Systems and methods for optimizing infrastructure

Abstract

According to several embodiments, this disclosure relates to systems and methods for optimizing infrastructure. A method is provided comprising the steps of: a. receiving one or more priority parameters for considering optimization, wherein each priority parameter is associated with a target aspect of the infrastructure related to one or more functions of the infrastructure, and includes weights; b. receiving an existing first optimization plan from an existing first optimization system, wherein the first optimization includes data related to the target aspect; c. generating a recommended second optimization plan by a second optimization system, wherein the second optimization plan is generated by modifying the first optimization plan to optimize the target aspect according to the weights using a prediction module including one or more models; d. sending the second optimization plan from the second optimization system to the first optimization system; and e. updating the first optimization plan based on the second optimization plan. A method, prediction module, and corresponding system for prioritizing one or more priority parameters are also provided. A system and corresponding method for optimizing parking space planning are also provided.

Description

Technical Field

[0001] This invention relates to systems and methods for optimizing infrastructure, such as transportation infrastructure. Specifically, but not exclusively, it relates to optimizing various interconnected functions of transportation infrastructure, such as airports, seaports, railway stations, other transportation hubs, or travel territories. Background Technology

[0002] Traditional practices in the travel and transportation industry rely heavily on human expertise and experience in management and planning. Decision-making processes are primarily influenced by operators' past experiences with similar situations and planners' common sense. Decisions are susceptible to planner biases and subjective judgment. Manual processing is time-consuming and often fails to consider all relevant factors simultaneously, resulting in imperfect planning. Human-centric systems struggle to adapt quickly to changing conditions or unexpected events. Automation is limited, meaning planning may not be optimized according to specific criteria.

[0003] Operations staff often rely on their experience to allocate parking spaces and manage transfers, frequently lacking a data-driven approach. For example, parking space allocation decisions may lack optimization for the passenger experience because human intuition may not always lead to the most efficient allocations, such as minimizing walking time, maximizing revenue generation, or reducing fuel consumption. Similarly, reallocation decisions used to increase retail sales and reduce the likelihood of missed transfers may not be optimized. Relying on human judgment can lead to missed opportunities to enhance the passenger experience through data-driven insights.

[0004] Manual parking stand allocation relies heavily on the expertise of airport staff, leveraging their understanding of stand characteristics. A lack of specific data on stand costs hinders the development of complex, data-driven solutions, particularly when indirect factors significantly impact costs. Manual stand allocation depends on domain expert knowledge and may struggle to adapt to changing circumstances. Rule-based systems may fail to adequately optimize stand utilization, leading to incomplete planning.

[0005] Existing airport management and optimization systems typically rely on manual planning and decision-making by airlines and airport experts, combined with historical data.

[0006] We recognize that existing systems and methods for optimizing various aspects of travel and transportation operations have shortcomings, resulting in incomplete assessments of past events and all relevant information. Furthermore, such systems are susceptible to subjective biases and inconsistencies. Therefore, a system and approach are needed to overcome these challenges by providing data-driven, intelligent solutions to optimize all aspects of travel and transportation operations, promote sustainability, and improve overall industry efficiency.

[0007] Although the problem has been described in the context of airport optimization systems and methods, one or more solutions provided by this invention can also be applied to any travel and transportation infrastructure, such as airports, seaports, railway stations, other transportation hubs or travel terminals, as well as other industries such as logistics. Summary of the Invention

[0008] According to several embodiments, this disclosure relates to systems and methods for optimizing one or more functions of a transportation infrastructure.

[0009] According to a first aspect of the invention, a method for optimizing parking stand planning is provided. The method includes the steps of: a. receiving one or more priority parameters for consideration in the parking stand planning, wherein each of these priority parameters includes a weight and is associated with one or more airport functions; b. receiving an existing first parking stand planning from an existing first parking stand planning optimization system, wherein the first parking stand planning includes data related to flights and their allocated parking stands; c. generating a recommended second parking stand planning from a second parking stand planning optimization system, wherein the second parking stand planning is generated by modifying the first parking stand planning to allocate parking stands to these flights based on the priority parameters using a prediction module including one or more models; d. transmitting the second parking stand planning from the second parking stand planning optimization system to the first parking stand planning optimization system; and e. updating the first parking stand planning based on the second parking stand planning.

[0010] Optionally, one or more existing rules related to the one or more airport functions can be obtained before the step of generating the second parking stand plan.

[0011] Optionally, the second parking space plan can be generated to avoid conflicts with these existing rules.

[0012] Optionally, the one or more existing rules related to the one or more airport functions can be obtained from an external database associated with the airport.

[0013] Optionally, the one or more priority parameters can be obtained from a rule-based database associated with the airport.

[0014] Optionally, the one or more priority parameters can be obtained by user input or from a database.

[0015] Optionally, the method may further include the following step: f. Sending a direction for controlling the aircraft to a parking position according to the second plan.

[0016] According to a second aspect of the invention, a method for prioritizing one or more priority parameters in a method according to a first aspect of the invention, wherein the weight of each priority parameter can be dynamically adjusted by a user.

[0017] Optionally, during the adjustment of this weight, the user may be shown changes to one or more airport functions corresponding to the parameter.

[0018] Optionally, the changes in one or more airport functions corresponding to the parameter can be displayed as sub-parameters with numerical values.

[0019] According to a third aspect of the invention, a system for optimizing parking stand planning includes: a. a receiver for receiving one or more priority parameters to be considered in the parking stand planning, wherein each of the priority parameters includes a weight and is associated with one or more airport functions, and for receiving an existing first parking stand planning from an existing first parking stand planning optimization system, wherein the first parking stand planning includes data related to flights and their allocated parking stands; b. a processor for generating a recommended second parking stand planning by a second parking stand planning optimization system, wherein the second parking stand planning is generated by modifying the first parking stand planning to allocate parking stands to the flights based on the priority parameters using a prediction module including one or more models; and c. a transmitter for transmitting the second parking stand planning from the second parking stand planning optimization system to the first parking stand planning optimization system for updating the first parking stand planning based on the second parking stand planning.

[0020] According to a fourth aspect of the invention, a prediction module is provided for use in a method for optimizing parking space planning according to a first aspect or in a system according to a third aspect of the invention. The prediction module includes multiple machine learning models. Each of these machine learning models is connected to one or more data sources and is trained to perform predictions on one or more consequences of one or more decisions based on historical data from said one or more data sources. The one or more consequences predicted by these machine learning models relate to multiple airport functions.

[0021] Optionally, the second parking space plan can be selected from one or more candidate second parking space plans that satisfy these priority parameters based on the one or more decisions and the consequences predicted by these machine learning models.

[0022] Optionally, the second parking space planning can be associated with the least negative consequences predicted by a machine learning model.

[0023] According to a fifth aspect of the invention, a method for optimizing infrastructure is provided. The method includes the steps of: a. receiving one or more priority parameters for considering optimization, wherein each of these priority parameters is associated with a target aspect of the infrastructure related to one or more functions of the infrastructure, and includes weights; b. receiving an existing first optimization plan from an existing first optimization system, wherein the first optimization includes data related to the target aspect; c. generating a recommended second optimization plan by a second optimization system, wherein the second optimization plan is generated by modifying the first optimization plan to optimize the target aspect according to the weights using a prediction module including one or more models; d. transmitting the second optimization plan from the second optimization system to the first optimization system; and e. updating the first optimization plan based on the second optimization plan.

