnerabilities, update frequency, and supplier information gap. By considering multiple navigational path metrics, a more comprehensive assessment of the overall navigational path of the delivery sequence chain can be achieved. The size of n depends on the number of navigational path factors included in the specific application scenario.

[0171] w j :w j is the weight of the jth navigation route metric; these weights reflect the relative importance of different navigation route metrics in the overall navigation route evaluation; by dynamically adjusting the importance of each navigation route metric, the model can better adapt to different application scenarios and ensure the accuracy and reliability of the navigation route evaluation results; the choice of weights can be optimized based on historical data or expert knowledge;

[0172] φ: φ is a nonlinear transformation function used to increase expressiveness and capture complex navigation route patterns. Common choices include the hyperbolic tangent function (tanh) and the sigmoid function. Nonlinear transformations enable the model to better handle complex nonlinear relationships, improving the flexibility and accuracy of navigation route evaluation. By introducing φ, the model can adjust the expression of navigation route characteristics over a wider range, enhancing its adaptability and robustness.

[0173] v·r j : Represents the dependency vector v and the feature vector r of the j-th navigation route indicator j The dot product between them measures the directional similarity of two vectors; a larger value indicates closer the two vectors are. By calculating the dot product, we can evaluate the correlation between the delivery sequence chain and specific navigation route indicators, thus providing a basis for quantifying navigation routes.

[0174] r j : is the j-th navigation route indicator R j The navigation route indicator is converted into a low-dimensional representation, which is convenient for mathematical operations and model processing. Each eigenvector contains multiple dimensions, each representing a feature or attribute, reflecting the specific information of the navigation route indicator. Through the eigenvector, the model can more accurately capture and quantify the impact of the navigation route indicator on the delivery sequence chain.

[0175] ||v||2||r j ||2: dependency vector v and feature vector r j The L2 norm (i.e., Euclidean norm) of the vector is calculated; the L2 norm measures the length or modulus of the vector and is used to standardize the dot product calculation to ensure the effectiveness of cosine similarity; by dividing by the L2 norm, the influence of the vector length can be eliminated, focusing on their directional similarity, thereby more accurately evaluating the correlation between the delivery sequence chain and the navigation route indicators.

[0176] Here's a specific example:

[0177] Assume there are three navigation route indicators R1, R2, and R3, and the corresponding eigenvectors are

[0178] r1=[0.9,0.8,0.7,0.6,0.5],r2=[0.5,0.6,0.7,0.8,0.9],r3=[0.1,0.2,0.3,0.4,0.5]

[0179] The weights are w1=0.4, w2=0.3, w3=0.3;

[0180] Assuming φ(x) = tanh(x), we can calculate:

[0181]

[0182] Assume that the calculation results are:

[0183] r quant,1 =0.35

[0184] r quant,2 =0.28

[0185] r quant,3 =0.17

[0186] Conclusion Explanation:

[0187] The quantitative navigation route characteristics r are obtained quant,i This value reflects the navigation route level of the delivery sequence chain in the i-th navigation route dimension. A lower quantitative navigation route characteristic value indicates that the navigation route of the delivery sequence chain in this navigation route dimension is relatively small and the device is relatively stable; a higher quantitative navigation route characteristic value means that there is a large potential navigation route, which requires further attention and measures to reduce the navigation route.

[0188] According to the importance of different navigation route indicators, the weight of each indicator is dynamically adjusted, and the adaptive learning rate and Bayesian optimization method are introduced to reflect the influence of different navigation route factors on the overall navigation route, and generate the navigation route weight; wherein, the navigation route weight w adj Calculated by the following formula:

[0189]

[0190] Among them, w adj is the navigation route weight, which is used to measure the difference between the predicted value and the actual value. is the mean of the loss function, σ L is the standard deviation of the loss function, W0 is the initial weight vector, W opt is the optimal weight configuration obtained by Bayesian optimization, N is the number of samples, is the predicted value of the i-th sample, y i is the true label or target value of the i-th sample, output by the model, is the square of the prediction error of each sample, and W is the weight matrix;

