Energy optimal flight control method of modular reconfigurable flight array

Abstract

The invention relates to the technical field of unmanned rotorcraft design and control, and provides an energy optimal flight control method of a modular reconfigurable flight array. Comprising the following steps: constructing a dynamic model corresponding to a modular reconfigurable flight array containing a module efficiency factor by introducing a mapping relation between a virtual control input quantity of an advanced controller and an actuator input quantity; wherein the dynamical model is a nonlinear dynamical model established by combining a force and moment model of a single module and a Newton-Euler formula based on an auxiliary coordinate system, and the module efficiency factor is used for determining fault location of the modular reconfigurable flight array; taking the mapping relation as a constraint condition corresponding to an energy minimum optimization problem selected by a module in the modular reconfigurable flight array, and obtaining an energy value function containing a module efficiency factor; estimating a module efficiency factor of a module in the modular reconfigurable flight array based on the mapping relation and the kinetic model to obtain an estimated value; and an estimated value obtained through estimation is substituted into an energy value function, and a set of energy optimal modules is determined in a finite set search and comparison mode.

Description

Technical Field

[0001] The present invention relates to the technical fields of unmanned rotorcraft design and control, and particularly to an energy-optimal flight control method for a modular reconfigurable flight array. Background Art

[0002] Modular reconfigurable flight arrays face a complex technical challenge in practical applications: how to achieve energy-optimal flight control when actuator fault information is unknown. The core of this problem lies in that the various modules in the flight array may have different performance due to faults or environmental factors, resulting in a decline in overall performance and an increase in energy consumption. Traditional flight control methods usually rely on accurate actuator state information, but in actual scenarios, this information is often difficult to obtain in real time or is uncertain. Therefore, how to dynamically evaluate the actual performance of each module under the condition of lacking accurate fault information and optimize the configuration and control strategy of the flight array accordingly has become a technical problem to be solved urgently.

[0003] This problem involves multiple technical aspects: firstly, how to estimate the module performance factor of each module in real time during flight to accurately reflect its current state; secondly, how to integrate these dynamically changing module performance factors into the energy optimization model to achieve a globally optimal control decision; furthermore, how to maximize the energy efficiency of the system while ensuring flight stability and reliability. These sub-problems are interrelated and jointly constitute a complex system optimization problem.

[0004] More deeply, this technical challenge also involves how to achieve the adaptive reconfiguration of the flight array in an uncertain and dynamically changing environment. This requires the control system to be able to quickly respond to changes in module performance, timely adjust the flight strategy to adapt to various possible fault situations and external disturbances. At the same time, how to efficiently execute these complex optimization algorithms under limited computing resources and real-time requirements is also a technical difficulty that cannot be ignored. Summary of the Invention

[0005] The present invention provides an energy-optimal module selection algorithm for a modular reconfigurable flight array when actuator fault information is unknown, mainly including:

[0006] Constructing a dynamic model corresponding to the modular reconfigurable flight array including module performance factors by introducing the mapping relationship between the virtual control input quantity of the high-level controller and the actuator input quantity;

[0007] Wherein the dynamic model is a non-linear dynamic model established based on an auxiliary coordinate system in combination with the force and moment model of a single module and Newton-Euler formula, and the module performance factor is used to determine the fault location of the modular reconfigurable flight array;

[0008] Take the mapping relationship as the constraint condition corresponding to the energy minimum optimization problem of module selection in the modular reconfigurable flight array, and obtain an energy value function including the module efficiency factor;

[0009] Estimate the module efficiency factor of the modules in the modular reconfigurable flight array based on the mapping relationship and the dynamic model to obtain an estimated value;

[0010] Substitute the estimated value obtained by estimation into the energy value function, and determine the set of energy-optimal modules by means of finite set search and comparison.

