Intelligent aviation oil optimization method and related equipment

Abstract

The invention discloses an intelligent aviation fuel optimization method and related equipment, and the method comprises the steps: constructing an aviation fuel consumption prediction model which is used for predicting the fuel consumption of each leg, and providing basic data for the subsequent refueling optimization calculation; a mathematical optimization method is adopted, the minimum fuel cost and the minimum fuel consumption serve as optimization objectives, and the optimal refueling amount of each leg is calculated; iterative optimization is carried out through a gradient descent algorithm, and an optimal refueling scheme is solved on the premise that preset constraint conditions are met. According to the method, a feasible fuel optimization scheme can be provided with higher precision and lower calculation cost by combining data-driven intelligent prediction and a mathematical optimization algorithm. Compared with a traditional method, the method can adapt to changes of different air routes, weather and market environments, and an airline company can make an aviation oil filling plan more flexibly and efficiently. The method can be widely applied to the aviation oil management field.

Description

Technical Field

[0001] The present invention relates to the field of aviation fuel management, and in particular to an intelligent aviation fuel optimization method and related equipment. Background Art

[0002] In the modern air transport industry, fuel costs are a significant component of airline operating expenses, typically accounting for 30% to 40% of total operating costs. Due to the long-term fluctuations in international oil prices, as well as the influence of factors such as national tax policies, supply chain conditions, and market competition, effectively optimizing fuel consumption has become a major challenge for airlines. Especially with the increasing complexity of global route networks and the increase in flight segments, precise fuel management can not only reduce operating costs but also improve overall flight scheduling efficiency.

[0003] Traditional aviation fuel management relies primarily on rules of thumb and static calculation models. For example, some airlines employ fixed refueling strategies, such as filling tanks as much as possible at airports with lower fuel prices to reduce refueling needs at airports with higher fuel prices. However, this approach may not be optimal, as carrying excess fuel on an aircraft results in additional weight, which increases fuel consumption. Additionally, some companies use linear regression or simple optimization models based on historical data to predict fuel consumption, but these methods often struggle to accurately adapt to variations in fuel consumption under varying flight conditions, such as wind speed, temperature, and altitude.

[0004] In recent years, data-driven optimization methods have been gradually applied to aviation fuel management. Some research and practice have begun to introduce machine learning models to predict fuel consumption and combine them with mathematical optimization methods to formulate refueling strategies. For example, models trained on historical flight data can provide more accurate fuel consumption predictions, thereby guiding aviation fuel management decisions. However, in practical applications, these methods still have certain limitations, such as high computational complexity, slow model iteration convergence, and difficulty in quantifying the impact of fuel price differences between different airports. Therefore, how to combine advanced data analysis methods to improve the accuracy of aviation fuel consumption predictions and develop more efficient and cost-effective refueling plans remains an urgent problem to be solved in the industry. Summary of the Invention

[0005] In order to solve at least one of the technical problems existing in the prior art to a certain extent, the purpose of the present invention is to provide an intelligent aviation fuel optimization method and related equipment based on neural network prediction, regression model replacement calculation, and gradient descent optimization.

[0006] The first technical solution adopted by the present invention is:

[0007] An intelligent aviation fuel optimization method comprises the following steps:

[0008] A data-driven deep learning model is used to build a fuel consumption prediction model to predict fuel consumption for each flight segment, providing basic data for subsequent refueling optimization calculations.

[0009] Using mathematical optimization methods, with minimum fuel cost and minimum fuel consumption as optimization goals, calculate the optimal refueling amount for each flight segment;

[0010] Through iterative optimization using the gradient descent algorithm, the optimal refueling plan is solved while satisfying the preset constraints.

[0011] Furthermore, factors influencing fuel consumption include: flight segment characteristics, flight conditions, aircraft performance parameters, and load and fuel status. By analyzing historical flight data, a mapping relationship between fuel consumption and various influencing factors is established;

[0012] The expression of the oil consumption prediction model is:

[0013] Y=f(X)+∈

[0014] Where X represents the input variables, i.e., the factors affecting fuel consumption; Y represents the predicted fuel consumption; and ∈ is the unavoidable error term.

[0015] Furthermore, the oil consumption prediction model consists of multiple fully connected layers, each of which extracts and transforms data features through a nonlinear activation function (ReLU); the loss function of the model uses the mean square error:

[0016]

[0017] in, is the model prediction value, Y i is the actual fuel consumption, and N is the number of samples. During the model training process, the gradient descent algorithm is used to optimize the parameters to ensure that the loss function is minimized.

[0018] Furthermore, the intelligent aviation fuel optimization method further includes a step of evaluating the accuracy of the model:

[0019] Cross-validation is used to evaluate the accuracy of the model; the evaluation indicators include:

[0020] 1) Mean absolute error: measures the average difference between the predicted value and the true value;

[0021] 2) Mean square error: evaluates the square of the prediction error to avoid positive and negative errors canceling each other out;

[0022] 3) R 2 Coefficient of determination: measures the model's ability to explain the data variance. The closer the value is to 1, the better the fitting effect.

