Shenzhen Airport Adds Intelligence to Stand Allocation
As they handle massive passenger flow every single day, airports must effectively and efficiently manage their airport stands — the areas where planes park before departure. At Shenzhen Airport, which has an annual passenger flow exceeding 50 million, this is clearly a priority. To address this issue, it has deployed Huawei's Intelligent Stand Allocation Solution, which applies big data and Artificial Intelligence (AI) to help the airport maximize its stand usage. The solution increases the usage of contact stands — areas where passenger boarding bridges can be used — by 5% to 10%, reducing airside transfer bus passenger flow by 2.6 million every year. By deploying Huawei's intelligent stand allocation system, Shenzhen Airport has become a pioneer in exploring and innovating Information Communications Technology (ICT) for the civil aviation industry.
The Airport Operation Center (AOC) prioritizes stand allocation, because the allocation directly impacts airport operations — including flight takeoff and landing, airfield safety, and passenger services — as well as determines the proper allocation of airport resources, such as boarding gates, airside transfer buses, and baggage carousels. The AOC must quickly adjust stand allocation plans when exceptions — such as flight delays, cancellations, or returns — occur.
The entire airport stand allocation process includes information collection, allocation based on static restrictions and priority rules, overall adjustment of constraints, and solution generation and fine-tuning. Currently, stands are manually allocated in the Operation Resource Management System (ORMS), requiring airport staff to obtain flight plans, confirm the flights to be allocated by aircraft no., query air route directions and aircraft models, and analyze passenger information.
Airports in China typically assign operational commanders for stand allocation on the Gantt chart — such as stand allocation modules in the ORMS and Airport Collaborative Decision Making System (A-CDM) — depending on airport constraints. This manual allocation mode is inefficient and involves multiple allocation rules, as well as the need to address various unplanned issues. More importantly, it leads to a higher risk of conflicts during aircraft taxiing, stand arrival, and pushback.
With busy operations and limited stand resources, Shenzhen Airport's conventional stand allocation mode needed to cope with the following challenges:
Shenzhen Airport has a total of 225 stands, of which 62 are contact stands. Using conventional methods, commanders needed to memorize numerous basic rules — such as flight plans, aircraft models, flight types, air route types, as well as priority rules — because of different stand features. This time-consuming method relied heavily on expert experience, and was completely inefficient.
To address these pain points, Shenzhen Airport assigned dedicated resource-allocation agents to dynamically allocate and adjust stands around the clock, and allocated extra staff resources to handle overnight parking flights for three to four hours per day. Although real-time data was available, it was still difficult to evaluate the stand allocation effects. Manual allocation couldn't meet urgent, unexpected allocation requirements in the event of exceptions, such as failure to take off or land because of bad weather, flight delays, and diversions.
Difficult optimization of core indicators
Aircraft docking rate, the ratio between stands allocated to different airlines, passengers' walking distance, conflicts between stand arrival and pushback, and task type all affect the optimization of stand allocation. The conventional allocation mode couldn't implement a global optimized allocation solution to improve a specific indicator (such as aircraft docking rate or conflict rate in different weather conditions) according to various scenarios.
Low level of intelligence
Shenzhen Airport needed more effective and prompt process-based flight information transfer because the surrounding information — regarding airfield planning layout, taxiing program design, and terminal business layout — failed to interact with the manual stand allocation system. Complicating matters, data from different sources — including airlines, ground service departments, and Air Traffic Control (ATC) authorities — couldn't be efficiently aggregated and synchronized, hindering the deep mining and comprehensive analysis of the allocation data.
Restrictions on global allocation efficiency improvement
The allocation effects of a single stand or airport-wide stands couldn't be evaluated, nor could they be quantified and reflected using real-time data. Failure to quickly and accurately adapt to changing allocation rules created a bottleneck, making it harder to increase the airport's operational efficiency.
In recent years, the vast amount of passenger traffic has led to multiple safety, operational, and service challengers to airport operations. In these circumstances, an efficient stand allocation system was urgently needed, to scientifically use and properly allocate key scenario-specific resources of the airport.
The rapid development of intelligent technologies, notably AI, has driven relevant research units and technology companies to delve into algorithms for stand allocation. The intelligent stand allocation system assists commanders in intelligent pre-allocation and real-time allocation of stands, and builds a solid foundation for intelligent distribution of check-in counters, boarding gates, and baggage carousels, further improving the airport operational efficiency and passenger experiences.
