A rapid diagnosis and fault-tolerant stability control system and method for rotor failure in multi-rotor unmanned aerial vehicles (UAVs).

Abstract

This invention discloses a rapid diagnosis and fault-tolerant stability control system and method for multi-rotor UAV rotor failure, belonging to the field of UAV flight control technology. It aims to solve the problems of high cost and unreliability of traditional rotor failure detection methods, insufficient robustness of power compensation, and lack of return-to-home logic. Based on high-order nonlinear observation theory, the system and method construct a multi-modal signal prediction and residual evaluation system to achieve rapid and accurate identification of abnormal states in the UAV's power system. This proactively addresses runaway issues, effectively preventing UAVs from losing control, tumbling, or even crashing due to power asymmetry. It solves the problem that traditional fixed control strategies are difficult to adapt to fault conditions, adapting to dynamic changes in complex conditions such as hovering and cruise. Ultimately, it achieves closed-loop control from fault detection to safe return, ensuring the UAV's autonomous and safe return capability in the event of propulsion system failure. It is suitable for UAV applications with high reliability requirements such as logistics transportation and inspection monitoring.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) flight control technology, specifically relating to a rapid diagnosis and fault-tolerant stable control system and method for multi-rotor UAV rotor breakage faults based on high-order nonlinear observation theory. Background Technology

[0002] Multirotor drones have been widely used in aerial photography, logistics, agricultural plant protection, and emergency rescue due to their flexibility, vertical takeoff and landing capabilities, and low cost. However, their flight stability is highly dependent on the integrity of the rotor system. Once a rotor failure occurs, it will lead to an imbalance in lift distribution, which in turn will cause loss of attitude control or even crash, resulting in equipment damage and safety hazards.

[0003] In recent years, although some progress has been made in the detection and protection technology for propeller breakage faults, the following key technical bottlenecks still exist:

[0004] (1) Traditional propeller breakage detection relies heavily on the speed feedback signal of the electronic speed governor to identify faults by judging whether the motor speed is lower than a preset threshold. However, this method has a significant drawback: when a part of the propeller blade breaks or falls off, the motor may still maintain a high speed, causing the system to misjudge normal operation;

[0005] (2) Traditional pseudo-inverse method or PID control algorithm does not consider the motor speed saturation limit when redistributing power, which can easily lead to the remaining rotor running under overload and cause secondary failures;

[0006] (3) The logic of emergency return and safe landing is missing. If the return operation is carried out directly after the propeller breaks without combining the remaining power to optimize the path in real time, it will cause the car to crash due to motor overload during the return journey.

[0007] (4) In order to improve fault tolerance, some solutions adopt dual-motor redundancy or folding spare blade design. However, such hardware modification will greatly increase the weight and power consumption of the UAV, offsetting the lightweight advantage of the multi-rotor platform, and significantly increasing the manufacturing cost, making it difficult to popularize in the civilian market.

[0008] Therefore, how to enable multi-rotor aircraft to continue flying safely after some blades fail, while being compatible with existing drone architectures and avoiding the cost burden caused by hardware redundancy, is a technical problem that urgently needs to be solved in the current technology. Summary of the Invention

[0009] The purpose of this invention is to address the aforementioned problems in the prior art by providing a rapid diagnosis and fault-tolerant stability control system and method for multi-rotor drones with broken propellers. When one or more propellers of a multi-rotor drone fail, the system's rapid response and adjustment can maintain the drone's stable flight, enabling a safe return or landing.

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] This invention first discloses a rapid diagnosis and fault-tolerant stability control system for rotor breakage faults in multi-rotor unmanned aerial vehicles (UAVs), comprising:

[0012] Monitoring module: Constructs a multimodal signal prediction and residual evaluation system to accurately identify abnormal states of the UAV power system and transmits the detected fault diagnosis signals to the control module;

[0013] Control module: Dynamically adjusts the control allocation matrix based on airframe parameters and received fault diagnosis signals, generates motor compensation signals, and sends propeller failure fault flags to the safe return module;

[0014] Safe return-to-home module: Based on a multi-constraint optimized autonomous navigation and stability assurance strategy, it automatically plans a safe return-to-home route by combining the current position, the return-to-home point position, and the set return-to-home altitude, and controls the aircraft to return to home along the route.