[0011] The technical solution provided by the embodiment of the present invention may include the following beneficial effects:

[0012] The present invention discloses an energy-optimal flight control method for a modular reconfigurable flight array. The method constructs an energy value function including the module efficiency factor by obtaining high-level control input information, module efficiency factors, a non-linear dynamic model, and a mapping relationship matrix. Using an adaptive parameter estimation algorithm, the module efficiency factors of each module are estimated online and substituted into the energy value function to solve for the set of energy-optimal modules in the case where the actuator fault information is unknown. Finally, the module actuator instructions corresponding to this set are used as the input instructions for the flight array to achieve energy-optimal flight control. The present invention can, in the case where the actuator fault is unknown, dynamically adjust the configuration of the flight array by real-time estimating the module efficiency factor and optimizing the energy value function, so as to achieve efficient and reliable flight control, and improve the adaptability and fault tolerance of the system. Description of the Drawings

[0013] Figure 1 It is a schematic diagram of the structure of a single module of the modular reconfigurable flight array in the present invention;

[0014] Figure 2 It is a schematic diagram of the auxiliary coordinate system used by the modular reconfigurable flight array in the present invention;

[0015] Figure 3 It is a flowchart of an energy-optimal flight control method for a modular reconfigurable flight array of the present invention;

[0016] Figure 4 It is a schematic diagram of the energy-optimal module selection framework of the modular reconfigurable flight array of the present invention. Detailed Embodiments

[0017] To enable those skilled in the art to better understand the technical solutions in this specification, the following will clearly and completely describe the technical solutions in the embodiments of this specification in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this specification.

[0018] The present invention provides an energy-optimal flight control method for a modular reconfigurable flight array, which can achieve the minimum energy consumption of the flight array under the same task and is compatible with normal and faulty actuator conditions.

[0019] The modular reconfigurable flight array (MRFA) is a system based on modular unmanned aerial vehicles or flight units. Its core feature is the ability to dynamically adjust the physical structure (such as shape, scale, function) or logical cooperation mode according to task requirements, realizing an intelligent cluster system that can flexibly adapt to complex environments and diverse tasks. It not only has the collaborative ability of traditional unmanned aerial vehicle swarms, but also endows the system with stronger functional expandability and environmental adaptability through the autonomous connection, separation, and recombination of modules.

[0020] As Figure 1 shown, the modular reconfigurable flight array is composed of several identical modules and can be spliced into different configurations according to requirements. Each module includes a propeller, an electronic speed controller, a motor, a battery, and a flight controller. As Figure 2 shown, among them, the propeller, the electronic speed controller, and the motor form a set of power systems, serving as the actuators of the module.

[0021] As Figure 3 shown, the energy-optimal flight control method of the modular reconfigurable flight array in this embodiment may specifically include:

[0022] S101. Construct a dynamic model corresponding to the modular reconfigurable flight array including module efficiency factors by introducing the mapping relationship between the virtual control input quantity of the high-level controller and the actuator input quantity.

[0023] The dynamic model is a non-linear dynamic model established based on the auxiliary coordinate system in combination with the force and moment models of a single module and Newton-Euler formula. The module efficiency factor is used to determine the fault location of the modular reconfigurable flight array.

[0024] As Figure 1 shown, first establish three auxiliary coordinate systems: the inertial coordinate system the configuration coordinate system and the module coordinate system Among them, the inertial coordinate system is fixed, the z-axis always points vertically to the sky, and three Euler angles, roll φ, pitch θ, and yaw ψ, can be obtained by rotating around the three axes; the configuration coordinate system has its origin coinciding with the center of gravity of any configuration of the flight array. The z-axis always points upward perpendicular to the array configuration plane. The lift position generated by module i is the geometric center (x i , y i ) of each module in the inertial coordinate system; the module coordinate system has its origin located at the geometric center of each module. i = 1, 2,..., N represents the module number, and N is the number of modules in this configuration.

[0025] As Figure 2 shown, each module of the flight array contains a set of power systems, namely, motors, electronic speed controllers, and propellers. Each module provides a lift in the direction consistent with the z-axis of the configuration coordinate system (the lift direction is consistent with the z-axis direction of the module coordinate system ) and the corresponding rotational torque M i . The force and torque models of each module are as follows: Among them, T i and M i represent the lift and torque that the i-th module can provide respectively, u i is the control input of the actuator corresponding to the i-th module, k t represents the lift coefficient of the i-th module (the lift coefficients of all modules are regarded as the same), represents the torque coefficient of the i-th module, among which are of the same magnitude, and their positive and negative depend on the rotation direction of the propeller in the module.