[0023] Furthermore, during the iterative calculation process of optimizing the refueling plan, calling the neural network model to calculate fuel consumption at each step would result in excessive computational overhead. Therefore, the next step is to introduce polynomial regression fitting, replacing the neural network with a mathematical function to improve computational efficiency and ensure rapid convergence of the optimization process:

[0024] Assume that the fuel consumption Y is determined by the following two variables: 1) U: the amount of fuel (tons) for the flight segment; 2) F: the total weight of fuel (tons) at takeoff of the flight segment;

[0025] Using bivariate polynomial regression, the following approximate function is constructed:

[0026] Y=a0+a1U+a2F+a3U 2 +a4UF+a5F 2 +∈

[0027] Where Y is the predicted fuel consumption; a0, a1, …, a5 are regression coefficients obtained by data fitting; ∈ is the random error term.

[0028] Furthermore, the prediction results of the neural network are used as training data to fit the polynomial regression. The training steps are as follows:

[0029] Sampling data points: Fixed flight environment variables, selected different combinations of refueling amount U and initial fuel weight F, and input them into the neural network to calculate fuel consumption Y;

[0030] Regression fitting: Use the least squares method to solve the regression coefficient a i , so that the fitting error is minimized;

[0031] Accuracy verification: by calculating the mean square error and the coefficient of determination R 2 , to ensure that the fitting accuracy is high enough.

[0032] Furthermore, the optimization objectives of minimizing fuel cost and fuel consumption include:

[0033] Assume that the flight consists of N segments and the fuel price of segment i at the departure airport is P i The amount of fuel in this section is U i , the total fuel cost is expressed as:

[0034]

[0035] The optimization goal of minimum fuel cost is:

[0036]

[0037] Assume that the fuel consumption of flight segment i is calculated by the fitted polynomial regression equation, that is:

[0038] Yi =f(U i ,F i )

[0039] Among them, Y i is the fuel consumption of flight segment i, U i is the amount of fuel at the departure point of the flight segment, F i is the weight of the aircraft's fuel at takeoff;

[0040] The optimization goal of minimizing total fuel consumption is:

[0041]

[0042] The optimization objectives of minimizing fuel cost and minimizing total fuel consumption are weighted and solved to obtain the final optimization objective.

[0043] Furthermore, the preset constraints include:

[0044] 1) Fuel reserve constraints:

[0045] At the destination airport of each flight segment, the aircraft must retain at least a predetermined amount of fuel as a reserve, namely:

[0046] F i+1 ≥R

[0047] Among them, F i+1 is the fuel weight at the end of segment i; R is the minimum reserve fuel;

[0048] 2) Takeoff weight constraints:

[0049] The aircraft's gross takeoff weight cannot exceed its maximum takeoff weight, which is:

[0050] W takeoff,i =W dry +P i +U i +F i ≤MTOW

[0051] Among them, W dry is the empty weight of the aircraft; P i is the effective load of the flight segment; U i is the amount of fuel for this segment; F i is the fuel weight at takeoff;

[0052] 3) Landing weight restrictions

[0053] The total weight of the aircraft upon arrival at the airport cannot exceed its maximum landing weight, that is:

[0054] W landung,i =W dry +Pi +(U i +F i -Y i )≤MLW

[0055] 4) Maximum fuel capacity constraints

[0056] The amount of fuel carried by an aircraft cannot exceed its maximum fuel capacity C, that is:

[0057] F i +U i ≤C

[0058] 5) Fuel mobility constraints

[0059] Fuel consumption must meet flight requirements, namely:

[0060] F i+1 =F i +U i -Y i .

[0061] Furthermore, the iterative optimization using the gradient descent algorithm to solve the optimal refueling plan under the premise of satisfying the preset constraints includes:

[0062] Adjust the refueling amount U for each segment by gradient descent method i , so that the objective function is gradually reduced while ensuring that all constraints are satisfied;

[0063] In each iteration, the refueling amount is adjusted according to the gradient of the objective function, and the formula is as follows:

[0064]

[0065] in, is the amount of refueling at the tth iteration; α is the learning rate; J is the objective function; is the partial derivative of the objective function with respect to the refueling amount, which indicates the effect of the change in refueling amount on the objective function;

[0066] After reaching the iteration termination condition, the optimal refueling plan is output.

[0067] The second technical solution adopted by the present invention is:

[0068] An electronic device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, at least one program, the code set, or the instruction set is loaded and executed by the processor to implement an intelligent aviation fuel optimization method as described above.

[0069] The third technical solution adopted by the present invention is:

[0070] A computer-readable storage medium stores at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, at least one program, the code set, or the instruction set is loaded and executed by a processor to implement an intelligent aviation fuel optimization method as described above.

[0071] The fourth technical solution adopted by the present invention is:

[0072] A computer program product or computer program includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, so that the computer device performs the above method.

[0073] The beneficial effects of this invention are: it combines data-driven intelligent prediction with mathematical optimization algorithms to provide a feasible fuel optimization solution with higher accuracy and lower computational cost. Compared with traditional methods, this invention can adapt to changes in different routes, weather conditions, and market conditions, enabling airlines to formulate jet fueling plans more flexibly and efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following introduction is made to the drawings of the embodiments of the present invention or the related technical solutions in the prior art. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative work.

