An airport catering scheduling system based on big data analysis

Abstract

The present invention relates to the technical field of airport catering scheduling, and in particular to an airport catering scheduling system based on big data analysis, including a real-time data perception layer, a decision optimization core layer and an execution monitoring layer; the real-time data perception layer collects flight dynamics, resource status and environmental variable data with the help of airport operation database, Internet of Things sensors and RFID tags; the fusion prediction module of the decision optimization core layer adopts an improved deep learning model to predict meal demand and resource consumption; the dynamic scheduling module uses a multi-objective genetic algorithm to generate a resource allocation plan; the path planning module uses an improved ant colony algorithm to determine the optimal transportation path; the execution monitoring layer monitors the execution of the plan through a digital twin visualization platform and traces abnormal events through a blockchain traceability unit. The present invention can effectively improve the resource allocation rationality, distribution efficiency and management reliability of airport catering, effectively guarantee the quality of airport catering services, and adapt to the needs of modern airport operations.

Description

Technical Field

[0001] The present invention relates to the technical field of airport catering scheduling, and in particular to an airport catering scheduling system based on big data analysis. Background Art

[0002] In the current airport operating environment, with the continuous growth in the number of flights and the increasing demand for service quality from passengers, airport catering faces many severe challenges. The traditional catering scheduling method can no longer meet the efficient and precise operation needs of modern airports.

[0003] Traditional airport catering scheduling often relies on experience and simple data statistics, and lacks the ability to deeply mine and analyze large amounts of complex data. In terms of meal demand forecasting, it is unable to fully consider the combined impact of factors such as flight dynamics (such as delays, advances, aircraft model changes, etc.), differences in passenger composition (food preferences of different groups such as tour groups and business travelers), and special events (such as major sports events, holidays, etc., which cause sudden changes in passenger flow) on meal demand, resulting in frequent deviations in meal preparation, either causing waste or failing to meet demand.

[0004] In terms of resource allocation, the allocation of resources such as manpower, food ingredients and equipment lacks scientific basis and systematic planning. The resource coordination between various catering operation units is not smooth, which easily leads to idle resources in some links and tight resources in key links, which seriously affects the efficiency and quality of catering. For example, during a busy period, meals may not be delivered to the boarding gate in time due to unreasonable allocation of transportation vehicles, affecting passengers' dining experience.

[0005] Transportation route planning also has flaws, and does not fully incorporate factors such as real-time airport traffic conditions, road construction information, and access restrictions in different areas. The delivery route may be improperly selected, resulting in longer transportation time, increasing the risk of food spoilage or flight delays. Moreover, there is a lack of effective monitoring and traceability in the management of the entire catering process. Once a problem occurs, such as food quality problems or delivery errors, it is difficult to quickly and accurately identify the cause and determine the person responsible, which is not conducive to timely problem solving and continuous improvement of service quality.

[0006] Therefore, in response to the above problems, an airport catering scheduling system based on big data analysis is proposed. Summary of the invention

[0007] The purpose of the present invention is to provide an airport catering scheduling system based on big data analysis to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] An airport catering dispatching system based on big data analysis includes a real-time data perception layer, a decision optimization core layer, and an execution monitoring layer:

[0010] The real-time data perception layer collects flight dynamic data, resource status data and environmental variable data in real time through the airport operation database, IoT sensors and RFID tags;

[0011] The core layer of decision optimization includes:

[0012] Fusion prediction module: uses an improved deep learning model to predict meal demand and resource consumption patterns;

[0013] Dynamic scheduling module: Generate resource allocation scheme based on multi-objective genetic algorithm;

[0014] Path planning module: Apply improved ant colony algorithm to calculate the optimal transportation path;

[0015] The execution monitoring layer includes a digital twin visualization platform and a blockchain traceability unit to implement solution execution monitoring and abnormal event tracing.

[0016] As a preferred solution, the improved deep learning model includes:

[0017] Spatiotemporal attention mechanism: Its calculation process is expressed as:

[0018] ,in, for Moment The attention weight of each historical feature; It is the intermediate variable used in calculating attention; is the number of historical features; ,in, is the weight vector used to calculate the attention score; is the same as the LSTM hidden state The multiplied weight matrix; is the LSTM hidden layer state; is the real-time environment feature vector The multiplied weight matrix; is the real-time environment feature vector; is the bias vector; Represented by natural numbers For the bottom, is the exponential function value of the exponent;

[0019] Multi-task output layer: output flight delay probability at the same time 、Special meal requirements and resource gap warning values .

[0020] As a preferred solution, the multi-objective genetic algorithm includes innovative operators:

[0021] Dynamic chromosome encoding structure: ,in, Represents the complete chromosome code; A code representing a resource allocation scheme; The code representing the path planning; A code indicating a time schedule;

[0022] The resource allocation scheme is represented by matrix coding: ,in, Indicates Class resources are in The number of operating units, , is the number of resource categories, , is the number of operating units;

[0023] Non-dominated sorting strategy: Its fitness function is defined as: ,in, Represents chromosome The fitness vector of , , Respectively represent the fitness functions of different dimensions; among them:

[0024] , is the total number of tasks, For the The actual completion time of a task, For the The deadline for each task;

[0025] , is the number of resource types, For the The cost coefficient of the class resource, For the The amount of class resources allocated, For the Class resource usage;

[0026] , is the number of paths, For the The importance coefficient of each path, It is to judge A function of the relationship between each path and obstacles;

[0027] Elite retention strategy, set the elite library update rules:

[0028] ,in, for The elite library of the moment, for The elite library of the moment, for The set of individuals newly generated at each moment, is a non-dominated sorting function, The size of the elite library.

