CN113093794A - Multimode accurate partitioning method for wide-area flight - Google Patents

Abstract

The invention relates to a wide-area flight-oriented multi-modal accurate division method. The key factors affecting the multi-modal division of an aircraft's wide-area climbing process are determined through demand analysis, which is used as a modal indicator variable; stability is carried out according to the demand analysis and prior knowledge. Preliminary determination of the modes to ensure that the modes are arranged according to the actual time sequence; modal division and identification of the offline data of the aircraft, combined with the preliminary division results of stable modes and similarity analysis to achieve accurate division of stable modes and transition modes; based on The multi-modal accurate division result is used to design a modal switching strategy, and finally a flight-oriented multi-modal switching system is given; the present invention realizes the accurate division of multi-modals in the wide-area climbing process of the aircraft by means of experience/data dual drive, and according to The multi-mode process switching strategy is designed for the attribution of the stable mode category and the start and end time of the transition mode, which is conducive to the smooth switching of multi-mode in wide-area flight, improves flight safety, and is suitable for engineering applications.

Description

A Multimodal Accurate Partitioning Method for Wide-area Flight

technical field

The invention belongs to the technical field of aerospace, in particular to a multi-modal accurate division method for wide-area flight.

Background technique

With the rapid development of aerospace technology, the envelope of the aircraft is getting wider and wider, making it possible for the aircraft to take off horizontally from the ground and fly in a wide area. During the wide-area climb, the speed of the aircraft continues to increase to hypersonic speeds, which will play an important role in long-distance rapid transportation, space travel, and global rapid strikes.

During wide-area flight, the aircraft may face different power modes, aerodynamic configurations and flight tasks, resulting in the existence of multiple flight modes. Multi-modality is a common feature of the aircraft's wide-area climb process. It is necessary to establish different models and design different controllers for different modes. Therefore, the aircraft's wide-area climb process is a switching process of the multi-modal flight controller. Precise division and identification are critical to the design of aircraft switching control systems. The wide-area flight process includes multiple stable flight modes, and there are different transition modes between different stable modes. The existing work mostly divides the wide-area flight modes based on experience and ignores the transition modes, so multi-mode cannot be realized. Accurate division and identification of , and smooth transition between adjacent stable modes, the safety is poor, which is not conducive to engineering implementation. Therefore, it is of great significance and urgent need to study the accurate multi-mode division method for wide-area flight for the research of flight multi-mode switching control.

SUMMARY OF THE INVENTION

technical problem to be solved

In order to overcome the shortcomings of poor practicability of the existing flight multi-modal division methods, the present invention provides a multi-modal accurate division method for wide-area flight.

Technical solutions

A multi-modal accurate division method for wide-area flight is characterized in that the steps are as follows:

Step 1: Based on the flight mission and command knowledge base, analyze the requirements of the aircraft in the wide-area climb process, determine the key factors affecting the multi-modal division of flight, and use it as a modal indicator variable;

Step 2: Fully consider the empirical risk of modal identification, and preliminarily determine the stable modes based on prior knowledge, and give the number of stable modes N ₀ for wide-area climb of the aircraft and the running time S _n of each stable mode, n=1,2,...,N ₀ , further analyze the logical relationship between multiple modes to ensure that the modes are arranged according to the actual time sequence;

Step 3: Based on the improved K-means clustering algorithm, the modal division and identification of the offline data of the aircraft are carried out to realize the accurate division of the stable modal and the transition modal;

Step 4: Establish a multi-modal model set based on the divided multiple stable modes, design a modal switching strategy based on the transition mode, and combine the multi-modal model set and the modal switching strategy to construct a wide-area climb-oriented aircraft switching system;

The aircraft system for wide-area climb is written as a multi-input multi-output switching system as shown below

Among them, X _i ,i=1,2,...,n are state variables, f _i,σ(t) and gi _,σ(t) are nonlinear functions, u _σ(t) is the control input, σ(t) ):[0,∞)→M={1,2,...,m} is the switching signal;

For the switching signal, the number m is equal to the number of divided stable modes N ₀ , and the switching strategy is specifically described as: the switching signal is switched according to the multi-modal process sequence, and in the stable mode, the corresponding mode is called Model and design the controller; in the transition mode, because the transition mode time is very short, if the transition mode start time is within the last window of the pre-order stable mode, the model corresponding to the pre-order stable mode is called, If the starting time of the transition mode is within the first window of the transition mode, the model corresponding to the post-sequence stable mode is called, and the soft switching control is designed at the same time.

