Airport shuttle bus optimization scheduling method and device based on improved capacity network model considering driver status

Abstract

The invention discloses an airport shuttle bus optimization scheduling method and device of an improved capacity network model considering the driver's state, wherein the method comprises: determining the flights that need shuttle bus demand according to the real-time information data of airport flights, setting the departure and return positions of the flights and the shuttle bus as nodes, setting the path of the shuttle bus's continuous service between the two nodes as directed edges, and constructing a novel capacity network model; at the same time, detecting the driving state of the shuttle bus driver in real time through the camera in the cockpit, and recalling the driver with abnormal driving state in time to stop the task of picking up and dropping off passengers; constructing a mixed integer programming model based on the capacity network with the minimum required number of shuttle buses as the goal, and setting the real-time driving state of the driver as a constraint condition, and equivalently converting it into a linear programming, and solving the minimum number of airport shuttle buses that ensure the safety of scheduling and the optimal path for scheduling between flights. The invention creatively constructs a capacity network, takes the driver's state as a reference factor for the scheduling plan, and completes the optimal scheduling of airport shuttle buses efficiently and at low cost, thereby ensuring the safety of the airport, reducing the airport operation cost, and improving the airport operation efficiency.

Description

Technical Field

[0001] The invention relates to the field of airport special vehicle optimization dispatching method, and in particular to an airport shuttle bus optimization dispatching method and device of an improved capacity network model taking driver status into consideration. Background Art

[0002] In recent years, with the accelerating process of economic globalization, the aviation transportation industry has developed rapidly. At the same time, the pressure on airport transportation has also increased. For example, according to the Airport Council International, the global airport passenger throughput in 2022 exceeded 6.6 billion, an increase of 43.8% over 2021. Among them, the top 20 airports in the world in terms of passenger throughput accounted for 18% of the total throughput, reaching 1.2 billion. As a transportation hub and service node, the operating efficiency of the airport is directly related to the smooth operation of the aviation industry and the travel experience of passengers. Therefore, airport optimization and scheduling has become an important strategic measure to improve the efficiency of the overall aviation system, reduce costs, increase profits, and improve service quality.

[0003] In the daily operation of the airport, the insufficient number of airport shuttle buses and improper scheduling are the main reasons for flight delays and affecting the efficiency of airport operations. According to statistics, in addition to uncontrollable factors such as weather, flight delays caused by improper scheduling by airlines account for 34% of the total. It can be seen that the scheduling efficiency of the airport directly affects the overall operating efficiency of the airport. The scheduling of shuttle buses is one of the most important links in airport scheduling. At the same time, the driving status of airport shuttle bus drivers also greatly affects the scheduling efficiency of airport shuttle buses. The driver's abnormal driving behavior (such as blinking, yawning, answering and making phone calls, etc.) also greatly threatens the safety of airport operations. Therefore, an efficient airport shuttle bus optimization scheduling solution that takes into account the driver's driving status can greatly reduce flight delays, improve the capacity utilization of the entire airport, and improve passenger satisfaction.

[0004] At present, the dispatch of airport shuttle buses is mostly completed by manual operation. Many complex tasks such as flight schedule arrangement, airport special vehicle arrangement and passenger service usually require dispatchers to monitor, coordinate and make decisions in real time. Manual dispatch can usually complete the scheduling tasks, but there are certain problems. 1) When faced with a large amount of information and complex decisions, manual dispatchers may be limited by processing capacity and find it difficult to effectively cope with high-intensity scheduling tasks, which may easily lead to information omissions and errors. 2) When encountering emergencies and requiring quick decisions, the response time of manual dispatch is relatively long and may not be able to meet urgent scheduling needs, resulting in flight delays and other operational problems. 3) Manual dispatchers have fixed working hours, while airport operations are usually 24 hours a day. Long working hours may cause fatigue to manual dispatchers, resulting in human errors such as misunderstanding of information and input errors, thereby affecting the accuracy and efficiency of decision-making. 4) The dispatch of airport shuttle buses often ignores the impact of the driving status of the shuttle bus driver on the safety of airport dispatch. The driver's abnormal driving behavior poses a great safety hazard to the safety of the airport. Summary of the invention

[0005] The present invention aims to overcome the above-mentioned shortcomings of the manual dispatching method and provides an airport shuttle bus optimization dispatching method and device based on an improved capacity network model taking into account the driver's status.

[0006] Aiming at the problems of frequent errors and low efficiency in manual dispatching at airports, the present invention, with the help of artificial intelligence and big data analysis technology, uses advanced optimization algorithms and models, takes the driving status of shuttle bus drivers as the key factor in shuttle bus dispatching, and proposes an airport shuttle bus optimization dispatching method based on an improved capacity network model that considers the driver's status. On the basis of the capacity network, a mixed integer programming model with the goal of minimizing the number of shuttle buses is constructed, and the driver's driving status is used as a constraint condition to complete the optimal dispatching of airport shuttle buses efficiently and at low cost, reduce airport operating costs, and improve airport operating efficiency.

