Civil aviation element configuration optimization method based on mixed integer dynamic programming

Abstract

The invention discloses a civil aviation element configuration optimization method based on mixed integer dynamic programming. The method comprises the following steps: S1, constructing a Chinese civil aviation core element and carbon emission spatial-temporal characteristic database; s2, establishing a mixed integer dynamic programming model with the purpose of minimizing the total cost of the system in the programming period; and S3, establishing decision variables, constraint conditions and a fusion solution algorithm framework required by the mixed integer dynamic programming model. And S4, based on the fusion solution algorithm framework, solving the mixed integer dynamic programming model by using a mathematical programming solver integrated with a robust optimization module so as to obtain a civil aviation core element dynamic configuration scheme which is optimal in total system cost in multiple periods in the future and meets the uncertainty scene. According to the method provided by the invention, a multi-source data fusion algorithm is optimized, a carbon emission measuring and calculating method is deepened, and solver parameters are adjusted, so that the collaborative requirements of improving the operation efficiency and reducing the carbon emission from the perspective of carbon cost are met.

Description

Technical Field

[0001] This invention generally relates to the field of carbon emission technology. More specifically, this invention relates to a method for optimizing the configuration of core civil aviation elements based on mixed-integer dynamic programming. Background Technology

[0002] As a core component of the national comprehensive transportation system, civil aviation plays an irreplaceable role in promoting economic and social development and improving travel efficiency. However, the civil aviation industry is also a key area of ​​energy consumption and carbon emissions, characterized by its wide coverage, complex spatial and temporal distribution, and significant challenges in emission reduction. With the advancement of global goals for "carbon neutrality" and "carbon peaking," China's civil aviation industry faces severe pressure to reduce emissions. According to the "14th Five-Year Plan for Green Development of Civil Aviation," it is necessary to ensure the growth of transportation demand while achieving a continuous decline in carbon emission intensity, ultimately reaching the industry-wide carbon neutrality target by 2060.

[0003] Carbon emissions from the civil aviation system are closely related to the configuration of its core elements (airports, routes, fleet, flight operations, etc.). For example, the matching efficiency between aircraft type and routes directly affects fuel consumption per flight, airport capacity and route network layout determine flight turnaround efficiency, and flight frequency is related to the total energy consumption. However, existing technologies for optimizing the configuration of core civil aviation elements lack deep integration of multi-source heterogeneous data (including refined carbon emissions and passenger demand fluctuation data), fail to quantify carbon costs with traditional operations and passenger time costs into a unified total system cost, lack multi-period dynamic correlation and uncertainty robust constraints, and the solution algorithms struggle to efficiently handle the coupling of high-dimensional discrete variables and complex constraints. This results in the inability to achieve global dynamic optimization of core civil aviation elements and fails to meet the synergistic requirements of improving operational efficiency and reducing carbon emissions from a carbon cost perspective.

[0004] In view of this, there is an urgent need to provide a dynamic optimization method for the configuration of core civil aviation elements based on mixed integer dynamic programming in order to reduce carbon emissions. Summary of the Invention

[0005] In order to at least solve one or more of the technical problems mentioned above, this invention proposes a dynamic optimization method for civil aviation element configuration based on mixed integer dynamic programming in several aspects.

[0006] In a first aspect, the present invention provides a dynamic optimization method for civil aviation element allocation based on mixed-integer dynamic programming, comprising the following steps: S1: Constructing a database of core elements and spatiotemporal characteristics of carbon emissions in Chinese civil aviation. The database integrates multi-source heterogeneous data through a spatiotemporal data fusion algorithm. The data includes at least airport geographic information, route network, fleet data, flight operation data, refined route-aircraft type carbon emission data, and passenger travel demand fluctuation data. S2: Establishing a mixed-integer dynamic programming model with the objective of minimizing the total system cost within the planning period. The total system cost includes traditional operating costs and embedded carbon costs, passenger time costs, and policy compliance costs. The carbon cost is calculated by multiplying carbon emissions by a dynamic carbon price, and the dynamic carbon price is correlated with the carbon market volatility coefficient. S3: Establish the decision variables, constraints, and fusion solution algorithm framework required for the mixed-integer dynamic programming model. The decision variables include integer-type aircraft type-route flight frequency allocation variables and airport resource time slot allocation variables. The constraints include demand satisfaction constraints, airport capacity constraints, fleet availability constraints, carbon emission calculation constraints, robustness constraints for uncertain demands, and dynamic correlation constraints between adjacent periods. The fusion solution algorithm framework is a coupled framework of mixed-integer dynamic programming and adaptive genetic algorithm, where mixed-integer dynamic programming handles deterministic constraints, and adaptive genetic algorithm optimizes the search efficiency of discrete decision variables. S4: Based on the fusion solution algorithm framework, use a mathematical programming solver integrating a robust optimization module to solve the mixed-integer dynamic programming model to obtain a dynamic configuration scheme for core civil aviation elements that optimizes the total system cost over future multiple periods and satisfies uncertain scenarios.

[0007] In some embodiments, the spatiotemporal data fusion algorithm in step S1 includes: using a spatial interpolation algorithm to correct coordinate deviations for airport geographic information, using a temporal smoothing algorithm to eliminate outliers for flight operation data, and using a feature alignment algorithm to achieve consistent matching of time and space dimensions for multi-source data.

[0008] In some embodiments, the refined route-aircraft type carbon emission data in step S1 is generated through a bottom-up dynamic calculation method. The bottom-up dynamic calculation method includes using the real-time fuel efficiency of the aircraft type and the actual distance of the route, calculating the carbon emission coefficient using the segmented carbon emission coefficient method, and assigning it spatiotemporal attributes of latitude and longitude-time stamp through a geographic information system.

[0009] In some embodiments, the dynamic carbon price in step S2 is calculated by multiplying the benchmark carbon price by the carbon market volatility coefficient and the policy adjustment coefficient, wherein the carbon market volatility coefficient is predicted by an autoregressive integral moving average time series model of historical carbon trading data.

[0010] In some embodiments, the uncertainty demand robustness constraint in step S3 includes: for any passenger demand fluctuation scenario, the model decision result must satisfy that the deviation between actual capacity and demand does not exceed a preset value, and the constraint is transformed into a solvable linear inequality through interval mathematics.

[0011] In some embodiments, the specific operation mode of the fusion solution algorithm framework in step S3 includes: initialization stage: generating an initial solution space for decision variables through an adaptive genetic algorithm; iterative optimization stage: applying deterministic constraints to the initial solution space using mixed integer dynamic programming to screen feasible solutions; using an adaptive genetic algorithm to perform crossover and mutation operations based on the objective function values ​​of feasible solutions to optimize the solution space; convergence judgment: when the fluctuation of the objective function value after a preset number of consecutive iterations is less than a preset fluctuation value, the optimal solution is output.

[0012] In some embodiments, the carbon emission calculation constraints in step S3 include: establishing a mathematical relationship between the decision variable and the carbon emission E: E=k×f(i,j), where f(i,j) is the carbon emission of aircraft type i on route j per flight; embedding the mathematical relationship as an equality constraint into the mixed integer dynamic programming model, and adding an upper limit constraint on the total carbon emission.

[0013] In some embodiments, the dynamic configuration scheme for core civil aviation elements in step S4 further includes: introducing aircraft type update decision variables in the medium- and long-term planning cycle to enable the scheme to have dynamic iterative adaptability.

[0014] In a second aspect, the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the method as described in any one of the first aspects.

[0015] By employing the dynamic optimization method for civil aviation element allocation based on mixed-integer dynamic programming provided above, this invention optimizes multi-source data fusion algorithms, deepens carbon emission calculation methods, and adjusts solver parameters. This reduces the "system cost prediction error" in the model output, improves the accuracy of carbon emission calculation, and provides more precise data for airline decision-making. Furthermore, in some embodiments, spatiotemporal data fusion algorithms allow for targeted categorized fusion algorithms and civil aviation scenario-based parameter settings, solving the problems of low accuracy and spatiotemporal inconsistency in existing multi-source data fusion technologies, thereby improving the spatiotemporal matching accuracy of multi-source data. Even further, in some embodiments, by refining carbon emission data at the route-aircraft dimension, performing phased calculations, coupling meteorological parameters, and mapping spatiotemporal attributes, the differences in carbon emissions for the same aircraft type on different routes and at different times can be accurately characterized.

[0016] Furthermore, in some embodiments, by coordinating the robust constraints of uncertainty requirements with other constraints of the model, the multi-period element configuration scheme is ensured to not only meet the global optimization objective but also have strong anti-interference capabilities, significantly improving the reliability and operability of the scheme in complex real-world operation scenarios.

[0017] Furthermore, in some embodiments, by integrating the hierarchical constraint processing and dynamic parameter adaptation mechanism of the solution algorithm framework, the feasibility and optimality of solutions in high-dimensional complex scenarios are improved.

[0018] Furthermore, in some embodiments, carbon emission calculation constraints can refine mathematical modeling, scenario-based total constraints, and dynamic adaptation mechanisms, solving the problems of "loose correlation, static rigidity, and poor coordination" of traditional carbon emission constraints. This ensures that the optimized scheme can accurately achieve emission reduction targets while meeting operational needs and economic efficiency, thereby enabling the allocation of civil aviation resources to achieve a three-dimensional balance of "efficiency-cost-emission reduction". Attached Figure Description

[0019] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent upon reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the invention are illustrated by way of example and not limitation, and like or corresponding reference numerals denote like or corresponding parts, wherein: Figure 1 An exemplary flowchart of a dynamic optimization method for civil aviation element configuration based on mixed integer dynamic programming according to some embodiments of the present invention is shown; Figure 2 An exemplary structural block diagram of an electronic device according to an embodiment of the present invention is shown. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0022] Figure 1 An exemplary flowchart of a dynamic optimization method 100 for civil aviation element configuration based on mixed integer dynamic programming according to some embodiments of the present invention is shown.

[0023] like Figure 1As shown, in method 100, a dynamic optimization method for civil aviation element allocation based on mixed-integer dynamic programming includes the following steps: S1: Construct a database of core elements and spatiotemporal characteristics of carbon emissions in China's civil aviation industry. The database integrates multi-source heterogeneous data through a spatiotemporal data fusion algorithm. The data includes at least airport geographic information, route network, fleet data, flight operation data, refined route-aircraft type carbon emission data, and passenger travel demand fluctuation data.