[0075] Figure 1 This is a flowchart of the steps of an intelligent aviation fuel optimization method according to an embodiment of the present invention;

[0076] Figure 2 4 is a flow chart of the intelligent aviation fuel optimization method in an embodiment of the present invention. DETAILED DESCRIPTION

[0077] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application and are not to be construed as limiting the present application. For the step numbers in the following embodiments, they are provided only for the convenience of explanation and are not intended to limit the order of the steps. The order of execution of the steps in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0078] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments and are not intended to limit the embodiments of the present application. The singular forms of "a", "said", and "the" used in the embodiments of the present application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. In addition, unless otherwise clearly defined, words such as setting, installing, and connecting should be understood in a broad sense, and those skilled in the art can reasonably determine the specific meanings of the above words in the present invention in combination with the specific content of the technical solution.

[0079] In the description of this application, it should be understood that descriptions involving orientations, such as up, down, front, back, left, right, etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, they cannot be understood as limitations on this application.

[0080] In the description of this application, "several" means one or more, "many" means more than two, "greater than," "less than," and "exceed" are understood to exclude the number itself, while "above," "below," and "within" are understood to include the number itself. The terms "first" and "second" are used solely to distinguish technical features and are not to be construed as indicating or implying relative importance, or as implicitly specifying the number or order of the technical features indicated.

[0081] In the description of this application, "and / or" describes the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0082] In general, the existing technical solutions have the following technical problems: 1) Insufficient dynamic environment modeling capabilities: Existing fuel consumption prediction models (such as multivariate linear regression, time series analysis) rely on fixed input variables (flight distance, nominal load, etc.) and cannot effectively integrate dynamic meteorological parameters such as gusts and turbulence, resulting in limited fuel consumption prediction accuracy, which in turn affects the global optimization effect of the refueling strategy. 2) The contradiction between computational efficiency and complexity: When solving multi-segment refueling strategies, traditional optimization algorithms (linear programming, dynamic programming) need to frequently call the complete prediction model for iterative calculations, resulting in an exponential increase in the calculation time of large-scale route networks, forcing the system to simplify constraints (such as ignoring airport supply capacity restrictions), ultimately reducing the practicality and economy of the strategy. 3) Dependence on static data: The existing system uses fixed oil price data and fuel consumption coefficients, which makes it difficult to respond to fuel market price fluctuations and dynamic changes in route weather in real time, causing the optimization strategy to lag behind the actual operating environment. 4) Lack of Multi-Objective Collaborative Optimization: Traditional approaches focus solely on minimizing cost or fuel consumption as a single optimization objective. They are unable to achieve a coordinated trade-off between cost and fuel consumption through a unified algorithmic framework, limiting the flexibility and applicability of the strategy. 5) Dependence on Human Intervention: Relying on pre-set segmentation rules (such as fixed segment splitting) to reduce computational load can lead to deviations from the global optimal solution due to human bias, and are difficult to adapt to complex and changing route network structures.

[0083] Based on this, the present invention proposes an intelligent aviation fuel optimization scheme based on neural network prediction, regression model alternative calculation, and gradient descent optimization. This scheme comprehensively utilizes data-driven methods to solve the shortcomings of traditional methods in prediction accuracy and computational efficiency.

[0084] Example 1

[0085] like Figure 1 and Figure 2 As shown, this embodiment provides an intelligent aviation fuel optimization method, including the following steps:

[0086] S1. Build a fuel consumption prediction model to predict the fuel consumption of each flight segment and provide basic data for subsequent refueling optimization calculations.

[0087] By learning historical flight data through deep neural networks, a high-precision fuel consumption prediction model is established, which can predict the fuel consumption of each flight segment based on factors such as the flight's take-off and landing airports, flight time, load, and weather.

[0088] It's important to note that in addition to using deep neural networks for modeling, other machine learning models, such as gradient boosting decision trees or XGBoost, can also be used in fuel consumption prediction. These methods offer strong explanatory power when processing high-dimensional data and have relatively low computational overhead. Furthermore, Bayesian regression or support vector machines can be considered for fuel consumption modeling. These methods can provide better fits, especially when the data volume is small, and reduce the reliance on large amounts of training data.

[0089] As a further preferred embodiment, in the optimization calculation process, in order to improve the calculation efficiency, the neural network model is not called in each iteration, but the polynomial regression model is used to fit the fuel consumption prediction to reduce the calculation burden while maintaining a high prediction accuracy.

[0090] If the accuracy of the polynomial regression fit is insufficient, other dimensionality reduction methods, such as principal component analysis or kernel regression, can be used to convert high-dimensional input variables into more physically meaningful, low-dimensional variables to reduce computational complexity. Furthermore, Gaussian process regression can be used for modeling. This approach generalizes well to small sample sizes and can account for prediction uncertainty, enabling the optimization algorithm to dynamically adjust its computational strategy.

[0091] S2. Use mathematical optimization methods to calculate the optimal refueling amount for each flight segment with minimum fuel cost and minimum fuel consumption as optimization goals.

[0092] S3. Perform iterative optimization through the gradient descent algorithm to find the best refueling plan while satisfying the preset constraints (such as reserve fuel, maximum takeoff weight, and maximum landing weight).

[0093] In some embodiments, in addition to using gradient descent algorithms for optimization calculations, genetic algorithms or particle swarm optimization can also be employed. These swarm intelligence-based optimization algorithms can more comprehensively search for optimal solutions, particularly when the objective function is complex and has a multi-peak distribution. They may offer greater global search capabilities than gradient descent algorithms. Furthermore, mixed integer linear programming or dynamic programming methods can be utilized to ensure that all constraints are satisfied during the optimization process while improving computational efficiency.

