A method and system for aircraft ground infrared de-icing based on multi-parameter collaborative optimization

Abstract

This invention belongs to the field of aircraft operation support systems and equipment technology, and discloses a method and system for aircraft ground infrared de-icing based on multi-parameter collaborative optimization. The invention constructs an aircraft ground infrared de-icing system based on multi-parameter collaborative optimization, collaboratively optimizes infrared de-icing parameters, and builds a multi-objective optimization model. An improved quantum genetic algorithm is used to solve the collaborative multi-objective optimization model and generate parameters, obtaining the optimal set of de-icing parameters. Based on real-world conditions, repeated experiments are conducted in the parameter validity verification module to clarify whether the de-icing effect meets the de-icing requirements. The obtained experimental data is processed and analyzed in the historical data module. This invention realizes the transformation of the aircraft ground infrared de-icing process from experience-based operation to intelligent, optimized decision-making, providing an advanced technical solution for efficient, energy-saving, and safe green de-icing operations.

Description

Technical Field

[0001] This invention belongs to the field of aircraft operation support systems and equipment technology, and particularly relates to an aircraft ground infrared de-icing method and system based on multi-parameter collaborative optimization. Background Technology

[0002] Air travel, with its convenience and efficiency, has become one of the most popular modes of transportation. As the fastest means of transport, airplanes offer immense convenience for both passengers and cargo. However, aircraft safety has also become a major concern. In northern winters, when aircraft operate in cold and damp conditions, frost, ice, or a mixture of ice can easily accumulate on critical aerodynamic surfaces such as wings and tail sections. Icing severely disrupts the aircraft's aerodynamic shape, increases weight and drag, reduces lift, and poses a significant threat to flight safety. Therefore, thorough and efficient ground de-icing before takeoff is an indispensable part of ensuring aviation safety.

[0003] The mainstream aircraft ground de-icing methods at airports mainly include liquid de-icing (thermal de-icing fluid) and infrared radiation de-icing. Liquid de-icing technology is currently widely used, but it suffers from problems such as high fluid consumption, high cost, environmental pollutant emissions (ethylene glycol), and the potential impact of de-icing fluid residue on subsequent aerodynamic performance. In contrast, infrared radiation de-icing, as a novel physical de-icing technology, utilizes infrared radiation energy to directly penetrate and heat the ice layer at the skin interface, achieving ice melting and peeling. It has potential advantages such as high energy utilization, no chemical pollution, and good directionality, and has become a research hotspot in this field. However, the efficient and safe application of infrared radiation de-icing technology to complex and ever-changing aircraft ground de-icing scenarios still faces a series of key technical challenges, which directly reflect the shortcomings and deficiencies of existing methods in terms of intelligence and optimization.

[0004] The current approach to parameter control is crude and lacks systematic optimization: Infrared de-icing is a complex process involving multi-physics coupling. Its efficiency and safety are simultaneously affected by environmental conditions (temperature, humidity, wind speed), ice accumulation (type, thickness, adhesion strength), and radiation parameters (power, intensity, distance, irradiated area, and time). Existing technologies and research often focus on the static control of a single or a few parameters, such as de-icing experiments at fixed power, or treat the environment as a constant condition. This "island-like" parameter processing approach fails to construct a dynamic synergistic optimization model among the environment, ice accumulation, and radiation parameters. As a result, the de-icing scheme cannot adapt to real-time changing external conditions and is difficult to achieve a globally optimal balance among multiple objectives such as efficiency, energy consumption, and safety.

[0005] Inadequate handling of multi-objective conflicts and safety constraints: An ideal de-icing scheme needs to simultaneously pursue the shortest de-icing time, the lowest energy consumption, and ensure that the aircraft's composite skin is not damaged by overheating (temperature rise constraint). These objectives are inherently conflicting (e.g., increasing power can shorten the time but increases energy consumption and the risk of overheating). Existing methods often use simple methods such as weighted summation to transform multiple objectives into a single objective, or only consider a single main objective, which obscures the trade-offs between objectives. Decision-makers cannot obtain a set of selectable, balanced Pareto optimal solutions. At the same time, for the hard constraint of skin temperature safety, there is a lack of an effective mechanism to organically embed it into the optimization process, often relying on post-event verification, which poses safety risks or leads to optimization failure.

[0006] Optimization algorithms often encounter local optima and convergence problems: When attempting to establish complex multi-parameter, multi-objective, constrained optimization models, the solution space is high-dimensional, nonlinear, and discontinuous. Traditional optimization algorithms (such as traditional genetic algorithms and gradient descent) are prone to premature convergence and getting trapped in local optima when solving such problems, making it difficult to obtain globally satisfactory parameter combinations. While traditional quantum genetic algorithms (QGA) have improved global search capabilities, their fixed quantum gate operations (such as fixed quantum rotation gate angles) lack adaptability and do not fully utilize the physical relationships between optimization parameters. Therefore, when facing specific problems in this field, both search efficiency and solution quality still need improvement. Summary of the Invention

[0007] To overcome the problems existing in related technologies, the present invention discloses an aircraft ground infrared de-icing method and system based on multi-parameter collaborative optimization, specifically involving an aircraft ground infrared de-icing parameter collaborative optimization method and parameter generation system based on an improved quantum genetic algorithm (IQGA) to ensure optimal collaborative optimization of multiple de-icing parameters.

[0008] The technical solution is as follows: A ground-based infrared de-icing method for aircraft based on multi-parameter collaborative optimization, comprising the following steps:

[0009] S1. Construct an aircraft ground infrared de-icing system based on multi-parameter collaborative optimization. The aircraft ground infrared de-icing system based on multi-parameter collaborative optimization includes an environmental acquisition module, a parameter generation module, a parameter validity verification module, and a historical data module.

[0010] S2. Based on the constructed aircraft ground infrared de-icing system based on multi-parameter collaborative optimization, the infrared de-icing parameters are collaboratively optimized, and a multi-objective optimization model is constructed.

[0011] S3 utilizes an improved quantum genetic algorithm to solve a collaborative multi-objective optimization model and generate parameters. The entanglement state between parameters is determined by the correlation between environmental parameters, ice accumulation parameters, and parameters in the current state. The solution process combines the balance between optimization objectives to finally obtain the optimal set of de-icing parameters.

[0012] S4. After generating the optimal set of de-icing parameters, repeat experiments are conducted in the parameter validity verification module based on the real environment to determine whether the de-icing effect meets the de-icing requirements.

[0013] S5, in the historical data module, performs data processing and analysis on the obtained experimental data.

[0014] In step S2, the infrared de-icing parameters are optimized collaboratively, and a multi-objective optimization model is constructed, including: taking ambient temperature, humidity, wind speed and the duration corresponding to each parameter as input layer I, taking ice type, ice thickness and ice evolution state as input layer II, and taking infrared radiation current, infrared radiation voltage, infrared radiation relative area, infrared radiation intensity, infrared irradiance and infrared source distance from ice point as output layer.

[0015] The optimization objectives are the shortest de-icing time, the most gradual change in skin temperature, and the lowest total energy consumption. These objectives are considered the core of the collaborative optimization, and safety constraints are added, with the constraint that the temperature rise of the aircraft surface skin does not exceed the safety threshold that would damage the aircraft skin material. A hierarchical nested multi-objective optimization model is then constructed.

[0016] Furthermore, in input layer I, the environment dynamic parameter vector is:

[0017]

[0018] In the formula, For ambient temperature, This refers to the relative humidity of the air. Wind speed, For the duration of each parameter, and ;

[0019] In input layer II, the icing state vector Perform quantitative modeling:

[0020]

[0021] In the formula, The encoding for the ice accumulation type, where, 1 represents raw ice, 2 represents clear ice, and 3 represents mixed ice; This refers to the thickness of the ice buildup, in mm. Let be the ice accumulation evolution state coefficient, where 0 represents a pure solid state, and 1 represents ice that has completely melted into a liquid water film;

[0022] The ice evolution state coefficient Used to calculate the equivalent thermal resistance of ice accumulation layer for:

[0023]

[0024] In the formula, In order to be in The equivalent thermal resistance of the ice layer at any given time, in units of ; and These are the thermal conductivity coefficients of ice and water, respectively. for The equivalent thermal conductivity of the ice layer at any given time This refers to the proportion or weighting coefficient of the liquid phase (water) in the ice layer;

[0025] In the output layer, the infrared radiation parameter matrix , It is a 6-dimensional real parameter matrix space; defined as:

[0026]

[0027] In the formula, It is infrared radiation current. Infrared radiation voltage, This refers to the relative radiation area. Infrared radiation intensity, Infrared irradiance, The distance from the infrared source to the ice accumulation point;

[0028] Temperature field of aircraft skin The change is controlled by a one-dimensional unsteady-state heat conduction equation, that is, along the skin thickness direction. The expression is:

[0029]

[0030] In the formula, These are the density, specific heat capacity, and thermal conductivity of the skin material, respectively; the key boundary conditions are located on the outer surface of the skin. , The direction of skin thickness; coupled radiative heating, convective heat dissipation in the air, and heat conduction through the ice layer:

[0031]

[0032]

[0033]

[0034] In the formula, For surface convection heat dissipation, The convective heat transfer coefficient is related to the wind speed. related; This refers to the heat flow conducted through the ice layer; The temperature distribution function inside the skin. The net radiative heat flow of the surface. The surface convective heat transfer coefficient is... for Time skin outer surface (i.e. Temperature at (location) Ambient air temperature, The temperature of the outer surface of the ice layer. The equivalent thermal resistance of the ice layer;

[0035] The mathematical expression for the safety constraint is the time during the entire de-icing process. The temperature of the inner skin's outer surface and any internal location must not exceed the maximum safe temperature allowed by its material. :

[0036]

[0037] In the formula, Skin thickness; In order to defrost time Maximum temperature at any location on the outer surface and inside of the inner skin;

[0038] During the optimization process, the partial differential equation system is solved rapidly using the finite difference method, integral approximation method, or response surface model based on historical data to predict... The evolution path involves intrinsically verifying security constraints while optimizing the objective function;

[0039] objective function De-icing completion time;

[0040]

[0041] In the formula, The time from the start of radiation to the fulfillment of the de-icing completion criterion, which is based on the state of icing. Or visual monitoring signals;

[0042] objective function Skin temperature stability;

[0043]

[0044] In the formula, for The temperature of the outer surface of the skin at all times;

[0045] objective function Total energy consumption;

[0046]

[0047] In the formula, For real-time power consumption, for Infrared system current at all times for Infrared system voltage at all times;

[0048] Constructing the Pareto optimal multi-objective vector from the objective function of the multi-objective optimization model Specifically:

[0049]

[0050]

[0051] In the formula, for Infrared system current at all times for Infrared system voltage at all times.

[0052] In step S3, the improved quantum genetic algorithm is used to solve the collaborative multi-objective optimization model and generate parameters, including:

[0053] S301, Quantum Encoding and Population Initialization;

[0054] S302, Quantum Observation and Parameter Decoding;

[0055] S303, Fitness Assessment and Pareto Ranking, will decode each set of parameters. Substitute these parameters into the multi-objective optimization model and calculate the three objective function values ​​for each parameter. At the same time, it is also necessary to pay attention to the constraints, and check whether the constraint condition of no damage to the surface skin of the aircraft is met; then, based on the Pareto dominance relation, perform non-dominated sorting on all feasible solutions in the current population, and calculate the crowding distance of each individual, so as to comprehensively evaluate the fitness of the individual.