[0024] Optionally, one or more existing rules related to the one or more functions of the infrastructure can be obtained before the step of generating the second optimization plan.

[0025] Optionally, this second optimization program can be generated to avoid conflicts with these existing rules.

[0026] Optionally, one or more existing rules related to one or more functions of the infrastructure can be obtained from an external database associated with the infrastructure.

[0027] Optionally, one or more priority parameters can be obtained from a rule-based database associated with the infrastructure.

[0028] Optionally, the one or more priority parameters can be obtained by user input or from a database.

[0029] According to a sixth aspect of the invention, there is a method for prioritizing one or more priority parameters in a method according to a fifth aspect of the invention, wherein each of these priority parameters can be dynamically adjusted by a user.

[0030] Optionally, during the adjustment of this weight, the user may be shown changes to one or more functions of the infrastructure corresponding to the parameter.

[0031] Optionally, the changes in one or more functions of the airport corresponding to the parameter can be displayed as sub-parameters with numerical values.

[0032] According to a seventh aspect of the invention, a system for optimizing infrastructure is provided. The system includes: a. a receiver for receiving one or more priority parameters for considering optimization, wherein each of the priority parameters is associated with a target aspect of the infrastructure related to one or more functions of the infrastructure, and includes weights; receiving an existing first optimization plan from an existing first optimization system, wherein the first optimization includes data related to the target aspect; b. a processor for generating a recommended second optimization plan by a second optimization system, wherein the second optimization plan is generated by modifying the first optimization plan to optimize the target aspect according to the weights using a prediction module including one or more models; and c. a transmitter for transmitting the second optimization plan from the second optimization system to the first optimization system for updating the first optimization plan based on the second optimization plan.

[0033] According to an eighth aspect of the invention, a prediction module is provided for use in a method according to a fifth aspect of the invention or in a system according to a seventh aspect of the invention. The prediction module includes a plurality of machine learning models, each of which is connected to one or more data sources and is trained to perform predictions on one or more consequences of one or more decisions based on historical data from the one or more data sources, and the one or more consequences predicted by the machine learning models relate to multiple functions of the infrastructure.

[0034] Optionally, the second optimal program can be selected from one or more candidate second optimal programs that satisfy these priority parameters based on the one or more decisions and the consequences predicted by the machine learning model.

[0035] Optionally, the second optimization program is associated with the minimum negative consequences predicted by the machine learning model.

[0036] Optionally, one or more of the methods in each of the above aspects may also include the step of sending instructions.

[0037] Optionally, the instruction may be used for any one or more of the following: controlling the aircraft to parking positions according to the second plan; controlling the allocation of security checkpoints according to the second plan; controlling the opening and / or allocation of resources in one or more areas of the infrastructure according to the second plan; controlling the timing of one or more information announcements according to the second plan; controlling the routes for one or more users of the infrastructure according to the second plan; controlling energy use in one or more areas of the infrastructure according to the second plan; and / or controlling the planning and / or allocation of resources for maintenance and / or cleaning programs according to the second plan.

[0038] Optionally, one or more of the methods in each of the above aspects can be implemented as a computer program product, which, when executed, implements the corresponding method. Attached Figure Description

[0039] Embodiments of the invention will now be described by way of example only and with reference to the accompanying drawings, wherein: Figure 1 This is a schematic diagram of an example system for parking space planning according to an embodiment; Figure 2 This is a schematic diagram of an example user interface according to an embodiment; Figure 3 This is a schematic diagram of an example optimized model according to an embodiment; Figure 4 The illustrations show (a) an example initial diagram and (b) an example final diagram according to the embodiments; Figure 5 An example diagram with fixed and secure nodes according to an embodiment is illustrated; Figure 6 The illustration shows example diagrams with a splitting degree of 9 to 4 according to an embodiment; Figure 7 The illustrations show (a) a final example drawing without fixed parts and (b) an example drawing without non-fixed parts; Figure 8 This is a schematic diagram of an exemplary system according to an embodiment; and Figure 9 With local process Figure 9A and Figure 9B The illustration shows a flowchart of an example optimization process according to an embodiment. Figure 9 The relative positions of the partial flowcharts are provided. Detailed Implementation

[0040] In complex systems, particularly in the aviation, travel, and transportation industries, navigational interconnected events often require intricate and comprehensive problem-solving approaches. For example, reducing aircraft emissions is a challenge that traditional machine learning prediction models often struggle to address holistically. Sustainability improvements preferably require a holistic perspective, as attempting to optimize a single factor in isolation can inadvertently trigger unintended consequences. For instance, efforts to improve fuel efficiency might unknowingly exacerbate security queues at airports and increase electricity consumption, leading to increased pollution. Understanding these complex relationships is crucial because correcting one problem can disrupt the delicate balance between multiple interconnected factors, impacting aspects such as retail sales, risk levels, flight delays, and the overall passenger experience. In other words, isolated improvements can lead to unintended negative cascading effects.

[0041] Considering another example of increased electricity consumption, interconnected factors could include pollution levels, as well as interconnected systems that could affect retail sales, risk mitigation, and passenger experience. Such interrelationships further suggest that focusing solely on a single improvement without considering its response within an interdependent network can lead to imperfect results and unintended trade-offs.

[0042] The following exemplary description is based on systems, apparatus, and methods used in the aviation industry to provide solutions to problems (such as one or more of the problems described above). However, it should be understood that the scope of the invention is not limited to the aviation industry, but also applies to any industry such as other tourism, packaging, or express delivery industries that transport items such as bags, suitcases, packaged goods, parcels, or mail. Therefore, embodiments of the invention are applicable throughout the tourism industry, such as rail, long-distance buses, automobiles, ships, and express delivery and distribution services.

[0043] Figure 1 The illustration shows a schematic diagram of an exemplary system for parking space planning according to an embodiment. Relative to Figure 1 Examples relating to parking space planning at airports are described below, along with systems and methods for optimizing transport infrastructure according to the present invention. However, those skilled in the art will understand that the principles described are applicable to optimizing any suitable function and / or aspect of any type and form of transport infrastructure, such as airports, seaports, railway stations, other transport hubs, or travel territories.

[0044] like Figure 1 As shown, interface 1102 can be provided for a user to set one or more priorities (1001) before initiating parking space optimization planning (1002). Alternatively, one or more priorities (1001) can be set simultaneously with initiating parking space optimization planning. The one or more priorities can be set by a user (e.g., an operations director, manager, or other stakeholders at the airport or one or more areas of the airport). For example, airport stakeholders who wish to optimize airport parking space planning for a schedule can set one or more priorities accordingly.

[0045] One or more priorities can be set via interface 1102. Figure 2 An exemplary interface 1102 for setting one or more priorities is shown. This interface allows one or more priorities to be balanced. Optionally, priorities can be balanced via the use of a slider. For example, in Figure 2 In the partial view of the airport director user interface 1102 shown, the further the slider is set to the right, the higher the priority level the user assigns to that priority.