[0015] Preferably, the aforementioned monitoring module is based on high-order nonlinear observation theory. The module has built-in adaptive convergence factor and nonlinear saturation function, which are used to filter and preprocess the original thrust command of the motor to suppress abnormal amplitude disturbances and improve fault identification accuracy.

[0016] Preferably, the autonomous navigation and stability assurance strategy of the aforementioned safe return module includes:

[0017] (1) Calculate the maximum allowable tilt angle of the UAV after the propeller breaks based on the number of remaining motors and the attitude envelope linear function. ;

[0018] (2) Built-in vertical motion dynamics constraint model to ensure attitude stability under low speed conditions.

[0019] This invention also discloses a rapid diagnosis and fault-tolerant stable control method for propeller breakage faults in multi-rotor unmanned aerial vehicles based on the aforementioned system, comprising the following steps:

[0020] S1. The monitoring module preprocesses the command signals of each motor based on a discrete-time nonlinear recursive estimator to obtain the filtered thrust command estimate:

[0021] ,

[0022] in, The filtered thrust command estimate for the i-th motor in the k-th sampling period; This represents the original thrust command for the i-th motor during the k-th sampling period; The adaptive convergence factor. ; The operating frequency of the monitoring module; It is a nonlinear saturation function used to suppress anomalous amplitude disturbances. , The slope parameter;

[0023] S2. Calculate the sum of filtered thrust commands for all currently activated motors. and maximum thrust command Simultaneously record the motor index corresponding to the maximum thrust command. :

[0024] ,

[0025] in, Enable motor index set; The sum of the filtered thrust commands of all activated motors; Define the upper bound for the filtered thrust command of all activated motors; `sup(.)` is the motor index corresponding to the maximum thrust command; `sup(.)` is the supremum operator; `argsup(.)` is the supremum parameter index operator. Thrust for each motor;

[0026] S3. Calculate the unbalanced degree measure based on information entropy weighting. :

[0027] ,

[0028] in, It is a dimensionless measure of non-equilibrium. For set The base number, i.e., the number of motors in use; The average thrust command. ; For the thrust distribution variance, ; It is a natural exponential function;

[0029] S4. Computer system fault status It is determined by the following time-varying differential inclusion relation:

[0030] ,

[0031] in, It is a fault status indicator. =1 indicates a fault. =0 indicates normal; It is the time derivative of the fault status indicator; It is a differential inclusion operator; It is the lag coefficient; Describe the characteristic function of set A; This represents the upper threshold for the fault, set to 1.5. The lower threshold for the fault is set to 1.25. Indicates the number of motors in use;

[0032] S5. The control module detects a system fault. When the value is 1, the idle speed of the drone is first set to zero to ensure that if the drone loses power from one or more propellers, the other motors can respond quickly and supplement the power. Then, the power output is redistributed through the control matrix to ensure that the aircraft maintains its current altitude and attitude stability.

[0033] S6, the safe return-to-home module is based on a multi-constraint optimized autonomous navigation and stability assurance strategy. It automatically plans a safe return-to-home route by combining the current position, the return-to-home point position, and the set return-to-home altitude, and controls the aircraft to return to home along the route.