[0056] S304, Elite Selection and Population Renewal, employs an elite retention strategy, directly preserving the top few individuals with the highest Pareto level in the current population to the next generation; the remaining individuals are selected from the parent generation through a tournament selection method, in preparation for subsequent quantum evolution operations;

[0057] S305, Dynamically Associated Quantum Entanglement Gate Cooperative Mutation, based on current environmental input. Ice accumulation status input And the distribution of parameters in the population, dynamically calculating the physical correlation coefficient between different infrared radiation parameters. And thereby construct or adjust quantum entanglement gates. Furthermore, this entanglement gate is used to perform co-mutation operations on the individuals selected in step S304 elite selection and population update, so that the parameters with strong correlations can be further co-optimized.

[0058] S306, Improved Adaptive Quantum Rotation Gate for Guiding Optimal Parameter Selection, including: designing an adaptive quantum rotation gate, the direction of the rotation angle of the adaptive quantum rotation gate is jointly determined by the probability amplitude of the current qubit and the guidance of elite individuals in the current Pareto front, and the size of the rotation angle of the adaptive quantum rotation gate is adaptively adjusted with the number of generations of evolution to achieve a balance between fast global search and fine-grained local search in the later stages, guiding the population to converge toward a more optimized Pareto front;

[0059] S307, Iteration terminates and Pareto optimal solution set is output; repeat steps S302-S306 above for iterative evolution until the preset maximum number of generations is reached or the convergence condition is met; finally, output all solutions on the Pareto non-dominated layer in the last generation population to form the optimal de-icing parameter set for use by the parameter validity verification module.

[0060] In step S301, quantum encoding and population initialization include:

[0061] Based on the input layer I environment dynamic parameter vector of the aforementioned multi-objective optimization model and the input layer II icing state vector The determined boundary conditions and security constraints affect the output layer infrared radiation parameter matrix. The six parameter solutions are encoded into qubits; an initial size of is generated. quantum population Each individual It is represented by a set of qubit chromosomes and is used to characterize a complete set of infrared de-icing parameters;

[0062] Specifically, this includes: using qubits Encode the decision variables, one Defined as ,in, ; and These represent the probabilities of collapsing to classical states 0 and 1 at the time of observation, respectively.

[0063] For those with A problem with one decision variable, a chromosome composed of... indivual String representation; each decision variable Use a length of 1 If the string is encoded, then the total length of an individual is... ;No. The Q-bit strings corresponding to the variables are:

[0064]

[0065] This formula is the complete representation of a quantum chromosome. Indicates the first in the population Individual; This represents the total number of decision variables. This corresponds to 6 infrared radiation parameters; The first decision variable is represented by a qubit. This represents the qubit representing the second decision variable; and so on up to the qubit... One variable.

[0066] For population initialization, the population size is set to N. For each individual in the population, each... Random initialization and ,make , indicating that the initial population collapses to 0 or 1 with equal probability, ensuring that the initial population is uniformly distributed in the search space; this initialization process is expressed as:

[0067]

[0068] in, ;

[0069] In the formula, This is the quantum population of generation 0; For the first The quantum chromosome of an individual; These represent the quantum chromosomes of the 1st, 2nd, and so on up to the Nth individual, where N represents the population size; Representing the The first individual The probability amplitude of each qubit; Indicates for all individuals and all qubits Both are valid.

[0070] In step S302, quantum observation and parameter decoding include:

[0071] quantum population Each individual in the process undergoes quantum observation, causing its qubit state to collapse into a classical binary string; the binary string is then decoded into a corresponding set of infrared radiation parameter values. ;Decoding It refers to the first Each quantum individual corresponds to a specific set of infrared radiation parameter values; therefore... It is a candidate solution in the optimization search process, representing a complete set of infrared de-icing parameter schemes evaluated by the algorithm in the current generation; specifically including: at the quantum observation level: in each generation For quantum populations Each individual in Observations were conducted based on their probability amplitude. Generate classic binary strings Then, decoding is performed to convert the binary string into a binary string. Divide the data into segments according to variables and map them back to the corresponding physical values ​​using linear transformations:

[0072]

[0073] In the formula, For the first The actual values ​​of the physical parameters to be optimized. This is the minimum allowable value for this physical parameter. This is the maximum allowed value for this physical parameter. For the corresponding parameter of Bit-bit binary encoded string The number of binary bits (bit string length) used to encode this parameter. The probability of obtaining the classical state 0 when measuring this qubit; The probability of obtaining classical state 1 when measuring this qubit; For variables The decimal number corresponding to the l-bit binary string; thus, a set of candidate solutions is obtained. Then perform a fitness assessment and decode the results. Substitute the physical model established in step S2 into the equation to calculate the corresponding three optimization objective values. And verify the security constraints, specifically by substituting the following into the formula:

[0074] Radiation source temperature calculation:

[0075]

[0076] In the formula, for Infrared system current at all times for Infrared system voltage at all times For electrothermal conversion efficiency, The Stefan-Boltzmann constant is... The effective radiation surface area of ​​the radiation source;

[0077] Net radiative heat flux density:

[0078]

[0079] In the formula, This represents the net radiative heat flux density received by the skin surface. For effective emission rate, The radiation source temperature was calculated in the previous step. For the perspective factor (angle coefficient), for The fourth power of the absolute temperature of the skin surface at any given time. This refers to the relative radiation area. The square of the distance from the radiation source to the detection point;

[0080] Solving for the temperature field of the skin:

[0081]

[0082] In the formula, The density of the skin material, The specific heat capacity of the skin material. For skin in depth ,time Temperature function at that location, The thermal conductivity of the skin material is... is the symbol for the second derivative of the space term.

[0083] In step S303, fitness evaluation and Pareto ranking include: dividing each set of decoded parameters... Substitute these parameters into the multi-objective optimization model and calculate the three objective function values ​​for each parameter. The system checks whether the constraint of no damage to the aircraft surface skin temperature is met. Then, based on the Pareto dominance relation, it performs a non-dominated sort of all feasible solutions in the current population and calculates the crowding distance of each individual to comprehensively evaluate the fitness of the individual.

[0084] The fitness assessment and Pareto ranking specifically include:

[0085] Step 1, constraint handling: Invoke the fast thermal response model, simplify the finite difference solver or pre-trained surrogate model, and calculate the parameter set. Peak temperature of lower skin Define constraint violation degree ,like ,but This is an infeasible solution; in the formula, For aircraft surface The maximum temperature at that location, To ensure safe temperatures for aircraft skin;

[0086] Step 2: Calculate the objective function, simultaneously calculating the values ​​of the three objective functions. ;

[0087] Step 3, Fitness Assignment Based on Pareto Ranking: A constrained non-dominated ranking is used. First, feasibility is compared. Among feasible solutions, the standard Pareto dominance relation is used for ranking. All individuals are assigned to different non-dominated fronts. ,in, It is the optimal frontier;

[0088] The fitness value of an individual is defined as:

[0089]

[0090] In the formula, For individuals The Pareto nondominated layers are: 1 for the frontier, 2 for the secondary frontier, and so on. To prevent small constants from being divided by zero, The weighting coefficient for the distance to congestion level. for The distance of congestion in its Pareto layer.

[0091] In step S304, elite selection and population update include: employing Choose a strategy to preserve the current generation population. and offspring populations generated through quantum operations All individuals are merged and sorted according to the fitness scores mentioned above. The top N best individuals are selected to enter the next generation. ;

[0092] In step S305, the cooperative mutation of the dynamically correlated quantum entanglement gate includes:

[0093] Step a, construct the correlation matrix, based on the physical model in step S2, analyze the coupling relationship between parameters; define parameters. and physical correlation coefficient The simplified model is as follows:

[0094]

[0095] In the formula, To calculate the correlation degree of the parameters, For the first An optimization objective function, For the first One parameter to be optimized. For the first One parameter to be optimized. The expected ratio of the changes in the two parameters based on physical laws. To adjust the parameters;

[0096] The physical model is divided into three main types: the infrared radiation heat transfer model, the heat conduction and temperature field model, and the physical law constraint model. Together, they constitute the correlation of computational parameters. The mathematical foundation of quantum entanglement gates ensures that the direction of mutation conforms to real physical laws;

[0097] Step b, Dynamic Entanglement Gate Applications of correlation Exceeding the threshold For parameter pairs, when performing mutation operations on their corresponding qubits, instead of independent bit flipping, a controlled quantum entanglement gate is applied. The strength of this gate is related to... Proportional; acts on associated bit pairs The simplified entanglement mutation is represented as:

[0098]

[0099] In the formula, It is a dynamically controlled quantum entanglement gate operator (matrix). The imaginary unit, For entanglement angle, .

[0100] Step S306, improving the selection of optimal parameters for adaptive quantum rotating door guidance includes:

[0101] First, the rotation direction needs to be determined: Let the current qubit be... The ideal state of the corresponding bit in the currently observed optimal solution or an elite solution is: Rotation direction The sign function of the difference in probability magnitude between the two factors determines that... Towards near;

[0102] The direction of the rotation angle of the adaptive quantum rotation gate is guided by the current state of the qubit and the state of the elite individual, while the magnitude of the rotation angle is determined by the state of the qubit itself. A function that adapts to the number of generations of evolution and dynamically adjusts with the number of generations and individual performance:

[0103]

[0104] In the formula, For the first In the first generation of evolution, the first The rotation angle of each qubit; and These are the minimum and maximum settings for the rotation angle, respectively. For the current generation, For the maximum number of generations, This is the global attenuation coefficient. For the first The variance or standard deviation of the parameter corresponding to each qubit in the current population. As the normalization factor, This is the local adjustment coefficient;

[0105] Step S307, iteration termination and Pareto optimal solution set output specifically includes: repeating steps S302-S306 above, with the termination condition usually being reaching the preset maximum number of generations. Or the improvement over multiple consecutive Pareto fronts is less than a certain threshold. ;

[0106] When the improved quantum genetic algorithm terminates, it outputs all non-dominated feasible solutions in the final population, forming the Pareto optimal solution set. Each solution in this solution set represents a set of infrared de-icing parameters that achieve a specific optimal balance among the three objectives of de-icing time, skin thermal shock, and total energy consumption.

[0107] Another objective of this invention is to provide an aircraft ground infrared de-icing system based on multi-parameter collaborative optimization. This system implements the aforementioned aircraft ground infrared de-icing method based on multi-parameter collaborative optimization. The system includes:

[0108] The environmental acquisition module is used to acquire and preprocess the dynamic environmental parameters of the de-icing site in real time to form an input vector.

[0109] The parameter generation module, with its built-in multi-objective optimization model and improved quantum genetic algorithm solution engine, is responsible for receiving input and generating the optimal set of de-icing parameters.

[0110] The parameter validity verification module is used to perform physical experiments to verify the generated parameter set in simulated and real environments, collect measured data and compare it with the optimization target;

[0111] The historical data module stores all relevant historical data, verification results, and performance metrics.

[0112] Combining all the above technical solutions, the beneficial effects of this invention are as follows:

[0113] First, this invention aims to solve the problems of isolated parameter control, difficulty in balancing multiple conflicting objectives, and the tendency of optimization algorithms to get trapped in local optima in existing infrared de-icing technologies. The steps include: constructing a system platform comprising an environmental acquisition module, a parameter generation module, a parameter validity verification module, and a historical data module; establishing a multi-objective collaborative optimization model with environmental dynamic parameters and icing state as inputs, infrared radiation parameters as outputs, de-icing time, skin temperature variation integral, and total energy consumption as optimization objectives, and skin temperature safety as a constraint; solving this model using an improved quantum genetic algorithm (IQGA), which achieves parameter collaborative mutation through dynamically associated quantum entanglement gates and balances global and local searches through adaptive quantum rotation gates, ultimately outputting a Pareto optimal solution set; and verifying the optimized parameters through physical experiments using the parameter validity verification platform, storing the results in the historical data module to support continuous system learning. This invention achieves intelligent collaborative optimization and closed-loop verification of de-icing parameters, improving de-icing efficiency, energy saving, and safety.