The further technical scheme of the present invention is: the specific process of step 3 is as follows:

(a) Window cutting

Select a cutting window of length H to cut the two-dimensional flight data matrix X along the sampling direction, there are

N=K×H+d (1)

Among them, N is the number of sampling data, K is the number of cutting windows, d is the sampling data that is not cut and 0≤d≤H;

求取这些二维矩阵的均值向量

作为模态划分的基本单元；Denote the two-dimensional matrix of window data obtained by cutting as

find the mean vector of these 2D matrices

As the basic unit of modal division;

(b) Clustering processing

和两个子类聚类中心的最小距离阈值θ，算法的输出是每个窗口属于不同子类的隶属关系

和子类个数C_st；The window mean vector is clustered by the improved K-means clustering algorithm, and the input of the algorithm is the set of window mean vectors

and the minimum distance threshold θ between the cluster centers of the two subclasses, the output of the algorithm is the affiliation of each window belonging to a different subclass

and the number of subclasses C _st ;

Therefore, through this algorithm, the entire multimodal process can be clustered into C _st subclasses;

(c) Time division

According to the time sequence of the window mean vector arrangement, the window units that are continuous in time and belong to the same subclass are divided into the same time period;

其中M₀为划分的时段个数。则每个子时段属于不同子类的隶属关系为Denote the divided sub-period as

where M ₀ is the number of divided time periods. Then the affiliation of each sub-period belonging to different sub-categories is:

Among them, C is the clustered subclass;

用于飞行器稳定模态的确定；Denote the length of the subperiod as

It is used to determine the stable mode of the aircraft;

(d) Stable mode determination

Based on the running time of each mode in step 2, the shortest running time of the stable mode is determined as

S _min =min{S ₁ ,S ₂ ,...,S _n } (3)

和稳定模态最短运行时间S_min来确定稳定模态By comparing the length of sub-periods

and the shortest stable mode running time S _min to determine the stable mode

Among them, the stable modes belonging to the same subclass are defined as the same stable mode;

The stable mode time interval in step 2 is further introduced to analyze the identified stable mode, and if the identified stable mode is within the corresponding time interval, it is classified as the specific stable mode;

(e) Precise division of modes

In-depth analysis of the windows adjacent to the stable mode and the transition mode determines that the start time of the transition mode occurs in the second half of the last window of the previous stable mode or the first half of the first window of the transition mode. The end time of the state occurs in the second half of the last window of the transition mode or the first half of the first window of the post-sequential stable mode;

For the confirmation of the starting time of the transition mode, it is assumed that the mode starts to transition from the _k1th window, and it is necessary to re-analyze and judge with a shorter sliding window L starting from the last window of the previous stable mode, that is, the _k1-1th window. , the sliding step is h;

The average value of the small sliding window obtained by the analysis of the two windows is further calculated, and the average value of the small sliding window is recorded as

The similarity between the sliding window and the preorder stable mode is defined as

为前序稳定模态的变量均值，J为多模态过程变量个数；in,

is the variable mean value of the pre-order stable mode, and J is the number of multi-modal process variables;

The similarity threshold α is introduced as a boundary parameter, and the relationship between each similarity and threshold is analyzed. The identification rule for the starting moment of the transition mode is:

The above rule is expressed as when there are r continuous small windows starting from the t ₁ small window and all satisfy γ _t <α, then the multimodal process is considered to enter the transition mode from the t ₁ small window;

Therefore, taking the starting position of the t _1th small window as the start of the transition mode, the starting moment of this transition mode is (k ₁ -2)×H+(t ₁ -1)×h+1;

For the confirmation of the end time of the transition mode, it is assumed that the mode starts to transition from the _k2th window. It is necessary to re-analyze and judge with a shorter sliding window L starting from the last window of the transition mode, that is, the _k2-1th window. The step size is h;