[0029] As a preferred solution, the improved ant colony algorithm includes:

[0030] Pheromones dynamic update mechanism:

[0031] ,in, for Time from node To Node The pheromone concentration, is the pheromone volatility coefficient, for Time from node To Node The pheromone concentration, is the number of ants, For the Only ants are on slave nodes in this cycle. To Node the amount of pheromone left on the path;

[0032] in, , is the total amount of pheromones released by ants under normal circumstances, For the The length of the path traveled by the ants, As the inspiration factor, To save time, is the base time, is the total amount of pheromones released by ants in emergency situations, The time when the emergency occurs. are parameters related to emergency events;

[0033] The path selection probability calculation adopts a two-factor decision model:

[0034] ,in, for Moment Ant slave node Select to Node The probability of for Time from node To Node The pheromone concentration, is the pheromone importance factor, for Time from node To Node The heuristic function value of is the importance factor of the heuristic function, For the Only ants on the node An optional node collection, is the conflict factor, To determine the path Is there a conflict function? for Time from the current ant's node to the node The pheromone concentration on the path, for Time from the current ant's node to the node heuristic information.

[0035] As a preferred solution, the digital twin visualization platform includes:

[0036] The coordinate transformation model of the three-dimensional space-time situation mapping unit is:

[0037] ,in, is the original coordinate; is the transformed coordinate; Rotation angle for airport layout; Respectively in The amount of translation in the direction; is the height scaling factor;

[0038] Real-time deviation alarm unit, triggers alarm when the following conditions are detected:

[0039] ,in, is the monitoring period, for The actual position vector at the moment, for The planned position vector at time, is the deviation threshold.

[0040] As a preferred solution, the blockchain traceability unit includes:

[0041] The data encapsulation structure of the smart contract-driven data on-chain module is as follows:

[0042] ,in, Represents a block in the blockchain, is the block header, For the block body;

[0043] ,in, is the hash value of the previous block, is the timestamp, for Merkel Root;

[0044] ,in, is the task ID, For the operator, is the hash value of the data, For signature;

[0045] The time complexity of the abnormal event tracing algorithm satisfies:

[0046] ,in, is the total number of blocks, Number of related-party transactions, Represented by natural numbers The logarithmic function of the base.

[0047] It can be seen from the technical solution provided by the present invention that the airport catering scheduling system based on big data analysis provided by the present invention has the following beneficial effects:

[0048] Optimization of resource allocation: Through the precise analysis of the fusion prediction module, the demand for meals and resource consumption patterns can be accurately estimated; based on this, the dynamic scheduling module uses the resource allocation plan generated by the multi-objective genetic algorithm to achieve the reasonable allocation of resources such as manpower, ingredients and equipment; for example, during the peak tourist season or during the flight intensive period, sufficient chefs, kitchen assistants and various ingredients can be deployed in advance to ensure the smooth production of meals and avoid insufficient meal supply due to resource shortages; during the period when there are relatively few flights, resource input can be appropriately reduced to reduce costs, effectively improving resource utilization efficiency and reducing resource waste;

[0049] Improved delivery efficiency: The path planning module uses an improved ant colony algorithm to calculate the optimal transportation path, which fully considers the complex environment and traffic conditions of the airport. During the transportation process, it can guide delivery personnel to avoid congested roads and obstacles and choose the fastest and safest route. At the same time, the real-time data perception layer monitors flight dynamic data in real time, so that the delivery time can be flexibly adjusted according to the actual take-off and landing time of the flight. For example, when the flight is delayed, the delivery personnel can delay their departure accordingly to avoid premature delivery and the deterioration of food quality. When the flight is advanced, the delivery speed can be accelerated in time to ensure that the food is delivered on time, which greatly improves the timeliness and accuracy of delivery.

[0050] Enhanced management reliability: The digital twin visualization platform of the execution monitoring layer, with the help of the three-dimensional spatiotemporal situation mapping unit and the real-time deviation alarm unit, realizes all-round visual monitoring of the execution of the meal distribution plan. Once the actual meal distribution process deviates from the plan, it can quickly issue an alarm to facilitate timely adjustments and ensure the smooth progress of the meal distribution task. The blockchain traceability unit uses the smart contract-driven data chain module and the efficient abnormal event tracing algorithm to ensure the integrity and traceability of the meal distribution data. When problems arise, such as food quality problems or delivery errors, the root cause of the problem and the relevant responsible persons can be traced quickly and accurately, which enhances the transparency and reliability of management and effectively improves the overall quality and management level of the airport catering service. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 The figure is a schematic diagram of the overall structure of an airport catering scheduling system based on big data analysis according to the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0053] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0054] like Figure 1 As shown, an embodiment of the present invention provides an airport catering scheduling system based on big data analysis, including a real-time data perception layer, a decision optimization core layer and an execution monitoring layer.