[0114] Secondly, in view of the problems existing in the above-mentioned prior art, such as isolated parameter control, difficulty in multi-objective trade-offs, insufficient performance of optimization algorithms, and inadequate security guarantees, this invention proposes a collaborative optimization method and system for aircraft ground infrared de-icing parameters based on an improved quantum genetic algorithm.

[0115] First, a hierarchical nested multi-objective collaborative optimization model was constructed, taking dynamic environmental parameters and icing state as inputs and infrared radiation parameter set as optimization outputs. The model explicitly and specifically takes de-icing time, skin temperature variation integral and total energy consumption as multiple objectives, and uses skin safety temperature as a constraint, providing a precise mathematical description for systematic optimization.

[0116] Secondly, in order to solve the problem of solving this complex model, an improved quantum genetic algorithm was designed. The improvements of this algorithm are mainly reflected in: (1) introducing a dynamic correlation quantum entanglement gate, which dynamically adjusts the mutation strategy according to the environment, the state of ice accumulation and the physical coupling relationship between parameters, so that the correlation parameters co-evolve, improving the physical targeting and efficiency of the search; (2) adopting an adaptive quantum rotation gate, which dynamically adjusts the search step size and direction according to the evolution stage and the individual's advantages and disadvantages, effectively balancing the global exploration and local development capabilities, avoiding premature convergence, and ensuring convergence to a high-quality Pareto front.

[0117] Finally, by constructing a closed-loop parameter validity verification platform, the optimized parameter set was experimentally verified in a simulated real environment, and the results were fed back to the historical database and the optimization model. This achieved a complete iterative optimization process of "modeling-optimization-verification-feedback", ensuring the reliability, security and practicality of the parameter scheme.

[0118] Third, through the above-mentioned systematic method, this invention aims to transform the aircraft ground infrared de-icing process from experience-based operation to intelligent and optimized decision-making, providing an advanced technical solution for efficient, energy-saving, and safe green de-icing operations.

[0119] This invention, through multi-parameter collaborative optimization, can save approximately 15-25% of energy and shorten de-icing time by 10-20% compared to traditional empirical methods, directly reducing airport operating costs. Simultaneously, it embeds skin temperature safety constraints into the optimization core, completely eliminating the risk of overheating damage and reducing aircraft maintenance costs and unplanned downtime due to improper operation. From an environmental perspective, this solution, as a physical de-icing technology, completely replaces chemical de-icing fluids, eliminating the pollution of ethylene glycol-based substances and aligning with the trend of green airport development. In terms of commercial applications, it can be developed into a standardized configuration for ground support equipment for airport groups, provide intelligent upgrade technology licenses for existing infrared de-icing equipment to airlines, and provide technical support for the Civil Aviation Administration of China to formulate industry standards for intelligent de-icing operations, possessing diversified market transformation paths.

[0120] Fourth, at the model level, a dynamic parameter fusion environment was constructed for the first time. Ice accumulation state vector and infrared radiation parameter matrix The hierarchical nested multi-objective optimization model achieves a leap from isolated parameter control to system-wide collaborative optimization. At the decision level, Pareto front solving replaces traditional weighted summation, providing users for the first time with a set of optimization schemes that balance de-icing time, thermal shock intensity, and total energy consumption, solving the problem of difficulty in quantifying and weighing conflicting multi-objectives. At the algorithm level, a pioneering dynamic correlation quantum entanglement gate mechanism is used to dynamically calculate the physical correlation of parameters based on real-time operating conditions. This enables the correlation parameters to co-evolve at the quantum level, breaking the limitations of traditional quantum genetic algorithms that do not utilize domain physics knowledge, and significantly improving search efficiency and solution quality.

[0121] Fifth, this invention achieves adaptive optimization of parameters under complex environments by real-time data acquisition. and As an optimization input, the de-icing scheme can dynamically respond to real-time changes in temperature, humidity, and wind speed, solving the dilemma of traditional empirical parameter models lagging behind the environment. This invention achieves a quantitative balance among conflicting optimization objectives for the first time, using de-icing time, skin temperature variation integral, and total energy consumption as parallel optimization objectives. By solving the Pareto front using an improved quantum genetic algorithm, efficiency, safety, and economy can be balanced and coexist at the quantitative level. Thirdly, safety constraints are truly embedded in the core of optimization rather than post-event verification—by predicting the peak skin temperature in real time through the heat conduction equation, [the system achieves this]. As a hard constraint embedded in the fitness evaluation, it ensures that all optimization solutions meet the safety requirements at the time of generation, eliminating the safety risks of traditional methods that rely on post-event detection.

[0122] Sixth, this invention breaks the prejudice that infrared de-icing parameters are independent and can be optimized separately, and reveals the deep coupling relationship between parameters through a physical model (such as...). This invention demonstrates the essential principle that multiple parameters must be optimized collaboratively by using quantum entanglement gates to achieve coordinated mutation of related parameters. It overturns the conventional thinking that multiple objectives can only be transformed into a single objective through weighted summation. Through Pareto non-dominated sorting and crowding distance calculation, it proves that heterogeneous objectives such as efficiency, safety, and economy can be optimized in parallel and coexist in a balanced manner. Thirdly, it corrects the traditional view that physical verification is merely an auxiliary step after optimization, constructing a closed-loop feedback system of optimization-verification-learning-re-optimization, allowing verification data to feed back into the initial population guidance and model parameter calibration, achieving a deep integration of optimization and verification. Fourthly, it breaks through the limitation that "quantum genetic algorithm improvement is limited to the algorithm itself," encoding the infrared radiation heat transfer law as the correlation coefficient of the quantum entanglement gate, achieving deep coupling between the algorithm mechanism and domain physics knowledge, and expanding the application boundaries of quantum evolutionary algorithms. Attached Figure Description

[0123] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure;

[0124] Figure 1 This is a flowchart of the aircraft ground infrared de-icing method based on multi-parameter collaborative optimization provided by the present invention;

[0125] Figure 2 This is a schematic diagram illustrating the principle of the aircraft ground infrared de-icing method based on multi-parameter collaborative optimization provided by the present invention.

[0126] Figure 3 A flowchart of the improved quantum genetic algorithm (IQGA) provided for this invention;

[0127] Figure 4 This is a partial magnified schematic diagram of the infrared device and its principle structure for infrared de-icing provided by the present invention.

[0128] Figure 5 An infrared schematic diagram of the aircraft ground infrared de-icing method based on multi-parameter collaborative optimization provided by the present invention;

[0129] Figure 6 This is a schematic diagram illustrating the optimization of high-speed icing state in the aircraft ground infrared de-icing method based on multi-parameter collaborative optimization provided by the present invention. Detailed Implementation

[0130] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0131] The innovations of this invention are as follows: First, a multi-objective collaborative optimization model integrating dynamic environmental parameters and icing conditions is constructed, achieving parallel collaborative optimization of multiple parameters such as infrared radiation current, voltage, radiation area, intensity, and distance. Second, an improved quantum genetic algorithm (IQGA) is designed, which uses a dynamically correlated quantum entanglement gate to enable strongly correlated parameters to mutate collaboratively according to physical laws, and employs an adaptive quantum rotation gate to balance global exploration and local development, significantly improving the efficiency and accuracy of Pareto front solutions. Finally, an effectiveness verification platform including environmental simulation, parameter execution, and data acquisition is constructed, forming an "optimization-verification-feedback" closed loop, and knowledge accumulation and continuous learning are achieved through a historical data module. This invention provides an efficient, energy-saving, and safe intelligent solution for aircraft ground infrared de-icing.

[0132] Example 1, such as Figure 1 As shown, the aircraft ground infrared de-icing method based on multi-parameter collaborative optimization provided by the present invention includes the following steps:

[0133] S1, Construct an aircraft ground infrared de-icing system based on multi-parameter collaborative optimization (aircraft ground infrared de-icing parameter collaborative optimization and generation system platform), the aircraft ground infrared de-icing system based on multi-parameter collaborative optimization includes an environmental acquisition module, a parameter generation module, a parameter validity verification module, and a historical data module;

[0134] S2. Based on the constructed aircraft ground infrared de-icing system based on multi-parameter collaborative optimization, the infrared de-icing parameters are collaboratively optimized, and a multi-objective optimization model is constructed.

[0135] Specifically, ambient temperature, humidity, wind speed, and the duration of each parameter are used as input layer I; ice type, ice thickness, and ice evolution state are used as input layer II; and infrared radiation current (A), infrared radiation voltage (V), and relative infrared radiation area (m²) are used as input parameters. 2 Infrared radiation intensity (kw / m) 2 Infrared irradiance (W / m) 2 The infrared source distance to the icing point (mm) and other parameters are used as the parameters to be optimized, i.e., the output layer;

[0136] Among infrared de-icing time, aircraft surface skin temperature change, and infrared de-icing energy consumption, the optimization objectives are the shortest de-icing time, the most gradual change in skin temperature (measured by the integral of the temperature change rate), and the lowest total energy consumption. The optimization objectives are regarded as the core of collaborative optimization, and safety constraints are added, with the constraint that the temperature rise of the aircraft surface skin does not exceed the safety threshold that would damage the aircraft skin material. A hierarchical nested multi-objective optimization model is constructed.

[0137] S3 utilizes an improved quantum genetic algorithm to solve a collaborative multi-objective optimization model and generate parameters. The entanglement state between parameters is determined by the correlation between environmental parameters, ice accumulation parameters, and parameters in the current state. The solution process combines the balance between optimization objectives to finally obtain the optimal set of de-icing parameters.

[0138] This invention designs an improved quantum genetic algorithm (IQGA) to solve the multi-objective optimization model constructed in step S2 and generate the optimal set of de-icing parameters. The improvements to this algorithm are mainly reflected in:

[0139] Dynamically correlated quantum entanglement gates: Quantum entanglement gates can be dynamically constructed based on the physical coupling relationship between current environmental parameters, icing state, and infrared radiation parameters, enabling parameters with strong correlation to change synergistically during mutation, thereby improving the targeting and efficiency of the search.

[0140] Adaptive Quantum Rotation Gate: The direction of the rotation angle is guided by the states of the current individual and elite individuals. The size of the rotation angle adaptively adjusts with the number of generations. A larger angle is used initially for global exploration, while a smaller angle is used later for refined local search, effectively balancing global and local search capabilities and avoiding premature convergence caused by traditional genetic algorithms. The specific steps of this improved quantum genetic algorithm include: quantum encoding and population initialization, quantum observation and parameter decoding, fitness evaluation and Pareto sorting, elite selection and population update, dynamic association quantum entanglement gate co-mutation, improved adaptive quantum rotation gate-guided search, iterative evolution until convergence, and finally outputting the Pareto optimal solution set.

[0141] S4. After generating the optimal set of de-icing parameters, repeat experiments are conducted in the parameter validity verification module based on the real environment to determine whether the de-icing effect meets the de-icing requirements.

[0142] In step S4, a parameter validity verification module is introduced. The optimal de-icing parameter set generated in step S3 is loaded into this module. This module replicates the actual situation using a real-world experimental platform, driving the infrared radiation actuator to operate according to the parameters. It also collects real-time data on the skin temperature field, ice accumulation status images, and real-time power consumption. The actual measurement results are compared with the predicted targets of the optimization model to determine whether the preset tolerance requirements are met. The verification results are fed back to the system, forming a closed loop of optimization, verification, and feedback.

[0143] S5, in the historical data module, performs data processing and analysis on the obtained experimental data.