[0046] Therefore, for a specific time period, users can balance and / or weight one or more priorities using the user interface or any other suitable method that receives user input. Figure 2 In the example, the user has one or more balanced and / or weighted priorities for March 14, 2024, so that: Prioritize capacity supply above all other key performance indicators; Set operational performance, revenue generation, and passenger experience to smaller values ​​(compared to capacity supply), but on an equal footing; and Setting the sustainability value to the smaller of all values ​​indicates that the value of this KPI is less than that of other priorities during this phase.

[0047] Such interfaces can be provided via the web and / or as an application. This application can be, for example, a single-page application (SPA), suitable for providing dynamically updated content without refreshing the entire page or redirecting to another webpage for data retrieval. Parking space optimization planning can be initiated via a REST call (1002), which in turn initiates the parking space optimization service.

[0048] For example, upon confirmation (e.g., by clicking Start or Apply in the interface), the system can trigger a REST API request for the parking stand optimization service. Optionally, the Airport Director's web application can trigger a REST API request for the parking stand optimization service. An example JSON request is as follows:

[0049]

[0050] As shown in the example JSON request above, the numerical representation can be a necessary consequence of the slider position (e.g., a given range from a minimum of 0 to a maximum of 10). Optionally, the schedule elements used to indicate time periods can be arrays, and each node in the array can have a start time and an end time. Optionally, the schedule can contain more than one profile per day. For example, during peak traffic periods, an airport can prioritize operational performance over other KPIs, while during periods of lower traffic, sustainability and profitability can be prioritized. This can be implemented using multiple nodes within a schedule array in the JSON request.

[0051] The parking space optimization service can receive requests and initiate an orchestration process. This process can be initiated by generating and saving a context document to track the process. An example of such a document is shown below:

[0052]

[0053] In the example document above, the association ID is a key element for tracking the process, both during and after completion. Several documents can be generated during the process's validity period. The association ID can be a binding element throughout all documents. The entire artifact dictionary can be obtained by searching the document library using the association ID as a key. This ID can be embedded in the base and standby stop planning URLs. The status can be initially set to "Initiated" and updated by the service as the process progresses. The schedule and corresponding profiles and settings can be included in the documents.

[0054] The parking space optimization service can be implemented as an API service. It can be configured to record status messages for creating, recording, and / or updating flight records and related documents. Such creation, recording, and updating can be performed by sending messages (e.g., status messages) to a database. The parking space optimization service can also be configured to create, record, and / or update events in an event logging database. Therefore, the parking space optimization service can create, record, and / or update events for each completed step of the parking space optimization process in the database.

[0055] Then, the parking space optimization service sends parking space optimization requests 1004 and 1006 to the parking space optimization model. For example... Figure 1 As shown, parking space optimization requests can be sent via message broker modules (1004, 1006) to the parking space optimization model. The message broker can be configured to enable asynchronous communication between different services, models, modules, adapters, interfaces, and / or systems. Alternatively, parking space optimization requests from the parking space optimization service can be sent directly via message broker modules (1004, 1006) to the parking space optimization model.

[0056] The following is an example message created by the parking space optimization service and sent to the parking space optimization model via a 10-message broker:

[0057] As shown in this example, the message may contain links to baseline parking stand planning (e.g., planning generated by the airport itself) and a list of priority settings for time segments of every 15 dates, also indicated in the message.

[0058] Despite Figure 1 In the example shown, messages are exchanged via a message broker; however, it should be understood that a message broker is an optional feature, and any other suitable method for exchanging messages can be used. For example, messages can alternatively be exchanged using HTTP / REST (request-response pattern).

[0059] As described above, message requests and responses can be stored in message memory 1154 for later analysis.

[0060] A stop position optimization request can be sent to stop position optimization model (“this model”) 1008. This model 1008 can be configured to be used as a container to hold various machine learning models. (Relative to...) Figure 3 This model 1108 is further described.

[0061] During this step, the model can generate one or more recommended parking positions to accommodate each flight. The model can use a baseline parking position plan as a starting point. Each set of recommendations for each of the one or more flights (“connecting flights” can be arrival / departure pairs) includes a list of parking positions and associated weights, initial parking positions, and predicted delayed flights.

[0062] The result generated according to this model 1108 can be sent to message broker ('this broker') 1106 at 1007. However, it should be understood that alternatively or additionally, the result generated according to this model 1108 can be sent directly to the parking space optimization service 1104.

[0063] The result can be sent as a message. An example of such a message is provided below:

[0064]

[0065]

[0066]

[0067] As shown in this example, the message can specify that the aircraft is associated with both arrival and departure flights (both of which can be seen in the link ID and two elements detailing the arrival and departure flights). In addition to the arrival / departure times, the message can also include the parking position assigned by the airport and the time slot allocation time (the time from when the aircraft arrives at the parking position until its departure time).

[0068] In addition, the message may include a list of arrays. In this case, each array may have a list of parking stands and scores. The arrays may be used in the following steps to calculate rules that can be sent to the parking stand planning service 1126. The parking stand planning service 1126 may be an airport management system (AMS), a part of it, or include the AMS.

[0069] The `departure_delay` element (valued at 0 in the example above) can contain the number of minutes the departing flight is predicted to be delayed by this model. This data can also be used in the following steps.

[0070] Parking space optimization service 1104 can consume 1008 recommendations generated by this model 1108 and map the scores to soft rules configured in parking space planning service 1126. Examples of such mappings are provided in Table 1. Table 1. Example mapping of scores to soft rules

[0071] Each flight can be mapped accordingly and the list can be published to this agent 1106.

[0072] This agent 1106 can receive 1009 AMS flight rules.

[0073] This agent 1106 can send data including recommended parking positions and scores to parking position planning service 1126 via adapter 1116.

[0074] The parking space planning service adapter ('this adapter') 1116 can consume parking space planning service flight rule messages and, iteratively, can generate an updateFlight() request including soft rules for each flight.

[0075] However, it should be understood that in some cases, such an adapter 1011 may not be necessary. For example, if an existing airport management system is configured to communicate directly with this agent 1106, this model 1108, and / or the parking stand optimization service 1104, then such an adapter 1116 may not be necessary.

[0076] Adapter 1116 can send an updateFlight() request for each flight.

[0077] Adapter 1116 can obtain 1012 optimized parking space planning from parking space planning service 1126.

[0078] Adapter 1116 and / or parking space planning service 1126 can publish optimization planning availability events to this agent 1106.

[0079] The parking space optimization service 1104 can consume event 1014. The fractional planning message can also be sent to this agent 1106 at 1005.

[0080] This model 1108 can consume 1016 score planning messages. A score plan can contain URLs for both the baseline and alternative plans. The score plan can score both the baseline and the alternative plans.

[0081] This model can send message 1017 to the parking space optimization service 1104, which has been rated for planning (e.g., via proxy 1106).

[0082] Regarding receiving the 1018 planning score message, the parking space optimization service 1104 can update the parking space planning optimization record to indicate that the parking space planning score has been received.