[0191] The following is a detailed explanation of each parameter:

[0192] w adj : A dynamically adjusted navigation route weight vector; it optimizes the weight configuration of each navigation route indicator based on the historical performance of the loss function and current data, allowing the model to more accurately reflect the impact of different navigation route factors on the overall navigation route. By introducing adaptive learning rate and Bayesian optimization methods, the model can flexibly adjust the weights to improve the accuracy and robustness of navigation route evaluation;

[0193] The mean of the loss function represents the historical average of the difference between the model's predictions and the actual values. By comparing the current loss with the historical mean, the model can assess its performance changes and adjust weights accordingly to optimize future performance.

[0194] σ L : is the standard deviation of the loss function, which measures the magnitude of the change in loss value. A larger standard deviation indicates that the loss fluctuates greatly and the model may be unstable. A smaller standard deviation means that the loss is more consistent and the model performance is more stable. By introducing the standard deviation, the model can respond more sensitively to changes in loss and adjust the weight configuration in a timely manner.

[0195] W0: is the initial weight vector of the model; it is the starting point for weight adjustment, usually based on random initialization or pre-training results; through step-by-step optimization, the model can start from the initial weight vector and find the best weight configuration to minimize the loss function;

[0196] W opt : It is the optimal weight configuration obtained through Bayesian optimization or other optimization methods; it represents the weight combination that minimizes the loss function on a given data set; through continuous iterative optimization, the model can gradually approach this optimal configuration and improve the accuracy of the prediction;

[0197] N: is the total number of samples in the dataset; it is used to calculate the average loss, ensuring that the loss function reflects the performance of the entire dataset rather than the anomalies of individual samples; a larger number of samples can make the loss estimate more stable and reliable;

[0198] is the predicted value of the i-th sample, output by the model; it is the model's prediction result for the input data, which is used to compare with the true label to evaluate the model's prediction performance;

[0199] y i :y i is the true label or target value of the i-th sample; it is the target predicted by the model and is used to measure the difference between the model prediction value and the actual situation; by comparing the predicted value and the true label, the accuracy of the model can be evaluated;

[0200] is the square of the prediction error of the i-th sample; it amplifies the impact of larger errors, making the model pay more attention to samples with larger prediction deviations; by minimizing the sum of squared prediction errors, the model can optimize its own parameters and improve the accuracy of predictions;

[0201] W: is the weight matrix in the model, used to connect neurons in the input layer and hidden layer, or between hidden layers; it determines the learning and expression capabilities of the model; through optimization algorithms such as gradient descent, the model can continuously adjust the weight matrix to minimize the loss function and improve prediction performance.

[0202] Here's a specific example:

[0203] Assuming the loss function has the mean Standard deviation σ L =0.05, initial weight W0=[0.2,0.3,0.5], optimal weight configuration W opt =[0.4,0.3,0.3], sample size N = 100, prediction error sum of squares

[0204] The value of the input is:

[0205]

[0206] Assuming the gradient Calculated:

[0207]

[0208] w adj =[0.2,0.3,0.5]+0.6065·[0.04,0.03,0.03]

[0209] w adj =[0.2,0.3,0.5]+[0.024,0.018,0.018]

[0210] w adj =[0.224,0.318,0.518]

[0211] Conclusion Explanation:

[0212] Get the navigation route weight w adj = [0.224, 0.318, 0.518], these weights reflect the relative importance of different navigation route indicators in the overall navigation route evaluation. A lower navigation route weight indicates that the navigation route indicator has less impact on the overall navigation route and the device is relatively stable; a higher navigation route weight means that there is a greater potential for navigation routes, and further attention and measures need to be taken to reduce the navigation route.

