Distributed electric propulsion multistage pseudo-inverse control distribution method for electric aircraft

Abstract

The invention discloses a distributed electric propulsion multistage pseudo-inverse control distribution method for an electric aircraft, and the method comprises the steps: respectively calculating a torque coefficient instruction and an actuator constraint according to the aircraft state and original input information provided by an upper flight controller; a control surface deflection instruction and a thruster thrust instruction for controlling an actuator are obtained by controlling iteration solution of the distribution weight matrix, the instructions are output to the corresponding actuator of the aircraft, and corresponding deflection or thrusting is completed, so that the state of the aircraft is changed. According to the method, multi-stage pseudo-inverse control distribution based on weight is constructed for a reachable set of force and torque of an electric aircraft by distributed electric propulsion, an actuator saturation weight and an actuator fault weight are automatically updated, a user adjustment weight is set, and an angular acceleration instruction generated for a self-driving instrument and an inner and outer ring flight control law is calculated. And multi-stage pseudo-inverse control distribution is carried out, so that cooperative flight control of the electric propeller and the control surface is realized. The flight fault-tolerant capability is improved, and the flight safety and reliability of the electric aircraft are effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to a technology in the field of aircraft control, and specifically to a distributed electric propulsion multi-stage pseudo-inverse control allocation method for electric aircraft. Background Art

[0002] One of the challenges of existing distributed electric propulsion control allocation technologies for electric aircraft is the ability to utilize the system's maximum torque reachable set. Since traditional control allocation methods use rules (direct control allocation) or pseudo-inverses (generalized inverses) to solve for the actuator control inputs for the aircraft torque commands; when the given torque command is close to the maximum torque of the aircraft, the control input solution is likely to exceed the saturation value of the actuator, causing the corresponding actual torque direction to deviate from the command direction, which has an adverse impact on the aircraft attitude control. Summary of the Invention

[0003] In view of the above deficiencies of the prior art, the present invention proposes a distributed electric propulsion multi-stage pseudo-inverse control allocation method for electric aircraft. For the force and moment reachable sets of distributed electric propulsion for electric aircraft, a weighted multi-stage pseudo-inverse control allocation is constructed. By automatically updating the actuator saturation weight and actuator fault weight, and setting the user-adjustable weight, for the angular acceleration commands generated by the autopilot and the inner and outer loop flight control laws, a multi-stage pseudo-inverse control allocation is performed to achieve coordinated flight control of the electric thrusters and control surfaces. Improve flight fault tolerance and effectively ensure the flight safety and reliability of electric aircraft.

[0004] The present invention is realized through the following technical solutions:

[0005] The present invention relates to a distributed electric propulsion multi-stage pseudo-inverse control allocation method for electric aircraft. After calculating the moment coefficient command and actuator constraints respectively according to the aircraft state and original input information provided by the upper-level flight controller, the control surface deflection command and thruster thrust command for controlling the actuator are obtained through iterative solution of the control allocation weight matrix, and the commands are output to the corresponding actuators of the aircraft to complete the corresponding deflection or propulsion, thereby changing the state of the aircraft. Specifically, it includes:

[0006] Step 1, calculate the moment coefficient command Where: C l is the roll moment coefficient, C m is the pitch moment coefficient, C n is the yaw moment coefficient to be solved, B e is the known system control effectiveness matrix, representing the control effectiveness of each actuator of the aircraft at the current operating point; u ordis the original control input solution output by a known autopilot and the inner and outer loop flight control laws, which is a known quantity, indicating that the upper-level control law of the aircraft is known. According to different aircraft configurations, it includes different original control input commands for elevators, ailerons, rudders, thrusters, etc.

[0007] Step 2, calculate the actuator normalization constraints, UB = [1 - elv trim , 1 - ail trim , 1 - rud trim , 1 - thrust trim , LB = [-1 - elv trim , -1 - ail trim , -1 - rud trim , -thrust trim , where: the vectors UB and LB represent the constraint upper bound and the constraint lower bound respectively; elv trim , ail trim , rud trim , thrust trim are the normalized values of the known elevator, aileron, rudder, and thruster trim amounts of the aircraft at the current operating point respectively. For any control input u of the aircraft, it is required that LB(i) ≤ u(i) ≤ UB(i), where i is the actuator number.

[0008] Step 3, weight calculation, specifically: the control allocation weight matrix W = W s W f W a , where: W s is the saturation weight matrix, W f is the fault weight matrix, W a is the user-adjustable weight matrix, and all matrices are m×m diagonal matrices, where m is the total number of actuators.