[0024] In one embodiment, this step integrates multi-source heterogeneous data through a spatiotemporal data fusion algorithm, providing fine-grained support for subsequent optimization. Specific implementation scenarios are described below: First, data collection is required. Specific data types and collection scope include: airport geographic information (including the number of runways, terminal capacity, and latitude and longitude coordinates for over 300 airports, such as Beijing Daxing Airport at 39°26′N, 116°33′E); route network data (including take-off and landing points and flight distances for over 1500 domestic scheduled routes, such as the actual flight distance of the Shanghai-Guangzhou route at 1150km); and fleet data (including the number of seats and fuel efficiency curves for over 15 mainstream aircraft types, such as the cruise fuel efficiency of the B737-800 at 2.6L / seat·100km). The data includes flight operation data (including actual take-off and landing times and delay durations of daily flights over the past three years), refined carbon emission data (based on a bottom-up approach, combined with real-time fuel efficiency of the aircraft type, and meteorological parameters such as flight altitude (8000-12000m) and wind speed (±30m / s), and calculated using segmented carbon emission coefficients (2.8kgCO2 / L for take-off and 3.1kgCO2 / L for cruise), and passenger demand fluctuation data (based on historical booking data and holiday patterns, extracting ±20% demand fluctuation characteristics for periods such as Spring Festival and summer travel).

[0025] In one implementation scenario, in addition to the original data types, environmental meteorological data, policy and regulatory data, and airline operation support data may also be included. Specifically, this includes: Environmental meteorological data includes real-time meteorological parameters of the areas traversed by the flight route (such as temperature ±0.5℃ accuracy, humidity ±3% accuracy, air pressure ±1hPa accuracy, and turbulence intensity level) and seasonal wind field distribution data (such as the wind direction change pattern of the flight route under the influence of the East Asian summer monsoon). These data are used to correct the calculation deviations of the aircraft's fuel efficiency and carbon emissions. For example, fuel consumption of the B737-800 can increase by 5%-15% under headwind conditions, which needs to be dynamically adjusted through real-time wind field data.

[0026] Policy and regulatory data: covering national and local carbon emission reduction policies (such as the "Special Action Plan for Carbon Emission Reduction in Civil Aviation"), regional airport capacity control policies (such as the peak-hour takeoff and landing restrictions at Beijing Capital International Airport), and carbon tariff-related rules (such as the potential impact of the EU's CBAM on the civil aviation industry), providing a basis for quantifying policy compliance costs. For example, the penalty coefficient for failing to meet regional emission reduction targets needs to be updated dynamically according to the latest policies.

[0027] Airline operational support data includes aircraft maintenance plans (e.g., A320neo requires a deep overhaul every 2000 flight hours), crew scheduling rules (e.g., crew continuous duty time does not exceed 14 hours), and fuel purchase contract data (e.g., price fluctuations between long-term contract prices and spot prices), used to optimize the calculation of maintenance and fuel costs in traditional operating costs.

[0028] Next, data fusion is performed. Specifically, inverse distance weighting (IDW) can be used to spatially interpolate airport geographic information, correcting coordinate deviations to within ±50m. Alternatively, a combination of IDW and Kriging interpolation can be employed. For detailed spatial data in airport geographic information, such as runway slope and terminal evacuation routes, IDW is first used to correct coordinate deviations to within ±30m, and then Kriging interpolation is used to fill in missing spatial values ​​based on surrounding terrain data (such as elevation changes), ensuring the spatial accuracy of the airport resource allocation model.

[0029] Outliers in flight operation data (such as extreme delays caused by weather) can be smoothed using a moving average method (with a 24-hour window). A hybrid algorithm combining moving average and Kalman filtering can also be introduced. For time-series data such as delay duration and takeoff / landing times in flight operation data, a 24-hour moving average is first used to initially eliminate random outliers (such as extreme delays caused by sudden thunderstorms). Then, Kalman filtering is used to dynamically correct the data and predict the data trend for the next moment (such as the rate of increase in delay duration during the morning rush hour), thereby improving the consistency of the data's time series.

[0030] By utilizing a timestamp alignment algorithm (unified to UTC+8 time zone) and a 1km×1km spatial grid, consistent matching of multi-source data in the "time-space" dimension is achieved, ultimately forming a dynamic database updated per second. Furthermore, a three-dimensional alignment framework of "time-space-semantic" can be constructed. In the time dimension, all data is unified to millisecond-level timestamps in the UTC+8 time zone; in the spatial dimension, the national airspace is divided into 1km×1km grids, mapping flight routes and airport locations to corresponding grids; in the semantic dimension, a data dictionary is established (e.g., "Aircraft type code B787-9 corresponds to 330 seats, fuel efficiency 2.1L / seat"). (100km) to achieve semantic consistency matching of multi-source data, such as accurately associating the "aircraft type number" in flight operation data with the "aircraft type parameters" in carbon emission data.

[0031] Furthermore, regarding refined carbon emission data calculation methods, a "life-cycle carbon emission" perspective can be introduced on top of the "bottom-up" dynamic calculation method. The refined calculation process could include: Carbon emission calculation is broken down into five stages: takeoff, climb, cruise, descent, and taxiing. Each stage uses a different carbon emission coefficient. For example, the carbon emission coefficient for the takeoff stage (0-3000m altitude) is 3.2 kg CO2 / L, for the cruise stage (8000-12000m altitude) it is 2.9 kg CO2 / L, and for the descent stage (12000-0m altitude) it is 3.0 kg CO2 / L. This stage division improves the accuracy of the calculation and avoids the large errors caused by the traditional "single coefficient" calculation.

[0032] Fuel efficiency dynamic correction model: Construct a four-dimensional correction model based on "aircraft type-payload-weather-altitude". For example, the A350-900 has a fuel efficiency of 1.8L / seat under the conditions of a 200-ton payload, a cruising altitude of 10,000m, and a tailwind of 10m / s. 100km; if the load increases to 220 tons and the headwind is 5m / s, the fuel efficiency drops to 2.0L / seat. At 100km, fuel efficiency parameters are dynamically adjusted through real-time data input to ensure the real-time nature of carbon emission data.

[0033] Enhanced spatiotemporal attributes of carbon emission data: A GIS system is used to assign three-dimensional attributes to carbon emission data, including latitude, longitude, timestamp, and altitude. For example, the instantaneous carbon emissions at 31°23′N, 121°47′E (Shanghai airspace), 08:30, and 5000m altitude when a B737-800 operates the Shanghai-Beijing route are recorded, providing support for subsequent analysis of carbon emission hotspots along the route.

[0034] Further, the process moves to S2: Establish a mixed-integer dynamic programming model with the goal of minimizing the total system cost during the planning period. The total system cost includes traditional operating costs and embedded carbon costs, passenger time costs, and policy compliance costs. The carbon cost is calculated by multiplying carbon emissions by the dynamic carbon price, which is associated with the carbon market volatility coefficient.

[0035] The following example illustrates the construction of an objective function to minimize the total system cost, using a 10-year planning period (2026-2035, divided into 40 quarters). The specific components of the total system cost are as follows: 1) Traditional operating cost calculation: Fuel cost = real-time fuel price (e.g., 8 yuan / L) × fuel consumption per shift (based on the BADA model, combined with real-time weather correction, e.g., fuel consumption per shift increases by 8% for B787-9 when the headwind is 15m / s) × flight frequency; Aircraft ownership cost = aircraft purchase price (e.g., A350 approximately 300 million US dollars) × annual depreciation rate (5%) ÷ number of flights per year; Crew manpower cost = duty time (including flight segment duration + 45 minutes of ground preparation) × average hourly wage per person (800 yuan) × number of crew members (4 people for wide-body aircraft); Airport charges = gate time slot fee (12,000 yuan / hour during peak hours) × occupancy time.

[0036] 2) Carbon cost calculation: Dynamic carbon price × carbon emissions. Where, dynamic carbon price = benchmark carbon price (referencing the national carbon market price of 70 yuan / ton CO2) × carbon market volatility coefficient (ARIMA(1,1,1) model prediction, value 0.8-1.2) × policy adjustment coefficient (1.3 for the Beijing-Tianjin-Hebei region, 1.0 for other regions); carbon emissions = flight frequency × carbon emissions per flight (determined by refined S1 data, e.g., an A320neo operating the Chengdu-Beijing route emits 28 tons of CO2 per flight).

[0037] In one embodiment, the carbon market volatility coefficient prediction model can be upgraded, specifically by using a combined prediction method of "ARIMA model + LSTM neural network". First, the linear trend of historical carbon trading data is processed by the ARIMA(1,1,1) model, and then the LSTM neural network is used to capture the nonlinear fluctuations in the data (such as sudden increases / decreases in carbon prices caused by sudden policy adjustments), thereby further improving the prediction accuracy, which is better than the accuracy of the traditional single ARIMA model.

[0038] Furthermore, based on the aforementioned "benchmark carbon price × volatility coefficient × policy coefficient," a three-dimensional adjustment factor of "region-industry-time" is introduced in one implementation scenario to enhance the dynamic adaptability of carbon prices. Details are as follows: Regional adjustment factor: Different coefficients are set according to the carbon emission reduction pressure of different regions. For example, the coefficient is 1.4 for key "dual carbon" regions such as Beijing-Tianjin-Hebei and Yangtze River Delta, and 1.0 for non-key regions in the central and western regions, to adapt to regional policy differences.

[0039] Industry adjustment factor: Taking into account the differences in emission reduction difficulty between the civil aviation industry and other high-energy-consuming industries (such as steel and power), an industry coefficient is set. For example, the civil aviation industry has a coefficient of 1.2 due to its high emission reduction technology bottleneck, to ensure that carbon prices are in line with the actual situation of the industry.

[0040] Time adjustment factor: Considering the long-term development trend of the carbon market, an annual reduction coefficient is set, for example, the coefficient is 1.0 for 2026-2030, 0.95 for 2031-2035, and 0.9 for 2036-2040, to reflect the trend of gradually tightening carbon emission reduction targets.

[0041] 3) Additional cost calculation: Passenger time cost = average delay time (based on historical data, such as a 22-minute delay during the morning peak at Pudong Airport) × unit time value (80 yuan / hour) × number of passengers per flight (180 people); policy compliance cost (such as fines for failing to meet regional emission reduction targets, calculated as excess emissions × 1.5 times the carbon price).

[0042] In addition, in some implementation scenarios, "hidden costs" and "contingency costs" may also be included, specifically including: Quantifying implicit costs: These include route network redundancy costs (such as the fleet resource costs occupied by inefficient routes, calculated based on an average daily depreciation of 50,000 yuan for idle aircraft) and passenger loss costs (losses due to passenger ticket cancellations caused by flight delays or insufficient capacity, calculated based on an average ticket price of 800 yuan per passenger × a loss rate of 3%). By constructing a "route efficiency - passenger satisfaction" correlation model, implicit costs are transformed into quantifiable monetary costs. For example, for every 1% increase in route redundancy rate, implicit costs increase by 2 million yuan per year.