[0083] At this point, interface 1102 (e.g., Airport Director SPA) may have read the updated status and can obtain a summary of the parking stand planning. This summary can be displayed for the user's attention.

[0084] In addition, events can be sent by the parking position optimization service 1104 for consumption by the notification service 1112 to generate notifications and send them to the digital twin 1100.

[0085] The parking stand planning service 1126 may be or include an external parking stand allocation system associated with the airport and / or a rule-based parking stand allocation system associated with the airport. For example, the parking stand planning service 1126 may be or include an external rule-based parking stand allocation system with existing rules relating to parking stand allocation and other airport functions.

[0086] The following text is relative to Figure 3 The parking space optimization model 1108 described above may include multiple AI models (e.g., machine learning models). Therefore, it should be understood that the systems and methods according to various embodiments can be used in conjunction with one or more existing rule-based parking space allocation systems to provide one or more improved parking space allocation plans. It should also be understood that the systems and methods according to various embodiments may generate synergistic advantages due to the combination of AI models with existing rule-based parking space allocation systems. For example, combining AI models with existing rule-based systems can ensure that improved parking space allocation plans are provided according to priority, while ensuring that the improved parking space allocation plans do not conflict with existing rules. Furthermore, such combinations of AI models with rule-based systems can prevent or reduce the likelihood of typical drawbacks associated with AI model recommendations (such as AI bias amplification). In other words, if the AI ​​model proposes a flawed optimization plan, the rule-based system can act as a safeguard to ensure that such flawed plans are not implemented, thereby warning of potential negative consequences.

[0087] Figure 3 An exemplary parking space optimization model 1108 according to an embodiment is illustrated, wherein the parking space optimization model 1108 is implemented in the form of a prediction module 300. The data source 310 for the system may include historical records or live streams of any data type that may be directly or indirectly related to and affect transportation infrastructure (e.g., Figure 1 The operation and processes of an airport (as in the example). Providing more data sources 310 can advantageously enable the prediction module 300 to predict the impact of specific decisions on the provided data types (e.g., flight takeoff and landing data, retail revenue, CO2 consumption for each aircraft type, connecting flights, and missed flights and / or transfers). Therefore, as Figure 3 As shown, one or more data sources in data source 310 can be used to train one or more machine learning models 312 to predict 314 the consequences of one decision on another (i.e., assigning parking space A1 to flight 101 versus assigning parking space A2 to flight 101). Therefore, prediction module 300 can advantageously compute one or more decisions that satisfy selected criteria while ensuring that each decision has minimal negative impact on all other operations and processes in the environment. In other words, depending on the value selected for the priority of each metric, prediction module 300 can determine the globally optimized value that satisfies the selected priorities (optimizing all model predictions). For example, at the level of a one-way flight, where each decision could potentially affect passenger experience, passenger spending, satisfaction factors (such as missing a connecting flight or the familiarity of the route from one gate to another), and the quality of service provided, prediction module 300 can consider all relevant historical data and compute the impact of a single decision on all important criteria of interest to users (e.g., one or more operators and managers).

[0088] like Figure 3 As shown, the prediction module 300 can utilize multiple machine learning models 312 to predict 314 decisions (e.g., selection of landing runways, parking stands, or gates) for the output of all provided data types. For example, a first model 312-A can be used to predict how to allocate the impact of flights on retail sales based on historical data, considering factors such as aircraft type, airline, destination country, time of day / year, and passenger numbers. Additionally, a second model 312-B can be used to predict the impact of the same parking stand allocation on fuel consumption and electricity consumption or CO2 emissions. In such cases, if the airport wants to maximize retail revenue while minimizing CO2 emissions and is not concerned about other criteria such as delays, the prediction module 300 can accept all predictions and determine a plan that meets the selected values ​​for retail and CO2.

[0089] Optionally, the prediction module 300 can also enable adjustments to existing plans. For example, in the example above, if mitigating delays becomes more important at a later stage, the prediction module 300 can be used to adjust the plan. It should also be understood that the optimal plan generated from the prediction module 300 can be used to enhance existing manually implemented rules.

[0090] In some embodiments, the prediction module 300 can be configured to generate an optimal plan for maximizing positive impacts while minimizing negative impacts in a fully automated manner. However, such automatically generated optimal plans may not be consistent with airport preferences and / or existing rules. Therefore, the prediction module 300 can be set up or configured to receive input via an optimization priority panel 302, such as... Figure 3 As shown in the diagram, the optimization priority panel provides an interface through which users (e.g., one or more airport managers and / or operators) can select their preferences.

[0091] Optionally, the optimal plan can be sent to other existing systems (e.g., via an API) to ensure that the proposed optimal plan does not conflict with one or more rules of the existing systems. Optionally, the deployment of the proposed optimal plan can be delayed until it is determined that the proposed plan does not conflict with the rules of the existing systems. Optionally, if there is authorization to override the conflicting rules (received from the user; or retrieved from a database (such as a policy database), then the proposed optimal plan can be deployed despite the conflict with the existing systems.

[0092] As described above, the systems and / or methods described above can be used not only for parking space planning optimization, but also for optimizing one or more other aspects of infrastructure, such as transportation infrastructure (e.g., airports, seaports, railway stations, other transportation or logistics hubs or travel territories).

[0093] The following Examples 1 through 14 provide some examples of aspects that can be optimized using this system and / or method: Example 1. Security checkpoint opening and resource allocation: The system can analyze historical patterns, focusing on security check efficiency and passenger flow, to recommend the optimal allocation of security check lanes.

[0094] Example 2. Immigration Entry Opening and Resource Allocation: The system can assess historical data, focusing on immigration processing times and / or passenger volume, to recommend the most effective opening and resource allocation strategies for immigration checkpoints.

[0095] Example 3. Notifying passengers of their boarding gate time: Taking into account historical notification times to the gate, passenger route preferences, and / or operational constraints, the system can recommend precise planning to minimize wait times during boarding.

[0096] Example 4. Notifying baggage collection time: Taking into account historical data on baggage handling, passenger routes, and / or operational efficiency, the system can optimize the timing of baggage collection notifications, thereby ensuring timely delivery.

[0097] Example 5. Passenger Route: The system can analyze passenger preferences, historical route patterns, and / or real-time operational constraints to suggest precise planning, thereby minimizing missed transfers and improving overall satisfaction.

[0098] Example 6. Baggage carousel allocation: The system can assess historical baggage data, flight schedules, and / or passenger flow to recommend the most efficient baggage carousel allocation, thereby reducing congestion.

[0099] Example 7. Transfer resource allocation: Taking into account historical transfer data, passenger flow patterns, and / or operational needs, the system can recommend precise resource allocation for transfer areas, thereby ensuring effective passenger movement.

[0100] Example 8. PRM resource allocation: The system can analyze historical data about passengers with mobility impairments (PRMs) and recommend the best resource allocation to provide necessary assistance.

[0101] Example 9. Resource Allocation Check-in / Reception Area: The system can take into account historical check-in data, passenger volume, and / or operational efficiency to recommend the best resource allocation for check-in / reception areas.

[0102] Example 10. Energy Management: This system can assess historical energy consumption patterns, operational plans, and environmental considerations to optimize energy management.