[0055] In this embodiment, the real-time data perception layer collects flight dynamic data, resource status data and environmental variable data in real time through the airport operation database, IoT sensors and RFID tags;

[0056] Furthermore, the real-time data perception layer is the source of data for the airport catering dispatching system. It uses a variety of technical means to comprehensively collect various key data, laying a solid foundation for the accurate decision-making and efficient operation of the entire system, and ensuring that the catering service can closely meet the actual operational needs of airport flights; specifically:

[0057] Data collection method and source:

[0058] Airport operation database collection: Deeply mine the airport operation database resources to extract detailed flight information, including flight number, take-off and landing time, aircraft model, route, etc.; obtain resource allocation details, such as the number of manpower in each catering operation area, the inventory level of various ingredients, the performance parameters and usage status of equipment; collect historical operation data, such as the actual demand for catering on past flights, resource consumption fluctuations in different time periods, etc.; these data provide a comprehensive basic information framework for the system to help analyze the law of catering demand and the rationality of resource allocation;

[0059] IoT sensor collection: IoT sensors are widely deployed at key locations in the airport; on the apron, sensors accurately monitor the real-time docking status, docking duration, and arrival and departure dynamics of aircraft, providing a direct basis for the accurate arrangement of meal delivery time; in the terminal, closely track the real-time changes in passenger flow, passenger gathering areas, and peak and trough periods of passenger flow, so as to estimate the approximate scale of meal demand in different time periods; in the catering workshop, strictly monitor the operating temperature, humidity, pressure and other key parameters of the equipment to ensure the stable operation of the equipment, and monitor the temperature, humidity, gas concentration and other indicators of the food storage environment to ensure the quality and safety of the food;

[0060] RFID tag collection: RFID tags are equipped for food ingredients, meal packaging and transportation equipment; the RFID tags of food ingredients record in detail key information such as category, purchase batch, production date, shelf life, supplier, etc., which facilitates accurate traceability and quality control in the meal preparation process; the labels on meal packaging can track the meal delivery route, link and delivery time node throughout the process; the labels on transportation equipment record the equipment's usage frequency, maintenance records, transportation task history and other information, which facilitates the reasonable dispatch of transportation resources and timely maintenance of equipment;

[0061] Technical architecture and components:

[0062] Data collection hardware equipment: Use high-performance sensor equipment, such as high-precision temperature and humidity sensors, pressure sensors, position sensors, etc., to ensure the accuracy and stability of collected data; equipped with advanced RFID readers and writers, with the ability to quickly identify and read tag information at a long distance, to achieve efficient processing of a large number of tags; data collection equipment has good compatibility and scalability, and can be easily connected to the airport's existing network infrastructure to ensure smooth data transmission;

[0063] Data transmission network: Build a dedicated data transmission network that combines wired and wireless network technologies. In fixed facility areas within the airport, such as catering workshops and warehouses, wired network connections are used to ensure high-speed and stable data transmission. In mobile devices and areas that require flexible deployment, such as aprons and terminals, wireless network technologies such as Wi-Fi or 4G / 5G are used to achieve real-time data transmission. The network has strong anti-interference capabilities and data encryption functions to ensure the security and integrity of data during transmission.

[0064] Data processing software system: Develop specialized data collection and preprocessing software to perform real-time cleaning, denoising, format conversion and other operations on the collected data; the software has intelligent data verification function, which can automatically identify and correct errors and abnormal values ​​in the data collection process; through data caching and queue technology, ensure the orderliness and reliability of data during transmission and processing, and avoid data loss or confusion;

[0065] Importance of data collection:

[0066] Ensure the timeliness and accuracy of meal distribution: Real-time collection of flight dynamic data, such as flight delays, advances or cancellations, allows the meal distribution plan to be adjusted accordingly to ensure that meals are supplied to the corresponding flights at the right time; based on real-time data on passenger flow and aircraft docking time, meal demand can be accurately calculated to avoid waste caused by excessive meal supply or dissatisfaction caused by insufficient meal supply, greatly improving the quality and efficiency of meal distribution services;

[0067] Optimize resource allocation and management: Through real-time monitoring of resource status data, the system can dynamically allocate human, food and equipment resources according to the actual needs of each catering operation unit; for example, when a certain area has a heavy catering task, idle personnel or equipment in other areas can be deployed in a timely manner for support; according to the real-time changes in food inventory, reasonable arrangements can be made for procurement and replenishment plans to reduce inventory costs, improve resource utilization efficiency, and ensure the smooth progress of catering production;

[0068] Ensure food safety and quality: Continuously monitor the data of food storage environment variables, such as temperature and humidity. Once the data exceeds the safe range, an alarm will be immediately issued and corresponding measures will be taken to effectively prevent food spoilage or damage. Real-time monitoring of equipment operation status can timely detect hidden dangers of equipment failure, perform maintenance in advance, and ensure the normal operation of equipment, thereby ensuring that the processing and storage conditions of meals meet the standards and guarantee food safety and quality.

[0069] The real-time data perception layer has become an indispensable and important part of the airport catering scheduling system due to its multi-channel data collection method, complete technical architecture and key data support function, and has effectively promoted the intelligent and efficient development of airport catering services.

[0070] In this embodiment, the decision optimization core layer includes a fusion prediction module, a dynamic scheduling module and a path planning module;

[0071] Furthermore, the fusion prediction module uses an improved deep learning model to predict meal demand and resource consumption patterns. The improved deep learning model includes:

[0072] Spatiotemporal attention mechanism: Its calculation process is expressed as:

[0073] ,in, for Moment The attention weight of each historical feature; It is the intermediate variable used in calculating attention; is the number of historical features; ,in, is the weight vector used to calculate the attention score; is the same as the LSTM hidden state The multiplied weight matrix; is the LSTM hidden layer state; is the real-time environment feature vector The multiplied weight matrix; is the real-time environment feature vector; is the bias vector; Represented by natural numbers For the bottom, is the exponential function value of the exponent;

[0074] Multi-task output layer: output flight delay probability at the same time 、Special meal requirements and resource gap warning values ;

[0075] The fusion prediction module is a key component of the core layer of the airport catering scheduling system's decision-making optimization. With the help of advanced improved deep learning models, it deeply mines and analyzes multi-source data to accurately predict meal demand and resource consumption patterns, providing a core basis for subsequent resource allocation and scheduling decisions, and effectively ensuring the efficiency and accuracy of airport catering services. Specifically:

[0076] Model Architecture:

[0077] Spatiotemporal attention mechanism: In the spatiotemporal attention mechanism, the calculation process is rigorous and complex; for a given moment and Historical features, attention weights The calculation is done by the formula The intermediate variable Depend on Calculated; here, As the weight vector for calculating the attention score, it plays a key role in determining the importance weights of different historical features; is the same as the LSTM hidden state The multiplied weight matrix; Contains time series information in historical data; is the real-time environment feature vector The multiplied weight matrix, Reflects environmental factors at the current moment, such as weather, holidays, etc.; is the bias vector; Represented by natural numbers For the bottom, is the exponential function value of the exponent; through such calculation method, the model can adaptively focus on key information according to the importance of the environment and historical data at different times. When processing a large amount of complex flight and environmental data, it can effectively screen out factors that have a greater impact on meal demand and resource consumption. For example, in the peak tourist season or in bad weather, it focuses on the stimulating effect of relevant historical data and real-time environmental changes on meal demand, ignores minor information, and improves the accuracy of prediction;

[0078] Multi-task output layer: The multi-task output layer has powerful functions and can output the probability of flight delay at the same time Special dietary requirements and resource gap warning values ; In terms of predicting the probability of flight delays, the model comprehensively analyzes multiple factors such as the on-time rate of historical flights, the operating status of current flights (such as weather at the departure point, air traffic congestion, etc.), and the operating status of airport equipment, and calculates the possibility of flight delays through model parameters trained by deep learning algorithms; for the prediction of special meal demand, combined with flight passengers' booking information, historical special meal demand data, and cultural, religious and other background information of the flight destination, data mining and analysis techniques are used to accurately estimate the demand for special meals (such as vegetarian, halal, low-sugar foods, etc.) on different flights; in the calculation of resource gap warning values, based on the predicted results of meal demand and the current resource reserves and consumption rate, various resources such as manpower, ingredients, and equipment are considered, and a resource consumption model is established to calculate the possible resource shortages that may occur in a certain period of time in the future, providing a quantitative reference for early resource allocation;

[0079] Data processing and prediction process:

[0080] Data reception and preprocessing: The fusion prediction module first receives rich data from the real-time data perception layer, including massive historical flight catering data, real-time updated flight dynamic data (such as flight number, take-off and landing time, aircraft model, route changes, etc.) and environmental variable data (such as temperature, humidity, holidays, tourist peak season, etc.); after receiving the data, it performs preprocessing operations on it, such as data cleaning, removing outliers and erroneous data to ensure data accuracy; data normalization, unifying different types of data into the same dimension and numerical range, to facilitate subsequent model calculation and analysis; data encoding, converting some classified data into a numerical form suitable for model input, such as converting flight destinations into corresponding geographic codes, etc.; through these preprocessing steps, the quality and availability of data are improved, laying the foundation for the efficient operation of the model;

[0081] Model training and optimization: Use a large amount of preprocessed historical data to train the improved deep learning model; during the training process, use appropriate optimization algorithms (such as stochastic gradient descent, Adam optimizer, etc.) to continuously adjust the model parameters to minimize the error between the predicted results and the actual data; through multiple iterative training, the model gradually learns the hidden laws and patterns in the data to improve the accuracy and stability of the prediction; at the same time, use cross-validation and other technical means to evaluate and verify the performance of the model to ensure that the model can maintain good generalization ability on different data sets; during the training process, continuously adjust the weight vector in the spatiotemporal attention mechanism , weight matrix , and the bias vector Parameters such as , to optimize the model's attention to and processing capabilities of different data features;

[0082] Real-time prediction and result output: After the model training is completed and reaches a certain performance standard, the fusion prediction module can predict the newly received real-time data; based on the current flight information and environmental conditions, the model quickly calculates the predicted results of meal demand and resource consumption pattern, and outputs key information such as flight delay probability, special meal demand and resource gap warning value to the dynamic scheduling module and path planning module; in actual operation, for example, when new flight information is entered into the system or environmental factors change (such as sudden bad weather causing flight delays or passenger delays), the model can update the prediction results in a short time, and provide the latest reference basis for subsequent decision-making in a timely manner, ensuring that the airport catering service can respond quickly to various changes and ensure smooth airport operations;

[0083] Furthermore, the dynamic scheduling module generates resource allocation schemes based on a multi-objective genetic algorithm, which contains innovative operators:

[0084] Dynamic chromosome encoding structure: ,in, Represents the complete chromosome code; A code representing a resource allocation scheme; The code representing the path planning; A code indicating a time schedule;

[0085] The resource allocation scheme is represented by matrix coding: ,in, Indicates Class resources are in The number of operating units, , is the number of resource categories, , is the number of operating units;

[0086] Non-dominated sorting strategy: Its fitness function is defined as: ,in, Represents chromosome The fitness vector of , , Respectively represent the fitness functions of different dimensions; among them:

[0087] , is the total number of tasks, For the The actual completion time of a task, For the The deadline for each task;

[0088] , is the number of resource types, For the The cost coefficient of the class resource, For the The amount of class resources allocated, For the Class resource usage;

[0089] , is the number of paths, For the The importance coefficient of each path, It is to judge A function of the relationship between each path and obstacles;

[0090] Elite retention strategy, set the elite library update rules:

[0091] ,in, for The elite library of the moment, for The elite library of the moment, for The set of individuals newly generated at each moment, is a non-dominated sorting function, is the size of the elite library;