[0213] Using the navigation route weight w adj Quantitative navigation route characteristics r quant,i Perform weighted calculations, introduce exponential decay and nonlinear terms to increase the expressive power of the model, and comprehensively evaluate the navigation route level of the delivery sequence chain to obtain the initial navigation route score S initial ; Among them, the initial navigation route score S initial Calculated by the following formula:

[0214]

[0215] Among them, S initial is the initial navigation route score, n is the number of navigation route indicators, w adj,i is the adjusted weight of the ith navigation route indicator, and σ r are the mean and standard deviation of the quantitative navigation route characteristics, d i is the shortest path distance between nodes in the delivery sequence chain, and σ d are the mean and standard deviation of the shortest path distance, respectively.

[0216] The following is a detailed explanation of each parameter:

[0217] S initial: The initial navigation route score of the delivery sequence chain integrates the influence of multiple navigation route indicators and their weights; it increases the expressive power of the model by introducing exponential reduction and nonlinear terms, and reflects the overall navigation route level of the delivery sequence chain; a lower initial navigation route score indicates that the navigation route of the delivery sequence chain is relatively small and the device is relatively stable; a higher score means that there is a large potential navigation route, which requires further attention and measures to reduce the navigation route;

[0218] n: represents the number of navigation route indicators; each navigation route indicator R i Each of these metrics measures the navigational path of a specific aspect of the delivery sequence chain, such as the number of vulnerabilities, update frequency, and supplier information gap. By considering multiple navigational path metrics, a more comprehensive assessment of the overall navigational path of the delivery sequence chain can be achieved. The size of n depends on the number of navigational path factors included in the specific application scenario.

[0219] w adj,i : is the adjusted weight of the i-th navigation route metric; these weights reflect the relative importance of different navigation route metrics in the overall navigation route evaluation; by dynamically adjusting the importance of each navigation route metric, the model can better adapt to different application scenarios and ensure the accuracy and reliability of the navigation route evaluation results; the choice of weights can be optimized based on historical data or expert knowledge;

[0220] σ r : is all the quantitative navigation route characteristics r quant,i The mean of σ represents the average level of navigation route characteristics; r It is the standard deviation of the quantified navigation route characteristics, which measures the variation of the navigation route characteristic values. By introducing the mean and standard deviation, the model can standardize the quantified navigation route characteristics, eliminate the influence of outliers, and enhance the sensitivity to navigation route changes. This helps to more accurately capture the distribution of navigation route characteristics, thereby improving the accuracy of navigation route evaluation.

[0221] d i : is the shortest path distance between nodes in the delivery sequence chain; it measures the closeness of the connection between nodes in the delivery sequence chain; shorter distances generally mean stronger delivery sequences and higher likelihood of navigation route propagation, while longer distances may indicate weaker delivery sequences and lower likelihood of navigation route propagation; by considering the shortest path distance, the model can more accurately assess the structural complexity of the delivery sequence chain and potential navigation routes;

[0222] and σ d : is the distance d of all shortest paths iThe mean of , which represents the average connection density between nodes in the delivery sequence chain; σ d is the standard deviation of the shortest path distance, which measures the variation of the path distance. By introducing the mean and standard deviation, the model can standardize the shortest path distance, eliminate the influence of outliers, and enhance the sensitivity to path changes. This helps to more accurately capture the structural characteristics of the food delivery sequence chain, thereby improving the accuracy of navigation route evaluation.

[0223] Here's a specific example:

[0224] Assume that the mean of the quantified navigation route characteristics Standard deviation σ r =0.07, the mean of the shortest path distance Standard deviation σ d =1, the specific shortest path distances are d1=2, d2=3, d3=4;

[0225] Substituting specific values ​​into the above formula, we can get:

[0226]

[0227] The calculation results are:

[0228] S initial =0.224·0.35 0.6065 0.6065+0.318 0.28 0.985 1+0.518 0.17 0.6065 .0.6065

[0229] S initial ≈0.224·0.35 0.6065 0.6065+0.318 0.28 0.985 +0.518 0.17 0.6065 0.6065

[0230] S initial ≈0.049+0.092+0.054

[0231] S initial ≈0.195

[0232] Through the above calculation, we get the initial navigation route score S initial≈0.195, reflecting the overall navigation routing performance of the delivery chain. A low initial navigation routing score indicates a relatively small navigation route within the delivery chain and a relatively stable setup. A higher score indicates a significant potential navigation route, requiring further attention and action to mitigate it. This assessment can provide valuable support for areas such as route management and cybersecurity, helping decision-makers better understand and address potential navigation routes.