[0103] Example 11. Cleaning Plan: By analyzing historical lounge usage patterns and operational plans, the system can recommend effective floor, restroom, and equipment cleaning plans.

[0104] Example 12. Waiting room resource allocation: Taking into account historical waiting room utilization rates, passenger demographics, and service levels, this system can optimize the allocation of resources for airport waiting rooms.

[0105] Example 13. Resource Allocation in Amusement Areas: The system can analyze historical data on amusement park usage, passenger demographics, and operational constraints to recommend optimal resource allocation.

[0106] Example 14. Retail Inventory Management: By estimating historical retail sales data, passenger preferences, and inventory levels, the system can optimize retail inventory management.

[0107] Compared to Figure 1The above embodiments illustrate how the system and / or method according to the present invention can be used to optimize parking stand planning at airports. However, considering various other use cases, such as Examples 1 to 14 above, relative to... Figure 8 Examples of a broader perspective of the invention are described.

[0108] Figure 8 An exemplary system according to an embodiment is shown. Figure 8 A schematic diagram. In this example, a user (e.g., an airport stakeholder, operations director, manager, or other stakeholders in one or more areas of the airport) 852 and / or an automatic trigger 804 can initiate an orchestrated workflow 812. When a user initiates an orchestrated workflow 812, the initiation can be performed via a user interface 802, as described above. The orchestrated workflow 812 can be initiated to optimize one or more aspects of the infrastructure.

[0109] The machine learning model 806 can be deployed for one or more aspects of the infrastructure, feature service 820, and / or service 830. Optionally, it can be integrated. Figure 8 One or more of the functions shown. For example, in Figure 1 In the previous example shown, the parking space optimization service 1104 can function as both the orchestrator 812 and the feature service. However, it should be understood that... Figure 1 The example can be modified to allow the two functions (orchestration 812 and feature service 820) to be performed separately (e.g., physically separate).

[0110] Figure 9 With local process Figure 9A and Figure 9B The diagram illustrates a flowchart of an example optimization process. Figure 9 The relative positions of the local flowcharts are provided. For the purposes of discussion, we will now use combinations... Figure 9A and Figure 9B The flowchart shown in the partial flowchart is called Figure 9 .

[0111] Figure 9 The simplified example end-to-end process is described at a high level, without taking into account automatic triggering mechanisms 804 or asynchronous notifications to the digital twin 1100 (e.g., the digital twin of an airport).

[0112] exist Figure 9In the example, user (e.g., an operations stakeholder or other stakeholder) 852 can set priorities on one or more aspects of the infrastructure. It should be understood that in some cases, a single key stakeholder can be responsible for prioritizing the entire array of key performance indicators, while in other cases, multiple staff members can be responsible for different aspects of the site. The Airport Director's user interface 802 can be configured to enable users to initiate planning requests.

[0113] When the automatic trigger 804 (not in) Figure 9 When the workflow 812 (shown in the diagram) is used to initiate orchestration, the automatic trigger 804 can, for example, be configured to execute time-initiated processes. An example of the automatic trigger 804 is provided below: The optimization process should run at midnight each day to generate the schedule to be executed that day.

[0114] The optimization process should be run at 6 p.m. the day before the operation to allow managers sufficient time to study the new plan and adjust their production system.

[0115] The optimization process runs every 10 minutes to handle new input data and adapt to changes in real-time conditions.

[0116] Therefore, API 803 or automatically triggered request 804 can initiate orchestration process 812. Orchestration 812 can control the processing flow and the status in progress. This can be supported, for example, by one or more delivery protocols (e.g., callbacks (webhooks), synchronous API requests / responses, send and forget / store and forward).

[0117] Arrangement 812 can trigger: Search for current airport plans; Recommendations are generated by machine learning models; Infrastructure systems (such as airports) optimize current planning by using recommendations mapped to specific configurations or rules adopted by that system; and The airport planning and optimization plan is scored.

[0118] The airport director 802 can then monitor the process status and, when the process is complete, obtain a score and display it to the user (e.g., by presenting the score on a screen accessible to the user).

[0119] It should be understood that, relative to the example of optimized parking position, adapter 1116 and / or agent 1106 (such as...) Figure 1 (As shown in the figure) can be used to optimize aspects other than parking space allocation.

[0120] Therefore, the systems and methods according to various embodiments combine machine learning and optimization models for predictive analytics to recommend and implement operational plans that are consistent with specific standard rates, thereby providing comprehensive and intelligent solutions for optimizing travel and transportation operations.

[0121] Furthermore, systems and methods according to various embodiments can enable control of one or more portions of an infrastructure to execute an optimization plan generated by the system. For example, the system may include or be coupled to a transmitter for sending instructions for one or more of the following: controlling an aircraft to a parking position according to a second plan; controlling the allocation of security checkpoints according to a second plan; controlling the opening and / or allocation of resources in one or more areas of the infrastructure (e.g., allocating staff and / or facilities for immigration arrival, transit, baggage carousel, PRM assistance, check-in, play areas, and / or waiting areas) according to a second plan; controlling the timing of one or more information announcements according to a second plan (e.g., informing passengers of the time to proceed to the gate and / or the time to collect baggage); controlling routes for one or more users of the infrastructure according to a second plan; controlling energy use in one or more areas of the infrastructure according to a second plan; controlling the planning and / or allocation of resources for maintenance and / or cleaning programs according to a second plan; and / or managing retail inventory.

[0122] Such instructions may be sent, for example, according to an optimized plan, to one or more corresponding devices and / or vehicles for automated control. Such instructions may also be sent, for example, to one or more screens, projections, or audio devices that draw attention to and / or prompt action from one or more users, such as infrastructure personnel (e.g., airline staff (pilots, flight attendants, and / or ground personnel), airport staff, and / or passengers).

[0123] Because each decision can have numerous consequences, finding the globally optimal plan can be challenging in some situations. Therefore, systems and methods for evaluating the dynamic relationships between different factors are desired, allowing for more informed and strategic optimization of the entire system.

[0124] In view of this, the systems and methods according to some embodiments can utilize graph-based optimizations, as described below.

[0125] Proposed optimization algorithm: Interconnecting various data sources: In this step, the system uses any correlations that may exist in the tabular data to associate any relevant information with the initial data. These various data sources may include timestamps, physical location IDs, departure and arrival locations, gate numbers, vehicle and equipment tags or aircraft IDs, terminal codes, transfer details, and the system uses these IDs to create a large matrix that includes all relevant information.

[0126] Filtering less important data: Depending on the target criterion to be optimized, the system selects from a set of defined or learnable thresholds for each data type. As the target value is selected, the system removes the least relevant data and retains the most important relevant data, processing this data to achieve the maximum gain on the target value. For example, if the goal is to optimize passenger transfer risk, the system filters any irrelevant data based on criteria such as whether a passenger has no transfers or whether a passenger has transfers but with sufficient time between their flights exceeding a certain threshold.