[0094] In other embodiments, in addition to direct segment-based optimization, refueling optimization decisions can also employ a hierarchical optimization approach. This involves first developing an overall refueling strategy based on historical data from multiple flights, followed by local optimization for individual flights, thereby improving fuel utilization across the entire flight network. Furthermore, game theory approaches can be incorporated to optimize inter-flight fueling plans through bidding or auction mechanisms when multiple flights share fuel, further reducing costs.

[0095] The workflow of this embodiment is as follows: first, train the neural network model based on the flight plan and historical data, and use the regression equation to approximate the fuel consumption forecast for each flight segment. Set the objective function and constraints, and use the gradient descent algorithm to optimize the refueling plan. Calculate the fuel consumption of each flight segment, iteratively update the refueling decision, and finally solve the overall optimal refueling strategy. The biggest advantage of the method of this embodiment is that it can combine data-driven intelligent prediction and mathematical optimization algorithms to provide a feasible fuel optimization solution with higher accuracy and lower computing cost. Compared with traditional methods, this system can adapt to changes in different routes, weather, and market environment, allowing airlines to formulate aviation fuel refueling plans more flexibly and efficiently.

[0096] The present embodiment is explained in detail below in conjunction with specific implementation methods.

[0097] (1) Aviation fuel consumption prediction model

[0098] Accurately predicting an aircraft's fuel consumption during different flight segments is crucial for optimizing fuel management. This example uses a data-driven deep learning model to build a fuel consumption prediction system. By analyzing historical flight data, it establishes a mapping relationship between fuel consumption and various influencing factors. This prediction model provides foundational data for subsequent refueling optimization calculations, ensuring the accuracy and feasibility of optimization decisions.

[0099] (1.1) Factors affecting fuel consumption

[0100] Fuel consumption is affected by many factors, including but not limited to:

[0101] a) Flight segment characteristics: altitude, geographical location, weather conditions, etc. of the departure and destination airports.

[0102] b) Flight conditions: wind speed, wind direction, temperature, atmospheric pressure, flight altitude, etc.

[0103] c) Aircraft performance parameters: aircraft model, maximum take-off weight, maximum landing weight, maximum fuel capacity, etc.

[0104] d) Load and fuel status: fuel weight, passenger and cargo weight at takeoff, etc.

[0105] Traditional fuel consumption prediction methods usually use linear regression or empirical formulas, but these methods are difficult to accurately describe complex nonlinear relationships. Therefore, this embodiment uses a deep neural network model to construct a fuel consumption prediction function f(x), namely:

[0106] Y=f(X)+∈

[0107] Where X represents the input variables (i.e., the influencing factors mentioned above), Y represents the predicted fuel consumption, and ∈ is the unavoidable error term.

[0108] (1.2) Data processing and feature engineering

[0109] To train the neural network, we first process the historical flight data, mainly including:

[0110] a) Data cleaning: remove outliers, fill missing values, and standardize data.

[0111] b) Feature Engineering: Numericalize categorical data such as time and airports. For example, airports are encoded using one-hot encoding, and wind speed and direction are converted into vector form.

[0112] c) Data segmentation: Divide the training and test sets by aircraft type or flight segment, and use cross-validation to ensure model generalization ability.

[0113] (1.3) Training of neural network models

[0114] The neural network model of this embodiment consists of multiple fully connected layers, each of which extracts and transforms data features through a nonlinear activation function (ReLU). The loss function of the model uses the mean square error:

[0115]

[0116] in, is the model prediction value, Y i is the actual fuel consumption, and N is the number of samples. During model training, the gradient descent algorithm is used to optimize the parameters to ensure that the loss function is minimized.

[0117] (1.4) Cross-validation and evaluation

[0118] To evaluate the accuracy of the model, we use cross-validation to ensure that data from different flight segments and aircraft models are effectively learned. The main evaluation indicators include:

[0119] a) Mean Absolute Error: measures the average difference between the predicted value and the true value.

[0120] b) Mean Square Error: Evaluates the square of the prediction error to avoid positive and negative errors canceling each other out.

[0121] c)R 2 Coefficient of determination: measures the model's ability to explain data variance. The closer the value is to 1, the better the fit. The calculation formula is as follows:

[0122]

[0123] The trained and evaluated neural network model can accurately predict fuel consumption for each flight segment. However, during the iterative calculation process of optimizing the refueling plan, invoking the neural network model to calculate fuel consumption at each step can result in excessive computational overhead. Therefore, this embodiment introduces polynomial regression fitting in the next step, replacing the neural network with a mathematical function to improve computational efficiency and ensure rapid convergence of the optimization process.

[0124] (1.5) Polynomial regression fitting

[0125] During aviation fuel optimization calculations, the fuel consumption forecast for each flight segment is a core input to the optimization algorithm. However, invoking a deep neural network for predictions during each optimization iteration would incur significant computational overhead, impacting optimization efficiency. Therefore, to ensure prediction accuracy while improving computational speed, this embodiment employs a polynomial regression model to fit the fuel consumption forecasts from the neural network. This allows for rapid calculation of fuel consumption using mathematical equations during the iterative optimization process, thereby improving solution efficiency.