[0009] The described saturation weight matrix is used to measure the distance between the actuator control input and the actuator constraints. This matrix is updated in real time, and its diagonal element W s (i, i) = 1 / min 2 {u(i) - LB(i) + bias, UB(i) - u(i) + bias}, where: bias is a bias used to avoid the denominator being zero.

[0010] The described fault weight matrix is used to isolate the actuators with faults. This matrix is updated in real time, and the value rule of its diagonal elements is as follows: when the actuator has no fault, the corresponding diagonal element takes the value of 1; when an actuator fault is detected, the corresponding diagonal element is set to a relatively large value (such as 1e5) to make the allocation solution of the corresponding actuator as small as possible.

[0011] The user-adjustable weight matrix is preferably not updated in real time. This weight is the weight assigned to each actuator defined by the user. The diagonal element value rule is: the larger the value, the smaller the actuator allocation solution.

[0012] Step 4, iteratively calculate the multi-level weighted pseudoinverse, specifically: The corresponding control allocation solution calculation formula is: Where: The superscript cross represents the pseudoinverse of the matrix. Since B e represents the aircraft control effectiveness matrix. For a distributed electric propulsion configuration, B e is usually not a square matrix, so it cannot be directly inverted. In order to obtain a matrix that satisfies , the above formula can be used to calculate the pseudoinverse.

[0013] The iterative calculation described above needs to satisfy the actuator normalization constraint obtained in Step 2, that is, to judge whether the control allocation solution satisfies LB(i) ≤ u(i) ≤ UB(i). When there is no actuator saturation, directly output the control allocation solution to complete the control allocation u cmd ← u; if the actuator allocation solution is less than the lower bound, let the corresponding u(i) = LB(i), if it is greater than the upper bound, let u(i) = UB(i). The actuator number that is already saturated is called sat_index, and the corresponding saturation position is u sat , and the corresponding saturated moment coefficient The remaining moment coefficient is If all actuators are saturated, in order not to expand the control allocation error, use the initial control allocation solution u cmd ← u (0) ; if not all are saturated, then use the remaining moment coefficient as the new moment coefficient command. The actuator number that is not saturated is called nom_index, and the remaining control effectiveness matrix is B e ← B e (:, nom_index), the remaining weight matrix is W ← W(nom_index, nom_index), return the weighted pseudoinverse calculation formula described above and calculate the control allocation solution until there is no actuator saturation or all actuators are saturated.

[0014] The present invention relates to a control allocation system for implementing the above method, including: an autopilot module, an inner and outer loop, and a control allocation module. Among them: the autopilot module performs controller resolution processing according to altitude command and direction angle command information to obtain pitch angle command and roll angle command. The inner and outer loops perform controller resolution processing according to pitch angle and roll angle command information to obtain the original control input solution. The control allocation module performs multi-level weighted pseudoinverse control allocation processing according to the original control input solution information to obtain the true control input command result. Technical effects

[0015] The present invention utilizes a weighted multi-level pseudo-inverse control allocation method to measure the distance between the actuator and the saturation position by calculating the saturation and fault weights. When the allocation solution is close to the saturation position, the allocation amount of the corresponding actuator is reduced, and the system torque reachable set assigned to each actuator is fully utilized, effectively increasing the allocation accuracy of torque instructions close to the boundary of the torque reachable set, thereby improving flight reliability and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Flowchart of the present invention;

[0017] Figure 2 Schematic diagram of the system of the present invention;

[0018] Figure 3 Assigning block diagrams to controls;

[0019] Figure 4 This is a diagram of the simulation system architecture of the embodiment;

[0020] Figure 5 The three-view diagram of the controlled object and the control input diagram of the embodiment;

[0021] Figure 6 Schematic diagram of the control effectiveness matrix of the controlled object in the embodiment;

[0022] In the figure: blue is the distributed electric propulsion torque reachable set, black is the nominal system torque reachable set;

[0023] Figure 7 It is the task envelope diagram of the embodiment;

[0024] Figure 8 Assign system response diagrams for multi-level weighted pseudo-inverse control under multiple point failures;

[0025] Fig. 9 Assign system trajectory diagram for multi-level weighted pseudo-inverse control under multiple failures. DETAILED DESCRIPTION

[0026] like Figure 3 As shown, this embodiment involves a distributed electric propulsion multi-level pseudo-inverse control allocation method for electric aircraft. The torque coefficient instruction, actuator constraint and control allocation weight matrix are calculated based on the aircraft state and original input information provided by the upper-level flight controller. After the control surface deflection instruction and propeller thrust instruction for controlling the actuator are obtained through iterative solution, the instruction is output to the corresponding actuator of the aircraft to complete the corresponding deflection or propulsion, thereby changing the state of the aircraft.