[0007] The airport shuttle bus optimization scheduling method of the present invention determines the flights that need shuttle bus demand based on the real-time information data of airport flights, sets the departure and return positions of the flights and the shuttle bus as nodes, and sets the path of the shuttle bus's continuous service between the two nodes as directed edges, thereby constructing a novel capacity network model; at the same time, the driving status of the shuttle bus driver is detected in real time through the camera in the cockpit, and the driver with abnormal driving status is recalled in time to stop the task of picking up and dropping off passengers; based on the capacity network, a mixed integer programming model is constructed with the minimum required number of shuttle buses as the goal, and the real-time driving status of the driver is set as a constraint condition, and it is equivalent to a linear programming, and the minimum number of airport shuttle buses that ensure scheduling safety and the optimal path for scheduling between flights are solved.

[0008] The technical solution of the present invention is:

[0009] A first aspect of the present invention relates to an airport shuttle bus optimization scheduling method based on an improved capacity network model taking into account driver status, comprising the following steps:

[0010] Step 1: Airport real-time information data collection. According to the actual real-time situation of the airport, determine the departure location, end location, flight location that needs shuttle service, and the location distance relationship between each location, and calculate the earliest time and latest time that flight i is allowed to be served by shuttle service. The process is as follows:

[0011] Step 1.1 collects the real-time situation of the airport. The data collected by the present invention are as follows:

[0012] a) Shuttle bus departure and terminal location: The shuttle bus departs from the departure station, picks up passengers at the destination flight location, and then returns to the terminal. The distance between the departure station and the terminal will affect the time of the shuttle bus to and from the flight, so it is necessary to arrange and dispatch it reasonably based on the location distance information.

[0013] b) Estimated arrival time and estimated departure time of flight: Since the shuttle bus needs to serve the scheduled flights according to the flight demand, for flight i that needs shuttle bus service, its estimated arrival time (STA i )、Scheduled departure time(STD i ) and the flight stop location.

[0014] c) The time required for the shuttle bus to travel between flights: Since a shuttle bus usually needs to complete the task of picking up and dropping off passengers for multiple flights, and the time spent between flights will affect the start time of the next flight service, the present invention takes the time required for the shuttle bus to travel between flights into account, so that the takeoff of the next flight is not affected by the previous flight.

[0015] d) The earliest waiting time T1 and the latest waiting time T2 of the shuttle bus before the scheduled arrival of the flight: According to regulations, the shuttle bus should wait at the berth in advance before the flight arrives to avoid delaying passengers' boarding and flight takeoff.

[0016] e) Required service time t ii : In order not to affect the take-off of flights and complete the shuttle bus's task of picking up passengers in time, the shuttle bus's mission duration for picking up flights should also be taken into account to make the shuttle bus's scheduling more reasonable and efficient.

[0017] Step 1.2 Calculate the earliest time e that the shuttle service is allowed for flight i based on the collected information data i and the latest time i The calculation process is as follows:

[0018] For flight i, the earliest time e at which the shuttle service is allowed i The calculation formula is:

[0019] e i =STD i -T2(1)

[0020] The latest time shuttle service is allowed i The calculation formula is:

[0021] l i =STA i -T1(2)

[0022] Step 2: Establishment of the capacity network model. The present invention sets the departure position, end position, and flight position of the shuttle bus as nodes, sets the position distance between nodes and the relationship between the shuttle bus's travel time between flights as scheduling arcs, and generates a real-time directed acyclic graph, i.e., a capacity network, through an algorithm using real-time data. The process is as follows:

[0023] Step 2.1 Determine the location and number of nodes based on the shuttle bus’s departure and end locations and the locations of flights that require shuttle bus service, set the shuttle bus’s routes to and from the flights as scheduling arcs, and determine the dependencies of the directed edges.

[0024] Step 2.2 generates a directed acyclic graph of all possible scheduling routes, namely, the capacity network, through the algorithm, such as Figure 1 shown.

[0025] Step 3: Based on the above capacity network, i.e., directed acyclic graph, generate the time adjacency matrix through Python and save the directed acyclic graph, such as Figure 2 shown.

[0026] Step 4: Detection of abnormal driving status of the driver. Due to the nature of their work, airport shuttle drivers usually need to work continuously for several hours. Long-term driving can easily lead to abnormal driving behaviors such as fatigue and distraction, which can easily cause major accidents at the airport. Moreover, long-term work greatly reduces the driver's work efficiency, and it is easy to make mistakes in driving decisions. Therefore, it is necessary to capture the driver's driving information in real time through the on-board camera. When the driver is found to have abnormal driving behavior, the driver will be recalled in time and will no longer perform the next dispatch task. The specific steps are as follows:

[0027] Step 4.1 Driver driving information collection: The driver's information is collected and identified through the on-board camera in the cockpit, including common fatigue behaviors such as yawning and closing eyes, as well as abnormal behaviors such as making phone calls and looking around while driving.