[0126] The relationship between fuel consumption and factors such as the aircraft's refueling volume and initial fuel weight is not a simple linear relationship, but includes nonlinear effects. For example: 1) Fuel consumption increases nonlinearly with the change in takeoff gross weight (the greater the weight, the higher the fuel consumption). 2) The impact of environmental factors at different airports (such as wind speed and temperature) on fuel consumption is more complex. 3) Fuel consumption optimization is an iterative solution process, and computational efficiency must be guaranteed. Polynomial regression can approximate the prediction results of the neural network with a lower computational cost and can retain nonlinear characteristics. Therefore, this embodiment selects polynomial regression for optimization solution.

[0127] (1.5.1) Polynomial regression model construction

[0128] To replace the neural network for fast fuel consumption calculation, we assume that the fuel consumption Y is mainly determined by the following two variables: 1) U: the amount of fuel (tons) for the flight segment; 2) F: the total weight of fuel (tons) at takeoff of the flight segment.

[0129] Using bivariate polynomial regression, the following approximate function is constructed:

[0130] Y=a0+a1U+a2F+a3U 2 +a4UF+a5F 2 +∈

[0131] Where Y is the predicted fuel consumption; a0, a1, …, a5 are regression coefficients obtained by data fitting; ∈ is the random error term.

[0132] This model can better fit the fuel consumption data predicted by the neural network, and its calculation speed is much faster than the neural network reasoning, so it is suitable for fast calculation in the gradient descent optimization process.

[0133] (1.5.2) Polynomial regression training and fitting

[0134] In order to obtain the appropriate regression coefficient, this embodiment uses the prediction results of the neural network as training data to fit the polynomial regression. The training steps are as follows:

[0135] 1) Sampling data points: Fixed the flight environment variables, selected different combinations of refueling amount U and initial fuel weight F, and input them into the neural network to calculate the fuel consumption Y.

[0136] 2) Regression fitting: Use the least squares method to solve the regression coefficient a i , so that the fitting error is minimized.

[0137] 3) Accuracy verification: calculate mean square error and coefficient of determination R 2 and other indicators to ensure that the fitting accuracy is high enough.

[0138] (1.5.3) Error control and optimization

[0139] In order to ensure the fitting accuracy of the polynomial regression model and enable it to replace the neural network for fuel consumption calculation during the optimization process, this embodiment adopts the following optimization measures:

[0140] 1) Order selection: Determine the optimal polynomial order through cross-validation. Generally, second-order or third-order polynomials can achieve a better balance.

[0141] 2) Data augmentation: Increasing the diversity of training data enables the regression model to cover a wider range of inputs and improve generalization capabilities.

[0142] 3) Model evaluation: Calculate the mean square error to ensure that the errors of polynomial regression and neural network are within an acceptable range.

[0143]

[0144] Among them, Y i is the predicted value of the neural network, is the polynomial regression fitting value, and N is the number of data points.

[0145] After determining the polynomial regression model, this embodiment replaces the neural network as the method for calculating fuel consumption. The regression equation is used to quickly calculate the fuel consumption Y during the gradient descent optimization process, and the refueling amount U is updated at each iteration until the optimal refueling strategy is found.

[0146] The following sections will detail the refueling optimization strategy, including objective function setting (minimum fuel cost, minimum fuel consumption), constraints (such as fuel reserves, maximum takeoff weight, etc.), and how to use the gradient descent algorithm to solve the optimal refueling plan.

[0147] (2) Refueling optimization strategy

[0148] The previous section described how to build a fuel consumption prediction model using deep neural networks and polynomial regression fitting, using polynomial regression instead of neural networks for fast calculations. Based on this fuel consumption prediction model, this embodiment can optimize flight refueling to determine how to allocate fueling amounts across different flight segments, enabling airlines to achieve the lowest fuel costs or fuel consumption while ensuring safety.

[0149] The core goal of refueling optimization is to determine how much fuel should be added at the departure airport for each flight segment to meet all flight requirements while minimizing fuel cost or fuel consumption. This optimization problem can be formulated as a constrained optimization problem and solved using gradient descent.

[0150] (2.1) Objective function setting

[0151] There are generally two goals for aviation fuel optimization:

[0152] 1) Lowest fuel cost

[0153] Under the premise of ensuring the normal operation of the flight, we hope to minimize the total refueling cost. Assume that the flight consists of N segments, and the fuel price of segment i at the departure airport is P i (Unit: RMB / ton), the amount of fuel in this section is U i (Unit: ton), then the total fuel cost can be expressed as:

[0154]

[0155] The optimization goal is to minimize the total fuel cost, that is:

[0156]

[0157] 2) Minimum fuel consumption

[0158] Another optimization goal is to minimize the total fuel consumption, that is, to find a refueling plan that consumes the least fuel for the entire flight. Assume that the fuel consumption of segment i is calculated by the fitted polynomial regression equation, that is:

[0159] Y i =f(U i ,F i )

[0160] Among them, Y i is the fuel consumption of flight segment i, U i is the amount of fuel at the departure point of the flight segment, F i is the weight of the aircraft's fuel at takeoff.

[0161] The optimization goal is to minimize the total fuel consumption:

[0162]

[0163] In practical applications, airlines may make a trade-off between minimizing fuel costs and minimizing fuel consumption, and adopt a weighted solution approach to make the impact of both work together on the optimization goal.