The scenario-specific intelligent stand allocation solution co-developed by Shenzhen Airport and Huawei is a groundbreaking project in the industry — the first application of AI in an airport's core production system. Guided by the 4A enterprise (business, application, data, and technology) architecture, the two parties scientifically analyzed each of these aspects in detail, and used AI technologies featured on Huawei's digital platform, to implement automatic and intelligent stand allocation, led by IT operations and facilitated by manual operations.
Many optimized scheduling rules and AI algorithms support intelligent stand allocation and dynamic adjustment. By applying these, core indicators — such as the rate of aircraft docking with jet bridges and bridge turnover rate — are improved. Meanwhile, stand or taxiing conflicts are minimized, enabling airport-wide optimal resource allocation and improving ground handling efficiency, as well as passenger satisfaction, all while ensuring safety.
The intelligent stand allocation system builds an AI platform based on Shenzhen Airport's unified big data platform. Various engine systems are constructed on the AI platform for intelligent stand scheduling. A unified, open, and interactive platform is also established at the top layer — realizing intelligent and visualized stand management.
The big data platform extracts data from peripheral systems and processes it in real time. The platform forms a specialized industry library based on industry experience, and pushes the data to the data module of the upper-layer application system through a standard interface. The integrated data is then used as an AI algorithm input.
The AI platform provides bottom-layer algorithms for the upper-layer engine system, including various algorithm libraries — such as the common basic algorithm library and operational planning optimization algorithm library. Appropriate AI algorithms, as well as mathematical programming and metaheuristic algorithms can be applied to meet stand allocation requirements.
The stand allocation algorithm engine is built based on mathematical optimization and heuristic algorithms. It supports batch allocation and dynamic adjustment, and provides decision-making support — such as intelligent rule engine, intelligent scheduling optimization engine, and simulation engine — for a variety of practical business applications.
As the core of the system, the intelligent rule engine uses operational planning optimization algorithms — such as metaheuristic, heuristic, and accurate solution algorithms — to perform modeling based on over 60 basic stand allocation rules. By factoring in multiple indicators — such as the aircraft docking rate, conflict rate, and passenger experience — the rule engine delivers a set of intelligent stand allocation results to help the airport operate more efficiently.
The industry's first scenario-specific, AI-based intelligent stand allocation solution has reformed and innovated the conventional manual allocation mode; it has also paved a path for the airport industry to digitalize operations. Shenzhen Airport has made the following improvements:
Intelligent stand allocation that is led by IT operations and facilitated by manual operations. Compared with manual allocation, the intelligent stand allocation system — dealing with more than 1000 flights per day — slashes the time required for allocating stands from four hours to less than one minute, with solution rolling updates performed every 10 minutes (each update lasts 10 seconds). The system also allows for manual intervention at any time.
Significant improvement in core indicators. The application of the intelligent stand allocation system has increased the aircraft docking rate at Shenzhen Airport from 71% to 76%. As the airport transports over 50 million passengers annually, this 5% increase translates to more than 2.6 million passengers annually that no longer require airside transfer buses.
Once the satellite hall has been built and is in operation, there will be more contact stands. This will maximize the efficiency of the intelligent stand allocation system and increase the aircraft docking rate by 10% to 15%.
The system can also improve the bridge turnover rate, increasing the number of flights docking to a bridge per day from 10.24 to 11.
Best use of airport-wide core resources. The system uses AI algorithms and operational research strategies to optimize the allocation of airports' core operational resources. As well as enabling intelligent stand allocation, the system can define new objectives and develop them quickly through flexible expansion without conflicting with the existing systems. With boarding gates, check-in counters, baggage carousels, and security checkpoints gradually applying the system, the global coordinated resource allocation will be more efficient.
In-depth integration of technologies and businesses. The intelligent stand allocation system incorporates the existing business rules (over 60 basic rules) to the AI algorithm engine, as well as converges and processes massive amounts of data from various IT system and departments — such as flights, stands, ATC data, and runway information. To cope with complex business procedures, system developers and business personnel worked together to conduct in-depth research on the internal stand allocation rules using AI and other innovative technologies. To date, this project has applied for three patents.
By deploying Huawei's intelligent stand allocation system, Shenzhen Airport has become a pioneer in innovating and exploring ICT for the civil aviation industry. The airport's management team has gained a deeper understanding and forward-looking perspectives in advancing the industry with technological innovations.
Following these successes, Huawei will continue expanding the application of AI technologies in airport resource allocation. Based on the existing system, we will use AI technologies to add intelligence to the allocation of airport-wide resources. This will maximize the benefits of sharing airport resources, achieve optimal resource usage, and eliminate increasingly prominent resource support bottlenecks — helping Shenzhen Airport become a forward-thinking and world-leading airport.