[0028] Step 4.2 uses the CenterNet model to detect the image collected by the camera. The CenterNet model structure is as follows: Figure 2 Comprehensively identify the three dangerous driving behaviors of the driver, blinking, yawning, and using the mobile phone, assign a driving state value E to the driver who is performing the task, and divide the driver's abnormal state into three states: normal, slightly abnormal, and severely abnormal. The reference image data label formulation method of the present invention is as follows:

[0029] a) Blinking: Under normal circumstances, blinking usually takes 0.2 to 0.4 seconds to complete. Therefore, the driver’s eye-opening time t is collected from real-time video data. If the driver closes his eyes for more than 1 second during a blink, it is determined that the driver has experienced fatigue blinking.

[0030] The number of times the driver blinks when fatigued is recorded through video image information, recorded as n1. Since the length of time to complete the blinking action is accidental, the error in judging the driver's fatigue state by only one blinking action is large, that is, the driver is not in a fatigued state in other situations, but accidentally in one blink, the blinking action time exceeds 0.5 seconds. Therefore, in order to prevent the influence of accidental situations on the results and improve the rationality of the judgment, the present invention indirectly reflects whether the driver is in a fatigue driving state by recording the fatigue blinking frequency f1, and the calculation formula is as follows:

[0031]

[0032] Where n1 is the number of fatigue blinks, that is, the number of blinks lasting more than 1 second; t is the time, and the recording unit is 1 minute.

[0033] b) Yawning: Since yawning is a process of yawning from a closed mouth to an open mouth, and then from an open mouth state to a closed mouth state, the state of the mouth area is easily confused with the state of normal speech and communication. Based on this, the present invention considers that the difference between the yawning state and the normal speech state is that the aspect ratio of the mouth area during yawning has a unique ratio. Therefore, the present invention indirectly determines whether the driver is yawning by calculating the aspect ratio M of the mouth. The calculation formula is as follows:

[0034]

[0035] Where, L a is the width of the mouth, that is, the distance between the upper and lower edges of the mouth; L b is the length of the mouth, that is, the distance between the left edge and the right edge of the mouth, as shown in Figure 3.

[0036] According to the real-time video data extracted by the camera, when it is detected that the mouth feature of the driver M≥1, it is considered that the driver has yawned once. In addition, it has a large error to determine that the driver is in a fatigued driving state only through the driver's M≥1 once, that is, the driver may have a mouth feature aspect ratio M≥1 due to other reasons, but the driver is not in a fatigued driving state. Similar to the above blink label, in order to prevent accidental events from affecting the experimental results, the present invention sets the number of times that the video data detects the driver's mouth feature M≥1 as n2. The present invention indirectly reflects whether the driver is in a fatigued driving state by recording the yawn frequency f2, and the calculation formula is as follows:

[0037]

[0038] In the formula, n2 is the number of fatigued blinks, that is, the number of times the driver's mouth feature M≥1; t is the time, and the recording unit is 1 min.

[0039] c) Answering or making a phone call: Since answering or making a phone call during the process of driving a shuttle bus is extremely likely to cause the driver to be distracted and is an important factor leading to airport accidents, it is necessary to detect in real time whether the driver is answering or making a phone call through the camera video data. When a mobile phone is recognized in the video image, it is considered that the driver is using the mobile phone during the driving process, and then E = 1.

[0040] Based on the above detection of the driver's driving state, when E = 0, it is regarded that the driver's driving state is normal; when 0 < E ≤ 0.5, it is regarded that the driver's driving state is slightly abnormal; when 0.5 < E ≤ 1, it is considered that the driver has a severe abnormal driving state and is very likely to pose a danger to the operation of the airport.

[0041] Step 4.3 Recall the driver in a severe abnormal state. When it is detected that the driver is in a severe abnormal state, that is, 0.5 < E ≤ 1, notify the driver in time and recall him / her to the terminal station without continuing to perform the scheduling task.

[0042] Step 5 Establishment of a mixed integer programming model. In order to quickly and accurately obtain the minimum number and optimal path of shuttle buses, a mixed integer programming model is established according to the airport flight time and location information data, and the objective function and constraints are set as follows:

[0043] Step 5.1 Set the objective function: Minimize the number of shuttle buses operating in the entire airport and reduce the airport operation cost. The objective function is

[0044]

[0045] Step 5.2 Set the constraints: In order to achieve optimal scheduling, the constraints are set as follows:

[0046] a) Usually a flight only needs one shuttle bus service, and the constraints are:

[0047]

[0048]

[0049] b) Bind the shuttle service start time to the service start time window, with the following constraints:

[0050]

[0051] c) The driver's driving status is detected in real time through the camera in the cockpit, and a driving status safety value E is assigned to the driver who is performing the task of picking up passengers. Only drivers with normal or slightly abnormal driving status are allowed to continue the task. The constraints are:

[0052] 0≤E≤0.5(10)

[0053] d) If flight i and flight j are served consecutively by the same shuttle bus, and flight i precedes flight j, then the service start time of flight i plus the connection time from flight i to flight j shall not be later than the shuttle bus service start time of flight j, subject to the following constraints:

[0054]

[0055] Where M is a positive integer and A3 is the set of scheduling arcs.