[0026] Finally, through the rotation transformation from the configuration coordinate system to the inertial coordinate system, combined with the force and torque models and the Newton-Euler equations, the nonlinear dynamic model of the flight array in the inertial coordinate system is established as:

[0027] Among them represents the known constant system matrix, represents the state information corresponding to the flight array. z, φ, θ, and ψ represent the altitude, roll angle, pitch angle, and yaw angle respectively, is the known constant input matrix, that is represents the nonlinear model uncertainty terms caused by factors such as air resistance, non-complete rigidity of the structure, and assembly errors in the known model. x0 represents the initial state information of the flight array. It should be noted that the constant system matrix indicates that the flight array is regarded as a rigid body during flight and the near-earth effect is not considered.

[0028] v is a virtual input quantity, including the vertical channel control quantity v1, the roll channel control quantity v2, the pitch channel control quantity v3, and the yaw channel control quantity v4, so that v = [v1 v2 v3 v4] T 。The virtual input quantity and the input quantity u = [u1 u2 … u N T of the actuator of the module efficiency factor have the following mapping relationship:

[0029]

[0030] where the diagonal matrix Λ = diag[α1 α2 … α N , 0 ≤ α i ≤ 1 represents the efficiency factor of module i, and α i = 1 means that the actuator of module i is in a completely healthy state; α i = 0 means that the actuator of module i is in a completely faulty state, and at this time, it can no longer provide lift and moment; 0 < α i < 1 indicates that there is a partial fault in module i, and the smaller the value, the more serious the fault, and the larger the value, the less serious the fault.

[0031] The mapping relationship matrix is:

[0032]

[0033] The mapping relationship matrix has four elements in each column, respectively representing the four virtual control quantities of the actuator corresponding to a single module of the flight array. Therefore, each column of the mapping relationship matrix uniquely corresponds to the components of a module in the array on the four virtual control variables. Among them, k t and are the lift coefficient and moment coefficient of the i-th module respectively, and (x1, y1), (x2, y2)...(x N , y N ) are the module coordinates of a single module in the modular reconfigurable flight array respectively.

[0034] S102. Use the mapping relationship as the constraint condition corresponding to the energy minimum optimization problem of module selection in the modular reconfigurable flight array to obtain an energy value function including the module efficiency factor.

[0035] Such as Figure 4 ​The modular reconfigurable flight array energy-optimal module selection framework shown is specifically to combine the virtual control input with this energy-optimal module selection algorithm to obtain the optimal module combination. The advantage of this method is that an effectiveness factor is introduced into the existing value function to reflect the fault conditions in the actual model, making the model closer to the actual working conditions, and finally being able to select the flight array modules with the optimal energy.

[0036] First, by solving the Moore-Penrose generalized inverse, the relationship between the virtual control input v and the actuator input u in formula (3) is transformed to obtain: where

[0037] Second, combining the mapping relationship in the above formula (3), the minimum energy problem of module selection can be described as satisfying the virtual control input v = [v1 v2 v3 v4] T and the constraint condition of the mapping relationship with the actuator input considering the module effectiveness factor (i.e., taking formula (3) as the constraint condition). The minimum energy problem is described as: where W is a positive definite diagonal matrix for weighting the actuators of each module. Obviously, the optimization problem (6) requires selecting the actuator input u to satisfy the mapping relationship shown in formula (3), that is, while satisfying the input mapping relationship, minimizing the energy of the actuators of each module.

[0038] Thus, the solution u0 of the optimization problem (6) can be expressed as:

[0039] Furthermore, since the system usually has the characteristic of actuator redundancy when the number of flight array modules is greater than 5, the inverse matrix (5) always exists. Therefore, substituting the optimal actuator input (7) into the optimization problem (6), the value function L can be reconstructed as follows: The reconstructed value function L can be adapted to any advanced controller algorithm combined with the flight array dynamics model (2). In addition, for formula (8), the key of the present invention is to select k modules from all the flight array modules of this configuration to minimize the energy under the constraint of k < N, rather than using all modules to participate in the flight mission of the flight array.