[0127] Graph processor: Extracting repairable nodes based on minimum / maximum thresholds involves the system extracting rows of data that can be technically repaired, such as those connecting flights with a minimum time of X minutes between arrival and departure. The minimum minutes depend on the walking time from point A to point B, and the size and walking time in travel or transportation hubs (airports, train stations, bus terminals, seaports, space centers, or intermodal terminals) where the solution applies to this threshold can be automatically calculated using computer vision and people tracking based on the average walking time between every two points.

[0128] The system assigns a distinct ID to each data point, and this labeled ID is used to highlight relevant data points. This unique label then helps filter out data with minimal or no relevance. In addition to the current optimization criteria, the method introduces additional edges into the graph to represent interconnected data that might be affected by modifications to specific data points. In the final step, transfers that cannot be disconnected are filtered out. For example, in the context of risky transfers, passengers with sufficient buffer time for transfers exceeding the maximum walking distance between their arrival and departure gates are excluded.

[0129] In the next step, the graph processor creates the graph structure by assigning nodes to each unique ID. Edges on the graph are meaningful associations between nodes (e.g., walking time between node A and node B). In some graphs, edges may contain more summary information about the data, such as the total number of passengers between flight 100 and flight 200, the minimum / maximum time for connecting passengers, or other information to provide better service based on age, gender, disability, or other factors. In real-world applications, there are constraints in place that limit the options within the system, preventing the selection of any arbitrary planning that might maximize performance. For example, some boarding bridges may be limited by aircraft size, or specific gates may be designated for specific flights. Therefore, the system establishes fixed nodes, either by directly implementing rules or by leveraging software with embedded rules, which cannot be relocated. If a fixed flight is associated with seven connecting (edges), the system creates seven new nodes by adding "_index" to the link_id with a unique identifier. Connectings are then established between these newly indexed nodes and their neighboring nodes from the initial graph, after which the initial fixed nodes are removed. Next, nodes with a degree (number of connected edges) greater than or equal to the `SPLIT_MIN_DEGREE` parameter are selected. The next step involves traversing all possible node degrees between the maximum degree and `SPLIT_MIN_DEGREE` in descending order. Assuming the maximum degree is 10, this process allows the graph to split into nodes of degree 10, followed by nodes of degree 9, and so on. The splitting process mirrors the number of artificial nodes corresponding to their degrees. For example, a node of degree 10 results in the creation of 10 subgraphs associated with a specific flight. Finally, the initial nodes are removed from the graph.

[0130] Following the above process, components with weak connections that consist only of fixed nodes are further excluded from the graph. This exclusion is necessary because optimizing the layout becomes impractical when all flights within the graph are fixed and cannot be repositioned as alternative parking positions. Subsequently, the system implements a function to assign weights to the graph, a crucial step for visualization. This function enhances visualization by proportionally increasing the thickness of lines on the graph based on relevant data, such as the number of connecting passengers. In the final step, the initial graph containing multiple connected components is divided into a series of smaller graphs. Each of these smaller graphs serves as input data for the optimization algorithm.

[0131] Optimizer: This step operates on the output from the graph processor, which consists of a series of small, independent graphs. The process requires traversing each graph one by one. Within each graph, the goal is to address all potential problems in optimizing the graph, ensuring that no edges have below-expected standards and avoiding or minimizing negative impacts on other edges in the graph. For example, in the context of risky transfers, edges define the time between two flights, and the aim is to make all risky transfers risk-free while maintaining transfers that might become risky due to gate changes as risk-free. The optimization process begins by strengthening the graph by adding a new column indicating the longest permissible value for the edges (e.g., the maximum time between gates for each transfer), thus ensuring flight safety. It then removes any transfers with insufficient values ​​(such as the time needed to fix all possible gate configurations), indicating negative walking times. After this, it assigns initial infrastructure values, such as gates, to all fixed flights to prevent them from moving. For each transfer in the graph, the system calculates all possible configuration pairs that would cause the value to fall below a threshold, starting with the nearest pair and moving to more distant pairs.

[0132] There are two optimization approaches. The first approach focuses on a single graph at a time. It begins by identifying transfers that can be improved (e.g., not fixed, carrying passengers, and with sufficient time for gate-to-gate transfers). It then assigns initial infrastructure values ​​(e.g., gates) to all transfers and generates transfer combinations to optimize, starting with all transfers and gradually eliminating others if necessary. For example, let's name three transfers A, B, and C. The system initially attempts to optimize all three transfers (A, B, C) together. If this is unsuccessful, it attempts combinations such as (A, B), (A, C), (B, C), and finally (A), (B), and (C). If no solution is feasible, it concludes that optimization for that set of nodes (e.g., flights) is impossible. Therefore, this process only applies to transfers that can be optimized in pairs.

[0133] The second optimization method is based on a recursive depth-first search (DFS) algorithm. It starts with the first transfer and considers all possible gate pairs with a distance of one. For subsequent transfers, it filters out gate pairs used in previous steps to avoid conflicts. The goal is to minimize the overall boundary values ​​between nodes (e.g., walking time between gates) for all passengers.

[0134] Example: For instance, if the first connecting flight obtains gate pair A1-A2, the second connecting flight cannot use these two gates unless its arrival or departure coincides with that of the first connecting flight. Therefore, the system passes a `chosen_gates` structure to the next optimization step, including the newly selected gate pair for the two flight segments of a given connecting flight. It then filters out any gate pairs used in previous optimization steps. This ensures that the same gate is not assigned to two different flights, and that the parking positions already obtained by the fixed flight remain unchanged. The fixed flight should maintain its assigned parking position without change. Additionally, the algorithm ignores solutions with shorter walking distances than the shortest solution found so far, thus speeding up computation by avoiding unnecessary searches. Further exploration stops if the total walking distance of the currently selected gates exceeds the total walking distance of the best solution found.

[0135] Code explanation: prepare_nodes: Nodes for preparing the graph based on unique inbound and transit flights.

[0136] create_graph: Creates a graph based on the input DataFrame.

[0137] visualize_graph: Use the Pyvis library to visualize graphs.

[0138] get_nodes_by_degree: Retrieves nodes by a degree equal to or greater than a specified degree.

[0139] split_components_by_degree: Split components in the graph based on node degree.

[0140] `sort_helper`: A helper method for sorting a DataFrame based on flight details. `_format_node_id`: Formats the node_id when using `link_id`.

[0141] split_grouped_connections: Split and group connections based on the processed graph.

[0142] get_fixed_nodes: Retrieves fixed nodes from the graph.

[0143] fix_nodes: Fix nodes in the graph based on given attributes.

[0144] remove_fixed_components: Removes components from a graph where all nodes are fixed.

[0145] remove_unfixable_components: Remove components from a graph where all edges have the minimum negative time between flights.

[0146] Using generative AI and large language model enhancement systems (LLM): The system can leverage Large Language Models (LLMs) based on generative AI. These models help analyze complex data and make decisions based on that data. The system's purpose is not merely to handle route maintenance tasks, but also to make resource allocation more efficient, meaning to improve how things operate by using resources such as time, staff, and equipment in the most efficient way possible.