[0043] Emergency cost calculation: This includes costs associated with flight cancellations / rescheduling due to extreme weather (such as typhoons and blizzards) (e.g., rental fees for temporary aircraft of 200,000 RMB per flight, passenger accommodation costs of 500 RMB per person) and emergency repair costs for equipment failures (e.g., emergency repair costs for sudden engine failures of 1 million RMB per incident). Based on historical emergency event data (statistics of national civil aviation emergency events over the past 5 years), Monte Carlo simulation is used to generate probability distributions of emergency scenarios, incorporating emergency costs into the total system cost in the form of expected costs.

[0044] In addition to traditional operating cost estimation, to improve the accuracy of cost estimation, it can also include... Fuel cost: Introducing a "jet fuel density-temperature" correction factor. For example, in the high temperature (35℃) environment of summer, the jet fuel density drops to 0.78kg / L. The fuel consumption calculation needs to be adjusted according to the actual density to avoid the 2%-3% error caused by the traditional "standard density" calculation.

[0045] Aircraft ownership costs: differentiate between the cost structures of "owned aircraft" and "leased aircraft". Owned aircraft are calculated as "purchase price × annual depreciation rate (5%-8%) + insurance premium (0.5% of purchase price)" while leased aircraft are calculated as "monthly rental fee (e.g., $300,000 per month for a B737-800) + maintenance deposit (10% of rental fee)" to improve the targeting of cost calculation.

[0046] Furthermore, the objective function for minimizing the total system cost can be constructed as a multi-objective weighted optimization function to balance cost minimization with emission reduction and efficiency objectives. The specific objective function formula is as follows: Min Z = α×C_total + β×E_total + γ×T_delay Where C_total is the total system cost (ten thousand yuan), and α is the cost weighting coefficient (valued at 0.6-0.8, adjusted according to the airline's cost sensitivity). E_total represents the total carbon emissions during the planning period (10,000 tons of CO2), and β is the emission reduction weighting coefficient (valued at 0.1-0.2, adjusted according to the requirements of the "dual carbon" policy). T_delay is the total passenger delay time (in ten thousand hours), and γ is the efficiency weighting coefficient (with a value of 0.1-0.2, adjusted according to the passenger satisfaction target). Furthermore, α+β+γ=1 ensures multi-objective collaborative optimization.

[0047] Further, the process moves to S3: establishing the decision variables, constraints, and fusion solution algorithm framework required for the mixed-integer dynamic programming model. Specifically, the decision variables may include integer-type aircraft type-route flight frequency allocation variables (e.g., 3-20 flights per week, with a 1-flight interval) and airport resource time slot allocation variables (e.g., 12 wide-body aircraft stands at Guangzhou Baiyun Airport from 7:00 to 9:00).

[0048] In one implementation scenario, decision variables may further include "continuous" and "binary" decision variables to enrich the decision dimensions, specifically including: Continuous decision variables include aircraft cruising speed (values ​​ranging from 450-550 km / h, optimized based on route distance and fuel efficiency) and flight departure / arrival time deviations (values ​​ranging from -30 to +30 minutes, used to alleviate airport congestion during peak hours). For example, on the Shanghai-Guangzhou route, adjusting the cruising speed from 500 km / h to 520 km / h can shorten flight time by 15 minutes and reduce passenger delays, but fuel consumption increases by 3%, requiring optimization through model trade-offs.

[0049] Binary decision variables include "Route opening / closing" (1=open, 0=closed), "Aircraft maintenance plan execution" (1=perform maintenance, 0=normal operation), and "Emergency capacity allocation" (1=activate standby aircraft, 0=not activate). For example, for routes with passenger traffic consistently below 50%, the model can decide to "close the route (variable = 0)" to free up fleet resources for more efficient routes.

[0050] The constraints include demand satisfaction constraints (e.g., single-route capacity ≥ 1.1 times the predicted demand), airport capacity constraints (takeoffs and landings per hour ≤ 85% of airport saturation capacity, such as 60 flights per hour at Beijing Capital International Airport), fleet availability constraints (total flight time of a certain aircraft type ≤ 3000 hours of annual available time per aircraft × number of fleets), carbon emission calculation constraints (the formula can be: E = k × f(i,j), where k is the frequency, f(i,j) is the emission per flight, and cumulative emissions ≤ regional annual quota), robustness constraints for uncertain demand (through Monte Carlo simulation of 1000 demand fluctuation scenarios, it is transformed into a linear inequality of "capacity-demand deviation ≤ 5%)", and dynamic correlation constraints between adjacent periods (e.g., fleet size in period t+1 = fleet size in period t + number of new additions - number of retirements, with retirement condition being aircraft age ≥ 20 years).

[0051] In some implementation scenarios, the constraints may also include: Crew qualification constraints: Different aircraft types require crews with corresponding qualifications. For example, crews operating A380 aircraft must hold "wide-body aircraft pilot qualifications". The constraint is in the form of "∑(x_ij×q_i)≤Q_j", where x_ij is the flight frequency of aircraft type i on route j, q_i is the number of qualified crews required for aircraft type i, and Q_j is the total number of qualified crews available for route j. This avoids operational risks caused by mismatched crew qualifications.

[0052] Aviation fuel supply constraints: The aviation fuel reserves of an airport must meet the fuel demand of all flights at the airport. The constraint is in the form of "∑(x_ij×f_ij)≤F_k", where f_ij is the fuel consumption of aircraft type i on route j per flight, and F_k is the daily aviation fuel reserves of airport k, to ensure a stable aviation fuel supply.

[0053] Carbon emission intensity constraints: The model must meet the carbon emission intensity reduction targets for the civil aviation industry. The constraint form is "E_total / P_total≤I_target", where P_total is the total passenger traffic (ten thousand passengers) during the planning period, and I_target is the carbon emission intensity target (kgCO2 / passenger, such as 80kgCO2 / passenger in 2030). This ensures that the model output scheme meets the emission reduction policy requirements.

[0054] The integrated solution algorithm framework is a coupled framework of mixed-integer dynamic programming (such as handling time-series constraints of 40 cycles) and adaptive genetic algorithm, in which mixed-integer dynamic programming handles deterministic constraints, and adaptive genetic algorithm optimizes the search efficiency of discrete decision variables.

[0055] It should be noted that the granularity of the period division in mixed-integer dynamic programming is adaptively adjusted according to the length of the planning period. Short-term planning (1-3 years) is divided into months, medium-term planning (3-10 years) into quarters, and long-term planning (10-20 years) into years, balancing the solution accuracy and efficiency.

[0056] In one implementation scenario, the genetic algorithm parameters can be a population size of 500, a crossover probability (dynamically adjusted based on fitness values, 0.6 for excellent individuals and 0.9 for average individuals), and a mutation probability of 0.05. Dynamic programming decomposes the problem periodically, calling the genetic algorithm in each period to output a local optimum, and finally integrating the global optimum through the Bellman equation. The crossover and mutation probabilities of the adaptive genetic algorithm can be dynamically adjusted according to population diversity. For example, when the population diversity (calculated using the Shannon diversity index) is higher than 0.8, the crossover probability is set to 0.7 and the mutation probability to 0.03 to avoid oversearch; when the diversity is lower than 0.5, the crossover probability is set to 0.9 and the mutation probability to 0.08 to increase population diversity and avoid getting trapped in local optima.

[0057] Furthermore, in one implementation scenario, a Particle Swarm Optimization (PSO) algorithm can be introduced as an auxiliary optimization algorithm. During the iterative optimization phase, the feasible solution output by the adaptive genetic algorithm is first fine-tuned locally using the PSO algorithm, and then mixed-integer dynamic programming is input to handle constraints, thereby improving the quality of the solution. For example, for the aircraft type-route frequency allocation solution, the PSO algorithm can fine-tune the frequency from 12 flights / week to 13 flights / week through "particle position update," thereby further reducing the system cost by 0.5%-1%.

[0058] Further, the process moves to S4: Based on the fusion solution algorithm framework, the mixed integer dynamic programming model is solved using a mathematical programming solver that integrates a robust optimization module, in order to obtain a dynamic configuration scheme for the core civil aviation elements that optimizes the total system cost over multiple future periods and satisfies the uncertainty scenario.

[0059] In one embodiment, a Gurobi solver with an integrated robust optimization module (e.g., setting MIPGap=0.01) is used to iteratively solve the model (e.g., convergence condition: cost fluctuation <0.1% for 50 consecutive iterations). The output includes: a quarterly aircraft type-route frequency table (e.g., 12 A330 flights per week and 18 A320neo flights per week on the Beijing-Shenzhen route in Q1 2027), an airport time slot allocation table (e.g., 8 narrow-body aircraft stands allocated at Hongqiao Airport from 14:00 to 16:00), a fleet renewal plan (e.g., retiring 5 B737-800s and adding 3 A321neo aircraft by 2030), and sensitivity analysis under different carbon price scenarios (e.g., when the carbon price rises to 100 yuan / ton, the frequency of wide-body aircraft on long-haul routes decreases by 15%).

[0060] Furthermore, in one implementation scenario, the solver module can integrate a "sensitivity analysis module" and a "scenario analysis module" on top of the original Gurobi solver. The sensitivity analysis module can calculate key indicators such as "the percentage increase in system cost for every 10 yuan / ton increase in carbon price" and "the magnitude of capacity adjustment for every 5% fluctuation in passenger demand," providing a basis for sensitivity assessment of the proposed solutions. The scenario analysis module can generate multiple scenario solutions, such as "baseline scenario (carbon price 70 yuan / ton)," "high carbon price scenario (carbon price 150 yuan / ton)," and "high demand scenario (passenger demand increases by 20%)," allowing airlines to choose according to their actual circumstances. Simultaneously, key parameters of the Gurobi solver can be adjusted to improve solution efficiency and accuracy. For example, adjusting the MIPGap (mixed integer programming gap) from 0.01 to 0.005 improves the accuracy of the optimal solution by 50%; setting the "number of parallel solution threads" to 80% of the CPU cores avoids resource waste and reduces solution time by 20%-30% (e.g., reducing the solution time for a 10-year planning period from 48 hours to 35 hours).