[0144] The complete set of optimal de-icing parameters, environmental conditions, verification results, and performance indicators are stored in the historical data module mentioned in step S1 for later querying, analysis, comparison, and to assist in algorithm evolution.

[0145] For example, Figure 2 This is a schematic diagram illustrating the principle of the aircraft ground infrared de-icing method based on multi-parameter collaborative optimization provided by the present invention.

[0146] For example, in step S2, such as Figure 2 The multi-objective optimization model adopts a hierarchical nested architecture.

[0147] The input layer of the model is physically quantified and preprocessed. The input layer of the model is divided into two parts: environmental dynamic parameters and icing state parameters. Both of them need to be transformed from raw data into physical quantities that can be used for quantitative calculation.

[0148] Input layer I: Environmental dynamic parameter vector;

[0149]

[0150] In the formula, This refers to the ambient temperature, expressed in °C. Relative humidity of air, in percentages (%) Wind speed, measured in m / s; The duration of each parameter is in seconds, and ;

[0151] Ambient temperature directly determines the initial temperature difference driving force and ambient heat sink temperature in the heat transfer process. Relative humidity and temperature together determine the dew point temperature. This is crucial for predicting the likelihood of icing and assessing the evaporative cooling effect. Wind speed through convective heat transfer coefficient This directly affects the rate of heat loss from the aircraft skin surface. The relationship between the two can be described by empirical correlations, such as for forced convection:

[0152]

[0153] In the formula, The coefficients, which are related to the physical properties of air and surface geometry, can be obtained through wind tunnel testing or computational fluid dynamics (CFD) calibration. It is a key disturbance source that must be considered in real time during dynamic optimization. Duration characterizes the stability of the environmental state and is used to determine whether the model needs to be reinitialized or historical data needs to be weighted by time windows.

[0154] Input Layer II: Icing State Vector Perform quantitative modeling:

[0155]

[0156] In the formula, The encoding for the ice accumulation type, where, 1 represents raw ice, 2 represents clear ice, and 3 represents mixed ice; This refers to the thickness of the ice buildup, in mm. Let be the ice accumulation evolution state coefficient, where 0 represents a pure solid state, and 1 represents ice that has completely melted into a liquid water film;

[0157] For example, for the above These need to be converted into physical parameters that can be directly used in thermodynamic calculations, among which... The ice accumulation type is coded as a discrete value (1: rough ice, 2: clear ice, 3: mixed ice). Different types of ice accumulation have different densities, thermal conductivity, and infrared radiation absorption / reflection characteristics. In the model, It is mainly used to select the corresponding set of material property parameters; The ice thickness, obtained using a micrometer, is measured in millimeters (mm). It is a key dimensional parameter for calculating thermal resistance and phase change heat. The ice evolution state coefficient is determined by... This indicates that it is a continuous variable. Used to describe the transition of ice layers from a purely solid state To a completely melted liquid water film The phase transition process. This invention introduces an equivalent composite medium model, treating the ice-water mixture in the melting process as a homogeneous medium with an equivalent thermal conductivity. Based on the thermal conductivity of ice Thermal conductivity of water Determined by linear mixing according to phase change ratio

[0158]

[0159] Therefore, the ice evolution state coefficient Used to calculate the equivalent thermal resistance of ice accumulation layers. for:

[0160]

[0161] In the formula, In order to be in The equivalent thermal resistance of the ice layer at any given time, in units of ; and These are the thermal conductivity coefficients of ice and water, respectively. for The equivalent thermal conductivity of the ice layer at any given time This refers to the proportion or weighting coefficient of the liquid phase (water) in the ice layer;

[0162] According to research, the thermal conductivity of ice is... Approximately The thermal conductivity of water Approximately This formula quantifies the state of ice accumulation (thickness, type, phase) into thermal resistance, and is the core bridge connecting the microscopic phase change process of ice accumulation with macroscopic heat transfer. The change over time is itself a function of radiative heating, but within the short timescale of a single optimization solution, it can be regarded as a quasi-static parameter based on the current detection value.

[0163] Output layer: Infrared radiation parameter matrix Defined as:

[0164]

[0165] In the formula, This refers to infrared radiation current, measured in amperes (A). Infrared radiation voltage, measured in V; The relative radiation area is expressed in meters (m²). 2 ; Infrared radiation intensity, unit: kw / m 2 ; Infrared irradiance, unit: W / m 2 ; The distance between the infrared source and the ice accumulation point is in mm.

[0166] For the infrared radiation parameter matrix In the collaborative optimization of key parameters, the infrared radiation current With infrared radiation voltage The electrical power input of the infrared radiation source is determined by the coordination and is the core execution variable for regulating the total radiation energy. By optimizing its combination, precise control of the heating rate of the target area can be achieved, and energy consumption can be minimized while meeting the de-icing requirements.

[0167] The relative area of ​​infrared radiation To optimize the effective coverage of infrared radiation energy on the aircraft skin surface, the geometric distribution of the icing area and the optical characteristics of the radiation source must be considered. The effective irradiation area can be changed by adjusting the position or array configuration of the radiation source in order to achieve efficient energy utilization and avoid local overheating or underheating.

[0168] The distance between the infrared source and the ice accumulation point It directly affects the energy flux density radiated to the skin surface, and its optimization requires a comprehensive balance of radiation transfer efficiency, space operation constraints, and skin temperature safety limits. In the multi-objective optimization model, this parameter is closely coupled with the radiation intensity parameter, and together they determine the net heat flux input to the ice-skin interface.

[0169] The infrared radiation intensity As a key physical quantity characterizing the radiative power per unit area, its optimization aims to ensure that the energy density acting on the skin-ice interface is always within the optimal range allowed by efficient de-icing and material safety under different environments and icing conditions.

[0170] For example, the coordinated optimization of key parameters in the infrared radiation parameter matrix P(t) specifically includes:

[0171] Based on the output infrared parameter matrix The effect of the infrared radiation on the aircraft skin-icing system is calculated using physical laws to obtain the optimized objective function value and constraint verification value. To address the infrared radiation energy input, the electrothermal conversion of the infrared radiation source (such as a silicon carbide plate) follows Joule's law. Its surface temperature... Determined by the balance between input electrical power and radiative heat dissipation, it can be derived from the Stefan-Boltzmann law:

[0172]

[0173] In the formula, The surface temperature (K) of the infrared radiation source. For electro-radiation conversion efficiency, The Stefan-Boltzmann constant is generally taken as 1. , The effective radiation area of ​​the radiation source, and For convection and other heat losses, usually for the sake of simplifying calculations, we can... Ignore it, or categorize it as efficiency. Considering the separation, we can conclude that:

[0174]

[0175] Deriving this expression greatly simplifies calculations, allowing for the processing of controllable electrical parameters in reality. Converted to the source temperature that determines the radiation spectrum distribution This is also the starting point for calculating radiative heat transfer.

[0176] Considering the geometric relationship between the infrared radiation source and the aircraft skin surface, atmospheric attenuation, and surface characteristics, the net radiative heat flux density per unit area reaching the skin is... The calculation is as follows:

[0177]

[0178] In the formula, In order to be in Net radiative heat flux density (W / m²) reaching the surface of the aircraft skin at any given time; The effective emissivity between the infrared radiation source and the skin surface is determined by considering the surface characteristics of both. That is, the viewing angle factor, which is related to the relative geometric position of the radiation source and the skin, and can be approximated as 1 for vertical radiation; for The surface temperature of the skin at any given time (K); The distance (in meters) between the infrared source and the ice accumulation point is given by this formula, which is the core energy input formula and clearly expresses the output parameters. How the thermal energy input to the skin-ice system is determined collectively is shown in the formula, which reflects the inverse squared attenuation effect over distance L.

[0179] Temperature field of aircraft skin (especially composite materials) The variation is controlled by a one-dimensional unsteady-state heat conduction equation (along the skin thickness direction). ):

[0180]

[0181] In the formula, These are the density, specific heat capacity, and thermal conductivity of the skin material, respectively; the key boundary conditions are located on the outer surface of the skin. , The direction of skin thickness; coupled radiative heating, convective heat dissipation in the air, and heat conduction through the ice layer:

[0182]

[0183]

[0184]

[0185] In the formula, For surface convection heat dissipation, The convective heat transfer coefficient is related to the wind speed. related; This refers to the heat flow conducted through the ice layer; The temperature distribution function inside the skin. The net radiative heat flow of the surface. The surface convective heat transfer coefficient is... for Time skin outer surface (i.e. Temperature at (location) Ambient air temperature, The temperature of the outer surface of the ice layer. The equivalent thermal resistance of the ice layer;

[0186] The mathematical expression for the safety constraint is the time during the entire de-icing process. The temperature of the inner skin's outer surface and any internal location must not exceed the maximum safe temperature allowed by its material. :

[0187]

[0188] In the formula, Skin thickness; In order to defrost time Maximum temperature at any location on the outer surface and inside of the inner skin;

[0189] During the optimization process, the partial differential equation system is solved rapidly using the finite difference method, integral approximation method, or response surface model based on historical data to predict... The evolution path is thus used to intrinsically verify security constraints while optimizing the objective function.

[0190] Based on the above physical model, the three optimization objective functions are defined as follows:

[0191] objective function De-icing completion time;

[0192]

[0193] In the formula, To start with radiation The time required to meet the de-icing completion criterion is based on the state of ice accumulation. Or visual monitoring signals; such as Or visual monitoring signals (complete ice removal).

[0194] objective function Skin temperature stability;

[0195]

[0196] In the formula, for The temperature of the outer surface of the skin at all times;

[0197] The goal is to minimize thermal shock, protect the structural integrity of the composite skin, and extend its fatigue life.

[0198] objective function Total energy consumption;

[0199]

[0200] In the formula, Real-time power consumption, in watts (W). for Infrared system current at all times for The voltage of the infrared system at all times; this target directly corresponds to the operating cost and energy efficiency of de-icing operations, and is an important indicator of green de-icing technology.

[0201] The objective function of the multi-objective optimization model is: to construct the Pareto optimal multi-objective vector. Specifically:

[0202]

[0203]

[0204] In the formula, for Infrared system current at all times for Infrared system voltage at all times.

[0205] In step S3, the improved quantum genetic algorithm (IQGA) is the core engine for solving the high-dimensional, nonlinear, multi-objective, constrained multi-objective optimization model established in step S2. Its improvement mechanism directly determines the efficiency of the optimization process and the quality of the solution. The infrared de-icing parameter optimization problem is formally mapped to the solution framework of IQGA, as shown in Table 1.

[0206] Table 1

[0207] IQGA, or Infinite Quantity Gaussian (IQGA), searches within the feasible region for... The set of solutions that achieves Pareto optimality.

[0208] Specifically, such as Figure 3 As shown, the use of an improved quantum genetic algorithm for solving a collaborative multi-objective optimization model and generating parameters includes:

[0209] S301, Quantum Encoding and Population Initialization;

[0210] Based on the aforementioned multi-objective optimization model, "Input Layer I: Environmental Dynamic Parameter Vector" "and Input Layer II: Icing State Vector" "Based on the determined boundary conditions and security constraints, the six parameter solutions of the output layer infrared radiation parameter matrix P(t) are encoded into qubits; an initial quantum population of size X is generated." Each individual It is represented by a set of qubit chromosomes and is used to characterize a complete set of infrared de-icing parameters.

[0211] Another concrete example is the use of qubits to leverage the parallelism advantage of quantum computing. Encode the decision variables. One Defined as ,in . and These represent the probabilities of collapsing to classical states 0 and 1 at the time of observation, respectively.