The average value of the small sliding window obtained by the analysis of the two windows is further calculated, and the average value of the small sliding window is recorded as

The similarity between the sliding window and the post-order stable mode is defined as

为后序稳定模态的变量均值；in,

is the variable mean of the post-order stable mode;

Analyzing the relationship between each similarity and threshold, the identification rule for the end of the transition mode is:

The above rule is expressed as: when there are r continuous small windows starting from the t2th small window and all satisfy γ _{t ≥α} , then the multimodal process is considered to enter the post-sequence stable mode from the _t2th small window;

Therefore, taking the starting position of the t ₂ th small window as the end of the transition mode, the end time of this transition mode is (k ₂ -2)×H+(t ₂ -1)×h+1.

beneficial effect

The invention proposes a multi-modal accurate division method for wide-area flight. This method analyzes the requirements of the aircraft's wide-area climbing process, determines the stable mode based on the prior knowledge, and further realizes the accurate division of the stable mode and transition mode based on the improved K-means clustering algorithm. Finally, combined with the multi-modal model The set and mode switching strategy is used to construct a multi-mode switching system for wide-area flight, which is convenient for engineering implementation. The beneficial effects are as follows:

(1) A clustering algorithm based on flight data is introduced on the basis of empirically determining the stable modal of the aircraft, which reduces the empirical risk of modal division;

(2) The stable mode time interval divided by experience is used to compare the clustered stable modes, and the classification of stable modes is realized;

(3) The wide-area climbing aircraft is transformed into a switching system, and the switching strategy of the switching signal is given through the precise division of multiple modes, which ensures smooth switching between multiple modes.

Description of drawings

FIG. 1 is a flow chart of the implementation of the present invention.

Detailed ways

The present invention will now be further described in conjunction with the embodiments and accompanying drawings:

Referring to Fig. 1, the specific steps of the multi-modal accurate division method for wide-area flight of the present invention are as follows:

Step 1: Consider a type of rocket based combined cycle (RBCC) aerospace vehicle. The main flight task of the aircraft is to take off horizontally and climb to a height of more than 30Km, while the speed reaches about Mach 20;

In the process of climbing, the aerospace vehicle passes through the dense atmosphere and the extremely wide airspace near the space. The environment in different flight intervals is quite different. The power system has extremely high requirements for adapting to wide-area work, low comprehensive fuel consumption, and high thrust-to-weight ratio. A single power It cannot be satisfied, so it is necessary to adopt a combined power method to relay climb to ensure that the aircraft can achieve the best thrust performance in each flight interval; different power systems correspond to different working modes and cause different flight modes. Therefore, aerospace aircraft The wide-area climb process is a multimodal process;

Through the above demand analysis, it is determined that the combined power working mode of the aerospace vehicle is a key influencing factor of multi-modal division, which can be used as a modal indicator variable;

Step 2: The RBCC aerospace vehicle combines the rocket engine and the ramjet engine in the same flow channel, and uses the rocket jet and the ramjet flow channel to form a new thermodynamic cycle; There are 4 stable modes in total, the injection mode, the sub-combustion ram mode, the scram mode and the rocket mode, and the running time of each stable mode is Sn , _n =1,2,3,4. These four stable modes work in sequence, among which the ejection mode is at Mach 0-2.5, the sub-ramming mode is at Mach 2.5-6, the scramjet mode is at Mach 6-10, and the ramming mode is at Mach 10-20. For the rocket mode, the velocity-time curve can be depicted according to the flight sample data, and then the time interval of each stable mode can be determined;

Step 3: Based on the improved K-means clustering algorithm, the modal division and identification of the offline data of the RBCC aerospace vehicle are carried out to realize the accurate division of the stable mode and the transition mode. The specific process is as follows

(a) Window cutting

Select a cutting window of length H to cut the two-dimensional flight data matrix X along the sampling direction, there are

N=K×H+d (1)

Among them, N is the number of sampling data, K is the number of cutting windows, d is the sampling data that is not cut and 0≤d≤H;

k＝1,2,…,K，求取这些二维矩阵的均值向量x_k作为模态划分的基本单元；Denote the two-dimensional matrix of window data obtained by cutting as

k=1,2,...,K, find the mean vector x _k of these two-dimensional matrices as the basic unit of modal division;