[0147] By using LLM (Limited Management System), the system can analyze different types of data, such as maintenance records, weather information, and flight patterns. This helps to predict potential problems with aircraft and airport facilities before they occur. For example, it can help with planned maintenance in advance to avoid unwanted issues and ensure the smooth operation of established processes, as well as managing things like gate assignments, check-in counters, and baggage handling areas at the airport. This approach reduces wasted time on idle aircraft, alleviates congestion at gates, and makes overall airport operations smoother. To manage flight crews, the system uses intelligent algorithms that consider crew fatigue levels, skill levels, and rules that need to be followed. This helps ensure that crews are assigned to flights in the most efficient way possible while still meeting all necessary regulations.

[0148] To help protect the environment, the system uses LLM (Limited Learning Model) to learn from other areas to make flight routes more fuel-efficient, thereby reducing carbon emissions. This helps to increase investment in making air travel more sustainable. By using LLM, the system can develop better plans to reduce flight delays and other disruptions, making passengers happier. It also works to optimize things like gate assignments and airport passenger flow to make the entire travel experience smoother. The system also includes a decision support system that uses data to help people make better decisions. This may involve identifying trends, identifying areas for improvement, and predicting future challenges. Overall, the system aims to move from reacting to problems using a single AI model to proactively preventing problems and helping to protect the environment by using the comprehensive knowledge of LLM, aiming to improve operational planning and processes, enhance the passenger experience, and provide better information support for decision-makers.

[0149] Compared to Figures 4 to 7Examples of graph-based algorithms applicable to various types of data are described. A graph with a large number of nodes of a given size may not be able to clearly represent all the important nodes. Therefore, one or more important things may be overlooked. Furthermore, solving such a large graph may be computationally difficult, time-consuming, and / or practically impossible (e.g., it may take thousands of years to solve). Graph-based algorithms according to some embodiments of the present invention can provide a solution to this problem by providing a simplified graph.

[0150] Figure 4 (a) shows an example initial graph consisting of Passenger Transfer Messages (PTMs) for a given day. In this example, nodes are the flights and edges that indicate the presence of connecting passengers between them.

[0151] In this example, the algorithm first prioritizes intercepting flights with multiple transfers, as these flights pose the greatest challenge, such as... Figure 5 As shown in the example. For simplicity, in this example, the algorithm prevents aircraft from moving from their current positions, and specifically fixes them in their positions. This allows the algorithm to decompose a large flight network into smaller, more manageable parts centered around these fixed points. Nodes are then processed in descending order of splitting degree, as shown in the example. Figure 6 As shown (i.e., from the highest degree of splitting to the lowest degree of splitting (e.g., in...) Figure 6 (In the example, from split degree 9 to split degree 4). These steps constitute the intermediate stages of the process.

[0152] Figure 7 Examples of diagrams are shown: (a) a diagram with no fixed components and (b) a diagram with no non-fixed components. The former refers to a diagram in which fixed elements have been considered but the final configuration has not been determined, while the latter indicates a diagram in which potential changes deemed impossible have been removed to mitigate risk. Figure 4 (b) shows an example of the final diagram.

[0153] Therefore, it should be understood that the systems and methods according to various embodiments can utilize pre-trained machine learning models as the basis for multiple aspects within the system, such as predicting the impact of decisions on standards such as revenue, fuel consumption, and environmental impact. The system can fine-tune the model with new data based on schedules (e.g., daily, weekly, monthly) and / or any time difference between predicted and actual values ​​exceeding a specific threshold to ensure adaptability and accuracy in reflecting the ever-changing nature of travel and transportation operations.

[0154] It should also be understood that the systems and methods according to various embodiments can implement graph-based solutions within and / or in combination with systems to identify and quantify correlations between different criteria. By constructing graphs representing the relationships and differences between various factors, the system can intelligently determine the most effective relationships. This approach can improve decision-making by providing a visual representation of the complex interconnections in a centralized dataset.

[0155] It should also be understood that the systems and methods according to various embodiments can be designed with a modular architecture that separates the prediction and optimization phases. Such a design enables the system to utilize machine learning models to predict the outcomes and impacts of different plans under specific criteria, and to provide an optimization processing model that selects the most suitable plan based on a predetermined standard rate. This separation allows for faster and more robust planning, thereby improving the overall efficiency of the system.

[0156] It should also be understood that the systems and methods according to various embodiments can be used to optimize various aspects of infrastructure independently or in conjunction with each other. Therefore, the system can allow users to prioritize specific criteria or consider multiple criteria simultaneously based on operational needs. This adaptability allows the system to respond to a variety of scenarios, thereby providing tailored solutions to the different challenges faced by travel and transportation customers.

[0157] It should also be understood that the systems and methods according to various embodiments can utilize continuous learning mechanisms by capturing the difference between predicted and actual values ​​over time. For example, when the difference between a prediction and an actual result exceeds a predefined threshold, the system can trigger an update to its machine learning model. This ensures that the model evolves and adapts to changing conditions, thereby improving its accuracy and responsiveness over time. Such continuous learning can also enhance the system's ability to make more informed predictions when it encounters new data and situations.

[0158] Therefore, systems and methods according to various embodiments can provide dynamic and adaptive solutions that leverage advanced machine learning techniques, along with one or more graph-based analytics, modular design, and continuous learning, to optimize various aspects of travel and transportation operations. The synergy of these components contributes to a complex and intelligent framework that addresses the limitations of traditional decision-making processes in various infrastructures, such as transportation infrastructure.

[0159] Communication between two or more components of the aforementioned system may be performed via one or more wired and / or wireless communication networks. One or more storage devices and / or systems are part of or connected to the item processing system, and these storage devices and / or systems may be remote storage devices or systems, such as cloud-based storage devices or systems.

[0160] The aforementioned wired or wireless communication networks can be public, private, wired, or wireless networks. Communication networks can include one or more of a local area network (LAN), a wide area network (WAN), the Internet, a mobile phone communication system, or a satellite communication system. Communication networks can include any suitable infrastructure, including copper cables, fiber optic cables or optical fibers, routers, firewalls, switches, gateway computers, and edge servers.

[0161] The system described above may include a graphical user interface (GUI). Embodiments of the invention may include an on-screen GUI. The GUI may be provided, for example, as a desktop app embedded in a website, an application as a device, or on a dedicated landing page. Computer-readable program instructions for implementing the GUI may be downloaded from a computer-readable storage medium to the client device via a network (e.g., the Internet, a local area network (LAN), a wide area network (WAN), and / or a wireless network). The instructions may be stored in a computer-readable storage medium within the client device.

[0162] As will be understood by those skilled in the art, the invention described herein can be embodied in whole or in part as a method, a data processing system, or a computer program product including readable computer instructions. Therefore, the invention can take the form of an entire hardware embodiment or an embodiment combining software, hardware, and any other suitable means or means.

[0163] Computer-readable program instructions may be stored on a non-transitory tangible computer-readable medium. Computer-readable storage media may include one or more of the following: electronic storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, portable computer floppy disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact optical disc read-only memory (CD-ROM), digital versatile optical disc (DVD), memory sticks, and floppy disks.

[0164] An exemplary embodiment of the present invention can be implemented as a circuit board, which may include a CPU, a bus, RAM, flash memory, one or more ports for operating connected I / O devices (such as printers, displays, keyboards, sensors, and cameras), ROM, a communication subsystem (such as a modem), and a communication medium.