[0027] like Figure 2 As shown, the control distribution system for implementing the above method includes: an autopilot module, inner and outer loops, and a control distribution module.

[0028] The described autopilot module includes: an altitude autopilot and a heading autopilot, where: the altitude autopilot generates a pitch angle command according to a given altitude command, and the heading autopilot generates a roll angle command according to a given heading command.

[0029] The described inner and outer loop includes: an inner loop and an outer loop, where: the inner loop performs controller resolution processing based on pitch, roll, and yaw angular velocity command information to obtain the original control surface deflection and thruster increment command results, and the outer loop performs controller resolution processing based on pitch angle and roll angle command information to obtain pitch, roll, and yaw angular velocity command results.

[0030] The described outer loop includes: a pitch angle controller, a roll angle controller, and a coordinated turn controller, where: the pitch angle controller performs controller resolution processing based on pitch angle command information to obtain a pitch angular velocity command result, the roll angle controller performs controller resolution processing based on roll angle command information to obtain a roll angular velocity command result, and the coordinated turn controller performs controller resolution processing based on roll angle command information to obtain a yaw angular velocity command result.

[0031] The described inner loop includes: a pitch angular velocity controller, a roll angular velocity controller, a yaw angular velocity controller, and a speed controller, where: the pitch angular velocity controller performs controller resolution processing based on pitch angular velocity command information to obtain the original elevator deflection increment command result, the roll angular velocity controller performs controller resolution processing based on roll angular velocity command information to obtain the original aileron deflection increment command result, the yaw angular velocity controller performs controller resolution processing based on yaw angular velocity command information to obtain the original rudder deflection increment command result, and the speed controller performs controller resolution processing based on speed command information to obtain the original thruster thrust increment command result.

[0032] The described control allocation module includes: an actuator saturation weight calculation unit, a fault weight calculation unit, and a user-defined weight calculation unit, where: the actuator saturation weight calculation unit performs saturation weight calculation processing based on actuator trim value information to obtain a saturation weight matrix result, the fault weight calculation unit performs fault isolation and weight setting processing based on fault information to obtain a fault weight matrix result, and the user-defined weight calculation unit performs diagonal element arrangement processing based on user-set weight information to obtain a user-defined weight matrix result.

[0033] After specific experiments, through the simulation architecture shown in Figure 4 in MATLAB / Simulink for Figure 5A high-precision simulation was performed on the control distributor in the flight control system shown in Table 1. That is, through the weighted multi-level pseudo-inverse control distribution algorithm of the present invention, the control distributor was able to maximize the utilization of the system's torque reachable set, distribute the control laws of the autopilot, outer loop, and inner loop, and obtain the input of the actuator. Figure 4 The flight controller of the simulation model Figure 2 The control distribution systems correspond one to one, such as the autopilot, outer loop, inner loop and control distribution.

[0034] The simulation architecture outputs altitude, heading, and speed commands through the autopilot input module and the pilot input module, and then transmits the commands to the autopilot, outer loop, and inner loop modules to output control allocation commands. The control distributor converts the control allocation commands into inputs for each actuator, i.e., control allocation solutions, based on the weighted multi-level pseudo-inverse control allocation algorithm described in the present invention. The actuators move according to the actuator commands and transmit position and speed signals to the propulsion module and the aerodynamic solution module. The generated forces and torques are then transmitted to the force and torque solution module for aircraft state solution. The solution results are transmitted to the data recording module for recording and simultaneously converted into signals required by the virtual instrument for graphical output.

[0035] like Figure 4 As shown, the simulation architecture includes: a flight controller and an aircraft dynamics model, wherein: the flight controller performs upper-level control law and control distribution law solution processing based on autopilot input and pilot input information to obtain actuator instruction results, and the aircraft dynamics model performs actuator dynamics and flight dynamics solution processing based on the actuator instruction information to obtain flight simulation results.

[0036] The aircraft dynamics simulation model includes: an actuator dynamics model, a cruise propeller and lift-enhancing propeller dynamics model, an aerodynamics model and a force and torque solution module, wherein: the actuator dynamics model performs actuator dynamics solution processing according to the control distribution instruction information to obtain the actual actuator (control surface and propeller) deflection position and thrust results; the cruise propeller and lift-enhancing propeller dynamics model performs propeller dynamics solution processing according to the actual actuator position and thrust information to obtain the force and torque results applied to the aircraft; the aerodynamics model performs aerodynamic coefficient table lookup according to the actuator position and thrust information and converts it into the force and torque results applied to the aircraft; the force and torque solution module performs aircraft state solution according to the force and torque information applied to the aircraft to obtain flight simulation results.