[0212] For those with A problem with one decision variable, a chromosome composed of... indivual String representation; each decision variable Use a length of 1 If the string is encoded, then the total length of an individual is... ;No. The Q-bit strings corresponding to the variables are:

[0213]

[0214] This formula is the complete representation of a quantum chromosome. Indicates the first in the population Individual; This represents the total number of decision variables. This corresponds to 6 infrared radiation parameters; The first decision variable is represented by a qubit. This represents the qubit representing the second decision variable; and so on up to the qubit... One variable.

[0215] For population initialization, the population size is set to N. For each individual in the population, each... Random initialization and , usually This means that the population collapses to 0 or 1 with equal probability, thus ensuring that the initial population is uniformly distributed within the search space. This initialization process is expressed as:

[0216]

[0217] in, ;

[0218] In the formula, This is the quantum population of generation 0 (initial generation); For the first The quantum chromosome of an individual; These represent the quantum chromosomes of the 1st, 2nd, and so on up to the Nth individual, where N represents the population size; Representing the The first individual The probability amplitude of each qubit; Indicates for all individuals and all qubits Both are valid.

[0219] S302, Quantum Observation and Parameter Decoding;

[0220] quantum population Each individual in the process undergoes quantum observation, causing its qubit state to collapse into a classical binary string; the binary string is then decoded into a corresponding set of infrared radiation parameter values. Decoded It refers to the first Each quantum individual corresponds to a specific set of infrared radiation parameter values; therefore... It is a candidate solution in the optimization search process, representing a complete set of infrared de-icing parameter schemes evaluated by the algorithm in the current generation.

[0221] For example, firstly, at the quantum observation level: in each generation For quantum populations Each individual in Observations were conducted based on their probability amplitude. Generate a classic binary string Then, decoding is performed to convert the binary string into a binary string. Divide the data into segments according to variables and map them back to their corresponding physical values ​​through linear transformations:

[0222]

[0223] In the formula, The probability of obtaining the classical state "0" when measuring this qubit; The probability of obtaining the classical state "1" when measuring this qubit.

[0224] in It represents a variable The decimal number corresponding to the l-bit binary string. This yields a set of candidate solutions. Then perform a fitness assessment and decode the results. Substitute the physical model established in step S2 into the equation to calculate the corresponding three optimization objective values. And verify the security constraints, specifically by substituting the following into the formula:

[0225] Radiation source temperature calculation:

[0226]

[0227] Net radiative heat flux density:

[0228]

[0229] Solving for the temperature field of the skin:

[0230]

[0231] S303, Fitness Assessment and Pareto Ranking: This involves evaluating the decoded parameters of each group. Substitute these parameters into the multi-objective optimization model and calculate the three objective function values ​​for each parameter. At the same time, it is also necessary to pay attention to constraints, and check whether the constraint condition of no damage to the surface skin temperature of the aircraft is met; then, based on the Pareto dominance relation, perform non-dominated sorting of all feasible solutions in the current population, and calculate the crowding distance of each individual, so as to comprehensively evaluate the fitness of the individual.

[0232] For example, fitness assessment and Pareto ranking specifically include:

[0233] Step 1, Constraint Handling (Feasibility Assessment): Invoke a fast thermal response model, such as a simplified finite difference solver or a pre-trained surrogate model, to calculate the parameter set. Peak temperature of lower skin Define the degree of constraint violation. .like 0, then This is an infeasible solution.

[0234] Step 2, Calculate the objective function: Calculate the values ​​of the three objective functions simultaneously. .

[0235] Step 3, Fitness Assignment Based on Pareto Ranking: A constrained-dominated ranking is used. First, feasibility is compared: feasible solutions always dominate infeasible solutions. Among feasible solutions, the standard Pareto dominance relation is used for ranking. All individuals are assigned to different non-dominated fronts. ,in, This is the optimal frontier. To maintain the diversity of solutions within the same frontier, the crowding distance for each individual is calculated, which measures the density between adjacent solutions on its frontier.

[0236] An individual's fitness value is determined by its priority in the optimization process; generally, individuals with smaller frontier indices (higher fitness) and larger crowding distances (sparser populations) have higher fitness. This can be defined as:

[0237]

[0238] In the formula, For individuals The Pareto nondominated layers are: 1 for the frontier, 2 for the secondary frontier, and so on. for The distance of congestion within its Pareto layer; This represents a small constant that prevents division by zero. This is the weighting coefficient for the crowding distance.

[0239] In elite selection and tournament selection, the fitness of an individual is evaluated by combining Pareto rank (preferring solutions at the forefront) and crowding (preferring solutions in sparse regions to maintain diversity), which is one of the core operations of multi-objective evolutionary algorithms.

[0240] S304, Elite Selection and Population Renewal: An elite retention strategy is adopted, in which the top few individuals with the highest Pareto level in the current population are directly retained to the next generation; the remaining individuals are selected from the parents through a tournament selection method to prepare for subsequent quantum evolution operations.

[0241] For example, adopting Choose a strategy. Retain the current generation population. and offspring populations generated through quantum operations All individuals are merged and sorted according to the fitness scores mentioned above. The top N best individuals are selected to enter the next generation. This ensures the preservation of elite individuals and prevents the loss of superior genes.

[0242] S305, Co-mutation of Dynamically Correlated Quantum Entanglement Gates, is the key difference between this invention and traditional QGA. Its purpose is to enable physically closely related parameters to change co-variantly during evolution, including:

[0243] Input based on the current environment Ice accumulation status input The distribution of parameters (physical coupling relationships between infrared radiation parameters) in the population, and the dynamic calculation of the physical correlation coefficients between different infrared radiation parameters (such as the aforementioned infrared radiation intensity and distance, voltage and current). And thereby construct or adjust quantum entanglement gates. The entanglement gate is used to perform co-mutation on the individuals selected in step S304 elite selection and population update, so that the parameters with strong correlation are further co-optimized, thereby improving the efficiency and globality of the search.

[0244] For example, the cooperative mutation of dynamically correlated quantum entanglement gates specifically includes:

[0245] Step a, Correlation Matrix Construction: Based on the physical model in step S2, analyze the coupling relationships between parameters. For example, radiation intensity... With distance Strong coupling via the inverse square law; current With voltage Together they determine the power. Define parameters. and physical correlation coefficient This coefficient can be obtained through offline analysis, calculating the sensitivity of the objective function to the cross-partial derivatives of the parameters, or through online learning, based on the correlation statistics of parameters in historical high-quality solutions. A simplified model is as follows:

[0246]

[0247] In the formula, To calculate the correlation degree of the parameters, For the first An optimization objective function, For the first One parameter to be optimized. For the first One parameter to be optimized. It is based on the expected ratio of the changes in two parameters according to physical laws (e.g., to keep the heat flow constant, L must be halved). If increased to 4 times, then , To adjust the parameters, the physical model can be broadly divided into three main types: the infrared radiation heat transfer model, the heat conduction and temperature field model, and the physical law constraint model. These three types together constitute the correlation of the calculation parameters. The mathematical foundation ensures that the mutation direction of the quantum entanglement gate conforms to the real physical laws.

[0248] Step b, Dynamic Entanglement Gate Application: For correlation Exceeding the threshold When performing a mutation operation on the parameter pairs corresponding to their qubits, instead of using independent bit flips, a controlled quantum entanglement gate (such as the controlled NOT gate CNOT or its generalized form) is applied. The operational strength of this gate is related to... Proportional. For example, an action on an associated bit pair The simplified entanglement mutation can be represented as:

[0249]

[0250] The entanglement angle This operation allows a change in the state of one bit to probabilistically affect the state of another bit, enabling cooperative exploration of parameters and greatly improving search efficiency in the direction of coupled parameters.

[0251] S306, an improved adaptive quantum rotation gate guides optimal parameter selection. The rotation gate is used to guide the probability magnitude of the qubit to update in a direction that yields better solutions. Traditional QGAs use rotation tables with fixed angles; this invention designs an adaptive mechanism, including:

[0252] An adaptive quantum rotation gate is designed. The direction of the rotation angle of the adaptive quantum rotation gate is jointly determined by the probability amplitude of the current qubit and the orientation of the elite individuals in the current Pareto front. The size of the rotation angle of the adaptive quantum rotation gate is adaptively adjusted with the number of generations of evolution to achieve fast global search in the early stage, avoid getting trapped in local optima, and refine the balance of fine local search in the later stage, guiding the population to converge toward a more optimized Pareto front.

[0253] For example, improving the selection of optimal parameters for adaptive quantum rotating gate guidance specifically includes:

[0254] First, the rotation direction needs to be determined: Let the current qubit be... The ideal state of the corresponding bit in the currently observed optimal solution or an elite solution is: Rotation direction The sign function of the difference in probability magnitude between the two is determined by the aim of making Towards Approaching. Once the approximate direction of rotation is determined, the rotation angle will be adaptively adjusted. It is the key to algorithm performance.

[0255] The adaptive quantum rotation gate's rotation angle is guided by both the current state of the qubit and the state of the elite individual, and its rotation angle is... Based on adaptive adjustment with the number of evolutionary generations, this invention proposes a function that dynamically adjusts with the number of evolutionary generations and individual performance:

[0256]

[0257] In the formula, For the first In the first generation of evolution, the first The rotation angle of each qubit; and Indicates the minimum and maximum set values ​​for the rotation angle; The current generation number; The maximum number of generations; The global decay coefficient controls the rate at which the angle decays exponentially with the algebraic progression, enabling "large-angle exploration in the early stages and fine-grained small-angle search in the later stages." For the first The variance or standard deviation of the parameter corresponding to each qubit in the current population; Normalization factor; This is a local adjustment coefficient, which is used when a parameter's distribution diverges in the population. When the area is large, the rotation angle is appropriately increased to promote exploration of the region; conversely, it is decreased to induce convergence. This integrates evolutionary stages and individual performance, achieving adaptive allocation of search intensity in time and space.

[0258] It can be seen that steps S304-S306 realize the adaptive search guided by physical knowledge.

[0259] S307, Iteration termination and Pareto optimal solution set output;

[0260] Repeat steps S302-S306 above for iterative evolution until the preset maximum number of generations is reached or the convergence condition is met; finally, output all solutions on the Pareto non-dominated layer in the last generation population to form the optimal de-icing parameter set for use by the parameter validity verification module.

[0261] Repeat steps S302-S306 above, with the termination condition typically being reaching the preset maximum number of generations. Or the improvement over multiple consecutive Pareto fronts is less than a certain threshold. .

[0262] When the algorithm terminates, it outputs all non-dominated feasible solutions in the final population, forming the Pareto optimal solution set. Each solution in this solution set represents a specific optimal balance among three objectives: de-icing time, skin thermal shock, and total energy consumption, achieved through infrared de-icing parameters. All solutions strictly meet skin temperature safety constraints and can be selected by the user based on their actual preferences (such as prioritizing time or energy consumption), or directly input into the parameter validity verification module for physical experimental verification.

[0263] For example, conducting repeated experiments in conjunction with the parameter validity verification module described in step S4 to determine whether the de-icing effect meets the de-icing requirements includes:

[0264] The parameter validity verification module is a semi-physical simulation experimental system. Its core function is to accurately reproduce actual or typical environmental and icing scenarios under controlled conditions, and load and run the optimized parameter set output by the parameter generation module to objectively evaluate its actual de-icing performance and safety.