[0061] Furthermore, based on the above dynamic configuration scheme, in one embodiment scenario, a "three-dimensional dynamic configuration scheme" can also be utilized, specifically including the "time-space-element" dimension: Time dimension: The plan is broken down into layers of "day-week-month-quarter-year". For example, the daily dimension outputs "airport resource allocation table for morning, noon and evening peak hours", the weekly dimension outputs "intraweek flight frequency fluctuation plan (such as a 10% increase in frequency on weekends)" and the monthly dimension outputs "monthly fleet maintenance plan" to ensure the feasibility of the plan at different time scales.

[0062] Spatial Dimension: Outputs "Airline Carbon Emission Hotspot Distribution Map" and "Airport Capacity Utilization Heat Map". The airline carbon emission hotspot distribution map marks routes with carbon emission intensity higher than 100 kg CO2 / passenger (such as some high-altitude routes), and proposes emission reduction measures such as "aircraft type replacement (such as replacing B737-800 with A320neo)" and "route optimization (such as shortening detour distance)". The airport capacity utilization heat map marks airport time periods with utilization rates higher than 90% (such as 8-10 am at Beijing Capital Airport), and proposes congestion relief measures such as "staggered takeoff and landing" and "planning for new runways".

[0063] Key elements: Two new sub-plans have been added: "Passenger Experience Optimization Sub-Plan" and "Emergency Response Sub-Plan." The Passenger Experience Optimization Sub-Plan includes "Flight Delay Compensation Standards (e.g., 100 RMB / person for a 2-hour delay)" and "Optimization of Transfer Connection Time (e.g., adjusting the minimum domestic transfer time from 45 minutes to 30 minutes)." The Emergency Response Sub-Plan includes "Flight Cancellation / Rescheduling Procedures During Typhoon Weather" and "Alternate Capacity Allocation Plan in Case of Aircraft Malfunction," enhancing the comprehensiveness of the plans.

[0064] Furthermore, in one embodiment scenario, the dynamic optimization method for civil aviation element configuration based on mixed-integer dynamic programming may further include step S5: scheme verification and iterative optimization, which may specifically include: 1. Solution Validation Method Construct a three-tiered verification system: "Data Verification - Simulation Verification - Field Verification". Data validation: Compare the "predicted carbon emissions" and "predicted costs" output by the solution with historical actual data. If the error is within 5%, the data validation is passed; if the error exceeds 10%, backtrack to steps S1-S4 and correct the data fusion algorithm or model parameters.

[0065] Simulation Verification: A simulation model is built using civil aviation system simulation software (such as AnyLogic). A dynamic configuration scheme is input, and the flight operation process during the planning period is simulated to verify whether indicators such as "flight punctuality rate," "passenger satisfaction," and "carbon emissions" meet the standards. For example, if the simulation results show that the flight punctuality rate increases from 80% to 88% and carbon emissions decrease by 12%, then the simulation verification is passed.

[0066] Field verification: Select 1-2 pilot areas (such as the Yangtze River Delta civil aviation circle) for a 3-month field trial operation, collect trial operation data (such as actual carbon emissions, passenger complaints), compare with the predicted values ​​of the plan, if the deviation is within 3%, the plan will be officially promoted; if the deviation is large, targeted optimization will be carried out, such as adjusting the airport capacity constraint parameters.

[0067] 2. Iterative optimization mechanism Establish a closed-loop iterative mechanism of "real-time data feedback - model parameter update - solution iterative output": Real-time data feedback: Collect "actual flight operation data", "real-time carbon emission monitoring data" and "passenger demand dynamic data" through the civil aviation industry's real-time data platform (such as the data platform of the Operation Monitoring Center of the Civil Aviation Administration of China) and update them to the database daily.

[0068] Model parameter updates: The model parameters are dynamically adjusted based on real-time data. For example, if the actual carbon price rises to 90 yuan / ton, the "dynamic carbon price" parameter in the model is automatically updated; if the actual fluctuation range of passenger demand is ±15%, the deviation preset value in the "uncertainty demand robustness constraint" is adjusted.

[0069] Solution Iteration Output: An iterative optimization solution is generated monthly, and the differences between the original solution and the new solution are compared (such as the percentage of cost reduction and the improvement of emission reduction effect). A "Solution Update Report" is sent to airlines to ensure that the solution always adapts to changes in the external environment.

[0070] The embodiments provided by the present invention utilize deep fusion of multi-source data to achieve refined characterization of carbon emissions, integrate all-factor costs from a unified carbon cost perspective, and combine fusion algorithms with dynamic robust constraints to solve the problems of poor global coordination and insufficient dynamic adaptability in existing technologies. This can reduce the total system cost and carbon emissions, providing a feasible optimization path for the civil aviation industry to balance transportation demand and low-carbon goals.

[0071] Furthermore, in one embodiment, the spatiotemporal data fusion algorithm in step S1 includes: using a spatial interpolation algorithm to correct coordinate deviations for airport geographic information, using a temporal smoothing algorithm to eliminate outliers for flight operation data, and using a feature alignment algorithm to achieve consistent matching of time and space dimensions for multi-source data.

[0072] The spatiotemporal data fusion algorithm in the above embodiments addresses the heterogeneity, bias, and spatiotemporal inconsistency issues of multi-source heterogeneous data (spatial, temporal, and attribute-based) in civil aviation. Through a three-level fusion architecture of "classified processing - multi-algorithm collaboration - precise alignment," it achieves a dual improvement in data quality and consistency. Specific technical details are as follows: Regarding spatial interpolation correction of airport geographic information, to address the issue of large original coordinate deviations (above ±100m) for some small and medium-sized airports, the inverse distance weighted method (IDW) is adopted for spatial interpolation optimization. The algorithm parameters are adapted to the civil aviation scenario: a search radius of 5km is set (covering the airport terminal, runway, and surrounding supporting areas), and a weighting coefficient p=2 (balancing the influence of neighboring data). Using high-precision GPS coordinates (deviation ≤10m) of 30 hub airports nationwide as benchmark control points, the coordinates of over 150 small and medium-sized airports are corrected, ultimately ensuring that the coordinate deviation of all airport geographic information is controlled within ±50m, meeting the requirement of "precise waypoint positioning" in route planning.

[0073] Regarding the handling of time-series anomalies in flight operation data, flight operation data (departure and landing times, delay durations, etc.) are easily affected by sudden factors such as weather and air traffic control, resulting in outliers. A combined technique of "3σ criterion detection + moving average smoothing" is adopted. First, outlier data (such as extreme values ​​of single delays exceeding 4 hours) is screened using the 3σ criterion (mean ± 3 standard deviations). Then, a moving average method with a window size of 24 hours is used for smoothing. Data within the window is weighted according to time (data from the most recent 12 hours has a weight of 0.7, and data from the previous 12 hours has a weight of 0.3). This preserves the time-series trend while eliminating sudden interference, resulting in an outlier removal rate of over 92% for flight operation data, providing stable basic data for subsequent fuel consumption and carbon emission calculations.

[0074] Regarding the spatiotemporal feature alignment of multi-source data, to address the inconsistency in the spatiotemporal dimensions of airport geographic information (spatial data), flight operation data (time-series data), and carbon emission data (attribute data), a feature alignment algorithm combining "timestamp standardization + spatial grid matching" is adopted. In the time dimension, all data is uniformly converted to 1-minute timestamps in the UTC+8 time zone (e.g., flight departure time accurate to "2026-03-15 08:32:00"). In the spatial dimension, a 1km×1km EqualEarth projection grid is used to divide the national airspace, mapping flight routes, airport locations, and carbon emission points to corresponding grid cells, achieving spatial association through grid IDs. In the attribute dimension, a cosine similarity algorithm (threshold ≥ 0.95) is used to match the correspondence between aircraft type, flight route, and carbon emission, ensuring unique association of multi-source data for the "same flight."

[0075] Finally, in terms of collaborative control of the fusion process, spatiotemporal data fusion is executed in the order of "data preprocessing → categorized optimization → spatiotemporal alignment → quality verification". The quality verification process adopts an "error threshold judgment" mechanism - when the spatial data deviation exceeds ±50m, the fluctuation of the time series data after smoothing exceeds 10%, or the attribute matching similarity is lower than 0.95, it automatically returns to the previous level for reprocessing to ensure the integrity (data coverage ≥98%) and consistency (spatiotemporal matching accuracy ≥99%) of the output database.

[0076] The above-described embodiments, through targeted categorized fusion algorithms and civil aviation scenario-based parameter settings, address the issues of low accuracy and spatiotemporal inconsistency in existing multi-source data fusion technologies. Compared to traditional single fusion methods, they not only improve airport coordinate correction accuracy but also enhance the outlier removal rate of flight data and increase the spatiotemporal matching accuracy of multi-source data. The constructed database provides fine-grained, highly consistent foundational data support for subsequent optimization models, directly ensuring the accuracy of carbon emission calculations and the rationality of decision variables, thus laying a core data foundation for the feasibility of the global optimization scheme.

[0077] Furthermore, in one embodiment, the refined route-aircraft dimension carbon emission data mentioned in step S1 is generated through a "bottom-up" dynamic calculation method; the "bottom-up" dynamic calculation method includes using the real-time fuel efficiency of the aircraft and the actual distance of the route, calculating using the segmented carbon emission coefficient method, and assigning it spatiotemporal attributes of latitude and longitude-time stamp through a geographic information system.

[0078] In the above embodiments, a dynamic calculation scheme of "multi-parameter coupling - phased calculation - spatiotemporal mapping" is proposed for generating refined carbon emission data at the route-aircraft dimension in step S1. This solves the problem that existing technologies rely on macroscopic statistical data and are difficult to characterize the spatiotemporal heterogeneity of carbon emissions. The specific explanation is as follows: First, basic data preprocessing and adaptation are performed. This involves collecting core input data and adapting it to specific scenarios, including aircraft technical parameters (engine thrust and wing area of ​​15+ mainstream aircraft models obtained from the Civil Aviation Administration's aircraft airworthiness database, such as the A320neo's engine thrust of 120kN and the B787-9's wing area of ​​325㎡), actual route operation parameters (based on the ADS-B Automatic Dependent Surveillance-Broadcast system, obtaining the actual flight distance of the flight segment and the flight duration of each stage, such as the Beijing-Shanghai route's actual flight distance of 1318km, with the cruise phase accounting for 75%), and real-time meteorological data (wind speed, temperature, and air pressure of the flight area obtained from the Civil Aviation Meteorological Center, updated every 15 minutes, with wind speed values ​​ranging from ±35m / s and temperature ranging from -55℃ to 35℃).