[0037] Table 1 Parameters of the aircraft dynamics model

[0038] like Figure 5As shown, the controlled object uses electrical energy as energy to power two wingtip cruise thrusters and 12 wing leading edge lift-enhancing thrusters, wherein the entire aircraft thrust is generated by the two wingtip cruise thrusters (dCT1 and dCT2), and the lift-enhancing thrusters (dT1 to dT2) are used to generate the thrust. 12 ) plays the role of increasing slipstream lift; the three sets of control surfaces include elevator (δelv), aileron (δail1 and δail2) and rudder (δrud), which provide turning torque for the aircraft. Its positive direction is as follows Figure 5 As shown by the arrows, the ailerons are configured for symmetrical differential deflection (i.e., δail1 = -δail2). It should be noted that due to the long moment arms of dCT1 and dCT2 about the center of mass of the controlled object, their differential thrust can generate a very considerable yaw moment; dT1 ~ dT 12 The slipstream effect can generate a certain amount of lift, and its thrust can be associated with a considerable nose-up moment. Therefore, if the 14 distributed thrusters of the controlled object are properly utilized, the flight quality of the aircraft can be effectively improved, and good fault-tolerant control can be achieved.

[0039] The aircraft dynamics model input is:

[0040] Where: delv, dail, drud are the deflection increments of the aircraft's elevator, aileron, and rudder, respectively, in radians; dCT1, dCT2, dT1, dT2, ..., dT 12 They are the thrust increments of the aircraft's left and right cruise thrusters and the lift-enhancing thrusters arranged from left to right, in N.

[0041] The output and state of the aircraft dynamics model are:

[0042] Where: wind axis velocity V, unit is m / s; wind axis attack angle α, sideslip angle β, unit is rad; three-axis attitude angle of the body axis relative to the earth axis: roll angle φ, pitch angle θ, yaw angle ψ, unit is rad; three-axis angular velocity: roll angular velocity p, pitch angular velocity q, yaw angular velocity r, unit is rad / s; the aircraft's earth axis position: x, y, z, unit is m.

[0043] Using the MATLAB / Simulink Linearization Toolbox, the trim state is (SI units) at an altitude of 2438 m and a true airspeed of 75 m / s. Specifically, [uvw φ θ ψ pqrxyz] T =[74.93 0 -0.25 0 -0.003 0 0 0 0 0 0 -2438.5] T, where the variables are: the three-axis velocity of the body axis, the three-axis attitude angle of the body axis relative to the Earth axis, the three-axis angular velocity, and the aircraft's Earth axis position. All units are in SI units.

[0044] The trim input at this time is: [elv ail rud CT T] T =[0.029 0 0 349.23 0.18] T , where: The variables are, in order: the absolute value of the elevator deflection angle, the absolute value of the aileron deflection angle, the absolute value of the rudder deflection angle, the absolute value of the cruise thruster thrust, and the absolute value of the high-lift thruster thrust. It should be noted that, since trimming is performed for steady level flight, the thrust during trimming is assumed to be symmetrical. 349.23 and 0.18 are the thrust trim values for a single cruise thruster and high-lift thruster thruster. However, in subsequent control allocation simulation experiments, the thrust of the thrusters is asymmetrical to address powered yaw and thruster failures, improving flight quality and reliability.

[0045] The model state equation and output equation corresponding to the working conditions of 2438m altitude and 75m / s true air speed are: Among them: the system matrix A is shown in Table 2, and the input matrix B is shown in Table 3.

[0046] Table 2 System matrix -0.0137 6.9700 -0.2557 0 -9.8000 0 0.0750 0.0077 0.0191 0 0 0 -0.0035 -1.0880 0.0015 0 0 0 0 0.9857 -0.0005 0 0 0 0 0 0.0934 0.1308 0 0 -0.0035 0 -1.0073 0 0 0 0 0 0 0 0 0 1.0000 0 -0.0034 0 0 0 0 0 0 0 0 0 0 1.0000 0 0 0 0 0 0 0 0 0 0 0 0 1.0000 0 0 0 0 -0.0067 -4.0149 0 0 0 -3.3811 0.0017 1.2556 0 0 0 0.0006 -11.7524 -5.3291 0 0 0 -0.0086 -1.8463 -0.0665 0 0 0 0 0.0045 2.2318 0 0 0 -0.0542 0 -1.8461 0 0 0 1.0000 0 0 0 0 0 0 0 0 0 0 0 0 0 74.9308 0.2547 0 74.9308 0 0 0 0 0 0 0 74.9308 0 0 -74.9308 0 0 0 0 0 0 0

[0047] Table 3 Input matrix (transposed for easier display)

[0048] According to the input matrix B and the control effectiveness matrix B e The relationship satisfies: The control effectiveness matrix B of the controlled object at an altitude of 2438m and a true air speed of 75m / s is obtained e , as shown in Table 4.