[0164] (2.2) Definition of constraints

[0165] When solving the optimal refueling plan, a series of safety and practical operation-related constraints need to be met, mainly including:

[0166] 1) Fuel reserve constraints

[0167] At the final airport of each leg, the aircraft must retain at least a certain amount of fuel as a reserve, namely:

[0168] F i+1 ≥R

[0169] Among them, F i+1 is the fuel weight at the end of segment i and R is the minimum reserve fuel quantity (usually determined by aviation safety regulations).

[0170] 2) Takeoff weight constraint (MTOW)

[0171] The aircraft's gross takeoff weight cannot exceed its maximum takeoff weight (MTOW):

[0172] W takeoff,i =W dry +P i +U i +F i ≤MTOW

[0173] Among them, W dry is the empty weight of the aircraft, W dry is the empty weight of the aircraft, U i is the fuel quantity for this segment, F i is the fuel weight at takeoff.

[0174] 3) Landing weight constraint (MLW)

[0175] The total weight of the aircraft when arriving at the airport cannot exceed its maximum landing weight (MLW):

[0176] W landing,i =W dry +P i +(U i +F i -Y i )≤MLW

[0177] 4) Maximum fuel capacity constraints

[0178] The amount of fuel an aircraft can carry cannot exceed its maximum fuel capacity C:

[0179] F i +U i ≤C

[0180] 5) Fuel mobility constraints

[0181] Fuel consumption must meet flight requirements, namely:

[0182] F i+1 =F i +U i -Y i

[0183] This formula means: After the current flight segment ends, the aircraft's fuel weight F i+1 Should be equal to the fuel weight F at the beginning of the segment i Add the refueling amount U i , minus the fuel consumption Y for this segment i .

[0184] (2.3) Gradient descent optimization solution

[0185] Since refueling optimization is a nonlinear constrained optimization problem, it is difficult to directly obtain an analytical solution. Therefore, this embodiment adopts a gradient descent algorithm to solve it.

[0186] 1) The goal of gradient descent

[0187] The amount of fuel U for each segment needs to be adjusted by the gradient descent method i , so that the objective function (fuel cost or fuel consumption) is gradually reduced while ensuring that all constraints are met.

[0188] 2) Iteratively update refueling strategy

[0189] The core idea of ​​gradient descent is to adjust the amount of fuel in each iteration according to the gradient of the objective function. The formula is as follows:

[0190]

[0191] in, is the fuel quantity at the tth iteration; α is the learning rate, which determines the step size of each adjustment; J is the objective function (fuel cost or fuel consumption); It is the partial derivative of the objective function with respect to the refueling amount, which indicates the impact of the change in refueling amount on the objective function.

[0192] 3) Constraint correction and dynamic adjustment

[0193] After each update, you need to check for new U i Check whether all constraints are met. If a constraint is violated (for example, exceeding MTOW or MLW), the following adjustments are required:

[0194] a) Violation of fuel reserve constraint: Increase refueling amount U i To ensure that you have enough fuel when you arrive at your destination.

[0195] b) Takeoff weight constraint violation: Reduce the amount of fuel U i , until the takeoff weight meets the requirements.

[0196] c) Landing weight constraint violation: Reduce the amount of fuel U i Or optimize fuel distribution between flight segments.

[0197] 4) Iteration termination condition

[0198] The gradient descent algorithm terminates when the following conditions are met:

[0199] a) Objective function convergence: If the change of the objective function value between two adjacent iterations is less than the set threshold (∈=10 -6 ), then it is considered that the optimal solution has been found.

[0200] b) Reaching the maximum number of iterations: To prevent the algorithm from falling into an infinite loop, a maximum number of iterations (100 times) is set.

[0201] (2.4) Final solution based on optimization strategy

[0202] Through the above optimization strategy, two optimal refueling solutions can be obtained:

[0203] 1) The lowest fuel cost solution is applicable to situations where fuel prices fluctuate significantly, helping airlines reduce costs.

[0204] 2) Minimum fuel consumption plan, suitable for situations where fuel consumption control is strict and carbon emissions need to be reduced.

[0205] (3) Advantages and beneficial effects

[0206] In summary, compared with the prior art, the present invention has at least the following advantages and beneficial effects:

[0207] (1) Solve the problem of insufficient dynamic environment modeling capabilities

[0208] Existing fuel consumption prediction models (such as linear regression and time series analysis) struggle to accurately model dynamic factors affecting flight, such as gusts, turbulence, and altitude changes. This limits the accuracy of traditional methods when faced with complex flight conditions. This paper uses a deep neural network as a fuel consumption prediction model. By learning from historical flight data, it establishes a nonlinear, high-dimensional fuel consumption prediction relationship. This model automatically extracts key features of the flight environment and, combined with meteorological data, flight status, and other information, provides more accurate fuel consumption estimates.

[0209] This improvement demonstrated significant technical effectiveness in actual testing: compared with traditional linear regression methods, the DNN model's prediction error (mean square error (MSE)) was reduced by approximately 20%, improving the accuracy of fuel consumption predictions. This in turn made refueling optimization plans more precise and reduced the risk of additional fuel waste or insufficient refueling due to prediction errors.

[0210] (2) Solve the contradiction between computational efficiency and complexity

[0211] When optimizing refueling strategies, traditional optimization methods require frequent calls to complete fuel consumption prediction models, significantly increasing computational complexity for large-scale route networks and leading to excessively long calculation times, making them difficult to apply in actual operations. This invention replaces neural network calculations with polynomial regression fitting, enabling the optimization algorithm to improve computational efficiency without sacrificing prediction accuracy.