[0034] Preferably, the specific steps for the aforementioned control matrix to reallocate power output are as follows:

[0035] S5.1 Input UAV body configuration parameters , , , Calculate the force and torque distribution matrix of the UAV. :

[0036] ,

[0037] in, For the first The pulling coefficient of each motor, For the first Torque coefficient of each motor For the first The distance from each motor to the y-axis of the machine body. For the first The distance from each motor to the x-axis of the machine body. ,

[0038] S5.2 Obtain the control allocation matrix of the UAV control system through pseudo-inverse calculation. :

[0039] ,

[0040] S5.3 Calculate the control signal quantity for each motor :

[0041] ,

[0042] ,

[0043] ,

[0044] in, , , These are the torques acting on each axis of the drone's body. , , These are the forces acting on each axis of the drone's body; , , , The relationship is:

[0045] ,

[0046] in, For the output of the UAV attitude angular rate and yaw rate controller, This is the output of the drone's linear velocity controller;

[0047] S5.4 When the monitoring module detects that a motor has stopped, it will transfer the force and torque allocation matrix from step S5.1. Set the column corresponding to the motor to zero, update matrix F, and then continue to execute steps S5.2 and S5.3 to obtain the motor signal control quantity after the control reassignment.

[0048] More preferably, specifically in a hexacopter UAV, when motor 2 stops, the updated force and torque distribution matrix... for:

[0049] .

[0050] This method can also be extended to multi-motor failure scenarios, but its practical feasibility depends on the UAV's configuration (such as the number and layout of motors) and the severity of the failure. For example, in UAVs with redundant motor designs (such as hexacopters or octocopters), even if multiple motors fail, as long as the remaining motors can provide sufficient force and torque range, the fault-tolerant control algorithm of this invention can still achieve stable control by updating the force and torque allocation matrix. However, if there are too many failed motors, it may lead to insufficient control capability and an inability to fully restore stability.

[0051] More preferably, step S6 includes the following sub-steps:

[0052] S6.1. Based on the remaining number of motors, calculate the maximum permissible tilt angle of the UAV after propeller failure using the attitude envelope linear function. ;

[0053] S6.2 Calculate the vertical motion dynamics under stability constraints to ensure attitude stability under low speed conditions;

[0054] S6.3. Transform the return path generation into constrained optimization. Through nonlinear constrained optimization and stability algorithms, ensure the UAV's autonomous and safe return capability in the event of propulsion system failure, and achieve a balance between fault tolerance and navigation safety.

[0055] Preferably, the aforementioned maximum permissible tilt angle The calculation process is as follows:

[0056] ,

[0057] in, , is the nominal maximum attitude angle; Nr represents the number of fault-free motors, and N represents the total number of motors; The power index characterizes the impact of system redundancy. This is the fault attenuation coefficient, reflecting the effects of multiple faults.

[0058] More preferably, the aforementioned vertical motion dynamics equation is:

[0059] ,

[0060] ,

[0061] ,

[0062] in, The adaptive convergent gain function is related to the angular rate. The rate of change of altitude is no more than 0.5 m / s. These are the basic convergence rate coefficients; It is the gain coefficient for angular velocity adjustment; It is an sigmoid saturation function, ensuring a smooth and bounded output; It is the body's angular velocity vector; It is the Euclidean norm; It is the minimum angular velocity threshold for maintaining attitude stability; This is the normalized reference value for angular velocity; It is a hyperbolic tangent function, providing smooth saturation characteristics; h is the current flight altitude; It is the target altitude of the return point; This is a parameter for the highly convergent region. The dynamic equation, through a nonlinear feedback mechanism, ensures that the vertical velocity always meets the stability constraint. When the angular velocity norm approaches the minimum threshold, the convergence gain is automatically adjusted to prevent the system from entering the unstable region.

[0063] More preferably, the aforementioned constrained optimization is as follows:

[0064] ,

[0065] in, , which represents the three-dimensional trajectory of the drone; This is the instantaneous heading angle; The heading angle at the moment the fault occurred; Maintain the weighting factor for the heading;

[0066] The sufficient condition for a drone to return safely is guaranteed by the following set of simultaneous inequalities:

[0067] ,

[0068] ,

[0069] ,

[0070] in, The maximum flight speed set for the drone The minimum torque required to maintain stability; The maximum torque provided to the remaining motors; These are parameters related to air density. For the rotational inertia tensor of the UAV; , To design tolerance parameters, this module, through the aforementioned nonlinear constraint optimization and stability algorithm, ensures the UAV's autonomous and safe return capability in the event of propulsion system failure, achieving a balance between fault tolerance and navigation safety.