[0049] Table 4 Control effectiveness matrix (transposed for easier display)

[0050] For the control effectiveness matrix B e ,pass Figure 6 The system torque reachable set intuitively shows the rotation capability of the corresponding system. Its input is the normalized actuator deflection and thrust, and its output is the aircraft torque coefficient; Figure 6As shown in the black area, it is the system rotation ability controlled only by the elevator, aileron, and rudder (the elements in the columns related to the thrusters (columns 4 - 17) of the corresponding control effectiveness matrix are all 0); as Figure 6 shown in the blue area, it is the system rotation ability using the coordinated control of control surfaces and thrusters, that is, B e shown in the matrix, where: C m represents the pitch ability of the aircraft, C n represents the yaw ability, C l represents the roll ability.

[0051] Based on Figure 4 the simulation framework shown, according to Figure 2 the autopilot and inner - outer loop control architecture, use the MATLAB / Simulink control system designer to design the autopilot and inner - outer loop controllers, that is, the upper - layer control law, for the above - mentioned controlled object, as shown in Table 5.

[0052] Table 5 Design Parameters

[0053] Based on the above - mentioned upper - layer control law, obtain the normalized result of the actuator control input without the participation of thrusters (direct control allocation). According to the simulation architecture and control system architecture of this embodiment, the actuator control input result solved by the upper - layer control law will be redistributed by the control allocator and transmitted to the real actuator. Next, the technical solution of the present invention will be applied to this control allocator.

[0054] For the high - precision simulation, use the Simulink S - function to embed the control allocation algorithm into the simulation environment, specifically including:

[0055] Step 1, calculate the moment coefficient command. In this embodiment, the inputs u of the S - function from the first element to the last element are respectively: u(1): elevator deflection increment input; u(2): aileron deflection increment input; u(3): rudder deflection increment input; u(4:5): cruise thruster thrust increment input; u(6:17): lift - augmentation thruster thrust increment input; u(18:21): actuator fault index and jam position; u(22): fault flag bit; u(23:24): absolute value of control input. Therefore, for the calculation of the moment coefficient command, let u_input = u(1:17); at the same time, according to the control effectiveness matrix Be_nom shown in the relationship between the input matrix and the control effectiveness matrix, in the S - function, obtain the moment coefficient Xdot_cmd = Be_nom * u_input.

[0056] Step 2, calculate the actuator constraints. According to the absolute value of the control input, define the normalized absolute value (trim amount) of the elevator as elv_trim, aileron: ail_trim, rudder: rud_trim, cruise thruster: Cruisethrust_trim, high-lift thruster: HLPthrust_trim. Thus, the upper bound of the actuator input at the current moment is: upper bound UB = [1 - elv_trim, 1 – ail_trim, 1 – rud_trim, 1 – Cruisethrust_trim, 1 – Cruisethrust_trim, 1 – HLPthrust_trim, …, 1 – HLPthrust_trim], and the lower bound LB = [-1 - elv_trim, -1 – ail_trim, -1 – rud_trim, –Cruisethrust_trim, –Cruisethrust_trim, -HLPthrust_trim, …, -HLPthrust_trim].

[0057] Step 3, calculate the weights. According to the control input definition, the weight matrix is a 17×17 square matrix. First, set the saturation weight deviation bias_sat = 0.0001. According to the constraint values calculated in Step 2, the diagonal elements of the saturation weight Weight_Sat are: Weight_Sat(i, i) = 1 / (min([(u_input(i) – LB(i) + bias_sat)(UB(i) - u_input(i) + bias_sat)]))^2, where: i refers to the actuator number, taking values from 1 to 17. Then calculate the fault weights. Set the fault weight deviation bias_fault = 100000. According to the actuator fault index fault_index, set the nominal (fault-free) fault weight Wf = diag(ones(17,1)) (the diagonal elements are 1) to Wf(sub2ind(size(Wf), fault_index, fault_index)) = bias_fault, indicating that the weight is set to 100000 at the fault position [fault_index, fault_index]. At this time, the deflection or thrust allocated by the control allocator to the faulty actuator is almost 0. Then calculate the user-adjustable weight Wa = diag(ones(17,1)). In this embodiment, it is desired that the distributed thrusters replace the work of the rudder as much as possible. Therefore, the user-defined weight of the rudder is set to Wa(3, 3) = 1000, and the user-defined weight of the distributed electric thrusters is set to Wa(i, i) = 0.001, where: i = 6 to 17. The final control allocation weight is: W = Weight_Sat * Wf * Wa.