[0233] Figure 2 The present application provides a structural diagram of a logistics navigation device (or device) for catering business, such as Figure 2 As shown, the device includes:

[0234] The parsing module 21 is used to parse the identification information of the delivery customer submitted by the logistics personnel and extract the attribute characteristics of the delivery customer from the identification information;

[0235] a positioning module 22 for locating, based on the attribute characteristics of the delivery customers, the node paths associated with the delivery customers to generate a delivery sequence chain in the node path, assigning weight values ​​to the delivery sequence chains, and generating a weighted delivery sequence chain list, wherein the delivery sequence chain includes a direct delivery sequence chain and an indirect delivery sequence chain;

[0236] A calculation module 23 is configured to calculate the navigation route duration of the delivery sequence chains in the weighted delivery sequence chain list using a Bayesian network model combined with a predefined navigation route indicator system, and to identify potential route accident points using an anomaly detection algorithm to generate a navigation route report including the navigation route duration results and the route accident points;

[0237] The drawing module 24 is used to draw a comprehensive navigation route map based on the navigation route report using a multidimensional visualization algorithm to obtain a navigation route view, and integrate a genetic algorithm and a simulated annealing algorithm to explore adjustment suggestions for high navigation route areas of the navigation route view to generate a navigation route mitigation strategy; wherein, the comprehensive navigation route map is grouped and displayed using a hierarchical clustering algorithm to display components with similar navigation route patterns, and the navigation route pattern is obtained by similarity judgment of the delivery sequence chain, the route accident points and the navigation route time. The navigation route mitigation strategy is used to reduce the navigation route time of the delivery customer.

[0238] Figure 2 The logistics navigation device for catering business can perform Figure 1The implementation principle and technical effects of the logistics navigation method for catering business described in the illustrated embodiment will not be repeated here. The specific manner in which each module and unit performs operations in the logistics navigation device for catering business in the above embodiment has been described in detail in the embodiment of the method and will not be elaborated here.

[0239] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0240] In addition, the functional units in the various embodiments of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units. The above is only an implementation method of the present application and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the description and drawings of this application, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

[0241] The above detailed description of the specific embodiments of the invention is intended only as an example, and the present application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions of the invention are also within the scope of the present application. Therefore, equivalent changes, modifications, and improvements made without departing from the spirit and scope of the present application should be included within the scope of the present application.

Claims

1. A logistics navigation method for catering business, characterized in that: include: Parsing the identification information of the delivery customer submitted by the logistics personnel, and extracting the attribute characteristics of the delivery customer from the identification information; According to the attribute characteristics of the delivery customers, the node paths associated with the delivery customers are located to generate a delivery sequence chain in the node path, and weight values ​​are assigned to the delivery sequence chains to generate a weighted delivery sequence chain list, wherein the delivery sequence chain includes a direct delivery sequence chain and an indirect delivery sequence chain; A Bayesian network model is used in combination with a predefined navigation route indicator system to calculate the navigation route time of the delivery sequence chain in the weighted delivery sequence chain list, and an anomaly detection algorithm is used to identify potential route accident points to generate a navigation route report including the navigation route time result and the route accident points; Based on the navigation route report, a comprehensive navigation route map is drawn using a multidimensional visualization algorithm to obtain a navigation route view, and a genetic algorithm and a simulated annealing algorithm are integrated to explore adjustment suggestions for high navigation route areas of the navigation route view to generate a navigation route mitigation strategy; wherein, the comprehensive navigation route map is grouped and displayed using a hierarchical clustering algorithm to display components with similar navigation route patterns, and the navigation route pattern is obtained by performing a similarity judgment on the food delivery sequence chain, the route accident points, and the navigation route time consumption, and the navigation route mitigation strategy is used to reduce the navigation route time consumption of the food delivery customer.