[0265] The parameter validity verification module can verify the authenticity of the parameters based on experimental data obtained from simulation and iteration. The platform consists of four main parts for real experiments and data processing. The first part is the environment simulation module, which is used to reproduce the input dynamic environmental parameters in the experimental environment. The second part is the infrared radiation execution module, which is used to load and execute the infrared radiation parameters in the optimal parameter set. The third part is the comprehensive data acquisition module, which is used to collect the skin temperature field, ice evolution state and actual power consumption data of the system in real time during the de-icing process. The last part is the verification and feedback module, which is used to compare the collected measured data with the optimization objectives and safety constraints of the multi-objective optimization model. If the measured results meet the preset tolerance requirements, the parameters are determined to be valid and the effect is as expected. Otherwise, the verification results are fed back to the system historical data module and the parameter re-optimization process is triggered.

[0266] For example, the environmental simulation module employs a walk-in high and low temperature humidity test chamber, equipped with a precision refrigeration system, a steam humidification system, and an adjustable speed fan array. This unit can adjust the input... Vectors dynamically reproduce specified temperatures (range: -30℃ to 10℃), relative humidity (range: 20% to 98%RH), and wind speeds (range: 0 to 15m / s) within the experimental chamber.

[0267] Infrared radiation actuation module: the core actuator, which includes:

[0268] Programmable DC power supply: used for precise output and regulation of infrared radiation current specified in the optimized parameters. With voltage ;

[0269] Infrared radiating plate array: Using silicon carbide as the radiating element, its dimensions are known, and it is used to achieve an optimized relative radiating area. .

[0270] Six-DOF robotic arm: The end effector is equipped with an infrared radiation plate, which is used to precisely control the relative distance (L) between the infrared source and the aircraft skin test piece fixed on the test bench, and can move along a preset path to achieve scanning and irradiation of a specific area.

[0271] The integrated data acquisition module includes:

[0272] Infrared thermal imager: Acquires two-dimensional temperature field distribution on the surface of the test piece at a frequency of not less than 5Hz. The spatial resolution meets the requirements for monitoring local hot spots;

[0273] High-speed camera: Records the entire process of ice melting and peeling under infrared radiation, used for visual determination of when de-icing is completed. And observe the evolution of the ice layer.

[0274] Power analyzer: Connected in series in the power supply circuit of the infrared radiation actuator, it collects and records the actual operating voltage, current, and instantaneous power of the system in real time. .

[0275] Environmental sensors: These are placed near the test specimen to monitor the actual temperature, humidity, and wind speed inside the chamber in real time, in order to verify the accuracy of environmental reproduction.

[0276] In the initial preparation phase, environmental parameters E are set according to the target operating conditions in an environmental simulation chamber. On the surface of the aircraft skin test piece (here, a plate made of real aircraft skin material), methods such as spray freezing are used to prepare materials that meet the requirements. Description (type) ,thickness ,state The ice layer. After preparation, the parameters are loaded and executed to verify that the platform control software receives a set of optimized parameters from the parameter generation module. The software automatically analyzes the parameters and synchronously controls the programmable power supply and robotic arm, bringing the system into the preset initial state. Upon starting the experiment, the infrared radiation actuator strictly follows... Define the timing and numerical operation. Data acquisition is performed synchronously. During the de-icing process, the infrared thermal imager, high-speed camera, power analyzer, and environmental sensors are triggered synchronously and continuously record until the system determines that de-icing is complete (e.g., the thermal imager shows that the temperature of the ice layer area has uniformly risen to close to 0℃ and the high-speed image shows that the ice layer has completely detached). After data acquisition is completed, the system saves all raw data and resets or replaces the test specimen to prepare for the verification of the next set of parameters.

[0277] The verification and feedback module calculates a quantitative evaluation index of the de-icing effect. To compare the performance of different parameter schemes:

[0278]

[0279] In the formula, These are measured values. For reference only. This is the safe temperature threshold.

[0280] For example, in step S5, the obtained experimental data is processed and analyzed in the historical data module, including the extraction of key performance indicators.

[0281] The moment when the ice layer is completely peeled off is automatically or semi-automatically identified from high-speed camera sequences using image processing algorithms and recorded as the measured de-icing time. .

[0282] Recorded by the power analyzer The curve from the start of the experiment to Integrate to obtain the measured total energy consumption. .

[0283] Extract the highest temperature value of the entire test specimen surface throughout the entire process from the infrared thermal imager data sequence. And the temperature-time curves for key locations (such as the heating center, edges, and material joints). Calculate them separately. , , Compared with the predicted values ​​of the optimization model , , (or The relative error , , .

[0284] The extracted data is filtered, judged, and labeled. ,and If all values ​​are less than the preset threshold, the parameter set is considered "valid". Otherwise, the parameter is considered invalid (safety constraint violation). If the temperature is safe but the time or energy consumption error exceeds the threshold, it can be marked as "partially valid" or "recalibrated" depending on the degree of exceedance.

[0285] The filtered structured data is then stored in the historical database. For each complete validation experiment, regardless of the validity of the result, a structured record is generated and stored in the historical data module. Each record mainly contains the following fields:

[0286] Task Identifier and Status: Unique Experiment ID, Timestamp, and Corresponding Environment Vector Ice accumulation state vector ;

[0287] Input parameters: The set of optimized parameters to be validated The optimized task ID and its source;

[0288] Raw and processed data: links to stored raw thermal images, videos, and power waveform files; and extracted data. Key indicator values;

[0289] Verification conclusions: validity determination results, calculated error values, and brief descriptions of phenomena or analysis of failure causes;

[0290] Performance tags: Tags that can be added manually or automatically, such as "high-efficiency solution", "energy-saving solution", "strong robustness", etc., for subsequent advanced queries.

[0291] The historical data module is not only used for archiving but also serves the iterative optimization of the system. It provides prior knowledge for the algorithm: before undertaking a new optimization task, the parameter generation module can query the historical database for successful cases similar to the current operating conditions. Effective parameter combinations from these cases can serve as high-quality initial seeds, injected into the initial population of the IQGA, thereby accelerating convergence and increasing the probability of finding the global optimum. Furthermore, the model can be calibrated and corrected based on historical data, and statistical analysis of historical data can be performed periodically to compare model predictions. ) and measured value ( Systematic biases. If a persistently large bias is found under specific operating conditions (such as extremely high humidity), a calibration process to optimize the model's internal parameters (such as the heat transfer coefficient) can be triggered, thereby improving the model's universality and prediction accuracy.

[0292] Example 2: This invention provides an aircraft ground infrared de-icing system based on multi-parameter collaborative optimization (aircraft ground infrared de-icing parameter collaborative optimization and generation system platform), the system including an environmental acquisition module, a parameter generation module, a parameter validity verification module, and a historical data module;

[0293] For example, in the environmental acquisition module, the module is used to acquire and preprocess the dynamic environmental parameters of the de-icing site in real time to form an input vector;

[0294] The parameter generation module, as the core computing unit, has a built-in multi-objective optimization model and an improved quantum genetic algorithm solution engine. It is responsible for receiving input and generating the optimal de-icing parameter set.

[0295] The parameter validity verification module is used to perform physical experiments to verify the generated parameter set in simulated and real environments, collect measured data and compare it with the optimization target.

[0296] The final step is in the historical data module, which stores all relevant historical data, verification results, and performance metrics, supporting data backtracking, analysis, and providing an experience-based learning foundation for algorithm optimization.

[0297] Specifically, the environmental acquisition module is responsible for collecting, monitoring, and preprocessing environmental parameters at the aircraft de-icing operation site in real time or as needed. By deploying a high-precision sensor network in the de-icing area, including temperature sensors, humidity sensors, anemometers, timing units, and signal conditioning circuits, it continuously acquires and generates a structured environmental dynamic parameter vector E(t), including but not limited to ambient temperature T(t), relative humidity H(t), and wind speed v(t), and records the duration τ(t) of stability or change for each parameter. Data is acquired at a fixed frequency (1Hz). This module has data filtering and outlier processing functions; after filtering, noise reduction, and other preprocessing, it generates a structured environmental dynamic parameter vector in real time.

[0298]

[0299] It provides accurate and real-time "Input I" to the parameter generation module via a communication interface. A simulation example is introduced: in an outdoor work site at -5℃, the module might upload vectors:

[0300]

[0301] This indicates that the current temperature is -5℃, humidity is 85%, and wind speed is 3m / s; this stable state has lasted for 200 seconds. To comprehensively evaluate the complex impact of the environment on de-icing, the module incorporates a comprehensive meteorological condition calculation index.

[0302]

[0303] In the formula, express The comprehensive meteorological conditions index at any given time indicates that the smaller the value, the more favorable the environment is for ice accumulation or the more difficult it is for ice removal. Ambient temperature (°C); Dew point temperature (°C) can be approximately calculated or measured by a sensor, reflecting the degree of air saturation; Wind speed (m / s); Relative humidity of air (%) The weighting coefficients for temperature difference, wind speed, and humidity are respectively obtained through regression analysis of historical data.

[0304] This index can be used for initial population generation or parameter correlation calculation in the IQGA algorithm, simplifying complex environments into a single influencing factor.

[0305] The parameter generation module is the core computation and decision-making unit of this system. This module incorporates a multi-objective collaborative optimization model and an improved quantum genetic algorithm solution engine. Its workflow is as follows: it receives input I from the environmental acquisition module and input II from the operating interface or external detection equipment regarding the icing status; using these inputs as conditions, and under the premise of satisfying the skin temperature safety constraints, it drives the improved quantum genetic algorithm to perform parallel collaborative optimization of output parameters such as infrared radiation current, voltage, relative area, intensity, irradiance, and radiation distance, realizing the complete algorithm flow from S1 to S7 as described in claim 3; after calculation, it outputs a set of Pareto non-dominated solutions, each solution corresponding to a set of infrared radiation parameters, for subsequent execution and verification.

[0306]

[0307] The parameter validity verification module is a verification platform that combines physical experiments with data comparison. This module receives the optimal parameter set provided by the parameter generation module and accurately reproduces the parameter scheme in a physical experimental setup using an infrared radiation actuator. Simultaneously, this module integrates data acquisition devices such as a thermal imager, high-speed camera, and power meter to collect real-time data on the skin temperature field distribution during the actual de-icing process. Images of ice accumulation evolution and actual power consumption of the system And conduct an effectiveness evaluation, calculating a quantitative evaluation index for de-icing effectiveness.

[0308]

[0309] In the formula, This indicates the de-icing performance index; the higher the value, the better the overall performance. These are the de-icing time and energy consumption of the reference scheme; The measured de-icing time and total energy consumption are for the experiment. This represents the highest temperature actually measured on the skin. The safe temperature threshold; This is the penalty range for exceeding temperature limits; The coefficients are weights, and their sum is 1.

[0310] The evaluation algorithm embedded in the module compares the collected measured data (such as actual de-icing time, peak temperature, and total energy consumption) with the predicted target values ​​of the optimization model. This is used to quantitatively score and compare the measured effects of different optimization schemes (such as multiple solutions on the Pareto front). The algorithm also judges the validity and reliability of the set of parameters based on the preset tolerance threshold, and the results are fed back to the historical data module.

[0311] For example, the historical data module serves as the system's database and knowledge base. This module uses a structured database to persistently store all relevant data, including: historical environmental data, icing status data, the optimization parameter set generated by the algorithm, a complete experimental report on parameter validity verification (including measured data and comparison results), and corresponding performance evaluation metrics. This module uses a time-series database to store all process data. For each stored optimization scheme P, the module asynchronously calculates its parameter scheme robustness (P).

[0312]

[0313] Robustness(P) is the robustness score for parameter P, where This refers to the parameter scheme under the nth disturbance (e.g., ambient temperature ±2℃, wind speed ±1m / s); This is an indicator function; it returns 1 if the condition is met, and 0 otherwise. For the first Each objective function allows a performance degradation percentage threshold. Through post-hoc robustness analysis, the most robust solution will be preferentially recommended under similar operating conditions.