[0056] e) Since one flight is usually served by only one shuttle bus, the capacity of the scheduling arc is set to 1 in the present invention, that is, all scheduling arcs in the capacity network are served by one shuttle bus, and the constraints are:

[0057]

[0058] Step 5.3: To facilitate the solution, the above mixed integer model is equivalent to a linear programming model, and the actual airport shuttle service disjoint paths are obtained, which improves the solution efficiency. Its objective function and constraints are as follows:

[0059] a) Objective function:

[0060]

[0061] b) In actual airport operations, a shuttle bus can usually travel between multiple flights and complete the passenger pick-up and drop-off services for multiple flights. In order to improve the efficiency of the shuttle bus to pick up passengers, a disjoint path is obtained with the following constraints:

[0062] 0 <E≤0.5(14)

[0063]

[0064]

[0065]

[0066]

[0067] Step 6 uses the time adjacency matrix of the capacity network generated in step 3 as the input of the linear programming model to solve the final result of the minimum number of airport shuttle buses and the optimal path, such as Figure 4 shown.

[0068] A second aspect of the present invention relates to an airport shuttle bus optimization scheduling device based on an improved capacity network model that takes into account the driver's status, comprising a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, they are used to implement the airport shuttle bus optimization scheduling method based on the improved capacity network model that takes into account the driver's status of the present invention.

[0069] A third aspect of the present invention relates to a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the airport shuttle bus optimization scheduling method of the present invention based on an improved capacity network model taking into account driver status.

[0070] The working principle of the present invention is as follows: combining airport flight information and location information, etc., to generate a capacity network for airport scheduling (including all shuttle bus scheduling routes), and storing the capacity network through an adjacency matrix, using a target detection algorithm to identify in real time the blinking, yawning, and phone calls of the shuttle bus driver, and using the result as one of the constraints of the linear programming model, so that the driver with abnormal driving status is recalled in time and the passenger pick-up and drop-off service is stopped. Finally, the adjacency matrix is ​​input into the linear programming model to obtain the final capacity network, that is, the minimum number of shuttle buses required for airport scheduling and the non-intersecting paths for shuttle bus scheduling are obtained.

[0071] The advantages of the present invention are: in view of the low efficiency of manual dispatching of airport shuttle buses, an airport shuttle bus capacity network is established, and the driver's driving status is detected in real time in combination with a target detection algorithm, and the driver's driving status is used as an important factor in the dispatching of airport shuttle buses, and non-intersecting paths of airport shuttle buses are obtained, making the dispatching of airport shuttle buses more efficient and safer. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 is a capacity network schematic diagram of the present invention;

[0073] Figure 2 It is a schematic diagram of the structure of the CenterNet model of the present invention;

[0074] Figure 3a and Figure 3b is a schematic diagram of the mouth features of the present invention, wherein Figure 3a This is the mouth feature recognition map under normal conditions. Figure 3b It is a mouth feature recognition map when yawning;

[0075] Figure 4 It is a schematic diagram of the capacity network result of the present invention. DETAILED DESCRIPTION

[0076] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0077] Example 1

[0078] This embodiment relates to an airport shuttle bus optimization scheduling method based on an improved capacity network model taking into account the driver's status, including the following steps:

[0079] Step 1: Real-time airport information data collection. Taking Xinchang Wanfeng General Airport as an example, according to the actual situation of Xinchang Wanfeng General Airport on December 26, 2022, determine the departure and end positions of the airport shuttle bus, the flight positions that require shuttle bus service, and the position distance relationship between the positions, and calculate the earliest and latest time that flight i is allowed to be served by the shuttle bus. The process is as follows:

[0080] Step 1.1 collects the real-time situation of Xinchang Wanfeng General Airport. The data collected by the present invention are as follows:

[0081] a) Shuttle bus departure and terminal location: The shuttle bus departs from the departure station, picks up passengers at the destination flight location, and then returns to the terminal. The distance between the departure station and the terminal will affect the time of the shuttle bus to and from the flight, so it is necessary to arrange and dispatch it reasonably based on the location distance information.

[0082] b) Estimated arrival time and estimated departure time of flight: Since the shuttle bus needs to serve the designated flights according to the flight demand, for flight i that needs shuttle bus service, its estimated arrival time (STA i )、Scheduled departure time(STD i ) and extract records of flight stop locations.