[0071] The advantages of this invention are:

[0072] (1) The propeller breakage fault rapid response and attitude stabilization system of the present invention is based on high-order nonlinear observation theory. By constructing a multimodal signal prediction and residual evaluation system, compared with the traditional single-mode detection, the present invention firstly uses a discrete-time nonlinear recursive estimator, combined with an adaptive convergence factor and a nonlinear saturation function, to effectively eliminate false signals such as electromagnetic interference and load fluctuations, ensuring that the thrust command estimation value is more accurate. Then, by calculating the total thrust, the maximum thrust and the corresponding motor characteristics, the abnormal motor can be quickly located, overcoming the problem of difficulty in identifying the faulty motor when multiple motors are in parallel. Furthermore, the information entropy weighted unbalanced degree measure can realize the quantitative assessment of the fault, ultimately providing a reliable fault basis for subsequent fault-tolerant control, realizing rapid and accurate identification of abnormal states of UAV power systems, avoiding fault missed detection / false detection, and adapting to dynamic changes under complex working conditions such as hovering and cruise.

[0073] (2) The monitoring module transmits the detected fault signals to the control module. When a specific motor failure is detected, the control module can immediately set the corresponding column in the allocation matrix to zero and recalculate the control signals of the remaining motors to quickly compensate for power loss and maintain attitude stability, laying the foundation for the subsequent safe operation (return to home / landing) of the UAV. The system can respond in a very short time when a motor or propeller fails, so that the power output can be optimally allocated according to the status of the remaining healthy motors, actively maintaining the required attitude and altitude. The whole process does not require manual intervention and is completed automatically by the control algorithm deployed in the flight control system, rather than passively responding to loss of control. It effectively prevents the UAV from losing control, rolling or even crashing due to power asymmetry. It is crucial for sudden failures, greatly reduces the risk of accidents, and solves the problem that traditional fixed control strategies are difficult to adapt to fault conditions.

[0074] (3) The present invention further realizes the safe return of the faulty UAV based on the autonomous navigation and stability assurance strategy of multi-constraint optimization, realizes the closed-loop control from fault detection to safe return, the process is coherent and highly operable, ensures the autonomous safe return capability of the UAV in the case of propulsion system failure, and realizes the unity of fault tolerance and navigation safety: it avoids the risk of crash caused by attitude over-limit or descent too fast during return, and realizes autonomous return after failure, which greatly improves the control stability and return safety under propeller failure. Attached Figure Description

[0075] Figure 1 This is a schematic diagram of the framework structure of the rapid diagnosis and fault-tolerant stability control system for multi-rotor UAV rotor breakage faults of the present invention;

[0076] Figure 2 This is a schematic diagram of the logic structure of the rapid diagnosis and fault-tolerant stable control method for multi-rotor UAV rotor breakage faults of the present invention;

[0077] Figure 3 This is a time history curve of the motor output PWM signal after a motor failure, as shown in the present invention.

[0078] Figure 4 This is a time history curve of the expected height and actual height after the motor of the present invention fails.

[0079] Figure 5 This is a time history curve of the expected vertical speed and the actual vertical speed after the motor of the present invention fails.

[0080] Figure 6 This is a time history curve of the expected roll angle and the actual roll angle after a motor failure according to the present invention.

[0081] Figure 7 This is a time history curve of the expected pitch angle and the actual pitch angle after a motor failure according to the present invention. Detailed Implementation

[0082] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0083] Example 1

[0084] See Figure 1 This embodiment discloses a rapid diagnosis and fault-tolerant stability control system for rotor breakage faults in multi-rotor unmanned aerial vehicles (UAVs). The system consists of a monitoring module, a control module, and a safe return-to-home module.