Claims

1. A method for optimizing parking space planning, comprising the following steps: a. Receive one or more priority parameters to be considered in the parking space planning. Each of the priority parameters includes a weight and is associated with one or more airport functions; b. Receive the existing first parking position plan from the existing first parking position planning and optimization system. The first parking space planning includes data related to flights and their allocated parking spaces; c. The second parking space planning optimization system generates a recommended second parking space plan. The second parking space plan is generated by modifying the first parking space plan so that a prediction module including one or more models can be used to allocate parking spaces to the flight according to the priority parameter. d. Send the second parking space plan from the second parking space planning optimization system to the first parking space planning optimization system; as well as e. Update the first parking space plan based on the second parking space plan.

2. The method as described in claim 1, The one or more existing rules related to the one or more airport functions mentioned above were obtained prior to the step of generating the second parking stand plan, and The second parking space plan is generated to avoid conflict with the existing rules.

3. The method as described in claim 1 or 2, The one or more existing rules related to the one or more airport functions are obtained from an external database associated with the airport.

4. The method as described in any of the preceding claims, wherein the one or more priority parameters are obtained from a rule-based database associated with the airport.

5. The method as described in any of the preceding claims, wherein the one or more priority parameters are obtained via user input or from a database.

6. A method for prioritizing one or more priority parameters in the method as described in any one of claims 1 to 5, wherein the weight of each priority parameter can be dynamically adjusted by a user.

7. The method of claim 6, wherein during the adjustment of the weights, the user is shown changes to one or more airport functions corresponding to the parameters.

8. The method of claim 6 or 7, wherein changes in one or more airport functions corresponding to the parameters are displayed in the form of sub-parameters with numerical values.

9. The method as described in any of the preceding claims further comprises the following steps: f. Send instructions to control the aircraft to the parking position according to the second plan.

10. A system for optimizing parking space planning, comprising: a. A receiver, the receiver being used for Receive one or more priority parameters to be considered in the parking space planning. Each priority parameter in the aforementioned priority parameters includes a weight and is associated with one or more airport functions, and Receive the existing first parking position plan from the existing first parking position planning and optimization system. The first parking space planning includes data related to flights and their allocated parking spaces; b. A processor, the processor being used to generate a recommended second parking position plan from the second parking position planning optimization system. The second parking space plan is generated by modifying the first parking space plan so that a prediction module including one or more models can be used to allocate parking spaces to the flight according to the priority parameter. as well as c. A transmitter for sending the second parking space plan from the second parking space planning optimization system to the first parking space planning optimization system for updating the first parking space plan based on the second parking space plan.

11. A prediction module, said prediction module being used in the method of any one of claims 1 to 9 or in the system of claim 10, The prediction module includes multiple machine learning models. Each machine learning model in the aforementioned machine learning models is connected to one or more data sources and is trained to perform predictions on one or more consequences of one or more decisions based on historical data from said one or more data sources. The one or more consequences predicted by the machine learning model are related to multiple airport functions.

12. The prediction module of claim 11, wherein the second parking space plan is selected from one or more candidate second parking space plans that satisfy the priority parameter based on the consequences of the one or more decisions and the predictions made according to the machine learning model.

13. The prediction module of claim 11 or 12, wherein the second parking space planning is associated with the least negative consequences predicted according to the machine learning model.

14. A computer program product that, when executed, performs the method as described in any one of claims 1 to 9.

15. A method for optimizing infrastructure, comprising the following steps: a. Receive one or more priority parameters for consideration in optimization. Each of the priority parameters is associated with a target aspect of the infrastructure that is related to one or more functions of the infrastructure, and includes a weight. b. Receive the existing first-optimization plan from the existing first-optimization system. The first optimization includes data related to the objective aspect; c. The second optimization system generates a recommended second optimization plan. The second optimization program is generated by modifying the first optimization program so that a prediction module including one or more models can be used to optimize the target aspect according to the weights. d. Send the second optimization plan from the second optimization system to the first optimization system; as well as e. Update the first optimization plan based on the second optimization plan.

16. The method of claim 15, One or more existing rules relating to the one or more functions of the infrastructure were obtained prior to the step of generating the second optimization plan, and The second optimization plan is generated to avoid conflict with the existing rules.

17. The method as described in claim 15 or 16, The one or more existing rules related to the one or more functions of the infrastructure are obtained from an external database associated with the infrastructure.

18. The method of any of the preceding claims, wherein the one or more priority parameters are obtained from a rule-based database associated with the infrastructure.

19. The method of any of the preceding claims, wherein the one or more priority parameters are obtained via user input or from a database.

20. A method for prioritizing one or more priority parameters in the method of any one of claims 15 to 19, wherein the weight of each priority parameter can be dynamically adjusted by a user.

21. The method of claim 20, wherein during the adjustment of the weights, the user is shown changes to one or more functions of the infrastructure corresponding to the parameters.

22. The method of claim 20 or 21, wherein changes in one or more functions of the airport corresponding to the parameter are displayed in the form of sub-parameters with numerical values.

23. The method as described in any of the preceding claims further comprises the following steps: f. Send instructions for any one or more of the following: According to the second plan, control the aircraft to the parking position; The allocation of security checkpoints is controlled according to the second plan; The second plan controls the opening and / or allocation of resources in one or more areas of the infrastructure; The second planning control is used for the timing of one or more information announcements; The second plan controls routes for one or more users of the infrastructure; Energy use in one or more areas of the infrastructure is controlled in accordance with the second plan; and / or The second planning control governs the planning and / or allocation of resources used for maintenance and / or cleaning programs.

24. A system for optimizing infrastructure, comprising: a. A receiver, the receiver being used for Receive one or more priority parameters for consideration in optimization. Each of the priority parameters is related to a target aspect of the infrastructure associated with one or more functions of the infrastructure, and includes a weight. Receive the existing first-optimization plan from the existing first-optimization system. The first optimization includes data related to the objective aspect; b. A processor, the processor being used to generate a recommended second optimization plan by the second optimization system. The second optimization program is generated by modifying the first optimization program so as to use a prediction module including one or more models to optimize the target aspect according to the weights; and c. A transmitter for transmitting the second optimization plan from the second optimization system to the first optimization system for updating the first optimization plan based on the second optimization plan.

25. A prediction module, said prediction module being used in the method of any one of claims 15 to 23 or in the system of claim 24, The prediction module includes multiple machine learning models. Each machine learning model in the aforementioned machine learning models is connected to one or more data sources and is trained to perform predictions on one or more consequences of one or more decisions based on historical data from said one or more data sources. The one or more consequences predicted by the machine learning model are related to multiple functions of the infrastructure.

26. The prediction module of claim 25, wherein the second optimization program is selected from one or more candidate second optimization programs that satisfy the priority parameter based on the one or more decisions and the consequences predicted according to the machine learning model.

27. The prediction module of claim 25 or 26, wherein the second optimization program is associated with the minimum negative consequences predicted according to the machine learning model.

28. A computer program product that, when executed, performs the method as described in any one of claims 15 to 23.

Images