[0314] This module also supports multi-dimensional querying, statistics and analysis of data, and can provide historical case data for the optimization algorithm in the parameter generation module for algorithm training, initial population guidance or experience learning, thereby realizing the continuous self-improvement and performance enhancement of the system.

[0315] It can be seen that the above modules work together to complete the entire process from environmental perception, intelligent optimization, experimental verification to knowledge accumulation.

[0316] The historical data module is further configured to perform the following functions:

[0317] Establish and maintain a structured case knowledge base to store and associate the following data entities: historical environment dynamic parameter vectors. Historical icing state vector The corresponding Pareto optimal de-icing parameter set and the complete verification results obtained through the parameter validity verification module, the verification results including measured performance indicators and validity judgment conclusions;

[0318] It provides a data interface and query engine to support multi-dimensional retrieval and statistical analysis of the case knowledge base based on working condition characteristics; when the parameter generation module starts a new optimization task, it can perform optimization based on the currently input working condition characteristics. Successful cases under similar historical working conditions are retrieved from the case knowledge base.

[0319] Based on retrieved historical success case data, initialization information is generated to guide the improved quantum genetic algorithm solution engine. This information includes at least: a candidate parameter set for constructing a high-quality initial population, and prior knowledge of the physical correlation between parameters for initializing the dynamic correlation quantum entanglement gate operation.

[0320] Continuously accumulate verification data and perform posterior analysis on the stored optimization parameter schemes based on a predefined robustness evaluation model to calculate their performance robustness score; when there is a systematic deviation between the model prediction value and the measured value under a specific working condition, the calibration process of the relevant physical parameters in the multi-objective collaborative optimization model is triggered.

[0321] As can be seen from the above embodiments, this invention firstly achieves a leap from isolated parameter control to system-wide collaborative optimization by constructing a multi-objective dynamic collaborative optimization model that integrates environmental, icing, and radiation parameters, enabling the de-icing scheme to adapt to complex and ever-changing external conditions. Secondly, based on an improved quantum genetic algorithm, it efficiently solves the Pareto front, providing users with a series of optimization schemes that achieve the best balance between de-icing time, energy consumption, and safety, supporting scientific and intelligent decision-making. Simultaneously, the algorithm significantly improves global search capability, convergence speed, and solution accuracy through a dynamically correlated quantum entanglement gate and an adaptive quantum rotation gate mechanism, effectively avoiding local optima. Furthermore, by embedding skin temperature safety constraints into the optimization core and relying on an independent physical verification platform to achieve closed-loop verification, the safety and reliability of de-icing operations are greatly enhanced. Finally, the system leverages the domain knowledge accumulated from historical data modules to achieve continuous algorithm guidance and self-evolution, forming an intelligent de-icing decision-making system with learning and optimization capabilities.

[0322] This invention provides a complete technical solution for achieving efficient, energy-saving, safe, and intelligent ground infrared de-icing operations for aircraft, and has broad prospects for engineering applications.

[0323] As can be seen from the above embodiments, Figure 4 This is a partial magnified schematic diagram of the infrared device and its principle structure for infrared de-icing provided by the present invention. Figure 5 An infrared schematic diagram of the aircraft ground infrared de-icing method based on multi-parameter collaborative optimization provided by the present invention; Figure 6 This is a schematic diagram illustrating the optimization of high-speed icing state in the aircraft ground infrared de-icing method based on multi-parameter collaborative optimization provided by the present invention.

[0324] The embodiments of this invention have achieved significant technological breakthroughs during research and application, demonstrating multi-dimensional and profound theoretical advantages compared to existing technologies. From a systems theory perspective, this invention, for the first time, constructs a hierarchical nested multi-objective collaborative optimization model that integrates dynamic environmental parameters and icing conditions. It takes six parameters—infrared radiation current, voltage, radiation area, intensity, and distance—as the overall optimization object, embedding the multi-parameter coupling relationship into the core of the model through radiation heat transfer equations and heat conduction equations. This achieves a paradigm shift from isolated parameter control to system collaborative optimization, enabling the de-icing scheme to adapt to complex and changing external conditions, and completing a fundamental transformation from experience-based operation to intelligent decision-making.

[0325] From the perspective of optimization theory, this invention addresses the multi-objective problem of inherent conflict among de-icing time, skin temperature variation integral, and total energy consumption. It employs non-dominated sorting and crowding distance calculation to solve for the complete Pareto front in a three-dimensional objective space, rather than a single optimal solution. This provides users with a set of optimization schemes that achieve the best balance between efficiency, safety, and economy. At the same time, skin temperature safety constraints are embedded as hard constraints in the optimization core to ensure that all Pareto solutions meet safety requirements. This achieves a quantitative trade-off of multi-objective conflicts and breaks through the theoretical limitations of traditional weighted summation that obscures the trade-offs between objectives.

[0326] From an algorithm design perspective, the improved quantum genetic algorithm proposed in this invention realizes a quantum evolution mechanism guided by physical knowledge. On the one hand, it encodes macroscopic physical laws into microscopic quantum operations through a dynamically associated quantum entanglement gate. Based on the physical model, it calculates the parameter correlation coefficient and constructs a quantum entanglement gate with an entanglement angle proportional to it, enabling the mutation direction of the associated parameters to co-evolve along an efficient path guided by physical laws. On the other hand, it uses an adaptive quantum rotation gate to dynamically adjust the rotation angle in both time and space dimensions. In the early stage, it explores the global environment with a large angle, and in the later stage, it develops the local environment with a small angle. The rotation direction is guided by elite individuals, which solves the long-standing exploration-development dilemma in evolutionary algorithms. Theoretically, it has been proven that the algorithm can converge to the global Pareto optimal front with probability 1.

[0327] From a cybernetics perspective, this invention constructs a closed-loop system of "optimization-verification-feedback." It accurately reproduces environmental conditions and loads optimization parameters through a semi-physical verification platform, collects measured data to calculate the de-icing effect index, and enables quantitative comparison of multiple solutions. The verification results are stored as structured cases to form a searchable knowledge base. Robust posterior analysis is used to screen high-quality solutions insensitive to environmental disturbances. Under new operating conditions, the system can retrieve similar historical cases, injecting effective solutions as high-quality seeds into the initial population. This achieves an evolutionary closed loop from "data" to "knowledge" to "intelligence," upgrading the system from a one-time optimizer to an intelligent agent with memory and learning capabilities, aligning with the cutting-edge direction of learning control in modern control theory.

[0328] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention and within the spirit and principles of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A ground-based infrared de-icing method for aircraft based on multi-parameter collaborative optimization, characterized in that, The method includes the following steps: S1. Construct an aircraft ground infrared de-icing system based on multi-parameter collaborative optimization. The aircraft ground infrared de-icing system based on multi-parameter collaborative optimization includes an environmental acquisition module, a parameter generation module, a parameter validity verification module, and a historical data module. S2. Based on the constructed aircraft ground infrared de-icing system based on multi-parameter collaborative optimization, the infrared de-icing parameters are collaboratively optimized, and a multi-objective optimization model is constructed. S3 utilizes an improved quantum genetic algorithm to solve a collaborative multi-objective optimization model and generate parameters. The entanglement state between parameters is determined by the correlation between environmental parameters, ice accumulation parameters, and parameters in the current state. The solution process combines the balance between optimization objectives to finally obtain the optimal set of de-icing parameters. S4. After generating the optimal set of de-icing parameters, repeat experiments are conducted in the parameter validity verification module based on the real environment to determine whether the de-icing effect meets the de-icing requirements. S5, in the historical data module, performs data processing and analysis on the obtained experimental data.

2. The aircraft ground infrared de-icing method based on multi-parameter collaborative optimization according to claim 1, characterized in that, In step S2, the infrared de-icing parameters are optimized collaboratively, and a multi-objective optimization model is constructed, including: taking ambient temperature, humidity, wind speed and the duration corresponding to each parameter as input layer I, taking ice type, ice thickness and ice evolution state as input layer II, and taking infrared radiation current, infrared radiation voltage, infrared radiation relative area, infrared radiation intensity, infrared irradiance and infrared source distance from ice point as output layer. The optimization objectives are the shortest de-icing time, the most gradual change in skin temperature, and the lowest total energy consumption. These objectives are considered the core of the collaborative optimization, and safety constraints are added, with the constraint that the temperature rise of the aircraft surface skin does not exceed the safety threshold that would damage the aircraft skin material. A hierarchical nested multi-objective optimization model is then constructed.

3. The aircraft ground infrared de-icing method based on multi-parameter collaborative optimization according to claim 2, characterized in that, In input layer I, the environment dynamic parameter vector is: ； In the formula, For ambient temperature, This refers to the relative humidity of the air. Wind speed, For the duration of each parameter, and ; In input layer II, the icing state vector Perform quantitative modeling: ； In the formula, The encoding for the ice accumulation type, where, 1 represents raw ice, 2 represents clear ice, and 3 represents mixed ice; This refers to the thickness of the ice buildup, in mm. Let be the ice accumulation evolution state coefficient, where 0 represents a pure solid state, and 1 represents ice that has completely melted into a liquid water film; The ice evolution state coefficient Used to calculate the equivalent thermal resistance of ice accumulation layer for: ； In the formula, In order to be in The equivalent thermal resistance of the ice layer at any given time, in units of ; and These are the thermal conductivity coefficients of ice and water, respectively. for The equivalent thermal conductivity of the ice layer at any given time This refers to the proportion or weighting coefficient of the liquid phase in the ice layer. In the output layer, the infrared radiation parameter matrix , It is a 6-dimensional real parameter matrix space; defined as: ； In the formula, It is infrared radiation current. Infrared radiation voltage, This refers to the relative radiation area. Infrared radiation intensity, Infrared irradiance, The distance from the infrared source to the ice accumulation point; Temperature field of aircraft skin The change is controlled by a one-dimensional unsteady-state heat conduction equation, that is, along the skin thickness direction. The expression is: ； In the formula, These are the density, specific heat capacity, and thermal conductivity of the skin material, respectively; the key boundary conditions are located on the outer surface of the skin. , The direction of skin thickness; coupled radiative heating, convective heat dissipation in the air, and heat conduction through the ice layer: ； ； ； In the formula, For surface convection heat dissipation, The convective heat transfer coefficient is related to the wind speed. related; This refers to the heat flow conducted through the ice layer; The temperature distribution function inside the skin. The net radiative heat flow of the surface. The surface convective heat transfer coefficient is... for The temperature of the outer surface of the skin is constantly monitored. Ambient air temperature, The temperature of the outer surface of the ice layer. The equivalent thermal resistance of the ice layer; The mathematical expression for the safety constraint is the time during the entire de-icing process. The temperature of the inner skin's outer surface and any internal location must not exceed the maximum safe temperature allowed by its material. : ； In the formula, Skin thickness; In order to defrost time Maximum temperature at any location on the outer surface and inside of the inner skin; During the optimization process, the partial differential equation system is solved rapidly using the finite difference method, integral approximation method, or response surface model based on historical data to predict... The evolution path involves intrinsically verifying security constraints while optimizing the objective function; objective function De-icing completion time; ； In the formula, The time from the start of radiation to the fulfillment of the de-icing completion criterion, which is based on the state of icing. Or visual monitoring signals; objective function Skin temperature stability; ； In the formula, for The temperature of the outer surface of the skin at all times; objective function Total energy consumption; ； In the formula, For real-time power consumption, for Infrared system current at all times for Infrared system voltage at all times; Constructing the Pareto optimal multi-objective vector from the objective function of the multi-objective optimization model Specifically: ； ； In the formula, for Infrared system current at all times for Infrared system voltage at all times.