[0058] Step 4, calculate the multi-level pseudo-inverse. To calculate the multi-level pseudo-inverse, it is first necessary to calculate the single-level pseudo-inverse. In this embodiment, a function u_output = WeightCA(W, Be_nom, Xdot_cmd) is defined separately. By inputting the weight, control effectiveness matrix, and torque command, the single-level pseudo-inverse distribution solution is output: u_output = W\Be_nom' / (Be_nom / W*Be_nom')*Xdot_cmd. At the same time, in this embodiment, a while statement is used to iterate the saturation state of the actuator, that is, to judge whether u_output(i) is greater than UB(i) or u_output(i) is less than LB(i). As long as the above situation is satisfied, the number sat_index of the saturated actuator and the number nom_index of the unsaturated actuator are recorded, and then the remaining moment coefficient is calculated: Xdot_cmd = Xdot_cmd - Be_nom(:, sat_index)*u_output(sat_index), and the multi-level pseudo-inverse is calculated using the single-level pseudo-inverse solution function: u_output(nom_index) = WeightCA(W(nom_index, nom_index), Be_nom(:, nom_index), Xdot_cmd). When no more actuators are saturated, the current u_output is output as the final control allocation solution; when all actuators are saturated, the first pseudo-inverse calculation result is used and an alarm is given.

[0059] The above control allocator and the upper-level control law together constitute the flight controller of this embodiment. To verify the flight fault tolerance ability based on this control allocation algorithm, this embodiment designs a flight mission including different flight phases, such as Figure 7 shown. The mission is as follows: Phase 1 (0 - 200 s): The aircraft cruises at 75 m / s at an altitude of 2438.5 m; Phase 2 (200 - 420 s): The aircraft descends 100 m at a descent rate of 5 m / s and then cruises at an altitude of 2338.5 m; Phase 3 (420 - 580 s): The aircraft coordinately turns from the 0 azimuth angle to the 90 deg azimuth angle and then cruises at an altitude of 2338.5 m; Phase 4 (580 - 800 s): The aircraft climbs 100 m at a climb rate of 5 m / s and then cruises at an altitude of 2438.5 m; Phase 5 (800 - 1200 s): The aircraft coordinately turns from the 90 deg azimuth angle to the 0 azimuth angle and then cruises at an altitude of 2438.5 m, where: Fault occurrence time (110 s): The left cruise thruster is stuck at a torque of 400 Nm, and the 1# lift augmentation thruster (the leftmost one) is stuck at the 0 valve position.

[0060] Combine the flight controller and the controlled object model of this embodiment, configure the above task instructions, set the simulation type to fixed step size, the solver to ode4 (Runge-Kutta), the basic sampling time to 0.01s, and the simulation duration to 1200s. Start the simulation and record the data. The system response is as follows Figure 8 As shown, the corresponding flight trajectory is as follows Fig. 9 shown.