[0092] The dynamic scheduling module is a key part of the core layer of the airport catering scheduling system. Based on a multi-objective genetic algorithm, it deeply processes the data provided by the fusion prediction module and generates a scientific and reasonable resource allocation plan to achieve efficient utilization of airport catering resources and accurate scheduling of catering tasks, ensuring the smooth operation of the entire catering process. Specifically:

[0093] Principle of multi-objective genetic algorithm:

[0094] Dynamic chromosome encoding structure: This module adopts a unique dynamic chromosome encoding structure ;in, The resource allocation scheme is represented by matrix coding, such as , here Clearly indicate the Class resources are in The number of operating units, ( is the number of resource categories, including manpower, food, equipment, etc.), ( is the number of operating units, including catering preparation areas, food storage areas, transportation teams, etc.); this coding method can intuitively reflect the distribution of resources in different operating units, facilitating the overall management and optimal allocation of resources; for example, in terms of manpower allocation, the number of chefs, kitchen assistants, and delivery personnel required for each catering link can be clearly defined; in terms of food allocation, the amount of food reserves of each operating unit can be accurately determined; Responsible for the coding of route planning, which can represent the transportation route information from the catering center to the destinations such as the boarding gate of each flight or the crew rest area; The time schedule is coded to determine the start time, duration and end time of each task link, ensuring the time coordination and order of the entire meal distribution process;

[0095] Non-dominated sorting strategy: Its fitness function is defined as ;in, , is the total number of tasks, For the The actual completion time of a task, For the The function is mainly used to measure the delay of tasks. By calculating the sum of the difference between the actual completion time and the deadline of all tasks, the generated scheduling plan can ensure that all tasks are completed on time as much as possible, reducing the risk of flight delays due to meal distribution. For example, in the actual meal distribution process, including food preparation, cooking, packaging, transportation and other task links, if a link is delayed, the function value will increase accordingly. In the iterative process of the genetic algorithm, this solution will gradually be eliminated or optimized.

[0096] , is the number of resource types, For the The cost coefficient of the class resource, For the The amount of class resources allocated, For the The usage of class resources; from the perspective of cost, this function considers the difference between resource allocation and usage multiplied by the cost coefficient, aiming to achieve efficient resource utilization and cost control; for example, in food resource management, if a certain food is allocated too much but actually used less, it will lead to increased costs. Through the evaluation of this function, the resource allocation plan will be adjusted during the algorithm evolution process to make resource allocation more reasonable;

[0097] , is the number of paths, For the The importance coefficient of each path, It is to judge A function that evaluates the relationship between paths and obstacles. This function focuses on the safety and efficiency of transportation paths. By evaluating different paths, it avoids choosing paths with obstacles or that are not suitable for transportation, ensuring that meals can be smoothly transported to the destination. For example, in the complex layout of an airport, some areas may have obstacles such as construction and traffic control. This function can guide the algorithm to select a better transportation path and improve transportation efficiency.

[0098] Elite Retention Strategy: Set the rules for updating the elite library ,in, for The elite library of the moment, for The elite library of the moment, for The set of individuals newly generated at each moment, is a non-dominated sorting function, is the size of the elite pool; in each iteration of the genetic algorithm, the newly generated individuals participate in the non-dominated sorting together with the elite individuals of the previous generation, and a certain number of elite individuals are selected to enter the next iteration; this strategy can retain excellent gene combinations, avoid losing excellent solutions during the evolution process, accelerate the convergence speed of the algorithm, and improve the quality of the solution; for example, in a certain round of iteration, some newly generated resource allocation solutions perform well in terms of task completion time, resource utilization efficiency, and path selection. After non-dominated sorting, these excellent solutions will be selected into the elite pool and continue to participate in evolution and optimization in subsequent iterations, so as to find a better resource allocation solution faster;

[0099] Resource allocation plan generation process:

[0100] Population initialization: First, a chromosome population of a certain size is initialized according to the scale and characteristics of the problem; each chromosome represents a possible combination of resource allocation, path planning, and time scheduling; during the initialization process, the code of the resource allocation plan is randomly generated , Path Planning Coding and time schedule coding , ensuring that the population has a certain degree of diversity; for example, in terms of resource allocation, different numbers of manpower may be randomly assigned to each catering operation unit, and the allocation of ingredients and equipment will also be randomly selected within a reasonable range, providing a rich foundation for subsequent evolution;

[0101] Fitness calculation and selection: For each chromosome in the initialized population, its fitness value is calculated according to the fitness function in the non-dominated sorting strategy; by comparing the size of the fitness value, a suitable selection method (such as roulette selection, tournament selection, etc.) is used to select a part of chromosomes with higher fitness as parents; these parents will participate in the subsequent crossover and mutation operations to produce new offspring individuals; in the selection process, chromosomes with higher fitness are more likely to be selected, thereby ensuring that excellent genes have a greater chance of being passed on to the next generation; for example, chromosomes with less task delay, high resource utilization efficiency and reasonable path selection have higher fitness values ​​and are more likely to be selected as parents for reproduction;

[0102] Crossover and mutation: Perform a crossover operation on the selected parent chromosomes, and exchange some gene fragments of the chromosomes according to a certain crossover probability to generate new gene combinations; for example, in the crossover process of resource allocation coding, the resource allocation quantities of some work units in the two chromosomes may be exchanged, thereby generating new resource allocation scheme possibilities; at the same time, perform a mutation operation on the newly generated daughter chromosomes to change certain gene values ​​on the chromosomes with a certain mutation probability; for example, in path planning coding, the connection path of a node may be randomly changed, or in time scheduling coding, the start time or duration of a task may be fine-tuned; through crossover and mutation operations, the diversity of the population is increased, the algorithm is prevented from falling into a local optimal solution, and a better resource allocation scheme space is explored;