[0079] Secondly, dynamic calculation of fuel consumption in stages is required. This includes using the BADA 4.0 (Base of Aircraft Data) civil aviation professional fuel consumption model, dividing the entire flight process into four stages: takeoff (0-3000m altitude), climb (3000-8000m altitude), cruise (8000-12000m altitude), and descent (8000-0m altitude), calculating fuel consumption for each stage. A weather correction factor is introduced for each stage: fuel consumption increases with headwinds (8% increase for every 10m / s increase in wind speed) and decreases with tailwinds (6% decrease for every 10m / s increase in wind speed); for every 10°C deviation of the outside temperature from the standard atmospheric temperature, fuel consumption is adjusted by 3%. For example, a B737-800 operating the Guangzhou-Chengdu route encountering a 20m / s headwind during the cruise phase results in a 16% increase in fuel consumption compared to standard operating conditions, ultimately calculating a single-flight fuel consumption of 5.2 tons.

[0080] Secondly, regarding the accurate calculation of segmented carbon emission coefficients, based on the Civil Aviation Industry Carbon Emission Accounting Standard (ICAO CAEP / 11) and combined with the quality of domestic aviation fuel (density 0.81 kg / L), segmented carbon emission coefficients are set as follows: 2.85 kg CO2 / L for takeoff, 2.95 kg CO2 / L for climb, 3.1 kg CO2 / L for cruise, and 2.9 kg CO2 / L for descent. Total carbon emissions = fuel consumption at each stage × corresponding carbon emission coefficient, achieving a precise mapping between "fuel consumption and carbon emissions" and avoiding calculation errors caused by traditional single coefficients.

[0081] Finally, spatiotemporal attributes need to be assigned and quality verified. Specifically, ArcGIS spatial analysis tools can be used to map carbon emission data for each flight phase to a 1km×1km Equal Earth projection grid. Combined with the timestamps from the ADS-B system (accurate to 1 minute), the carbon emission data is assigned a three-dimensional spatiotemporal attribute of "latitude and longitude coordinates - flight phase - time," such as "2026-04-20 10:15, 30.67°N, 104.06°E (Chengdu airspace), cruise phase, carbon emissions 0.8 tons." Finally, a "measured value comparison verification" mechanism is used to compare the calculated results with the measured data from the Aircraft Engine Emission Monitoring System (EEMS) to ensure that the calculation error is controlled within ±3%, meeting the requirements for refined optimization. The above-described embodiment overcomes the limitations of traditional macroscopic accounting methods by employing phased calculations, meteorological parameter coupling, and spatiotemporal attribute mapping. Compared to existing technologies, the calculation accuracy of carbon emissions at the route-aircraft type dimension can be improved by more than 40%, with a spatiotemporal resolution of 1km×1km×1 minute. This allows for precise characterization of carbon emission differences for the same aircraft type on different routes and at different times. The generated refined data provides core support for the accurate quantification of carbon costs and the effective implementation of carbon emission constraints in subsequent models, ensuring that the emission reduction effect of the optimized scheme is quantifiable and verifiable.

[0082] Furthermore, in one embodiment, the dynamic carbon price in step S2 is calculated as follows: based on a benchmark carbon price, multiplied by a carbon market volatility coefficient and a policy adjustment coefficient, wherein the carbon market volatility coefficient is predicted using an autoregressive integral moving average time series model of historical carbon trading data. The specific technical details are explained below: Regarding the precise anchoring of the benchmark carbon price, the core data source is the national carbon market spot trading data. Daily carbon trading closing prices over the past three years (data source: Shanghai Environment and Energy Exchange) are collected, and a weighted average method (with annual trading volume as the weight) is used to calculate the benchmark carbon price for the civil aviation industry, set at 70 yuan / ton CO2. An industry correction factor is introduced during the anchoring process: because the marginal emission reduction cost of civil aviation carbon emissions is higher than the industry average, a 5% positive correction is applied to the benchmark carbon price, ultimately determining it to be 73.5 yuan / ton CO2, ensuring alignment with the emission reduction characteristics of the civil aviation industry.

[0083] Furthermore, in predicting carbon market volatility coefficients, the ARIMA(1,1,1) time series model can be used. The model parameters are adapted to the characteristics of the carbon market: a first-order autoregressive term (capturing short-term price trends), a first-order differencing term (smoothing non-stationary trading data), and a first-order moving average term (smoothing random fluctuations). The training data consists of daily carbon market trading prices over the past three years. The prediction step size matches the planning period (quarterly prediction, outputting the volatility coefficient for the next quarter at a time), with the value strictly controlled within the range of 0.8-1.2. For example, predicting a tight supply and demand in the carbon market in Q2 2027 results in a volatility coefficient of 1.15; predicting a balanced supply and demand in Q1 2028 results in a volatility coefficient of 1.02.

[0084] Furthermore, regarding the scenario-based setting of policy adjustment coefficients, based on national and regional "dual carbon" policy documents, three scenarios can be categorized according to regional emission reduction pressure and policy intensity, quantifying the policy adjustment coefficients: 1.3 for key emission reduction regions such as the Beijing-Tianjin-Hebei region and the Yangtze River Delta; 1.2 for regions with coordinated development and emission reduction such as the Guangdong-Hong Kong-Macao Greater Bay Area and the Chengdu-Chongqing region; and 1.0 for other ordinary regions. During special policy periods (such as the critical period of emission reduction during the 14th Five-Year Plan and the period for ensuring the success of major international events), an upward adjustment mechanism will be triggered, up to a maximum of 1.5, ensuring a deep connection between carbon prices and policy guidance.

[0085] Finally, dynamic coupling calculations and real-time verification are performed. The specific formula for calculating the dynamic carbon price is: Dynamic Carbon Price = Benchmark Carbon Price × Carbon Market Volatility Coefficient × Policy Adjustment Coefficient. A monthly update mechanism is established: at the beginning of each month, carbon market trading data from the previous month is collected, and the ARIMA model is retrained to update the volatility coefficient; every quarter, the policy adjustment coefficient is simultaneously optimized based on regional policy adjustments. The verification process employs an "error backtesting" mechanism, comparing the calculation results with the actual carbon trading prices of the same period to ensure that the prediction error of the dynamic carbon price is controlled within ±8%, meeting the accuracy requirements of model optimization.

[0086] The above-described embodiments achieve a breakthrough in carbon pricing, moving from a "statically fixed" to a "dynamically adaptable" approach, through the synergy of time series forecasting and policy factor quantification. Compared to existing technologies, the market response speed of carbon pricing can be improved by 70%, and the regional policy adaptability can be increased by 50%. The accurate quantification of dynamic carbon pricing enables the model to truly reflect carbon costs under different market environments and policy scenarios, avoiding the optimization bias caused by static carbon pricing. This ensures that the allocation of civil aviation resources meets both economic requirements and aligns with emission reduction policy guidelines.

[0087] Furthermore, in one embodiment, the conventional operating costs mentioned in step S2 include one or more of the following: fuel costs, aircraft ownership costs, crew manpower costs, airport charges, and maintenance costs.

[0088] The above embodiments focus on the refined accounting of traditional operating costs in step S2. Addressing the problems of crude operating cost accounting in existing technologies and the lack of integration with the dynamic characteristics and element relationships of civil aviation scenarios, a calculation scheme of "precise itemized modeling - dynamic coupling of multiple parameters - scenario-based adaptation" is proposed. This scheme can cover fuel costs, aircraft ownership costs, crew labor costs, airport fees, and maintenance costs. Specific technical details are as follows: Firstly, regarding dynamic fuel cost calculation, based on the refined fuel consumption data in step S3, and combined with real-time dynamic adjustments to fuel prices (obtained by connecting to the international crude oil futures API and the domestic aviation fuel sales database to obtain the daily aviation fuel price, with fluctuations ranging from 6-10 yuan / L), a route characteristic correction coefficient is introduced: for long routes (>2000km), due to the high proportion of cruising, fuel efficiency is improved by 5%, and the cost coefficient is taken as 0.95; for short routes (<800km), due to frequent takeoffs and landings, fuel efficiency is reduced by 8%, and the cost coefficient is taken as 1.08. For example, an A321neo operating the Beijing-Urumqi long route (2400km) consumes 6.8 tons of fuel per flight. With a daily fuel price of 7.5 yuan / L, the fuel cost = 6800L × 7.5 yuan / L × 0.95 = 48450 yuan.

[0089] Secondly, regarding the scenario-based calculation of aircraft ownership costs, the life cycle cost (LCC) analysis method can be used, categorized by aircraft type: Narrow-body aircraft (such as the A320 series) have a purchase price of approximately US$110 million and an annual depreciation rate of 5.5% (0-10 years) and 4% (11-20 years); wide-body aircraft (such as the A350 series) have a purchase price of approximately US$310 million and an annual depreciation rate of 4.8% (0-10 years) and 3.5% (11-20 years). Simultaneously, the cost of capital tied up (calculated at an annualized interest rate of 3.85%) and residual value recovery (the residual value after 20 years is 15% of the purchase price) are included. Finally, the cost is amortized to a single flight based on the number of flights per year. For example, if a 5-year-old A330 operates 300 flights per year, the ownership cost per flight = (310 million × 4.8% + 310 million × 3.85%) ÷ 300 ≈ 82,000 yuan.

[0090] Secondly, regarding the accurate measurement of crew manpower costs, in accordance with the Civil Aviation Administration's "Rules for Flight Crew Duty," crews are configured and costs are calculated based on route duration: narrow-body aircraft operate short-haul routes (<4 hours) with 2 pilots and 3 flight attendants, while long-haul routes (>6 hours) require 3 pilots and 4 flight attendants. Pilots earn 800-1200 yuan per hour (depending on their rank), and flight attendants earn 300-500 yuan per hour. Ground standby costs (calculated at 20% of flight time) and overseas allowances (800 yuan per person per day) are also included. For example, for a B787 operating the Shanghai-Los Angeles route (12 hours), the crew manpower cost = (3×1000 + 4×400)×12 + 7×800 = 62,400 yuan.

[0091] Furthermore, regarding the detailed accounting of airport charges and maintenance costs, airport charges are calculated differently based on "time period + aircraft type": during peak hours (7:00-9:00, 16:00-18:00), the gate fee for narrow-body aircraft is 12,000 yuan / time, and for wide-body aircraft it is 25,000 yuan / time, with a 30% reduction during off-peak hours; this is compounded by ground service charges (baggage handling 0.8 yuan / piece, jet bridge use 0.3 million yuan / time). Maintenance costs are accrued based on flight hours: 800 yuan per flight hour for narrow-body aircraft (e.g., A320neo), and 1,500 yuan per flight hour for wide-body aircraft (e.g., B777), and are linked to Aircraft Health Management System (AHMS) data; maintenance costs increase by 20% if parts are overdue for replacement.