[0061] like Figure 8 As shown in FIG, the response diagram of the multi-level weighted pseudo-inverse control distribution system under multi-point failure faults, where: Hcmd and H in sub-figure (1) are the altitude command and the actual altitude of the aircraft (positive when pointing to the center of the earth); Directioncmd and psi in sub-figure (2) are the heading command and the actual yaw angle of the aircraft; Vset and V in sub-figure (3) are the speed command and the actual speed of the aircraft; sub-figure (4) is the absolute value of the thrust of the 12 lift-enhancing thrusters, where: the blue solid line is the absolute value of the thrust of the faulty lift-enhancing thruster, the green solid line and the other 10 lines covered by it (a total of 11 lines) are the absolute values of the thrust of the normal lift-enhancing thruster; CT1 and CT2 in sub-figure (5) are the absolute values of the thrust of the left and right cruise thrusters respectively; sub-figure (6) is the fault identification bit, 0 represents no fault, 1 represents a multi-point failure; sub-figures (7)(8)(9) are the absolute values of the control surface deflection of the aileron, elevator and rudder respectively. From the fault identification position, we can see that the fault occurred at 110 seconds and lasted until the end. At this time, the left cruise thruster was stuck at a constant torque of 400Nm. The thrust changed slightly in real time with other conditions such as true airspeed. At the same time, the leftmost 1# lift thruster produced almost no thrust. Figure 5 It can be seen that at this time, the two thrusters at the farthest end and with the longest lever arms on the left side of the aircraft have all failed. However, by observing sub-graphs (1) and (2) at this time, it can be found that the aircraft's altitude and direction are almost unaffected. This is because under the technical solution of multi-level weighted pseudo-inverse control distribution proposed by the present invention, after the multi-point failure occurs, the 11 fault-free lift thrusters T2 to T12 shown in sub-graph (4) gradually reduce the thrust to the level of the faulty thruster, and the fault-free cruise thruster CT2 shown in sub-graph (5) immediately generates differential thrust to compensate for the unbalanced yaw moment caused by the multi-point failure, while maintaining the thrust command of the entire aircraft based on the speed controller as much as possible; the control surfaces shown in sub-graphs (7) (8) (9) also participate in the coordinated control after the failure occurs, restoring the system torque to the torque command without saturation. Finally, under the technical solution proposed by the present invention, the altitude command and the direction command are well tracked, while the speed command has an inevitable deviation due to the multi-point failure problem. The corresponding actual flight trajectory diagram is as follows: Fig. 9 It can be found that the unbalanced torque generated by the failed thruster can be reasonably distributed to the remaining thrusters and control surfaces through the multi-level weighted pseudo-inverse control distribution method, which effectively compensates for the multi-point failure of the thruster.

[0062] Compared with the prior art, through the multi-level weighted pseudo-inverse control allocation simulation experiment on the controlled object in this embodiment, the present invention fully proves the full utilization of the system torque reachable set and the improvement of fault tolerance by the proposed algorithm, realizes the differential yaw of the distributed electric propulsion aircraft, and improves the safety and reliability of the system.

[0063] Those skilled in the art can make local adjustments to the above specific implementation in different ways without departing from the principles and purposes of the present invention. The protection scope of the present invention is subject to the claims and is not limited by the above specific implementation. All implementation solutions within its scope are subject to the constraints of the present invention.

Claims

1. A distributed electric propulsion multi-stage pseudo-inverse control allocation method for electric aircraft, characterized in that After calculating the moment coefficient command and actuator constraints respectively based on the aircraft state and original input information provided by the upper flight controller, the control surface deflection command and thruster thrust command for controlling the actuators are obtained through iterative solution of the control distribution weight matrix, and the commands are output to the corresponding actuators of the aircraft to complete the corresponding deflection or propulsion, thereby changing the state of the aircraft.

2. The distributed electric propulsion multi-stage pseudo-inverse control allocation method for electric aircraft according to claim 1, characterized in that specifically Including: Step 1, calculate the moment coefficient command Where: C l is the roll moment coefficient, C m is the pitch moment coefficient, C n is the yaw moment coefficient to be determined, B e is the known system control effectiveness matrix, representing the control effectiveness of each actuator of the aircraft at the current operating point; u ord is the original control input solution output by the known autopilot and the inner and outer loop flight control laws; Step 2: Calculate the actuator normalization constraint, UB = [1-elv trim ,1-ail trim ,1-rud trim ,1-thrust trim ],LB=[-1-elv trim ,-1-ail trim ,-1-rud trim ,-thrust trim ], where: vectors UB and LB represent the upper bound and lower bound of the constraint respectively; elv trim ,ail trim ,rud trim ,thrust trim are the normalized values of the known elevator, aileron, rudder, and propeller trim values of the aircraft at the current operating point. For any control input u of the aircraft, LB(i)≤u(i)≤UB(i), where i is the actuator number; Step 3, weight calculation, specifically: control the distribution weight matrix \(W = W\) s W f W a , where: \(W\) s is the saturation weight matrix, \(W\) f is the fault weight matrix, \(W\) a is the user-adjustable weight matrix, and all matrices are \(m\times m\) diagonal matrices, where \(m\) is the total number of actuators; Step 4, iteratively calculate the multi-level weighted pseudo-inverse, specifically as follows: The corresponding control allocation solution calculation formula is: where: the cross superscript represents the pseudo-inverse of the matrix, and B e represents the control effectiveness matrix of the aircraft.