[0103] Iteration and convergence: After multiple iterations of the above selection, crossover and mutation operations, the population continues to evolve; during the iteration process, the quality of each chromosome is continuously evaluated according to the fitness function, and the excellent individuals are retained and the poor individuals are eliminated; as the number of iterations increases, the population gradually converges to a set of optimal or near-optimal resource allocation plans; when certain termination conditions are met (such as reaching the preset number of iterations, the fitness value no longer changes significantly, etc.), the algorithm stops iterating and outputs the final resource allocation plan, which is used to guide the actual airport catering resource allocation and task scheduling process;

[0104] Furthermore, the path planning module uses an improved ant colony algorithm to calculate the optimal transportation path. The improved ant colony algorithm includes:

[0105] Pheromones dynamic update mechanism:

[0106] ,in, for Time from node To Node The pheromone concentration, is the pheromone volatility coefficient, for Time from node To Node The pheromone concentration, is the number of ants, For the Only ants are on slave nodes in this cycle. To Node the amount of pheromone left on the path;

[0107] in, , is the total amount of pheromones released by ants under normal circumstances, For the The length of the path traveled by the ants, As the inspiration factor, To save time, is the base time, is the total amount of pheromones released by ants in emergency situations, The time when the emergency occurs. are parameters related to emergency events;

[0108] The path selection probability calculation adopts a two-factor decision model:

[0109] ,in, for Moment Ant slave node Select to Node The probability of for Time from node To Node The pheromone concentration, is the pheromone importance factor, for Time from node To Node The heuristic function value of is the importance factor of the heuristic function, For the Only ants on the node An optional node collection, is the conflict factor, To determine the path Is there a conflict function? for Time from the current ant's node to the node The pheromone concentration on the path, for Time from the current ant's node to the node Heuristic information;

[0110] The path planning module is an important part of the core layer of the airport catering scheduling system. It is based on the improved ant colony algorithm and aims to plan the optimal path for catering transportation to improve transportation efficiency, reduce costs, and ensure that meals can be delivered to the destination in a timely and safe manner to meet the timeliness and reliability requirements of airport catering. Specifically:

[0111] Improved ant colony algorithm mechanism:

[0112] Pheromone dynamic update mechanism: The pheromone dynamic update mechanism adopted by this module is ,in, for Time from node To Node The pheromone concentration, is the pheromone volatility coefficient, for Time from node To Node The pheromone concentration, is the number of ants, For the Only ants are on slave nodes in this cycle. To Node The amount of pheromone left on the path; in calculating When the path is normal, it is divided into two cases: normal path and emergency path. For the normal path, ,here is the total amount of pheromones released by ants under normal circumstances, For the The length of the path traveled by the ants, As the inspiration factor, To save time, As the benchmark time; this calculation method makes ants tend to choose paths with high pheromone concentrations, short distances, and time savings when choosing paths, because ants that have walked these paths will leave more pheromones, thereby guiding subsequent ants to choose better paths; for example, in the regular transportation channels of an airport, if a path is short and there are no obstacles during transportation, the pheromones left by ants will gradually accumulate, and the probability of subsequent ants choosing this path will increase; in terms of emergency paths, ,in, is the total amount of pheromones released by ants in emergency situations, The time when the emergency occurs. Parameters related to emergency events; when an emergency event occurs (such as a sudden traffic jam or equipment failure in a certain area), the pheromones on the emergency path will be dynamically updated according to the time relationship with the emergency event, guiding the ants to quickly find alternative paths to ensure the continuity of transportation;

[0113] Path selection probability calculation: Path selection probability calculation adopts a two-factor decision model ,in, for Moment Ant slave node Select to Node The probability of for Time from node To Node The pheromone concentration, is the pheromone importance factor, for Time from node To Node The heuristic function value of is the importance factor of the heuristic function, For the Only ants on the node An optional node collection, As the conflict factor, To determine the path Is there a conflict function? for Time from the current ant's node to the node The pheromone concentration on the path, for Time from the current ant's node to the node Heuristic information;

[0114] The pheromone concentration and the heuristic function value jointly affect the path selection of ants. The path with high pheromone concentration is more likely to be selected. At the same time, the heuristic function value also provides ants with prior knowledge. For example, the heuristic function value is calculated based on factors such as the distance between nodes and the smoothness of the road, guiding the ants to choose a more reasonable path. The conflict factor prevents ants from choosing conflicting paths and ensures the feasibility and rationality of path planning. For example, in the complex transportation network of an airport, if a path is under maintenance or occupied by other transportation tasks, the conflict function will reduce the probability of selecting the path to avoid transportation conflicts.

[0115] Optimal transportation route calculation process:

[0116] Environmental modeling and data preparation: First, the catering transportation environment of the airport is modeled, and key locations such as the catering center, boarding gates for each flight, and crew rest areas are abstracted as nodes. The connecting channels between nodes are regarded as paths, and the attribute information such as the length and traffic capacity of the path is determined; at the same time, relevant data provided by the real-time data perception layer is collected, such as the current traffic conditions of the airport, the flow of people in each area, the operating status of equipment, etc. These data will serve as the basic information for the operation of the algorithm and affect the initial distribution of pheromones and the calculation of the heuristic function; for example, if there is a large flow of people in a certain area, the heuristic function value of the path in that area may be reduced, prompting the ants to avoid that area as much as possible to reduce transportation obstacles;

[0117] Ant search and pheromone update: A certain number of ants are released in the constructed environment model. The ants start searching from the catering center to various destinations according to the path selection probability formula. During the search process, the ants will leave pheromones on the paths they pass through, and continuously update the pheromone concentration on the paths according to the pheromone dynamic update mechanism. After each ant completes a search from the catering center to the destination, it will update the pheromone on the path it passes through, so that the pheromone concentration of the better path gradually increases, guiding subsequent ants to choose a better path. For example, after multiple search iterations, the pheromone concentration on the path connecting the catering center and the boarding gate with dense flights and good transportation conditions will increase significantly, becoming the path that the ants prefer.