[0092] The above-described embodiment solves the problem of traditional operating cost accounting being crude and having large deviations from actual operations by coupling detailed modeling of each item with civil aviation scenario parameters. It can significantly improve the accuracy of cost accounting and further reduce the error range of single-shift operating cost calculation (e.g., it can be controlled within ±5% according to calculations).

[0093] Furthermore, in one embodiment, the uncertainty demand robustness constraint mentioned in step S3 includes: for any passenger demand fluctuation scenario, the model decision result must satisfy that the deviation between actual capacity and demand does not exceed a preset value, and the constraint is transformed into a solvable linear inequality through interval mathematics.

[0094] The uncertainty demand robust constraint in step S3 of the above embodiment addresses the problem that existing technologies do not fully consider the fluctuations in civil aviation passenger demand and that the constraint design lacks anti-interference capabilities, leading to the failure of the optimization scheme in actual implementation. Therefore, a full-process constraint construction scheme of "demand fluctuation characteristic quantification - robust set modeling - linear transformation solution - scenario adaptation" is proposed. Specific technical details are as follows: In terms of accurately extracting demand fluctuation characteristics, based on nearly five years of civil aviation passenger transport data, we integrated OTA platform booking data, airline CRM system travel records, and the Civil Aviation Administration's annual transport statistics report, splitting the demand sample by route type (trunk lines, feeder lines, regional express lines) and time period (weekdays, weekends, holidays, Spring Festival / summer travel season). We used the coefficient of variation (CV) method and the 95th percentile method to quantify fluctuation characteristics: demand for trunk lines (such as Beijing-Shanghai) is stable, with a fluctuation range of ±15%; demand for feeder lines (such as Lijiang-Shangri-La) fluctuates by ±25%; during peak periods such as Spring Festival / summer travel season, the fluctuation range for all types of routes increases to ±30%. Simultaneously, we extracted demand trend factors (such as a weekly demand fluctuation coefficient of 0.8-1.2 for business routes and a weekend peak coefficient of 1.5 for tourist routes) to provide data support for constraint modeling.

[0095] Furthermore, in terms of robust constraint quantitative modeling, an interval-type uncertainty set is constructed, defining the demand interval for a single route in the t-th planning period as [D_min(t), D_max(t)], where D_min(t) is the lower limit of the 95th quantile of demand in that period, and D_max(t) is the upper limit of the 95th quantile. The core objective of robust constraints is that, regardless of how demand fluctuates within the interval, the capacity configuration output by the model (aircraft type and number of seats × flight frequency × maximum load factor of 85%) must satisfy "the relative deviation between capacity and demand ≤ 5%", avoiding capacity shortages or waste caused by traditional constraints only adapting to a single demand scenario. For example, if the demand interval for the Guangzhou-Sanya tourist route in Q2 2027 is [12,000-18,000 passengers / week], then the capacity needs to be controlled within [11,400-18,900 passengers / week] to ensure coverage of extreme fluctuation scenarios.

[0096] Furthermore, regarding the linear transformation and solution adaptation of constraints, considering the nonlinear characteristics of robust constraints, interval mathematics and duality theory are used to transform them into linear inequalities solvable by the mixed-integer dynamic programming model. The transformation logic is as follows: for any demand d∈[D_min(t), D_max(t)], it must satisfy 0.95d≤C_i×f_ij(t)≤1.05d (where C_i is the number of seats for aircraft type i, and f_ij(t) is the flight frequency of route j in period t). Through extreme value analysis, this is simplified to two linear constraints: C_i×f_ij(t)≥0.95×D_max(t) (covering the maximum demand scenario) and C_i×f_ij(t)≤1.05×D_min(t) (covering the minimum demand scenario), ensuring that the constraint transformation has no accuracy loss and is compatible with other constraints in the model (airport capacity, fleet availability).

[0097] Finally, scenario-based constraint parameter adaptation is performed. A dynamic adjustment mechanism for constraint parameters is established, with parameters set differently based on route attributes and time period characteristics: for business trunk routes (such as Shanghai-Beijing) with high demand stability, the deviation threshold is tightened to 3%, with fluctuations calculated at ±15%; for feeder tourist routes, the deviation threshold can be relaxed to 5%, with fluctuations calculated at ±25%; during special periods such as major events and holidays, a temporary threshold adjustment mechanism is triggered (up to 6%), while limiting the month-on-month change in capacity between adjacent periods to ≤10%, avoiding operational chaos caused by frequent adjustments. Monte Carlo simulation (1000 random demand scenarios) is used to verify the effectiveness of the constraints, ensuring a constraint satisfaction rate of ≥99%.

[0098] The above-described embodiment addresses the shortcomings of existing technologies, such as poor anti-interference capabilities and limited adaptability, by accurately quantifying demand fluctuations and linearly transforming robust constraints. In demand-fluctuation scenarios, this embodiment improves landing success rates, reduces capacity shortages, and avoids capacity waste. Through the coordinated adaptation of robust constraints with other model constraints, the multi-period element configuration scheme satisfies both global optimization objectives and possesses strong anti-interference capabilities, significantly enhancing its reliability and operability in complex real-world operational scenarios.

[0099] Furthermore, in one embodiment, the specific operation mode of the fusion solution algorithm framework described in step S3 includes: Initialization phase: The initial solution space of the decision variables is generated using an adaptive genetic algorithm.

[0100] Iterative optimization phase: Deterministic constraints are applied to the initial solution space using mixed-integer dynamic programming to screen feasible solutions; an adaptive genetic algorithm is used to perform crossover and mutation operations based on the objective function values ​​of feasible solutions to optimize the solution space.

[0101] Convergence criterion: When the fluctuation of the objective function value after a preset number of iterations is less than a preset fluctuation value, the optimal solution is output.

[0102] The above embodiments address the high-dimensional discreteness (15+ aircraft types, 1500+ routes, 40 planning cycles), multi-constraint coupling (temporal constraints, resource constraints, robust constraints), and global optimality requirements of the civil aviation element allocation problem by using a fusion solution framework of "hybrid integer dynamic programming and adaptive genetic algorithm" in step S3. A five-dimensional solution scheme is proposed, consisting of "scenario-based encoding - hierarchical constraint processing - deep collaboration between dual algorithms - dynamic parameter adaptation - global convergence verification," overcoming the bottlenecks of traditional single algorithms such as "slow convergence, susceptibility to local optima, and poor adaptability." Specific technical details are as follows: First, in the initialization phase: customized population generation and coding design for civil aviation. A hybrid coding method of "three-layer gene structure + industry rule guidance" is adopted to ensure the feasibility and quality of the initial solution. The chromosome length is designed according to the four dimensions of "aircraft type-route-airport-cycle", with a total length of 15×1500×8 (time period)×40 (cycle) = 7,200,000 bits, which is suitable for high-dimensional decision space.

[0103] The three-layer gene structure is as follows: The first layer (aircraft type selection gene) uses integer encoding (1-15 correspond to 15 mainstream aircraft types, such as 1=A320neo, 5=B787-9, 10=ARJ21); the second layer (flight frequency gene) uses real number encoding (3-20 flights / week, step size 1 flight, supports integer solution output); the third layer (airport time slot allocation gene) uses binary encoding (0-1 indicates whether a certain aircraft type occupies a specific gate during a certain time period, such as "1" representing that the A330 occupies wide-body gate No. 1 at Guangzhou Baiyun Airport from 7:00 to 8:00).

[0104] The initial population size is dynamically set according to the complexity of the planning cycle: 500 individuals for short-term cycles (1-2 years) and 800 individuals for medium- to long-term cycles (3-10 years). Three types of industry rules are incorporated during population generation: hub routes (e.g., Beijing-Shanghai) prioritize wide-body aircraft (initial gene values ​​biased towards 5-8), feeder routes (e.g., Ganzhou-Xiamen) prioritize narrow-body aircraft (initial gene values ​​biased towards 1-4 and 10-12), and during peak hours (7:00-9:00 and 16:00-18:00), the proportion of aircraft stand allocation with gene "1" does not exceed 85% of the airport's capacity. Simultaneously, the best historical configuration data from the past 3 years (accounting for 30%) is introduced for mutation optimization to avoid inefficiencies caused by random search, increasing the feasibility rate of the initial solution to over 85%.

[0105] Secondly, in the iterative optimization phase: hierarchical collaboration of the two algorithms and constraint handling. The specific steps are as follows: 1. Layered Constraint Processing in Mixed Integer Dynamic Programming: The constraint system is layered into "hard constraints - soft constraints - temporal constraints," and applied one by one through the state transition equation of dynamic programming. Hard constraints (airport capacity, fleet availability) are directly embedded in the state transition matrix, defining the state variables as "fleet size at the end of period t, airport gate occupancy status," and the state transition equation is S(t+1)=f(S(t),x(t)), where x(t) is the decision variable for period t, ensuring that the core constraints are satisfied in each iteration. Soft constraints (robust demand constraints) are transformed into objective function terms through a penalty function. Solutions that do not satisfy "capacity-demand deviation ≤ 5%" are penalized by adding a penalty cost equal to the deviation rate × 10% × the total system cost. Temporal constraints (fleet updates in adjacent periods, route network evolution) are integrated through the stage benefit function of the Bellman equation, where stage benefit = current period total cost - cross-period adjustment cost (fleet addition / retirement cost, route addition / reduction cost), ensuring the continuity of multi-period solutions.

[0106] 2. Discrete Optimization and Dynamic Parameter Adjustment of Adaptive Genetic Algorithm: Based on the feasible solution subset selected by dynamic programming (rejection rate 25%-35%), the algorithm parameters are adjusted hierarchically according to the fitness value (the inverse of the total system cost). The top 20% of excellent individuals have a crossover probability of 0.6 and a mutation probability of 0.03, employing a "route cluster crossover" strategy (dividing route clusters into 7 major regions such as North China and East China, with crossover only within the same cluster, preserving regional capacity allocation logic); the middle 60% of ordinary individuals have a crossover probability of 0.9 and a mutation probability of 0.07, employing a "random fragment crossover" strategy (crossover fragment length is 50-100 genes, covering multiple routes); the bottom 20% of inferior individuals have a crossover probability of 0.95 and a mutation probability of 0.1, forcibly introducing new gene fragments. The mutation operation incorporates civil aviation heuristic rules: priority is given to mutating the aircraft type or frequency genes of short routes (<800km) and low frequency (<5 flights / week) to avoid operational fluctuations on long routes and high frequency routes; when mutating the genes of fleet updates, the rule of "annual retirement rate ≤5% and fuel efficiency of new aircraft type ≥15% higher than that of retired aircraft type" must be met.