[0040] Finally, the final form of the energy value function is obtained by defining an auxiliary set, specifically as follows: Define the auxiliary set and Y set , where it satisfies: Formula (9) satisfies where is the set (with N elements) composed of all column vectors in the mapping relationship matrix and Yset It is a set composed of k column vectors in the mapping matrix Y, which physically corresponds to the actuators of the module.

[0041] For the selected actuator set Y set , the selection mapping matrix corresponding to the selected module is defined as Y. Therefore, the energy minimization problem in the above formula (6) can be re-described as:

[0042]

[0043] Formula (10) is the final form of the energy value function containing the module effectiveness factor, indicating that the energy magnitude is solved from all combinations of the selected modules, and the module combination with the minimum energy is selected. Among them, Card(Y) represents the basis of matrix Y. For the known virtual control input v, the selection mapping matrix Y, and the actuator weight matrix W of the module Y , where the selection mapping matrix Y corresponds to the module effectiveness factor Λ Y .

[0044] S103. Estimate the module effectiveness factor of the modules in the modular reconfigurable flight array based on the mapping relationship and the dynamic model to obtain the estimated value.

[0045] Based on the idea of online estimation of unknown parameters, a fault location algorithm for the modular reconfigurable flight array is constructed (that is, the effectiveness factor corresponding to each module is estimated, and this effectiveness factor uniquely corresponds to the health state of the actuator, realizing fault location). Its purpose is to determine which specific module in the flight array has a fault and obtain its corresponding module effectiveness factor. The specific steps are as follows:

[0046] Step 1: Combine formula (2) and (3), and linearly parameterize the dynamic model as:

[0047]

[0048] where is the rewritten form of the actuator input quantity, is the rewritten form of the fault factor (which is also a parameter to be estimated in this algorithm), so as to rewrite formula (11) into the following form: where and are both intermediate variables of the algorithm.

[0049] Step 2: Design an adaptive rate that can make the parameter to be estimated converge. First, introduce a first-order low-pass filter (·) on both sides of formula (12) f =(·) / (κ f s + 1), and the filtered variable is obtained as:

[0050]

[0051] Among them and are the filtering variables of x f , Φ f and respectively, and κ f is the filtering coefficient to be adjusted. Substituting formula (13) into (12) gives:

[0052]

[0053] Step 3: Introduce auxiliary matrices D and F as follows:

[0054]

[0055] where ι > 0 is an adjustable algorithm parameter. Solving formula (15) simultaneously gives:

[0056]

[0057] Combining formulas (14) and (16), through the constructed auxiliary matrix, we get At the same time, define vector Q as: Among them is the estimated value of the module efficiency factor, and

[0058] Finally, the fault location algorithm can be designed as: where Γ > 0 is the learning gain matrix. The corresponding formula (18) of the algorithm can online estimate the efficiency factors of each module actuator, thereby realizing the fault location of the modular reconfigurable flight array.

[0059] S104. Substitute the estimated value obtained by estimation into the energy value function, and determine the set of energy-optimal modules by means of finite set search and comparison.

[0060] Specifically, substitute the module efficiency factor into the energy value function (i.e., formula (10)), and by means of finite set search and comparison, that is, substitute all optional module combinations into the value function, obtain the energy consumption of different combinations, and compare them to finally obtain the module combination with the minimum energy drive.

[0061] As can be seen from the above technical solutions, for the energy value function of the actuator redundancy system such as the modular reconfigurable flight array, the energy consumption during flight is quantified. While considering all flight array modules, this value function also takes into account the efficiency factor of individual modules, further accurately reflecting the energy consumption under actual fault conditions. At the same time, the efficiency factor is abstracted as an unknown parameter, and an adaptive parameter estimation method is used to achieve fault location. The problem of infinitely many solutions of parameters caused by more unknown parameters than equations in the actuator redundancy system is solved. Finally, the optimal module selection of the energy of the modular reconfigurable flight array under unknown fault conditions is realized, improving the main performance index of the flight time of the flight array.