[0118] Iterative optimization and path determination: After multiple iterative searches and pheromone updates, as the number of ants increases and the number of iterations accumulates, the algorithm gradually converges and eventually determines the optimal transportation path or a set of approximate optimal paths from the catering center to each destination. During the iteration process, the pheromone importance factor is continuously adjusted. , heuristic function importance factor , conflict factors and other parameters to optimize the path selection strategy to adapt to the complex and changeable environment and transportation needs of the airport; when certain termination conditions are met (such as reaching the preset number of iterations, the pheromone concentration of the optimal path no longer changes significantly, etc.), the algorithm stops running and outputs the final transportation path plan to guide the actual meal distribution and transportation process, ensuring that the meals can be delivered to the destination efficiently and accurately.

[0119] In this embodiment, the execution monitoring layer includes a digital twin visualization platform and a blockchain traceability unit to implement solution execution monitoring and abnormal event tracing;

[0120] Furthermore, the execution monitoring layer plays a vital role in the airport catering dispatching system. Through the digital twin visualization platform and blockchain traceability unit, it realizes all-round monitoring of the execution of catering plans and effective tracing of abnormal events, ensuring the transparency, controllability and reliability of the entire catering process, and guaranteeing the quality and efficiency of airport catering services. The execution monitoring layer specifically includes:

[0121] Digital twin visualization platform, the digital twin visualization platform includes:

[0122] The coordinate transformation model of the three-dimensional space-time situation mapping unit is:

[0123] ,in, The original coordinates are usually the actual location coordinates of various catering-related facilities, transportation equipment, personnel, etc. in the airport; The coordinates are transformed. Through this coordinate transformation, the complex actual layout of the airport and the catering process can be displayed on the monitoring interface in a clearer and more intuitive way. The rotation angle of the airport layout can be adjusted according to the needs of monitoring perspective so that the catering process can be observed from different angles; Respectively in The amount of translation in the direction is used to precisely adjust the display position and highlight key areas or devices; The height scaling factor can be used to reasonably scale and display facilities or equipment at different heights, so that monitoring personnel can better grasp the overall layout and the spatial relationship between various elements. For example, when monitoring the operation of equipment and personnel in the catering workshop, by adjusting the coordinate transformation parameters, the position and working status of each device in three-dimensional space, as well as the movement path of personnel, can be clearly seen, which helps to timely discover potential safety hazards or irregular operations and improve production efficiency and safety.

[0124] Real-time deviation alarm unit, which sets strict deviation thresholds , used to monitor the deviation between the actual meal distribution process and the plan; the conditions for triggering the alarm are: ,in, The monitoring period covers the entire time range from meal preparation to delivery completion. for The actual position vector at the moment, including the actual processing progress of the meal, the actual position of the transport vehicle, the actual working position of the personnel, etc. for The planned position vector at the moment is the standard position and progress information pre-set according to the meal distribution scheduling plan. Once the Euclidean distance between the actual position vector and the planned position vector is found within the monitoring period ( ) exceeds the deviation threshold , the system immediately triggers an alarm; for example, if the processing progress of a batch of meals is delayed by a certain time than planned, causing it to be If the actual position of the flight deviates too much from the planned position, the system will promptly issue an alarm to remind relevant personnel to take measures to make adjustments, such as increasing manpower input, optimizing processing procedures, or adjusting transportation routes, to ensure that the catering task can be completed smoothly as planned and avoid affecting flight catering.

[0125] Blockchain traceability unit, the blockchain traceability unit includes:

[0126] The data encapsulation structure of the smart contract-driven data on-chain module is as follows:

[0127] ,in, Represents a block in the blockchain, The block header is the basic unit of data storage and transmission. For the block body;

[0128] is the hash value of the previous block. Through this chain structure, the integrity and non-tamperability of the data are guaranteed, so that each block is closely connected with the previous block to form a complete data chain. is the hash value of the previous block, It is a timestamp that accurately records the time when data is generated or an operation occurs, providing a time dimension basis for data tracing and auditing. Merkle root, which is a hash summary of the data in the block body, used to quickly verify the integrity and consistency of the data;

[0129] ,in, It is the task ID, which uniquely identifies each meal distribution task, making it easy to query and track specific tasks in the entire blockchain system. For operators, the executors of each operation link are clearly recorded, which enhances the accuracy of responsibility tracing. The hash value of the data is used to perform hash operations on various data in the meal preparation process (such as food source information, processing parameters, transportation records, etc.) to ensure the authenticity and security of the data. For signature, digital signature technology is usually used, and the operator signs the data to further ensure the legitimacy and non-repudiation of the data; for example, in the food procurement process, when a batch of food is put into storage, its supplier information, purchase time, quality inspection report and other data will be encapsulated in the block body, and the chain operation will be automatically triggered through the smart contract. These original data can be queried through the blockchain at any stage of meal preparation, and due to the non-tamperable nature of the data, the credibility and reliability of the information are ensured;

[0130] Abnormal event tracing algorithm: The time complexity of this algorithm satisfies: ,in, is the total number of blocks, Number of related-party transactions, Represented by natural numbers is a logarithmic function with a base of 1. This means that when tracing abnormal events, as the number of blocks increases, the algorithm's time growth rate is relatively slow, and it has high efficiency. When an abnormal event occurs (such as food quality problems, delivery delays leading to flight complaints, etc.), the algorithm can quickly locate the relevant blocks and transaction records in the blockchain, and trace the source of the event and the entire process. For example, if a passenger on a flight reports that the food has a strange smell, the abnormal event tracing algorithm can quickly find the source of the ingredients for the batch of meals, the data of each link in the processing process, the temperature and humidity records during transportation, and other information from the blockchain, accurately determine the link and person responsible for the problem, so as to take timely measures to rectify and punish, and also provide strong data support for subsequent quality improvement and process optimization.