[0107] Furthermore, regarding the escape from local optima and dynamic parameter adaptation mechanisms, a "triple escape mechanism" is designed to break the local optimum trap: ① When convergence fails after 1000 iterations, the mutation probability is automatically increased to 0.15, and a new population of 30% is regenerated (adjusting the initial gene distribution based on the latest market data); ② When the optimal solution remains unchanged for 30 consecutive iterations, a "constraint relaxation iteration" is triggered, temporarily relaxing the airport capacity constraint to 90%, allowing exploration of a better solution space, and restoring the original constraint after convergence; ③ During cross-cycle iterations, 50% of the gene fragments from the previous cycle's optimal solution are recombinated with 50% of randomly generated fragments to avoid local optima in the temporal dimension. Simultaneously, a "cycle-complexity-parameter" mapping table is established: short-term cycle (1-2 years) 1000 iterations, crossover probability baseline 0.75; medium-term cycle (3-5 years) 1500 iterations, crossover probability baseline 0.8; long-term cycle (6-10 years) 2000 iterations, crossover probability baseline 0.85. After each iteration, the population size is dynamically adjusted according to the feasible solution rate: when the feasible solution rate is <60%, the population size is increased by 20%; when the feasible solution rate is >90%, the population size is reduced by 10%, balancing the solution efficiency and solution quality.

[0108] Finally, global convergence verification and solution output optimization are performed. Three-dimensional convergence criteria are set: ① Objective function convergence: The total system cost fluctuation is ≤0.1% over 50 consecutive iterations, and the difference between the optimal and suboptimal solutions is ≤0.3%; ② Constraint satisfaction rate convergence: The proportion of feasible solutions that satisfy all constraints (100% satisfaction of hard constraints, and ≥99.5% satisfaction rate of soft constraints) is ≥99.8%; ③ Solution stability convergence: The deviation of the optimal solution in 10 independent repeated solutions is ≤0.5%. After convergence, the optimal solution undergoes "engineering adaptation optimization": the flight frequency is adjusted to an integer (e.g., 3.2 flights / week is corrected to 3 flights / week), the time slot allocation is corrected according to the actual airport gate layout (to avoid gate conflicts), and a full-dimensional configuration table including "aircraft type-route-frequency-time slot-fleet update" is output, along with a constraint satisfaction verification report for each period.

[0109] The above-described solution, through a five-dimensional fusion solution framework deeply adapted to civil aviation scenarios, improves solution efficiency by over 70% (reducing the solution time for a 10-year planning period from 6 hours to 1.2 hours) compared to traditional single mixed-integer programming algorithms. It also increases the probability of obtaining the global optimal solution by 50% (avoiding local optima by reducing the percentage from 40% to 15%) and raises the constraint satisfaction rate from 95% to 99.8%. The algorithm's hierarchical constraint processing and dynamic parameter adaptation mechanism ensure the feasibility and optimality of solutions in high-dimensional complex scenarios. Simultaneously, cross-cycle collaborative iteration guarantees the coherence of multi-stage configuration schemes, providing efficient and reliable solution support for the dynamic configuration of civil aviation elements and significantly reducing the engineering implementation cost of the optimization scheme.

[0110] Furthermore, in one embodiment, the carbon emission calculation constraint in step S3 includes: establishing a mathematical relationship between the decision variable and the carbon emission amount E: E = k × f(i,j), where f(i,j) is the carbon emission amount of aircraft type i on route j per flight. This mathematical relationship is embedded as an equality constraint into the mixed-integer dynamic programming model, and an upper limit constraint on the total carbon emission is added.

[0111] This embodiment addresses the issues in existing technologies regarding carbon emission calculation constraints, such as loose correlation between carbon emissions and decision variables, static constraint design, and failure to adapt to regional emission reduction differences and multi-period target progression. It proposes an integrated constraint construction scheme encompassing "refined mathematical modeling, scenario-based total constraint, dynamic adaptation and adjustment, multi-constraint synergy and compatibility, and full-process verification." This achieves deep coupling between carbon emission constraints and civil aviation resource allocation decisions, ensuring that emission reduction targets are quantifiable and implementable. Specific technical details are as follows: First, a refined mathematical model of carbon emissions and decision variables is developed.

[0112] A mathematical relationship based on "three-dimensional correlation + phased correction" is constructed to accurately characterize the impact of decision variables on carbon emissions. The core mathematical expression is: E(t,i,j) = f(i,j) × k(t,i,j) × α(t,i,j) × β(j) The civil aviation scenario-based definitions and calculation logic for each parameter are as follows: E(t,i,j): Carbon emissions (in tons of CO2) for the t-th planning period (divided by quarter, t=1-40), aircraft type i, and route j.

[0113] f(i,j): The baseline single-flight carbon emissions of aircraft type i on route j (unit: tons of CO2 / flight), determined by the refined data in step S1 (e.g., A320neo flying the Shanghai-Guangzhou route f=28 tons / flight, B787-9 flying the Beijing-Los Angeles route f=85 tons / flight).

[0114] k(t,i,j): Flight frequency of aircraft type i on route j in period t (decision variable, integer type, 3-20 flights / week).

[0115] α(t,i,j): The dynamic correction coefficient for the t-th period, which is related to real-time weather and aircraft status. α = 1 + 0.08 × (headwind speed / 10) - 0.06 × (tailwind speed / 10) + 0.03 × (actual temperature - standard atmospheric temperature) / 10, with a value range of 0.8-1.2. It is updated monthly based on data from the Civil Aviation Meteorological Center.

[0116] β(j): Regional correction coefficient for route j, set according to the emission reduction policies of the region where the route takes off and lands. For example, β=1.1 for key regions such as Beijing-Tianjin-Hebei and Yangtze River Delta, β=1.05 for coordinated regions such as Chengdu-Chongqing and Guangdong-Hong Kong-Macao, and β=1.0 for other regions.

[0117] Simultaneously, this mathematical relationship is transformed into an equality constraint embedded in a mixed-integer dynamic programming model, namely: E(t,i,j) - f(i,j) × k(t,i,j) × α(t,i,j) × β(j) = 0 Ensure that each set of decision variables (k(t,i,j)) can accurately map to the corresponding carbon emissions, and avoid the loss of control over emission reduction effects caused by the fuzzy "decision-emission" relationship in traditional constraints.

[0118] Furthermore, a limit on the total amount of carbon emissions for specific scenarios is set.

[0119] Based on the national "dual-carbon" policy and civil aviation industry planning, a three-dimensional upper limit for total carbon emissions is set according to "region-cycle-aircraft type" to avoid localized emission reduction imbalances caused by a single total constraint. The specific setting logic is as follows: 1. Regional Total Constraints: Referring to the "Provincial Greenhouse Gas Emission Quota Allocation Scheme", the country is divided into 7 major regions. The total carbon emission limit for each region in period t is Q_region(t) = regional baseline quota × (1 - emission reduction rate)^t. The regional baseline quota is based on the actual measured value of regional civil aviation carbon emissions in 2025 (e.g., the baseline quota for East China region in 2025 is 50 million tons of CO2). The emission reduction rate is set progressively according to the following rates: 5% in the 14th Five-Year Plan, 8% in the 15th Five-Year Plan, 10% in the 16th Five-Year Plan, and 12% in the 17th Five-Year Plan. 2. Total Aircraft Capacity Constraints: Differentiated upper limits are set for aircraft models with different emission reduction potential. For older aircraft models (≥15 years old), Q_age(t) = initial emissions × (1-0.15)^t (accelerated phase-out), and for new energy-efficient aircraft models (such as A320neo and B787 series), Q_new(t) = initial emissions × (1+0.05)^t (moderate expansion), ensuring that total capacity constraints are coordinated with fleet renewal planning. 3. Periodic Progression Constraint: Set a cross-period total amount progression rule. The upper limit of the total amount in the region in the (t+1)th period is Q_region(t+1) = Q_region(t)×(1-emission reduction rate), and the emission reduction rate in a single period shall not exceed 10% to avoid operational interruption caused by excessively rapid emission reduction.

[0120] The mathematical expression for the total quantity constraint is: Σ(i,j) E(t,i,j) ≤ Q_region(t) Σ(j) E(t,i_old,j) ≤ Q_age(t) Σ(j) E(t,i_new,j) ≤ Q_new(t) Where i_old represents the set of models with an age of ≥15 years, and i_new represents the set of energy-saving models.

[0121] Furthermore, dynamic constraint adaptation and multi-constraint synergy compatibility are implemented. This is achieved by establishing a dynamic constraint adjustment mechanism driven by three factors: policy, market, and technology. ① Policy adjustment: When regional emission reduction policies are strengthened (such as during the support period for major international events), the total emission limit will be temporarily reduced by 5%, and the constraints on fleet renewal will be relaxed (the annual retirement rate can be increased to 8%).

[0122] ② Market adjustment: When carbon prices rise by more than 100 yuan / ton, the total limit can be appropriately relaxed by 3% to balance emission reduction and economic efficiency.

[0123] ③ Technological Adjustment: When the adoption rate of new zero-carbon fuels (such as sustainable aviation fuel SAF) reaches 10%, the total emission limit will be reduced by 6% in tandem with the SAF emission reduction ratio (such as SAF emission reduction of 60%).

[0124] Furthermore, to address the potential conflict between carbon emission constraints and constraints such as airport capacity and fleet availability, this embodiment of the invention designs a "constraint priority coordination mechanism": hard constraints (airport capacity and fleet availability) have the highest priority, while carbon emission constraints serve as the core soft constraint. When a conflict occurs (e.g., carbon emissions exceed the limit if airport capacity is met), the objective function is adjusted through a penalty function. The excess portion is penalized at twice the carbon price, forcing decision variables to optimize towards "low emissions and high efficiency." At the same time, the proportion of penalty costs to the total system cost is limited to no more than 15%, avoiding excessive emission reduction that could lead to uncontrolled operating costs.

[0125] Furthermore, full-process constraint verification and error control are carried out.

[0126] Construct a full-process verification system of "modeling-solving-output": ① Modeling stage: Use Monte Carlo simulation to simulate 1000 demand and weather fluctuation scenarios to verify the accuracy of the constraint mathematical relationship and ensure that the error is ≤3%; ② Solving stage: Verify the constraint satisfaction every 100 iterations. Solutions that exceed carbon emission standards are directly eliminated, and solutions that violate hard constraints are penalized according to the degree of deviation; ③ Output stage: Compare the carbon emission calculation value of the optimized solution with the regional quota, and cross-validate it with the measured data of the Aircraft Engine Emission Monitoring System (EEMS) to ensure that the actual carbon emission after the solution is implemented deviates from the constraint target by ≤5%.