[0062] The above description is only an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. An energy-optimal flight control method for a modular reconfigurable flight array, characterized in that, Including: Constructing a dynamic model corresponding to a modular reconfigurable flight array containing a module effectiveness factor by introducing the mapping relationship between the virtual control input quantity of the advanced controller and the actuator input quantity; The dynamic model is a nonlinear dynamic model established based on the auxiliary coordinate system in combination with the force and moment model of a single module and the Newton-Euler formula, and the module effectiveness factor is used to determine the fault location of the modular reconfigurable flight array; Taking the mapping relationship as the constraint condition corresponding to the energy minimum optimization problem of module selection in the modular reconfigurable flight array to obtain an energy value function containing the module effectiveness factor; Estimating the module effectiveness factor of the modules in the modular reconfigurable flight array based on the mapping relationship and the dynamic model to obtain an estimated value; Substituting the estimated value obtained by estimation into the energy value function, and determining the set of energy-optimal modules by means of finite set search and comparison.

2. The energy-optimal flight control method according to claim 1, wherein The expression of the dynamic model is: wherein is a known constant system matrix, is the state information corresponding to the modular reconfigurable flight array, where z, φ, θ and ψ respectively represent altitude, roll angle, pitch angle and yaw angle, is a known constant input matrix, represents the nonlinear model uncertainty caused by known aerodynamic drag, structural non-rigidity and assembly error in the model, x0 is the initial state information of the modular reconfigurable flight array; v = [v1 v2 v3 v4] T represents the virtual control input quantity, v1 is the control quantity of the vertical channel, v2 is the control quantity of the roll channel, v3 is the control quantity of the pitch channel, v4 is the control quantity of the yaw channel, u = [u1 u2 … u N T represents the actuator input quantity;​ The mapping relationship between the virtual control input quantity and the actuator input quantity is expressed as: where the diagonal matrix Λ = diag[α1 α2 … α N , 0 ≤ α i ≤ 1 represents the module efficiency factor of module i, α i = 1 represents that the actuator of module i is in a completely healthy state, and α i = 0 represents that the actuator of module i is in a completely failed state; Accordingly, the mapping relationship matrix corresponding to the mapping relationship between the virtual control input quantity and the actuator input quantity is expressed as: Among them, each column of the mapping relationship matrix B contains four elements, respectively representing the four virtual control quantities of the actuator corresponding to a single module of the flight array. Therefore, each column of the mapping relationship matrix uniquely corresponds to the components of a module in the array on the four virtual control variables, where k t and are respectively the lift coefficient and moment coefficient of the i-th module, and (x1, y1), (x2, y2)...(x N , y N ) are respectively the module coordinates of a single module in the modular reconfigurable flight array.

3. The energy-optimal flight control method according to claim 2, characterized in that, The energy value function is expressed as: where Card(Y) represents the basis of matrix Y. For the known virtual control input quantity v of L, the selection mapping matrix Y and the actuator weight matrix W of the module Y , where the selection mapping matrix Y corresponds to the module efficiency factor Λ Y .

4. The energy-optimal flight control method according to claim 3, wherein The estimating the module effectiveness factor of the modules in the modular reconfigurable flight array based on the mapping relationship and the dynamic model includes: Based on the mapping relationship and the dynamic model, the dynamic model is linearly parameterized as: In the formula is the rewritten form of the actuator input quantity, is the rewritten form of the fault factor, so that is rewritten as: In the formula and are both intermediate variables of the algorithm; By introducing a first-order low-pass filter (·) on both sides of the formula =(·) / (κ f s + 1), the filtered variable is obtained as follows: f In the formula and are the filtered variables of x f , Φ f and respectively, and κ f is the filter coefficient to be adjusted. Substituting the filtered variable into gives the formula: ​ Introduce the auxiliary matrices D and F, and combine with the formula to obtain At the same time, define the vector Q as: where is the estimated value of the module efficiency factor, Γ>0 is the learning gain matrix, and represents the estimation error.

Images