4. The aircraft ground infrared de-icing method based on multi-parameter collaborative optimization according to claim 1, characterized in that, In step S3, the improved quantum genetic algorithm is used to solve the collaborative multi-objective optimization model and generate parameters, including: S301, Quantum Encoding and Population Initialization; S302, Quantum Observation and Parameter Decoding; S303, Fitness Assessment and Pareto Ranking, will decode each set of parameters. Substitute these parameters into the multi-objective optimization model and calculate the three objective function values ​​for each parameter. At the same time, it is also necessary to pay attention to the constraints, and check whether the constraint condition of no damage to the surface skin of the aircraft is met; then, based on the Pareto dominance relation, perform non-dominated sorting on all feasible solutions in the current population, and calculate the crowding distance of each individual, so as to comprehensively evaluate the fitness of the individual. S304, Elite Selection and Population Renewal, employs an elite retention strategy, directly preserving the top few individuals with the highest Pareto level in the current population to the next generation; the remaining individuals are selected from the parent generation through a tournament selection method, in preparation for subsequent quantum evolution operations; S305, Dynamically Associated Quantum Entanglement Gate Cooperative Mutation, based on current environmental input. Ice accumulation status input And the distribution of parameters in the population, dynamically calculating the physical correlation coefficient between different infrared radiation parameters. And thereby construct or adjust quantum entanglement gates. Furthermore, this entanglement gate is used to perform co-mutation operations on the individuals selected in step S304 elite selection and population update, so that the parameters with strong correlations can be further co-optimized. S306, Improved Adaptive Quantum Rotation Gate for Guiding Optimal Parameter Selection, including: designing an adaptive quantum rotation gate, the direction of the rotation angle of the adaptive quantum rotation gate is jointly determined by the probability amplitude of the current qubit and the guidance of elite individuals in the current Pareto front, and the size of the rotation angle of the adaptive quantum rotation gate is adaptively adjusted with the number of generations of evolution to achieve a balance between fast global search and fine-grained local search in the later stages, guiding the population to converge toward a more optimized Pareto front; S307, Iteration terminates and Pareto optimal solution set is output; repeat steps S302-S306 above for iterative evolution until the preset maximum number of generations is reached or the convergence condition is met; finally, output all solutions on the Pareto non-dominated layer in the last generation population to form the optimal de-icing parameter set for use by the parameter validity verification module.

5. The aircraft ground infrared de-icing method based on multi-parameter collaborative optimization according to claim 4, characterized in that, In step S301, quantum encoding and population initialization include: Based on the input layer I environment dynamic parameter vector of the aforementioned multi-objective optimization model and the input layer II icing state vector The determined boundary conditions and security constraints affect the output layer infrared radiation parameter matrix. The six parameter solutions are encoded into qubits; an initial size of is generated. quantum population Each individual It is represented by a set of qubit chromosomes and is used to characterize a complete set of infrared de-icing parameters; Specifically, this includes: using qubits Encode the decision variables, one Defined as ,in, ; and These represent the probabilities of collapsing to classical states 0 and 1 at the time of observation, respectively. For those with A problem with one decision variable, a chromosome composed of... indivual String representation; each decision variable Use a length of 1 If the string is encoded, then the total length of an individual is... ;No. The Q-bit strings corresponding to the variables are: ； This formula is the complete representation of a quantum chromosome. Indicates the first in the population Individual; This represents the total number of decision variables. This corresponds to 6 infrared radiation parameters; The first decision variable is represented by a qubit. This represents the qubit representing the second decision variable; and so on up to the qubit... One variable; For population initialization, the population size is set to N. For each individual in the population, each... Random initialization and ,make , indicating that the initial population collapses to 0 or 1 with equal probability, ensuring that the initial population is uniformly distributed in the search space; this initialization process is expressed as: ； in, ; In the formula, This is the quantum population of generation 0; For the first The quantum chromosome of an individual; These represent the quantum chromosomes of the 1st, 2nd, and so on up to the Nth individual, where N represents the population size; Representing the The first individual The probability amplitude of each qubit; Indicates for all individuals and all qubits Both are valid.

6. The aircraft ground infrared de-icing method based on multi-parameter collaborative optimization according to claim 4, characterized in that, In step S302, quantum observation and parameter decoding include: quantum population Each individual in the process undergoes quantum observation, causing its qubit state to collapse into a classical binary string; the binary string is then decoded into a corresponding set of infrared radiation parameter values. ;Decoding It refers to the first Each quantum individual corresponds to a specific set of infrared radiation parameter values; therefore... It is a candidate solution in the optimization search process, representing a complete set of infrared de-icing parameter schemes evaluated by the algorithm in the current generation; specifically including: at the quantum observation level: in each generation For quantum populations Each individual in Observations were conducted based on their probability amplitude. Generate classic binary strings Then, decoding is performed to convert the binary string into a binary string. Divide the data into segments according to variables and map them back to the corresponding physical values ​​using linear transformations: ； In the formula, For the first The actual values ​​of the physical parameters to be optimized. This is the minimum allowable value for this physical parameter. This is the maximum allowed value for this physical parameter. For the corresponding parameter of Bit-bit binary encoded string The number of binary bits used to encode this parameter. The probability of obtaining the classical state 0 when measuring this qubit; The probability of obtaining classical state 1 when measuring this qubit; For variables The decimal number corresponding to the l-bit binary string; thus, a set of candidate solutions is obtained. Then perform a fitness assessment and decode the results. Substitute the physical model established in step S2 into the equation to calculate the corresponding three optimization objective values. And verify the security constraints, specifically by substituting the following into the formula: Radiation source temperature calculation: ； In the formula, for Infrared system current at all times for Infrared system voltage at all times For electrothermal conversion efficiency, The Stefan-Boltzmann constant is... The effective radiation surface area of ​​the radiation source; Net radiative heat flux density: ； In the formula, This represents the net radiative heat flux density received by the skin surface. For effective emission rate, The radiation source temperature was calculated in the previous step. As a perspective factor, for The fourth power of the absolute temperature of the skin surface at any given time. This refers to the relative radiation area. The square of the distance from the radiation source to the detection point; Solving for the temperature field of the skin: ； In the formula, The density of the skin material, The specific heat capacity of the skin material. For skin in depth ,time Temperature function at that location, The thermal conductivity of the skin material is... is the symbol for the second derivative of the space term.

7. The aircraft ground infrared de-icing method based on multi-parameter collaborative optimization according to claim 4, characterized in that, In step S303, fitness evaluation and Pareto ranking include: dividing each set of decoded parameters... Substitute these parameters into the multi-objective optimization model and calculate the three objective function values ​​for each parameter. The system checks whether the constraint of no damage to the aircraft surface skin temperature is met. Then, based on the Pareto dominance relation, it performs a non-dominated sort of all feasible solutions in the current population and calculates the crowding distance of each individual to comprehensively evaluate the fitness of the individual. The fitness assessment and Pareto ranking specifically include: Step 1, constraint handling: Invoke the fast thermal response model, simplify the finite difference solver or pre-trained surrogate model, and calculate the parameter set. Peak temperature of lower skin Define constraint violation degree ,like ,but This is an infeasible solution; in the formula, For aircraft surface The maximum temperature at that location, To ensure safe temperatures for aircraft skin; Step 2: Calculate the objective function, simultaneously calculating the values ​​of the three objective functions. ; Step 3, Fitness Assignment Based on Pareto Ranking: A constrained non-dominated ranking is used. First, feasibility is compared. Among feasible solutions, the standard Pareto dominance relation is used for ranking. All individuals are assigned to different non-dominated fronts. ,in, It is the optimal frontier; The fitness value of an individual is defined as: ； In the formula, For individuals The Pareto nondominated layers are: 1 for the frontier, 2 for the secondary frontier, and so on. To prevent small constants from being divided by zero, The weighting coefficient for the distance to congestion level. for The distance of congestion in its Pareto layer.

8. The aircraft ground infrared de-icing method based on multi-parameter collaborative optimization according to claim 4, characterized in that, In step S304, elite selection and population update include: employing Choose a strategy to preserve the current generation population. and offspring populations generated through quantum operations All individuals are merged and sorted according to the fitness scores mentioned above. The top N best individuals are selected to enter the next generation. ; In step S305, the cooperative mutation of the dynamically correlated quantum entanglement gate includes: Step a, construct the correlation matrix, based on the physical model in step S2, analyze the coupling relationship between parameters; define parameters. and physical correlation coefficient The simplified model is as follows: ； In the formula, To calculate the correlation degree of the parameters, For the first An optimization objective function, For the first One parameter to be optimized. For the first One parameter to be optimized. The expected ratio of the changes in the two parameters based on physical laws. To adjust the parameters; The physical model is divided into three main types: the infrared radiation heat transfer model, the heat conduction and temperature field model, and the physical law constraint model. Together, they constitute the correlation of computational parameters. The mathematical foundation of quantum entanglement gates ensures that the direction of mutation conforms to real physical laws; Step b, Dynamic Entanglement Gate Applications of correlation Exceeding the threshold For parameter pairs, when performing mutation operations on their corresponding qubits, instead of independent bit flipping, a controlled quantum entanglement gate is applied. The strength of this gate is related to... Proportional; acts on associated bit pairs The simplified entanglement mutation is represented as: ； In the formula, For dynamically controlled quantum entanglement gate operators, The imaginary unit, For entanglement angle, .

9. The aircraft ground infrared de-icing method based on multi-parameter collaborative optimization according to claim 4, characterized in that, Step S306, improving the selection of optimal parameters for adaptive quantum rotating door guidance includes: First, the rotation direction needs to be determined: Let the current qubit be... The ideal state of the corresponding bit in the currently observed optimal solution or an elite solution is: Rotation direction The sign function of the difference in probability magnitude between the two factors determines that... Towards near; The direction of the rotation angle of the adaptive quantum rotation gate is guided by the current state of the qubit and the state of the elite individual, while the magnitude of the rotation angle is determined by the state of the qubit itself. A function that adapts to the number of generations of evolution and dynamically adjusts with the number of generations and individual performance: ； In the formula, For the first In the first generation of evolution, the first The rotation angle of each qubit; and These are the minimum and maximum settings for the rotation angle, respectively. For the current generation, For the maximum number of generations, This is the global attenuation coefficient. For the first The variance or standard deviation of the parameter corresponding to each qubit in the current population. As the normalization factor, This is the local adjustment coefficient; Step S307, iteration termination and Pareto optimal solution set output specifically includes: repeating steps S302-S306 above, with the termination condition usually being reaching the preset maximum number of generations. Or the improvement over multiple consecutive Pareto fronts is less than a certain threshold. ; When the improved quantum genetic algorithm terminates, it outputs all non-dominated feasible solutions in the final population, forming the Pareto optimal solution set. Each solution in this solution set represents a set of infrared de-icing parameters that achieve a specific optimal balance among the three objectives of de-icing time, skin thermal shock, and total energy consumption.

10. An aircraft ground infrared de-icing system based on multi-parameter collaborative optimization, characterized in that, The system implements the aircraft ground infrared de-icing method based on multi-parameter collaborative optimization as described in any one of claims 1-9, and the system includes: The environmental acquisition module is used to acquire and preprocess the dynamic environmental parameters of the de-icing site in real time to form an input vector. The parameter generation module, with its built-in multi-objective optimization model and improved quantum genetic algorithm solution engine, is responsible for receiving input and generating the optimal set of de-icing parameters. The parameter validity verification module is used to perform physical experiments to verify the generated parameter set in simulated and real environments, collect measured data and compare it with the optimization target; The historical data module stores all relevant historical data, verification results, and performance metrics.

Images