(b) Clustering processing

和两个子类聚类中心的最小距离阈值θ，算法的输出是每个窗口属于不同子类的隶属关系

和子类个数C_st；The window mean vector is clustered by the improved K-means clustering algorithm, and the input of the algorithm is the set of window mean vectors

and the minimum distance threshold θ between the cluster centers of the two subclasses, the output of the algorithm is the affiliation of each window belonging to a different subclass

and the number of subclasses C _st ;

Therefore, through this algorithm, the entire multimodal process can be clustered into C _st subclasses;

(c) Time division

According to the time sequence of the window mean vector arrangement, the window units that are continuous in time and belong to the same subclass are divided into the same time period;

其中M₀为划分的时段个数；则每个子时段属于不同子类的隶属关系为Denote the divided sub-period as

where M ₀ is the number of divided time periods; then the affiliation of each sub-period belonging to different sub-classes is:

Among them, C is the clustered subclass;

用于飞行器稳定模态的确定；Denote the length of the subperiod as

It is used to determine the stable mode of the aircraft;

(d) Stable mode determination

Based on the running time of each mode in step 2, the shortest running time of the stable mode is determined as

S _min =min {S ₁ , S ₂ , S ₃ , S ₄ } (3)

和稳定模态最短运行时间S_min来确定稳定模态By comparing the length of sub-periods

and the shortest stable mode running time S _min to determine the stable mode

Among them, the stable modes belonging to the same subclass are defined as the same stable mode;

The stable mode time interval in step 2 is further introduced to analyze the identified stable mode, and if the identified stable mode is within the corresponding time interval, it is classified as the specific stable mode;

(e) Precise division of modes

In-depth analysis of the windows adjacent to the stable mode and the transition mode determines that the start time of the transition mode occurs in the second half of the last window of the previous stable mode or the first half of the first window of the transition mode. The end time of the state occurs in the second half of the last window of the transition mode or the first half of the first window of the post-sequential stable mode;

For the confirmation of the starting time of the transition mode, it is assumed that the mode starts to transition from the _k1th window, and it is necessary to re-analyze and judge with a shorter sliding window L starting from the last window of the previous stable mode, that is, the _k1-1th window. , the sliding step is h;

The average value of the small sliding window obtained by the analysis of the two windows is further calculated, and the average value of the small sliding window is recorded as

The similarity between the sliding window and the preorder stable mode is defined as

为前序稳定模态的变量均值，J为多模态过程变量个数；in,

is the variable mean value of the pre-order stable mode, and J is the number of multi-modal process variables;

The similarity threshold α is introduced as a boundary parameter, and the relationship between each similarity and threshold is analyzed. The identification rule for the starting moment of the transition mode is:

The above rule is expressed as when there are r continuous small windows starting from the t ₁ small window and all satisfy γ _t <α, then the multimodal process is considered to enter the transition mode from the t ₁ small window;

Therefore, taking the starting position of the t _1th small window as the start of the transition mode, the starting moment of this transition mode is (k ₁ -2)×H+(t ₁ -1)×h+1;

For the confirmation of the end time of the transition mode, it is assumed that the mode starts to transition from the _k2th window. It is necessary to re-analyze and judge with a shorter sliding window L starting from the last window of the transition mode, that is, the _k2-1th window. The step size is h;

The average value of the small sliding window obtained by the analysis of the two windows is further calculated, and the average value of the small sliding window is recorded as

The similarity between the sliding window and the post-order stable mode is defined as

为后序稳定模态的变量均值；in,

is the variable mean of the post-order stable mode;

Analyzing the relationship between each similarity and threshold, the identification rule for the end of the transition mode is:

The above rule is expressed as: when there are r continuous small windows starting from the t2th small window and all satisfy γ _{t ≥α} , then the multimodal process is considered to enter the post-sequence stable mode from the _t2th small window;

Therefore, taking the starting position of the t _2th small window as the end of the transition mode, the end time of this transition mode is (k ₂ -2)×H+(t ₂ -1)×h+1;