2. The method according to claim 1, characterized in that The Bayesian network model is combined with a predefined navigation route indicator system to calculate the navigation route time of the delivery sequence chain in the weighted delivery sequence chain list, including: According to the Bayesian network model, a low-dimensional vector representation of the weighted delivery sequence chain list is obtained by using a graph embedding algorithm to obtain a dependency vector; Based on the dependency vector, a predefined navigation route indicator system is used in combination with a multi-dimensional weight adjustment mechanism to calculate an initial navigation route score, and the initial navigation route score is assigned to the delivery sequence chain, wherein the navigation route indicator system includes the number of vulnerabilities, update frequency, and community activity; Using the delivery sequence chain assigned initial navigation route scores, the Bayesian network model is combined with the gradient boosting decision tree algorithm to optimize the navigation route probability distribution of the child nodes that the delivery customer directly or indirectly depends on under the condition of the parent node of the delivery customer, thereby obtaining the optimized conditional probability; According to the optimized conditional probability and in combination with an integrated learning method, the overall navigation route score of the food delivery sequence chain is evaluated to obtain a navigation route time result.

3. The method according to claim 1, characterized in that The method of using an anomaly detection algorithm to identify potential route accident points to generate a navigation route report including navigation route time results and the route accident points includes: Based on the navigation route time results, an anomaly detection algorithm is used in combination with an adaptive threshold setting strategy to analyze the navigation route time of the delivery sequence chain, identify data points that deviate from the normal range, and obtain route accident points; Using natural language processing technology to parse the safety notices and technical documents of the accident point on the route, automatically extracting information directly related to the accident point in the safety notices and technical documents, and generating details of the accident point; Summarize and organize all the navigation route time results and the accident point details, and apply time series analysis to predict future navigation route trends to generate comprehensive navigation route data; An interactive interface is created using a visualization tool to organize the comprehensive navigation route data, and a navigation route report including the navigation route time result and the accident point archive is generated.

4. The method according to claim 2, characterized in that The method utilizes the delivery sequence chain assigned the initial navigation route score, and optimizes the navigation route probability distribution of the child nodes that the delivery customer directly or indirectly depends on under the condition of the parent node of the delivery customer through the Bayesian network model combined with the gradient boosting decision tree algorithm, to obtain the optimized conditional probability, including: Collecting and organizing the low-dimensional vector representation of the food delivery sequence chain and the initial navigation route score to construct a training dataset; Initializing and processing the training data set based on the Bayesian network model to obtain a gradient boosting decision tree model, wherein the gradient boosting decision tree model is used to assist in estimating conditional probabilities in the optimized Bayesian network model; Using the training data set, jointly training the Bayesian network model and the gradient boosting decision tree model, optimizing the estimation of the conditional probability by iteratively calculating the loss function and adjusting the parameters by backpropagation, and obtaining a joint training result; According to the joint training results, the conditional probability distribution of each node in the Bayesian network model is refined, and the conditional probability values ​​of the child nodes that the delivery customer directly or indirectly depends on are updated under the condition of the parent node of a given delivery customer to obtain the optimized conditional probability.

5. The method according to claim 3, characterized in that Based on the navigation route time result, an anomaly detection algorithm is used in combination with an adaptive threshold setting strategy to analyze the navigation route time of the delivery sequence chain, identify data points that deviate from the normal range, and obtain route accident points, including: Selecting a suitable anomaly detection algorithm based on the navigation route time result, wherein the anomaly detection algorithm includes an isolation forest and a local outlier factor, and the anomaly detection algorithm can effectively identify abnormal patterns in the navigation route time data to generate an anomaly detection algorithm; Based on the historical data and domain knowledge of the food delivery sequence chain, an initial threshold is set for the anomaly detection algorithm to preliminarily screen outliers and obtain a preliminarily set threshold; Applying an adaptive threshold setting strategy and automatically updating the initially set threshold in combination with monitoring device performance indicators to improve accuracy and generate an optimized threshold, wherein the monitoring device performance indicators include accuracy and recall; The anomaly detection algorithm is combined with the optimized threshold to analyze the navigation route time results, identify data points that deviate from the normal range, and obtain route accident points.