[0083] c) The time required for the shuttle bus to travel between flights: Since a shuttle bus usually needs to complete the task of picking up and dropping off passengers for multiple flights, and the time spent between flights will affect the start time of the next flight service, the present invention takes the time required for the shuttle bus to travel between flights into account, so that the takeoff of the next flight is not affected by the previous flight.

[0084] d) The earliest waiting time T1 and the latest waiting time T2 of the shuttle bus before the scheduled arrival of the flight: According to regulations, the shuttle bus should wait at the berth in advance before the flight arrives to avoid delaying passengers' boarding and flight takeoff.

[0085] e) Required service time t ii : In order not to affect the take-off of flights and complete the shuttle bus's task of picking up passengers in time, the shuttle bus's mission duration for picking up flights should also be taken into account to make the shuttle bus's scheduling more reasonable and efficient.

[0086] Step 1.2 Calculate the earliest time e that the shuttle service is allowed for flight i based on the collected information data i and the latest time i The calculation process is as follows:

[0087] For flight i, the earliest time e at which the shuttle service is allowed i The calculation formula is:

[0088] e i =STD i -T2(1)

[0089] The latest time shuttle service is allowed i The calculation formula is:

[0090] l i =STA i -T1(2)

[0091] Step 2: Establishment of the capacity network model. The present invention sets the departure position, end position, and flight position of the shuttle bus as nodes, sets the position distance between nodes and the relationship between the shuttle bus's travel time between flights as scheduling arcs, and generates a real-time directed acyclic graph, i.e., a capacity network, through an algorithm using real-time data. The process is as follows:

[0092] Step 2.1 According to the departure and end positions of the shuttle bus, and the positions of the flights that need shuttle bus service, it is determined that there are seven flights waiting for service at Xinchang Airport at this time, namely n1, n2, n3...n7. The shuttle bus needs to depart from the departure point s, go to and from the seven flights to complete the passenger pick-up and drop-off service, and finally return to the parking point t. The route of the shuttle bus to and from the flight is set as the scheduling arc, and all possible dependencies of the directed edges are determined based on the time window relationship between the flights at this time.

[0093] Step 2.2 Generate a directed acyclic graph of all possible scheduling routes of Xinchang Airport at this time through Python, that is, the capacity network, such as Figure 1 shown.

[0094] Step 3: Based on the above capacity network, i.e., directed acyclic graph, a time adjacency matrix is ​​generated through an algorithm, and the directed acyclic graph is saved. The adjacency matrix is ​​shown below.

[0095]

[0096] Step 4: Detection of abnormal driving status of the driver. Due to the nature of their work, airport shuttle drivers usually need to work continuously for several hours. Long-term driving can easily lead to abnormal driving behaviors such as fatigue and distraction, which can easily cause major accidents at the airport. Moreover, long-term work greatly reduces the driver's work efficiency, and it is easy to make mistakes in driving decisions. Therefore, it is necessary to capture the driver's driving information in real time through the on-board camera. When the driver is found to have abnormal driving behavior, the driver will be recalled in time and will no longer perform the next dispatch task. The specific steps are as follows:

[0097] Step 4.1 Driver driving information collection: The driver's information is collected and identified through the on-board camera in the cockpit, including common fatigue behaviors such as yawning and closing eyes, as well as abnormal behaviors such as making phone calls and looking around while driving.

[0098] Step 4.2 uses the CenterNet model to detect the image collected by the camera. The CenterNet model structure is as follows: Figure 2 Comprehensively identify the three dangerous driving behaviors of the driver, blinking, yawning, and using the mobile phone, assign a driving state value E to the driver who is performing the task, and divide the driver's abnormal state into three states: normal, slightly abnormal, and severely abnormal. The reference image data label formulation method of the present invention is as follows:

[0099] a) Blinking: Under normal circumstances, blinking usually takes 0.2 to 0.4 seconds to complete. Therefore, the driver’s eye-opening time t is collected from real-time video data. If the driver closes his eyes for more than 1 second during a blink, it is determined that the driver has experienced fatigue blinking.

[0100] The number of times the driver blinks when fatigued is recorded through video image information, recorded as n1. Since the duration of the blinking action may vary by chance, the error in judging the driver's fatigue state based on only one blinking action is large, that is, the driver is not in a fatigued state in other cases, but accidentally in one blink, the blinking action time exceeds 0.5 seconds. Therefore, in order to prevent the influence of accidental situations on the results and improve the rationality of the judgment, the present invention indirectly reflects whether the driver is in a fatigue driving state by recording the fatigue blinking frequency, and the calculation formula is as follows:

[0101]

[0102] Where n1 is the number of fatigue blinks, that is, the number of blinks lasting more than 1 second; t is the time, and the recording unit is 1 minute.