[0085] The monitoring module employs a motor fault detection method based on high-order nonlinear observation theory. By constructing a multimodal signal prediction and residual evaluation system, it achieves accurate identification of abnormal states in the UAV's power system and transmits the detected fault diagnosis signals to the control module. The control module then dynamically adjusts the control allocation matrix based on the airframe parameters and the received fault diagnosis signals, generates motor compensation signals, and sends a propeller failure flag to the safe return-to-home module. Finally, the safe return-to-home module calculates the maximum attitude angle of the UAV during flight using the number of faulty motors and the total number of motors, while limiting the maximum descent rate. Combining the current position, the return-to-home point position, and the set return-to-home altitude, it automatically plans a safe return-to-home route and controls the aircraft to return along the route.

[0086] The typical workflow of this system is as follows: when the monitoring module detects a fault in the motor power system through flight status feedback, it sends a fault diagnosis signal to the control module; the control module automatically performs a control matrix reconstruction process based on the fault diagnosis signal, outputs compensation commands to maintain attitude stability, and notifies the safe return module to update the number of remaining motors; the safe return module calculates the maximum tilt angle and feeds back the attitude constraints to the control module, enabling the UAV to safely return and land, ending the flight mission.

[0087] Example 2

[0088] See Figure 2 This embodiment discloses the application of the rapid diagnosis and fault-tolerant stable control method for propeller breakage based on the aforementioned system in a hexarotor unmanned aerial vehicle (UAV). The specific method includes the following steps:

[0089] S1. The monitoring module preprocesses the command signals of each motor based on a discrete-time nonlinear recursive estimator to obtain the filtered thrust command estimate:

[0090] ,

[0091] in, The filtered thrust command estimate for the i-th motor in the k-th sampling period; This represents the original thrust command for the i-th motor during the k-th sampling period; The adaptive convergence factor. ; The operating frequency of the monitoring module; It is a nonlinear saturation function used to suppress anomalous amplitude disturbances. , This is the slope parameter.

[0092] S2. Calculate the sum of filtered thrust commands for all currently activated motors. and maximum thrust command Simultaneously record the motor index corresponding to the maximum thrust command. :

[0093] ,

[0094] in, Enable motor index set; The sum of the filtered thrust commands of all activated motors; Define the upper bound for the filtered thrust command of all activated motors; `sup(.)` is the motor index corresponding to the maximum thrust command; `sup(.)` is the supremum operator; `argsup(.)` is the supremum parameter index operator. The thrust for each motor.

[0095] S3. Calculate the unbalanced degree measure based on information entropy weighting. :

[0096] ,

[0097] in, It is a dimensionless measure of non-equilibrium. For set The base number, i.e., the number of motors in use; The average thrust command. ; For the thrust distribution variance, ; It is a natural exponential function.

[0098] S4. Computer system fault status It is determined by the following time-varying differential inclusion relation:

[0099] ,

[0100] in, It is a fault status indicator. =1 indicates a fault. =0 indicates normal; It is the time derivative of the fault status indicator; It is a differential inclusion operator; It is the lag coefficient; Describe the characteristic function of set A; This represents the upper threshold for the fault, set to 1.5. The lower threshold for the fault is set to 1.25. This indicates the number of motors in use.

[0101] S5. The control module detects a system fault. When the value is 1, the drone's idle speed is first set to zero to ensure that if the drone loses power from one or more propellers, the other motors can respond quickly and replenish power. Then, the power output is redistributed through the control matrix to ensure that the aircraft maintains its current altitude and attitude stability.

[0102] Specifically, the steps for the control matrix to reallocate power output are as follows:

[0103] S5.1 Input UAV body configuration parameters , , , Calculate the force and torque distribution matrix of the UAV. :

[0104] ,

[0105] in, For the first The pulling coefficient of each motor, For the first Torque coefficient of each motor For the first The distance from each motor to the y-axis of the machine body. For the first The distance from each motor to the x-axis of the machine body. .

[0106] Specifically, in this embodiment, if motor 2 malfunctions and stops, the updated force and torque distribution matrix will... for:

[0107] .

[0108] S5.2 Obtain the control allocation matrix of the UAV control system through pseudo-inverse calculation. :

[0109] .