6. The method according to claim 1, characterized in that The method of drawing a comprehensive navigation route map based on the navigation route report using a multi-dimensional visualization algorithm to obtain a navigation route view includes: According to the navigation route report, data of the food delivery sequence chain and navigation route information are collected and organized so that the data format is consistent and ready for visualization processing, thereby obtaining a navigation route data set; Using a multi-dimensional visualization algorithm, the navigation route data set is mapped, and the navigation route time and accident point location of the food delivery sequence chain are mapped into a multi-dimensional space to generate a visual representation; Based on the visual representation, a hierarchical clustering algorithm is applied to group components with similar navigation route patterns and perform similarity judgment and classify them into different navigation route pattern groups to obtain grouped navigation route patterns; In combination with geographic information or network topology, adjusting the visual layout of the navigation route modes displayed in groups, optimizing the visual effect, and generating a comprehensive navigation route map; Highlighting high navigation route areas of the integrated navigation route map through color coding and icon styles helps logistics personnel quickly identify key navigation route points, and utilizing interactive tools to enhance logistics personnel experience and obtain a navigation route view.

7. The method according to claim 1, characterized in that The method of locating the node paths associated with the delivery customers based on the attribute characteristics of the delivery customers to generate a delivery sequence chain in the node path, assigning weight values ​​to the delivery sequence chains, and generating a weighted delivery sequence chain list includes: Utilizing the parsed delivery customer identification information submitted by the logistics personnel, extracting attribute characteristics of the delivery customer; Based on the attribute characteristics of the delivery customers, the delivery sequence between the delivery customers is modeled through a graph database, the node paths associated with the delivery customers are identified and located, and a delivery sequence network is generated; Based on the meal delivery sequence network, the meal delivery sequence chain is refined and processed using static code analysis tools and manager log data to obtain the specific source and version information of the meal delivery sequence, and the information is organized into a meal delivery sequence chain record; Classifying the meal delivery sequence chain records, distinguishing between direct meal delivery sequence chains and indirect meal delivery sequence chains, assigning corresponding weight values ​​to each type of meal delivery sequence chain, and generating a preliminary weighted meal delivery sequence chain; Based on the comprehensive navigation route data, an expert device or a machine learning model is introduced to review and adjust the preliminary weighted meal delivery sequence chain, optimize the weight value, obtain the optimized weighted meal delivery sequence chain, and summarize and organize all the optimized meal delivery sequence chains to form a weighted meal delivery sequence chain list.

8. A logistics navigation device for catering business, characterized in that: include: A parsing module, configured to parse the identification information of the delivery customer submitted by the logistics personnel and extract attribute characteristics of the delivery customer from the identification information; a positioning module, configured to locate, based on the attribute characteristics of the delivery customers, the node paths associated with the delivery customers, to generate a delivery sequence chain in the node path, assign weight values ​​to the delivery sequence chains, and generate a weighted delivery sequence chain list, wherein the delivery sequence chain includes a direct delivery sequence chain and an indirect delivery sequence chain; a calculation module, configured to calculate the navigation route duration of the delivery sequence chains in the weighted delivery sequence chain list using a Bayesian network model in combination with a predefined navigation route indicator system, and identify potential route accident points using an anomaly detection algorithm to generate a navigation route report including the navigation route duration results and the route accident points; A drawing module is used to draw a comprehensive navigation route map based on the navigation route report using a multidimensional visualization algorithm to obtain a navigation route view, and to integrate a genetic algorithm and a simulated annealing algorithm to explore adjustment suggestions for high navigation route areas of the navigation route view to generate a navigation route mitigation strategy; wherein, the comprehensive navigation route map is grouped and displayed using a hierarchical clustering algorithm to display components with similar navigation route patterns, the navigation route patterns are obtained by similarity judgment of the delivery sequence chain, the route accident points and the navigation route time, and the navigation route mitigation strategy is used to reduce the navigation route time of the delivery customer.

Images