[0103] b) Yawning: Since yawning is a process of yawning from a closed mouth to an open mouth, and then from an open mouth state to a closed mouth state, the state of the mouth area is easily confused with the state of normal speech and communication. Based on this, the present invention considers that the difference between the yawning state and the normal speech state is that the aspect ratio of the mouth area during yawning has a unique ratio. Therefore, the present invention indirectly determines whether the driver is yawning by calculating the aspect ratio M of the mouth. The calculation formula is as follows:

[0104]

[0105] Where, L a is the width of the mouth, that is, the distance between the upper and lower edges of the mouth; L b is the length of the mouth, that is, the distance between the left and right edges of the mouth, such as Figure 3a and Figure 3b shown.

[0106] According to the real-time video data extracted by the camera, when the driver's mouth feature M≥1 is detected, it is considered that the driver has yawned once. In addition, the judgment error of judging that the driver is in a fatigue driving state only by the driver's M≥1 once is large, that is, the driver may have a mouth feature aspect ratio M≥1 due to other reasons, but the driver is not in a fatigue driving state. Like the above-mentioned blinking label, in order to prevent accidental events from affecting the experimental results, the present invention sets the number of times the video data detects the driver's mouth feature M≥1 to n2. The present invention indirectly reflects whether the driver is in a fatigue driving state by recording the yawning frequency. The calculation formula is as follows:

[0107]

[0108] Where n2 is the number of fatigue blinks, that is, the number of times the driver's mouth feature M≥1; t is time, and the recording unit is 1 minute.

[0109] c) Answering or making phone calls: Since answering or making phone calls while driving a shuttle bus can easily lead to distracted driving, which is an important factor causing airport accidents, the camera video data is used to detect in real time whether the driver is answering or making phone calls. When a mobile phone is recognized in the video image, it is considered that the driver is using the mobile phone while driving, and E=1.

[0110] Based on the above detection of the driver's driving state, when E = 0, it is considered that the driver's driving state is normal; when 0 < E ≤ 0.5, it is considered that the driver's driving state is slightly abnormal; when 0.5 < E ≤ 1, it is considered that the driver has a severe abnormal driving state and is very likely to pose a danger to the operation of the airport.

[0111] Step 4.3 Recall the driver in a severe abnormal state. When it is detected that the driver is in a severe abnormal state, that is, 0.5 < E ≤ 1, notify the driver in time and recall him / her to the terminal station without continuing to execute the dispatching task.

[0112] Step 5 Establishment of the mixed integer programming model. In order to quickly and accurately obtain the minimum number and optimal path of the shuttle buses, a mixed integer programming model is established according to the airport flight schedule and location information data. The objective function and constraints are set as follows:

[0113] Step 5.1 Set the objective function: Minimize the number of shuttle buses for the entire airport operation and reduce the airport operation cost. The objective function is

[0114]

[0115] Step 5.2 Set the constraints: In order to achieve the optimal dispatching, the constraints are set as follows:

[0116] f) Usually, only one shuttle bus is required for one flight service. The constraint is:

[0117]

[0118]

[0119] g) Bind the shuttle bus service start time to the service start time window. The constraint is:

[0120]

[0121] h) Real-time detect the driver's driving state through the in-cockpit camera, and assign a driving state safety value E to the driver who is performing the task of picking up and dropping off passengers. Only drivers with normal or slightly abnormal driving states are allowed to continue to perform the task. The constraint is:

[0122] 0 ≤ E ≤ 0.5(10)

[0123] i) If flight i and flight j are served by the same shuttle bus continuously, and flight i is before flight j, then the service start time of flight i plus the connection time from flight i to flight j should not be later than the shuttle bus service start time of flight j. The constraint is:

[0124]

[0125] Where M is a positive integer and A3 is the set of scheduling arcs.

[0126] j) Since one flight is usually served by only one shuttle bus, the capacity of the scheduling arc is set to 1 in the present invention, that is, all scheduling arcs in the capacity network are served by one shuttle bus, and the constraint condition is:

[0127]

[0128] Step 5.3: To facilitate the solution, the above mixed integer model is equivalent to a linear programming model, and the actual airport shuttle service disjoint paths are obtained, which improves the solution efficiency. Its objective function and constraints are as follows:

[0129] c) Objective function:

[0130]

[0131] d) In actual airport operations, a shuttle bus can usually travel between multiple flights and complete the passenger pick-up and drop-off services for multiple flights. In order to improve the efficiency of the shuttle bus to pick up passengers, a disjoint path is obtained with the following constraints:

[0132] 0<E≤0.5(14)

[0133]

[0134]

[0135]

[0136]

[0137] Step 6 uses the time adjacency matrix of the capacity network generated in step 3 as the input of the linear programming model to solve the final result of the minimum number of airport shuttles and the optimal path. The results show that during the dispatch of the Xinchang Airport shuttle, a driver was detected to yawn 8 times in one minute while traveling from the shuttle departure point to flight n3. His driving state value E = 0.8, which is within 0.5-1. The driver was judged to be driving abnormally and was recalled. The optimal dispatch plan for the seven flights at Xinchang Airport was obtained, as shown in the following figure: Figure 4 shown.