3. The distributed electric propulsion multi-stage pseudo-inverse control allocation method for electric aircraft according to claim 2, characterized in that, The described saturation weight matrix is used to measure the distance between the actuator control input and the actuator constraint. This matrix is updated in real time, and its diagonal element W s (i,i) = 1 / min 2 {u(i) - LB(i) + bias, UB(i) - u(i) + bias}, where: bias is the bias; The fault weight matrix is used to isolate the faulty actuators, and the matrix is updated in real time. The value rule of its diagonal elements is as follows: when the actuator is fault-free, the corresponding diagonal element takes the value of 1.

4. The distributed electric propulsion multi-stage pseudo-inverse control allocation method for electric aircraft according to claim 2, characterized in that For the distributed electric propulsion configuration, B e cannot be directly inverted. To obtain a matrix that satisfies , calculate the pseudoinverse 5. The distributed electric propulsion multi-stage pseudo-inverse control allocation method for electric aircraft according to claim 1 or 2, characterized in that The iterative calculation needs to satisfy the actuator normalization constraint, that is, to judge whether the control distribution solution satisfies LB(i) ≤ u(i) ≤ UB(i). When there is no actuator saturation, directly output the control allocation solution to complete the control allocation u cmd ← u; When the actuator allocation solution is less than the lower bound, let the corresponding u(i) = LB(i). When it is greater than the upper bound, let u(i) = UB(i), corresponding to the saturation torque coefficient The remaining torque coefficient is where: the actuator number that has been saturated is sat_index, and the corresponding saturation position is u sat ; When all actuators are saturated, the initial control allocation solution u cmd ←u (0) ; When not fully saturated, the remaining moment coefficient is used as the new moment coefficient command, and the remaining control effectiveness matrix is B e ←B e (:, nom_index), the remaining weight matrix is w←w(nom_index, nom_index), return the weighted pseudo-inverse calculation formula and calculate the control allocation solution until there is no actuator saturation or all actuators are saturated, where: the unsaturated actuator number is nom_index.

6. A control allocation system for implementing the method according to any one of claims 1-5, characterized in that, Including: An autopilot module, inner and outer loop circuits, and a control distribution module. Among them: the autopilot module performs controller calculation and processing according to the altitude command and direction angle command information to obtain the pitch angle command and roll angle command; the inner and outer loop circuits perform controller calculation and processing according to the pitch angle and roll angle command information to obtain the original control input solution; the control distribution module performs multi-level weighted pseudo-inverse control distribution processing according to the original control input solution information to obtain the real control input command result.

7. The control distribution system according to claim 6, characterized in that, The autopilot module includes: an altitude autopilot and a heading autopilot. Among them: the altitude autopilot generates a pitch angle command according to the given altitude command, and the heading autopilot generates a roll angle command according to the given heading command.

8. The control distribution system according to claim 6, characterized in that: The inner and outer loop circuits include: an inner loop and an outer loop. Among them: the inner loop performs controller calculation and processing according to the pitch, roll, and yaw angular velocity command information to obtain the original control surface deflection and thruster increment command results; the outer loop performs controller calculation and processing according to the pitch angle and roll angle command information to obtain the pitch, roll, and yaw angular velocity command results.

9. The control distribution system according to claim 8, characterized in that, The outer loop includes: a pitch angle controller, a roll angle controller, and a coordinated turn controller. Among them: the pitch angle controller performs controller calculation and processing according to the pitch angle command information to obtain the pitch angular velocity command result; the roll angle controller performs controller calculation and processing according to the roll angle command information to obtain the roll angular velocity command result; the coordinated turn controller performs controller calculation and processing according to the roll angle command information to obtain the yaw angular velocity command result. The inner loop includes: a pitch angular velocity controller, a roll angular velocity controller, a yaw angular velocity controller, and a speed controller. Among them: the pitch angular velocity controller performs controller calculation and processing according to the pitch angular velocity command information to obtain the original elevator deflection increment command result; the roll angular velocity controller performs controller calculation and processing according to the roll angular velocity command information to obtain the original aileron deflection increment command result; the yaw angular velocity controller performs controller calculation and processing according to the yaw angular velocity command information to obtain the original rudder deflection increment command result; the speed controller performs controller calculation and processing according to the speed command information to obtain the original thruster thrust increment command result.

10. The control distribution system according to claim 6, characterized in that The described control allocation module includes: an actuator saturation weight calculation unit, a fault weight calculation unit, and a user-defined weight calculation unit, where: the actuator saturation weight calculation unit performs saturation weight calculation processing based on the actuator trim value information to obtain a saturation weight matrix result; the fault weight calculation unit performs fault isolation and weight setting processing based on the fault information to obtain a fault weight matrix result; and the user-defined weight calculation unit performs diagonal element arrangement processing based on the weight information set by the user to obtain a user-defined weight matrix result.

Images