[0127] The above-described embodiments, through refined mathematical modeling, scenario-based total constraints, and dynamic adaptation mechanisms, address the problems of "loose correlation, static rigidity, and poor coordination" in traditional carbon emission constraints. Compared to existing technologies, they improve the accuracy of the correlation between carbon emissions and decision variables and the adaptability to regional emission reduction targets, thereby significantly reducing the incidence of multi-constraint conflicts. The deep synergy between carbon emission constraints, model objective functions, and other constraints ensures that the optimized solution accurately achieves emission reduction targets while meeting operational needs and economic efficiency. This enables a three-dimensional balance of "efficiency-cost-emission reduction" in civil aviation resource allocation, ensuring the phased progress of the industry's carbon neutrality goals.

[0128] Furthermore, in one embodiment, the dynamic configuration scheme for core civil aviation elements in step S4 further includes: introducing aircraft type update decision variables in the medium- and long-term planning cycle to enable the scheme to have dynamic iterative adaptability.

[0129] The above embodiments address the shortcomings of existing optimization schemes, such as insufficient consideration of the impact of medium- and long-term technological progress on resource allocation and difficulty in adapting to new aircraft model iterations and low-carbon technology upgrades. To address these shortcomings, an optimization design is proposed: "quantification of technological progress factors - embedding of aircraft model update decisions - multi-cycle adaptation and collaboration." Specific technical details are as follows: Regarding the scenario-based quantification and parameter definition of technological progress factors, the core technological progress factors are clearly defined as improved fuel efficiency of new aircraft models, compatibility with Sustainable Aviation Fuel (SAF), and technological maturity. Quantitative parameters are set in conjunction with the technological iteration patterns of the civil aviation industry: the improvement rate of fuel efficiency of new aircraft models compared to existing older aircraft models (≥15 years old) is progressively based on planning stages: ≥15% during the 15th Five-Year Plan period (2026-2030), ≥25% during the 16th Five-Year Plan period (2031-2035), and ≥35% during the 17th Five-Year Plan period (2036-2040); the SAF compatibility ratio of new aircraft models is ≥50% (meeting the SAF application standards of the Civil Aviation Administration of China); and the technological maturity is divided into thresholds according to the EASA / FAA airworthiness certification stages, with new aircraft models only included in the configuration scheme candidate set after passing TC (Type Certificate) + PC (Production Certificate) certification.

[0130] Secondly, regarding the embedded design of aircraft model update decision variables, integer aircraft model update variables have been added to the existing decision variable system, including "number of new aircraft introduced," "number of old aircraft retired," and "priority of new aircraft-route matching." Update trigger rules are set: when the carbon emissions per shift of an old aircraft model are more than 20% higher than those of a new aircraft of the same class, or when the aircraft's age reaches 18 years and its maintenance costs are 30% higher than the industry average, mandatory retirement is triggered. New aircraft models are prioritized for long-haul routes (>2000km), high-frequency routes (≥15 flights / week), and routes in key emission reduction areas. Matching priority is quantified through "emission reduction contribution + operating revenue ratio" (priority = 0.6 × emission reduction rate + 0.4 × revenue improvement rate).

[0131] Secondly, regarding multi-cycle technology adaptation and constraint coordination, aircraft upgrade decisions are deeply integrated with multi-cycle planning, setting a three-stage adaptation cycle of "introduction-adjustment-expansion": In the introduction stage (year 1), the proportion of new aircraft deployed is ≤5% of the total fleet, with 10% reserved for crew training and airport adaptation (such as jet bridge renovation and maintenance equipment upgrades); in the adjustment stage (years 2-3), the deployment proportion is adjusted to 10%-15% based on actual operational data (such as on-time performance ≥92%, maintenance costs ≤105% of budget); in the expansion stage (year 4 and beyond), it is gradually increased to 20%-30% based on technological maturity and market demand. Simultaneously, airport adaptation constraints are added to ensure that target airports meet hardware requirements such as runway strength and parking space size before the introduction of new aircraft, avoiding conflicts in resource allocation.

[0132] Furthermore, regarding the dynamic adjustment and feasibility verification of the plan, a dynamic tracking mechanism for technological progress will be established. Every six months, the Civil Aviation Administration's new aircraft type airworthiness announcements and SAF industry development reports will be collected to update parameters such as fuel efficiency improvement rate and technology maturity. Through system dynamics simulation of the operation scenario after the new aircraft type is put into use, the matching degree of route capacity, airport support capacity and emission reduction target achievement will be verified to ensure the feasibility and optimization of the plan under the technological iteration scenario.

[0133] The above embodiments address the pain points of traditional solutions, namely "static rigidity and poor technological adaptability," by quantitatively embedding technological advancements and using multi-cycle adaptation design. This enables the configuration scheme to proactively adapt to new model iterations and low-carbon technology upgrades, avoiding the risk of operational interruptions caused by technological iterations.

[0134] In summary, the embodiments of the present invention achieve dynamic optimization of civil aviation element configuration by optimizing multi-source data fusion algorithms, deepening carbon emission calculation methods, and adjusting solver parameters. This reduces the "system cost prediction error" of the model output, improves the accuracy of carbon emission calculation, and effectively controls carbon emissions.

[0135] Figure 2 An exemplary structural block diagram of an electronic device 200 according to an embodiment of the present invention is shown.

[0136] like Figure 2 As shown, the electronic device 200 specifically includes a memory 201, a processor 202, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it can implement the method described in any of the embodiments of method 100 described above. It is understood that the electronic device is a specific implementation of the aforementioned method 100, therefore the foregoing text is combined with... Figure 1 The characteristics described can be similarly applied here, and will not be repeated here.

[0137] While numerous embodiments of the invention have been shown and described herein, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of the invention. The appended claims are intended to define the scope of protection of the invention and therefore cover equivalents or alternatives within the scope of these claims.

Claims

1. A method for optimizing the allocation of civil aviation elements based on mixed-integer dynamic programming, characterized in that, Includes the following steps: S1: Construct a database of core elements and spatiotemporal characteristics of carbon emissions in China's civil aviation industry. The database integrates multi-source heterogeneous data through a spatiotemporal data fusion algorithm. The data includes at least airport geographic information, route network, fleet data, flight operation data, refined carbon emission data at the route-aircraft type dimension, and passenger travel demand fluctuation data. S2: Establish a mixed-integer dynamic programming model with the goal of minimizing the total system cost during the planning period. The total system cost includes traditional operating costs and embedded carbon costs, passenger time costs, and policy compliance costs. The carbon cost is calculated by multiplying carbon emissions by the dynamic carbon price, and the dynamic carbon price is associated with the carbon market volatility coefficient. S3: Establish the decision variables, constraints, and fusion solution algorithm framework required for the mixed-integer dynamic programming model. The decision variables include integer aircraft type-route flight frequency allocation variables and airport resource time slot allocation variables. The constraints include demand satisfaction constraints, airport capacity constraints, fleet availability constraints, carbon emission calculation constraints, uncertain demand robustness constraints, and adjacent period dynamic correlation constraints. The fusion solution algorithm framework is a coupled framework of mixed-integer dynamic programming and adaptive genetic algorithm, wherein mixed-integer dynamic programming handles deterministic constraints, and adaptive genetic algorithm optimizes the search efficiency of discrete decision variables. S4: Based on the aforementioned fusion solution algorithm framework, the mixed-integer dynamic programming model is solved using a mathematical programming solver that integrates a robust optimization module, in order to obtain a dynamic configuration scheme for the core civil aviation elements that optimizes the total system cost over multiple future periods and satisfies the uncertainty scenario.

2. The method according to claim 1, characterized in that, The spatiotemporal data fusion algorithm described in step S1 includes: using a spatial interpolation algorithm to correct coordinate deviations for airport geographic information, using a temporal smoothing algorithm to eliminate outliers for flight operation data, and using a feature alignment algorithm to achieve consistent matching of time and space dimensions for multi-source data.

3. The method according to claim 1, characterized in that, The refined route-aircraft carbon emission data mentioned in step S1 is generated through a "bottom-up" dynamic calculation method. The "bottom-up" dynamic calculation method includes using the real-time fuel efficiency of the aircraft and the actual distance of the route, calculating the carbon emission coefficient using the segmented carbon emission coefficient method, and assigning it spatiotemporal attributes of latitude and longitude and timestamp through a geographic information system.

4. The method according to claim 1, characterized in that, The dynamic carbon price mentioned in step S2 is calculated by multiplying the benchmark carbon price by the carbon market volatility coefficient and the policy adjustment coefficient. The carbon market volatility coefficient is predicted by an autoregressive integral moving average time series model of historical carbon trading data.

5. The method according to claim 1, characterized in that, The traditional operating costs mentioned in step S2 include one or more of the following: fuel costs, aircraft ownership costs, crew labor costs, airport charges, and maintenance costs.

6. The method according to claim 1, characterized in that, The robust constraint on uncertain demand mentioned in step S3 includes: for any scenario of fluctuating passenger demand, the model decision result must satisfy that the deviation between actual capacity and demand does not exceed a preset value, and this constraint is transformed into a solvable linear inequality through interval mathematics.

7. The method according to claim 1, characterized in that, The specific operation mode of the fusion solution algorithm framework described in step S3 includes: Initialization phase: The initial solution space for the decision variables is generated using an adaptive genetic algorithm; Iterative optimization phase: Deterministic constraints are imposed on the initial solution space using mixed-integer dynamic programming to screen feasible solutions; an adaptive genetic algorithm is used to perform crossover and mutation operations based on the objective function values ​​of feasible solutions to optimize the solution space; Convergence criterion: When the fluctuation of the objective function value after a preset number of iterations is less than a preset fluctuation value, the optimal solution is output.

8. The method according to claim 1, characterized in that, The carbon emission calculation constraints mentioned in step S3 include: Establish the mathematical relationship between decision variables and carbon emissions E: E=k×f(i,j), where f(i,j) is the carbon emissions of aircraft type i on route j per flight; The mathematical relationship is embedded as an equality constraint in the mixed-integer dynamic programming model, and an upper limit constraint on total carbon emissions is added.

9. The method according to claim 1, characterized in that, The dynamic configuration scheme for core civil aviation elements described in step S4 also includes: introducing aircraft type update decision variables in the medium- and long-term planning cycle to make the scheme dynamically iterative and adaptable.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1-7.

Images