[0138] Example 2

[0139] The present embodiment relates to an airport shuttle bus optimization scheduling method based on an improved capacity network model taking into account the driver's status, including a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, they are used to implement the airport shuttle bus optimization scheduling method based on an improved capacity network model taking into account the driver's status of Example 1.

[0140] The present invention creatively constructs a capacity network, takes the driver status as a reference factor for the scheduling plan, and efficiently and low-costly completes the optimized scheduling of airport shuttle buses, thereby ensuring the safety of the airport, reducing the airport operation cost, and improving the airport operation efficiency.

[0141] Example 3

[0142] This embodiment relates to a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, the airport shuttle bus optimization scheduling method of the improved capacity network model considering the driver's status of Embodiment 1 is implemented.

[0143] Each embodiment of the present invention is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0144] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. The airport shuttle bus optimization scheduling method based on the improved capacity network model considering the driver's status specifically includes the following steps: Step 1: Airport real-time information data collection: According to the actual real-time situation of the airport, determine the shuttle bus departure location, end location, flight location that needs shuttle bus service, and the location distance relationship between each location, and calculate the earliest time and latest time that flight i is allowed to be served by shuttle bus; Step 2: Establishment of the capacity network model; set the shuttle departure location, end location, and flight location that requires shuttle service as nodes, set the location distance between nodes and the relationship between the shuttle bus’s travel time between flights as scheduling arcs, and pass the real-time data through the algorithm to generate a real-time directed acyclic graph, i.e., the capacity network; Step 3: Based on the above capacity network, i.e., the directed acyclic graph, a time adjacency matrix is ​​generated through an algorithm, and the directed acyclic graph is saved; Step 4: Detect the driver's abnormal driving state; Specifically include: Step 4.1: Collecting the driver's driving information; collecting and identifying the driver's information through the onboard camera in the cockpit, including the driver's fatigue behaviors such as yawning and closing eyes, and abnormal behaviors such as making phone calls and looking around while driving; Step 4.2 uses the CenterNet model to detect the images collected by the camera; comprehensively identify the three dangerous driving behaviors of the driver, blinking, yawning, and using the mobile phone, assign a driving state value E to the driver who is performing the task, and divide the driver's abnormal state into three states: normal, slightly abnormal, and severely abnormal; Based on the above detection of the driver's driving state, when E=0, the driver's driving state is considered normal; when 0E≤0.5, the driver's driving state is considered slightly abnormal; when 0.5<E≤1, the driver is considered to have a severely abnormal driving state, which poses a danger to the operation of the airport; The reference image data label formulation method is as follows: a) Blinking: Under normal circumstances, blinking usually takes 0.2 to 0.4 seconds. Therefore, the driver’s eye-opening time t is collected from the real-time video data. If the driver closes his eyes for more than 1 second during a blink, the driver is deemed to have experienced fatigue blinking. The number of times the driver blinks when fatigued is recorded through video image information, recorded as n1; due to the fact that the duration of blinking action may vary by chance, the error in judging the driver's fatigue state based on only one blink action is large, that is, the driver is not in a fatigued state in other cases, but occasionally in one blink, the blinking action time exceeds 0.5 seconds; therefore, in order to prevent the influence of accidental situations on the results and improve the rationality of the judgment, the fatigue blinking frequency f1 is recorded to indirectly reflect whether the driver is in a fatigue driving state, and the calculation formula is as follows: In the formula, n1 is the number of fatigue blinks, that is, the number of blinks lasting more than 1 second; t is the time, and the recording unit is 1 minute; b) Yawning: Since yawning is a process of yawning from closed mouth to open mouth and then from open mouth to closed mouth, the state of the mouth area is easily confused with the state of normal speech communication. Based on this, the difference between the yawning state and the normal speech state is that the aspect ratio of the mouth area has a unique ratio when yawning; therefore, the aspect ratio M of the mouth is used to indirectly determine whether the driver is yawning. The calculation formula is as follows: Where, L a is the width of the mouth, that is, the distance between the upper and lower edges of the mouth; L b is the length of the mouth, i.e. the distance between the left and right edges of the mouth; According to the real-time video data extracted by the camera, when the driver's mouth feature M≥1 is detected, it is considered that the driver yawned once; in addition, the judgment error of judging that the driver is in a fatigue driving state only by the driver's M≥1 is large, that is, the driver may have a mouth feature aspect ratio M≥1 due to other reasons, but the driver is not in a fatigue driving state; similar to the above blinking label, in order to prevent accidental events from affecting the experimental results, the number of times the video data detects the driver's mouth feature M≥1 is set to n2, and the yawning frequency f2 is recorded to indirectly reflect whether the driver is in a fatigue driving state. The calculation formula is as follows: Where n2 is the number of fatigue blinks, that is, the number of times the driver's mouth feature M≥1; t is the time, and the recording unit is 1 minute; c) Making or receiving phone calls: Since making or receiving phone calls while driving a shuttle bus can easily lead to driver distraction, which is an important factor causing airport accidents, the camera video data is used to detect in real time whether the driver is making or receiving phone calls. When a phone is identified in the video image, it is considered that the driver is using the phone while driving, and E=1; Step 4.3 Recalling drivers in severe abnormality states: When it is detected that a driver is in severe abnormality state, that is, 0.5<E≤1, the driver is promptly notified and recalled to the terminal station, and the dispatching task is no longer continued; Step 5: Establish a mixed integer programming model; specifically include: Step 5.1 Set the objective function: minimize the number of shuttle buses in the entire airport operation and reduce the airport operation cost. The objective function is: Step 5.2 Set constraints: In order to achieve optimal scheduling, set the constraints as follows: a) A flight only needs one shuttle bus service, and the constraints are: b) Bind the shuttle service start time to the service start time window, with the following constraints: c) The driver's driving status is detected in real time through the camera in the cockpit, and a driving status safety value E is assigned to the driver who is performing the task of picking up passengers. Only drivers with normal or slightly abnormal driving status are allowed to continue the task. The constraints are: 0E≤0.5(10)； d) If flight i and flight j are served consecutively by the same shuttle bus, and flight i precedes flight j, then the service start time of flight i plus the connection time from flight i to flight j shall not be later than the shuttle bus service start time of flight j, subject to the following constraints: Where M is a positive integer, and A3 is the set of scheduling arcs; e) Since a flight is usually served by only one shuttle bus, the capacity of the scheduling arc is set to 1, that is, all scheduling arcs in the capacity network are served by one shuttle bus. The constraints are: Step 5.3: To facilitate the solution, the above mixed integer programming model is equivalent to a linear programming model to obtain the actual airport shuttle service disjoint paths and improve the solution efficiency. The objective function and constraints are as follows: a) Objective function: b) In actual airport operations, a shuttle bus can usually travel between multiple flights and complete the passenger pick-up and drop-off services for multiple flights. In order to improve the efficiency of the shuttle bus to pick up passengers, a disjoint path is obtained with the following constraints: 0＜E≤0.5(14) Step 6: Use the time adjacency matrix of the capacity network generated in step 3 as the input of the linear programming model to solve the final result of the minimum number of airport shuttle buses and the optimal path.