[0110] S5.3 Calculate the control signal quantity for each motor :

[0111] ,

[0112] ,

[0113] ,

[0114] in, , , These are the torques acting on each axis of the drone's body. , , These are the forces acting on each axis of the drone's body; , , , The relationship is:

[0115] ,

[0116] in, For the output of the UAV attitude angular rate and yaw rate controller, This is the output of the drone's linear velocity controller.

[0117] S5.4 When the monitoring module detects that a motor has stopped, it will transfer the force and torque allocation matrix from step S5.1. Set the column corresponding to the motor to zero, update matrix F, and then continue to execute steps S5.2 and S5.3 to obtain the motor signal control quantity after the control reassignment.

[0118] S6, the safe return-to-home module is based on a multi-constraint optimized autonomous navigation and stability assurance strategy. It automatically plans a safe return-to-home route by combining the current position, the return-to-home point position, and the set return-to-home altitude, and controls the aircraft to return to home along the route.

[0119] Specifically, step S6 includes the following sub-steps:

[0120] S6.1. Based on the remaining number of motors, calculate the maximum permissible tilt angle of the UAV after propeller failure using the attitude envelope linear function. :

[0121] ,

[0122] in, , is the nominal maximum attitude angle; Nr represents the number of fault-free motors, and N represents the total number of motors; The power index characterizes the impact of system redundancy. This is the fault attenuation coefficient, reflecting the effects of multiple faults.

[0123] S6.2 Calculate the vertical motion dynamics under stability constraints to ensure attitude stability under low-speed conditions:

[0124] ,

[0125] ,

[0126] ,

[0127] in, The adaptive convergent gain function is related to the angular rate. The rate of change of altitude is no more than 0.5 m / s. These are the basic convergence rate coefficients; It is the gain coefficient for angular velocity adjustment; It is an sigmoid saturation function, ensuring a smooth and bounded output; It is the body's angular velocity vector; It is the Euclidean norm; It is the minimum angular velocity threshold for maintaining attitude stability; This is the normalized reference value for angular velocity; It is a hyperbolic tangent function, providing smooth saturation characteristics; h is the current flight altitude; It is the target altitude of the return point; It is a scale parameter of a highly convergent region.

[0128] S6.3. Transform the return path generation into constrained optimization. Through nonlinear constrained optimization and stability algorithms, ensure the UAV's autonomous and safe return capability in the event of propulsion system failure, and achieve a balance between fault tolerance and navigation safety.

[0129] Specifically, constrained optimization is as follows:

[0130] ,

[0131] in, , which represents the three-dimensional trajectory of the drone; This is the instantaneous heading angle; The heading angle at the moment the fault occurred; Maintain the weighting factor for the heading;

[0132] The sufficient condition for a drone to return safely is guaranteed by the following set of simultaneous inequalities:

[0133] ,

[0134] ,

[0135] ,

[0136] in, The minimum torque required to maintain stability; The maximum torque provided to the remaining motors; These are parameters related to air density. For the rotational inertia tensor of the UAV; , To design tolerance parameters.

[0137] The method of the present invention was verified as follows: Figure 3 This is the time history curve of the motor output PWM signal after a motor failure. Figure 4 This is a time history curve showing the expected height and actual height after a motor failure. Figure 5 The graphs show the time history curves of expected and actual vertical speeds after a motor failure. Combining these three graphs, we can see that in this embodiment, after the failure of motor number 2, the remaining motors of the multi-rotor UAV will quickly increase and adjust, and both altitude and vertical speed can be maintained stably. Figure 6 The curves show the time history of the expected roll angle and the actual roll angle after a motor failure. Figure 7 The graphs show the time history curves of the expected pitch angle and the actual pitch angle after a motor failure. Combining these two graphs, we can see that after a motor failure, the remaining motors of the multi-rotor UAV will adjust quickly, and both the pitch angle and roll angle can remain stable.