Step 4: Establish a multi-modal model set based on the divided 4 stable modes, design a mode switching strategy based on the transition mode, and combine the multi-modal model set and the mode switching strategy to construct an RBCC aerospace vehicle switching system;

Write the RBCC aerospace vehicle system as a multiple-input multiple-output switching system as shown below

Among them, the three-channel attitude angle X ₁ =[θ ψ φ] ^Τ and attitude angular velocity X ₂ =[ω _x ω _y ω _z ] ^Τ is the state variable, θ, ψ, φ, ω _x , ω _y and ω _z are respectively Pitch, yaw, roll, roll velocity, yaw velocity, and pitch velocity; u _σ(t) = [δ _x,σ(t) δ _y,σ(t) δ _z,σ(t) ] ^Τ is the control input, δ _i,σ(t) , i=x, y, z are the roll rudder deflection, yaw rudder deflection and pitch rudder deflection, respectively; σ(t)∈{1,2,3,4} In order to switch the signal, it corresponds to the ejection mode, the sub-combustion ram mode, the scramjet mode and the rocket mode in turn;

The nonlinear function looks like this:

f _1,σ(t) = 0

为气动力系数，Δ项包括了参数、模型不确定性以及线性化误差；Among them, J _i , i=x, y, z are the moments of inertia in the x, y and z directions respectively; q is the dynamic pressure, S=334.73m ² is the reference area; L _b = 18.288m is the lateral direction, L _c = 24.384m is the longitudinal reference length; α is the angle of attack, and β is the side slip angle;

is the aerodynamic coefficient, and the Δ term includes parameters, model uncertainty and linearization error;

The switching signal corresponds to the divided 4 stable modes. The specific description of the switching strategy is as follows: the switching signal is switched according to the sequence of the multi-modal process. In the stable mode, the corresponding mode model is called and the controller is designed; in the transition mode In the state, because the transition mode time is very short, if the transition mode start time is within the last window of the pre-order stable mode, the model corresponding to the pre-order stable mode is called. In the first window of the mode, the model corresponding to the post-sequence stable mode is called, and the soft switching control is designed at the same time.

The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited to this. Any person skilled in the art can easily think of various equivalents within the technical scope disclosed by the present invention. Modifications or substitutions should be included within the protection scope of the present invention.

Claims (2)

1. a multi-modal accurate division method for wide-area flight is characterized in that the steps are as follows: Step 1: Based on the flight mission and command knowledge base, analyze the requirements of the aircraft in the wide-area climb process, determine the key factors affecting the multi-modal division of flight, and use it as a modal indicator variable; Step 2: Fully consider the empirical risk of modal identification, and preliminarily determine the stable modes based on prior knowledge, and give the number of stable modes N ₀ for wide-area climb of the aircraft and the running time S _n of each stable mode, n=1,2,...,N ₀ , further analyze the logical relationship between multiple modes to ensure that the modes are arranged according to the actual time sequence; Step 3: Based on the improved K-means clustering algorithm, the modal division and identification of the offline data of the aircraft are carried out to realize the accurate division of the stable modal and the transition modal; Step 4: Establish a multi-modal model set based on the divided multiple stable modes, design a modal switching strategy based on the transition mode, and combine the multi-modal model set and the modal switching strategy to construct a wide-area climb-oriented aircraft switching system; The aircraft system for wide-area climb is written as a multi-input multi-output switching system as shown below

Among them, X _i ,i=1,2,...,n are state variables, f _i,σ(t) and gi _,σ(t) are nonlinear functions, u _σ(t) is the control input, σ(t) ):[0,∞)→M={1,2,...,m} is the switching signal; For the switching signal, the number m is equal to the number of divided stable modes N ₀ , and the switching strategy is specifically described as: the switching signal is switched according to the multi-modal process sequence, and in the stable mode, the corresponding mode is called Model and design the controller; in the transition mode, because the transition mode time is very short, if the transition mode start time is within the last window of the pre-order stable mode, the model corresponding to the pre-order stable mode is called, If the starting time of the transition mode is within the first window of the transition mode, the model corresponding to the post-sequence stable mode is called, and the soft switching control is designed at the same time. 2. a kind of multimodal accurate division method for wide-area flight according to claim 1 is characterized in that the concrete process of step 3 is as follows: (a) Window cutting Select a cutting window of length H to cut the two-dimensional flight data matrix X along the sampling direction, there are N=K×H+d (1) Among them, N is the number of sampling data, K is the number of cutting windows, d is the sampling data that is not cut and 0≤d≤H; Denote the two-dimensional matrix of window data obtained by cutting as