[0212] Test results show that using polynomial regression instead of neural networks can reduce computation time by over 63% with the same computing resources, meeting the real-time computing requirements of large-scale route optimization. Furthermore, polynomial regression's low computational complexity allows for rapid adjustments during flight scheduling, improving the operability of the optimization strategy.

[0213] (3) Solve the problem of static data dependency

[0214] Traditional systems typically rely on fixed fuel price data and fuel consumption coefficients, making them difficult to respond to market price fluctuations and real-time weather changes, resulting in optimization results lagging behind actual operations. The optimization system of the present invention dynamically receives real-time fuel prices, airport fuel supply data, and weather information, and incorporates these variables into the optimization calculations, ensuring that the results are always synchronized with the current market environment.

[0215] Tests show that when oil prices fluctuate significantly, the fuel cost deviation between the traditional fixed oil price optimization strategy and the real-time optimization strategy can reach 5%-15%. The system of the present invention can automatically adjust the refueling strategy based on the latest data, making refueling decisions more in line with the current economic environment and reducing unnecessary cost losses.

[0216] (4) Solve the problem of lack of multi-objective collaborative optimization

[0217] Traditional refueling optimization methods typically only consider minimum fuel cost or minimum fuel consumption, lacking a flexible trade-off mechanism and making it difficult to optimize both objectives simultaneously. This paper proposes a multi-objective optimization framework based on weight adjustment, allowing airlines to dynamically balance fuel cost and fuel consumption based on different operational requirements, making the optimization results more consistent with actual scheduling needs.

[0218] (5) Solve the problem of dependence on human intervention

[0219] Traditional systems typically require manual configuration of fixed refueling strategies or rules, such as pre-setting flight segment splits or limiting refueling capacity at certain airports. This approach can easily deviate from the global optimal solution and struggles to adapt to complex route structures. This invention utilizes a fully automated optimization calculation process, allowing the system to autonomously analyze all feasible refueling options without relying on manually pre-set rules, directly calculating the optimal solution.

[0220] Example 2

[0221] An embodiment of the present invention further provides an electronic device, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the following Figure 1 An intelligent aviation fuel optimization method is shown.

[0222] It is understood that the memory may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory includes a non-transitory computer-readable storage medium. The memory may be used to store instructions, programs, codes, code sets, or instruction sets. The memory may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function, instructions for implementing the various method embodiments described above, etc.; the data storage area may store data created based on the use of the server, etc.

[0223] The processor may include one or more processing cores. The processor utilizes various interfaces and circuits to connect various components within the server. It executes various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory, as well as accessing data stored in memory. Optionally, the processor may be implemented using at least one of the following hardware forms: digital signal processing (DSP), field-programmable gate array (FPGA), and programmable logic array (PLA). The processor may integrate one or a combination of a central processing unit (CPU) and a modem. The CPU primarily processes the operating system and application programs, while the modem handles wireless communications. It is understood that the modem may not be integrated into the processor and may be implemented separately via a single chip.

[0224] Since the electronic device is an electronic device corresponding to an intelligent aviation fuel optimization method in an embodiment of the present invention, and the principle of solving the problem by the electronic device is similar to that of the method, the implementation of the electronic device can refer to the implementation process of the above-mentioned method embodiment, and the repeated parts will not be repeated.

[0225] Example 3

[0226] An embodiment of the present invention further provides a computer-readable storage medium, wherein the storage medium stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, the at least one program, the code set or instruction set is loaded and executed by a processor to implement the following Figure 1 An intelligent aviation fuel optimization method is shown.

[0227] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program. The program can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0228] Since the storage medium is the storage medium corresponding to an intelligent aviation fuel optimization method of an embodiment of the present invention, and the principle of solving the problem by the storage medium is similar to that of the method, the implementation of the storage medium can refer to the implementation process of the above-mentioned method embodiment, and the repeated parts will not be repeated.

[0229] Example 4

[0230] In some possible implementations, various aspects of the methods of the embodiments of the present invention may also be implemented in the form of a program product, which includes program code. When the program product is executed on a computer device, the program code is used to cause the computer device to perform the steps of the intelligent aviation fuel optimization method according to various exemplary embodiments of the present application as described above in this specification. The executable computer program code or "code" used to perform the various embodiments may be written in a high-level programming language such as C, C++, Python, Smalltalk, Java, JavaScript, Visual Basic, Structured Query Language (e.g., Transact-SQL), Perl, or in various other programming languages.

[0231] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0232] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0233] The above embodiments are intended only to illustrate the technical concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. They are not intended to limit the scope of protection of the present invention. Any equivalent changes or modifications made based on the essence of the present invention are intended to be covered by the scope of protection of the present invention.

Claims

1. An intelligent aviation fuel optimization method, characterized in that: The following steps are involved: Build a fuel consumption prediction model to predict fuel consumption for each flight segment, providing basic data for subsequent refueling optimization calculations; Using mathematical optimization methods, with minimum fuel cost and minimum fuel consumption as optimization goals, calculate the optimal refueling amount for each flight segment; Through iterative optimization using the gradient descent algorithm, the optimal refueling plan is solved while satisfying the preset constraints.