[0131] Through the collaborative work of the digital twin visualization platform and the blockchain traceability unit, the execution monitoring layer achieves comprehensive monitoring and effective traceability of the airport catering scheduling system from the execution process to data management, greatly improving the reliability and transparency of the system, and providing a solid guarantee for the stable operation of the airport catering service.

[0132] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An airport catering dispatching system based on big data analysis, characterized by: It includes real-time data perception layer, decision optimization core layer and execution monitoring layer: The real-time data perception layer collects flight dynamic data, resource status data and environmental variable data in real time through the airport operation database, Internet of Things sensors and RFID tags; The decision optimization core layer includes: Fusion prediction module: uses an improved deep learning model to predict meal demand and resource consumption patterns; Dynamic scheduling module: Generates resource allocation scheme based on multi-objective genetic algorithm, which contains innovative operators: Dynamic chromosome encoding structure: ,in, Represents the complete chromosome code; A code representing a resource allocation scheme; The code representing the path planning; A code indicating a time schedule; The resource allocation scheme is represented by matrix coding: ,in, Indicates Class resources are in The number of operating units, , is the number of resource categories, , is the number of operating units; Non-dominated sorting strategy: Its fitness function is defined as: ,in, Represents chromosome The fitness vector of , , Respectively represent the fitness functions of different dimensions; among them: , is the total number of tasks, For the The actual completion time of a task, For the The deadline for each task; , is the number of resource types, For the The cost coefficient of the class resource, For the The amount of class resources allocated, For the Class resource usage; , is the number of paths, For the The importance coefficient of each path, It is to judge A function of the relationship between each path and obstacles; Elite retention strategy, set the elite library update rules: ,in, for The elite library of the moment, for The elite library of the moment, for The set of individuals newly generated at each moment, is a non-dominated sorting function, is the size of the elite library; Path planning module: Apply improved ant colony algorithm to calculate the optimal transportation path; The execution monitoring layer includes a digital twin visualization platform and a blockchain traceability unit to realize solution execution monitoring and abnormal event tracing.

2. The airport catering scheduling system based on big data analysis according to claim 1 is characterized by: The improved deep learning model comprises: Spatiotemporal attention mechanism: Its calculation process is expressed as: ,in, for Moment The attention weight of each historical feature; It is the intermediate variable used in calculating attention; is the number of historical features; ,in, is the weight vector used to calculate the attention score; is the same as the LSTM hidden state The multiplied weight matrix; is the LSTM hidden layer state; is the real-time environment feature vector The multiplied weight matrix; is the real-time environment feature vector; is the bias vector; Represented by natural numbers For the bottom, is the exponential function value of the exponent; Multi-task output layer: output flight delay probability at the same time 、Special meal requirements and resource gap warning values .

3. The airport catering scheduling system based on big data analysis according to claim 1 is characterized by: The improved ant colony algorithm comprises: Pheromones dynamic update mechanism: ,in, for Time from node To Node The pheromone concentration, is the pheromone volatility coefficient, for Time from node To Node The pheromone concentration, is the number of ants, For the Only ants are on slave nodes in this cycle. To Node the amount of pheromone left on the path; in, , is the total amount of pheromones released by ants under normal circumstances, For the The length of the path traveled by the ants, As the inspiration factor, To save time, is the base time, is the total amount of pheromones released by ants in emergency situations, The time when the emergency occurred. are parameters related to emergency events; The path selection probability calculation adopts a two-factor decision model: ,in, for Moment Ant slave node Select to Node The probability of for Time from node To Node The pheromone concentration, is the pheromone importance factor, for Time from node To Node The heuristic function value of is the importance factor of the heuristic function, For the Only ants are on the node An optional node collection, As the conflict factor, To determine the path Is there a conflict function? for Time from the current ant's node to the node The pheromone concentration on the path, for Time from the current ant's node to the node heuristic information.

4. The airport catering scheduling system based on big data analysis according to claim 1 is characterized by: The digital twin visualization platform includes: The coordinate transformation model of the three-dimensional space-time situation mapping unit is: ,in, is the original coordinate; is the transformed coordinate; Rotation angle for airport layout; Respectively in The amount of translation in the direction; is the height scaling factor; Real-time deviation alarm unit, triggers alarm when the following conditions are detected: ,in, is the monitoring period, for The actual position vector at the moment, for The planned position vector at time, is the deviation threshold.

5. The airport catering scheduling system based on big data analysis according to claim 1 is characterized by: The blockchain traceability unit includes: The data encapsulation structure of the smart contract-driven data on-chain module is as follows: ,in, Represents a block in the blockchain, is the block header, For the block body; ,in, is the hash value of the previous block, is the timestamp, for Merkel Root; ,in, is the task ID, For the operator, is the hash value of the data, For signature; The time complexity of the abnormal event tracing algorithm satisfies: ,in, is the total number of blocks, Number of related-party transactions, Represented by natural numbers The logarithmic function of the base.

Images