2. The airport shuttle bus optimization scheduling method based on the improved capacity network model considering the driver's status as claimed in claim 1 is characterized by: Step 1 specifically includes: Step 1.1 collects the real-time situation of the airport. The collected data includes: a) The location of the shuttle bus departure and terminal; b) Estimated flight arrival time and estimated flight departure time; c) The time required for shuttle buses to travel between flights; d) The earliest and latest waiting time for the shuttle bus before the scheduled flight arrival; e) Required service duration; Step 1.2: Calculate the earliest and latest times that shuttle service is allowed for flight i based on the collected information data. The calculation process is as follows: For flight i, the earliest time e at which the shuttle service is allowed i The calculation formula is: yes i =STD i -T2 (1) The latest time shuttle service is allowed i The calculation formula is: the i =STA i -T1 (2) 3. The airport shuttle bus optimization scheduling method based on the improved capacity network model considering the driver's status as claimed in claim 2 is characterized by: Step 2 specifically includes: Step 2.1 Determine the location and number of nodes based on the shuttle bus’s departure and end locations and the locations of flights that require shuttle bus service, set the shuttle bus’s routes to and from the flights as scheduling arcs, and determine the dependencies of directed edges; Step 2.2 generates a directed acyclic graph of all possible scheduling routes, namely, the capacity network, through an algorithm.

4. The airport shuttle bus optimization scheduling method based on the improved capacity network model considering the driver's status as claimed in claim 3 is characterized by: Step 3 specifically includes: According to the above capacity network, i.e., directed acyclic graph, a time adjacency matrix is ​​generated by python and the directed acyclic graph is saved.

5. An airport shuttle bus optimization scheduling device based on an improved capacity network model taking into account the driver's status, characterized in that it includes a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, it is used to implement the airport shuttle bus optimization scheduling method based on the improved capacity network model taking into account the driver's status as described in any one of claims 1-4.

6. A computer-readable storage medium, characterized in that: A program is stored thereon, and when the program is executed by a processor, the airport shuttle bus optimization scheduling method of the improved capacity network model considering the driver status described in any one of claims 1-4 is implemented.

Images