find the mean vector of these 2D matrices

As the basic unit of modal division; (b) Clustering processing The window mean vector is clustered by the improved K-means clustering algorithm, and the input of the algorithm is the set of window mean vectors

and the minimum distance threshold θ between the cluster centers of the two subclasses, the output of the algorithm is the membership u(k) that each window belongs to a different subclass:

and the number of subclasses C _st ; Therefore, through this algorithm, the entire multimodal process can be clustered into C _st subclasses; (c) Time division According to the time sequence of the window mean vector arrangement, the window units that are continuous in time and belong to the same subclass are divided into the same time period; Denote the divided sub-period as

where M ₀ is the number of divided time periods. Then the affiliation of each sub-period belonging to different sub-categories is:

Among them, C is the clustered subclass; Denote the length of the subperiod as

It is used to determine the stable mode of the aircraft; (d) Stable mode determination Based on the running time of each mode in step 2, the shortest running time of the stable mode is determined as S _min =min{S ₁ ,S ₂ ,...,S _n } (3) By comparing the length of sub-periods

and the shortest stable mode running time S _min to determine the stable mode

Among them, the stable modes belonging to the same subclass are defined as the same stable mode; The stable mode time interval in step 2 is further introduced to analyze the identified stable mode, and if the identified stable mode is within the corresponding time interval, it is classified as the specific stable mode; (e) Precise division of modes In-depth analysis of the windows adjacent to the stable mode and the transition mode determines that the start time of the transition mode occurs in the second half of the last window of the previous stable mode or the first half of the first window of the transition mode. The end time of the state occurs in the second half of the last window of the transition mode or the first half of the first window of the post-sequential stable mode; For the confirmation of the starting time of the transition mode, it is assumed that the mode starts to transition from the _k1th window, and it is necessary to re-analyze and judge with a shorter sliding window L starting from the last window of the previous stable mode, that is, the _k1-1th window. , the sliding step is h; The average value of the small sliding window obtained by the analysis of the two windows is further calculated, and the average value of the small sliding window is recorded as

The similarity between the sliding window and the pre-order stable mode is defined as

in,

is the variable mean value of the pre-order stable mode, and J is the number of multi-modal process variables; The similarity threshold α is introduced as a boundary parameter, and the relationship between each similarity and threshold is analyzed. The identification rule for the starting moment of the transition mode is:

The above rule is expressed as when there are r continuous small windows starting from the t ₁ small window and all satisfy γ _t <α, then the multimodal process is considered to enter the transition mode from the t ₁ small window; Therefore, taking the starting position of the t _1th small window as the start of the transition mode, the starting moment of this transition mode is (k ₁ -2)×H+(t ₁ -1)×h+1; For the confirmation of the end time of the transition mode, it is assumed that the mode starts to transition from the _k2th window. It is necessary to re-analyze and judge with a shorter sliding window L starting from the last window of the transition mode, that is, the _k2-1th window. The step size is h; The average value of the small sliding window obtained by the analysis of the two windows is further calculated, and the average value of the small sliding window is recorded as

The similarity between the sliding window and the post-order stable mode is defined as

in,

is the variable mean of the post-order stable mode; Analyzing the relationship between each similarity and threshold, the determination rule for the end of the transition mode is:

The above rule is expressed as: when there are r continuous small windows starting from the t2th small window and all satisfy γ _{t ≥α} , then the multimodal process is considered to enter the post-sequence stable mode from the _t2th small window; Therefore, taking the starting position of the t ₂ th small window as the end of the transition mode, the end time of this transition mode is (k ₂ -2)×H+(t ₂ -1)×h+1.

Images