2. The intelligent aviation fuel optimization method according to claim 1, characterized in that: Factors influencing fuel consumption include: flight segment characteristics, flight conditions, aircraft performance parameters, and load and fuel status. By analyzing historical flight data, a mapping relationship between fuel consumption and various influencing factors is established. The expression of the oil consumption prediction model is: Y=f(X)+∈ Where X represents the input variable, i.e., the factor affecting fuel consumption; Y represents the predicted fuel consumption; and ∈ is the error term.

3. The intelligent aviation fuel optimization method according to claim 2, characterized in that: The oil consumption prediction model consists of multiple fully connected layers. Each layer extracts and transforms data features through a nonlinear activation function. The loss function of the model uses the mean square error: in, is the model prediction value, Y i is the actual fuel consumption, and N is the number of samples. During the model training process, the gradient descent algorithm is used to optimize the parameters to ensure that the loss function is minimized.

4. The intelligent aviation fuel optimization method according to claim 2, characterized in that: The intelligent aviation fuel optimization method also includes a step of evaluating the accuracy of the model: Cross-validation is used to evaluate the accuracy of the model; the evaluation indicators include: 1) Mean absolute error: measures the average difference between the predicted value and the true value; 2) Mean square error: evaluates the square of the prediction error to avoid positive and negative errors canceling each other out; 3) R 2 Coefficient of determination: measures the model's ability to explain the data variance. The closer the value is to 1, the better the fitting effect.

5. The intelligent aviation fuel optimization method according to claim 1 or 2, characterized in that: During the iterative calculation process of optimizing the refueling plan, calling the aviation fuel consumption prediction model to calculate fuel consumption at each step will result in excessive computational overhead. Therefore, the next step is to introduce polynomial regression fitting, replacing the neural network with a mathematical function to improve computational efficiency and ensure rapid convergence of the optimization process: Assume that the fuel consumption Y is determined by the following two variables: 1) U: the amount of fuel used for the flight segment; 2) F: the total weight of fuel used at takeoff; Using bivariate polynomial regression, the following approximate function is constructed: <h2 style=";text-align:left;direction:ltr">Y = a0 + a1U + a2F + a3U<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> +a4UF+a5F<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> +∈ Where Y is the predicted fuel consumption; a0, a1, …, a5 are regression coefficients obtained by data fitting; ∈ is the error term.

6. The intelligent aviation fuel optimization method according to claim 5, characterized in that: Use the model prediction results as training data to fit the polynomial regression. The training steps are as follows: Sampling data points: Fixed flight environment variables, selected different combinations of refueling amount U and initial fuel weight F, and input them into the neural network to calculate fuel consumption Y; Regression fitting: Use the least squares method to solve the regression coefficient a i , so that the fitting error is minimized; Accuracy verification: by calculating the mean square error and the coefficient of determination R 2 , to ensure that the fitting accuracy is high enough.

7. The intelligent aviation fuel optimization method according to claim 1, characterized in that: The optimization objectives of minimizing fuel cost and fuel consumption include: Assume that the flight consists of N segments and the fuel price of segment i at the departure airport is P i The amount of fuel in this section is U i , the total fuel cost is expressed as: The optimization goal of minimum fuel cost is: Assume that the fuel consumption of flight segment i is calculated by the fitted polynomial regression equation, that is: Y i =f(U i ,F i ) Among them, Y i is the fuel consumption of flight segment i, U i is the amount of fuel at the departure point of the flight segment, F i is the weight of the aircraft's fuel at takeoff; The optimization goal of minimizing total fuel consumption is: 。 8. The intelligent aviation fuel optimization method according to claim 1, characterized in that: The preset constraints include: 1) Fuel reserve constraints: At the destination airport of each flight segment, the aircraft must retain at least a predetermined amount of fuel as a reserve, namely: F i+1 ≥R Among them, F i+1 is the fuel weight at the end of segment i; R is the minimum reserve fuel; 2) Takeoff weight constraints: The aircraft's gross takeoff weight cannot exceed its maximum takeoff weight, which is: W takeoff,i =W dry +P i +U i +F i ≤MTOW Among them, W dry is the empty weight of the aircraft; P i is the effective load of the flight segment; U i is the amount of fuel for this segment; F i is the fuel weight at takeoff; 3) Landing weight restrictions The total weight of the aircraft upon arrival at the airport cannot exceed its maximum landing weight, that is: W landing,i =W dry +P i +(U i +F i -Y i )≤MLW 4) Maximum fuel capacity constraints The amount of fuel carried by an aircraft cannot exceed its maximum fuel capacity C, that is: F i +U i ≤C 5) Fuel mobility constraints Fuel consumption must meet flight requirements, namely: F i+1 =F i +U i -Y i 。 9. The intelligent aviation fuel optimization method according to claim 1, characterized in that: The iterative optimization using the gradient descent algorithm to solve the optimal refueling plan under the premise of satisfying the preset constraints includes: Adjust the refueling amount U for each segment by gradient descent method i , so that the objective function is gradually reduced while ensuring that all constraints are satisfied; In each iteration, the refueling amount is adjusted according to the gradient of the objective function, and the formula is as follows: in, is the amount of refueling at the tth iteration; α is the learning rate; J is the objective function; is the partial derivative of the objective function with respect to the refueling amount, which indicates the effect of the change in refueling amount on the objective function; After reaching the iteration termination condition, the optimal refueling plan is output.

10. An electronic device, characterized in that: The electronic device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the method according to any one of claims 1 to 9.

Images