[0138] In summary, the propeller breakage fault rapid response and attitude stabilization system and method of this invention can respond in a very short time when a motor or propeller fails, enabling the power output to be optimally allocated according to the status of the remaining healthy motors, actively maintaining the required attitude and altitude. The entire process requires no manual intervention and is completed automatically by the control algorithm deployed in the flight control system, rather than passively responding to loss of control. This effectively prevents the UAV from losing control, tumbling, or even crashing due to power asymmetry, which is crucial for sudden faults, greatly reducing the risk of accidents and solving the problem that traditional fixed control strategies are difficult to adapt to fault conditions. Furthermore, it achieves closed-loop control from fault detection to safe return, with a coherent process and strong operability, ensuring the UAV's autonomous and safe return capability in the event of propulsion system failure. It achieves a balance between fault tolerance and navigation safety: avoiding the risk of crash due to attitude exceeding limits or excessive descent during return, and enabling autonomous return after a fault, significantly improving control stability and return safety under propeller breakage faults.

[0139] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any way, and all technical solutions obtained by equivalent substitution or equivalent transformation fall within the protection scope of the present invention.

Claims

1. A multi-rotor unmanned aerial vehicle broken propeller fault rapid diagnosis and fault-tolerant stable control method, characterized in that, The method comprises the following steps: S1, the monitoring module pre-processes each motor command signal based on a discrete-time nonlinear recursive estimator to obtain a filtered thrust command estimation value: ， wherein, is the filtered thrust command estimate for the ith motor in the kth sample period; is the raw thrust command for the ith motor in the kth sample period; is an adaptive convergence factor, ; is the monitoring module operating frequency; is a non-linear saturation function used to dampen amplitude anomalies, , is a slope parameter; S2, calculate the sum of all current filter thrust instructions of enabled motors and the maximum thrust instruction while recording the motor index corresponding to the maximum thrust instruction : ， wherein, is the set of enabled motor indices; is the sum of filtered thrust commands for the enabled motors; is the upper bound of filtered thrust commands for all enabled motors; is the motor index corresponding to the maximum thrust command; sup(.) is the upper bound operator; argsup(.) is the upper bound argument index operator; is the thrust of each motor; S3、calculating a non-equilibrium degree measure based on information entropy weighting : ， wherein is a dimensionless unbalance measure; is the cardinality of the set , i.e. the number of enabled motors; is the thrust command mean, ; is the thrust distribution variance, ; is the natural exponential function; S4, computing system failure state is determined by the following time-varying differential inclusion: ， wherein is a fault status flag, = 1 indicates a fault, = 0 indicates normal; is a time derivative of the fault status flag; is a differential inclusion operator; is a hysteresis coefficient; denotes an indicator function of the set A; denotes a fault upper threshold, taken as 1.5; denotes a fault lower threshold, taken as 1.25; denotes the number of active motors; S5、control module detects system fault state When the value is 1, first set the idle value of the unmanned aerial vehicle to zero to ensure that when the unmanned aerial vehicle loses the power of one or more propellers, other motors can quickly respond and supplement power, and then redistribute the power output through the control matrix to ensure that the aircraft maintains the current height and attitude stability; S6, the safe return module automatically plans a safe return route based on a multi-constraint optimization autonomous navigation and stability guarantee strategy, in combination with the current position and the return point position and the set return height, and controls the aircraft to return according to the route.

2. The method of claim 1, wherein, The specific steps of the control matrix re-distributing the power output are as follows: S5.1, inputting unmanned aerial vehicle body configuration parameters , , , , calculating the force and moment distribution matrix of the unmanned aerial vehicle : ， wherein, is the pull coefficient of the th motor, is the torque coefficient of the th motor, is the distance from the th motor to the body y-axis, is the distance from the th motor to the body x-axis, , S5.2, obtain the control allocation matrix of the UAV control system by pseudo-inverse calculation : ， S5.3, determining the control signal quantity for each motor : ， ， ， wherein, , , are the moments acting on the respective axes of the UAV body, , , are the forces acting on the respective axes of the UAV body; , , , the relationship is: ， wherein, is the output of the UAV attitude angle rate and yaw angle rate controller, is the output of the UAV linear velocity controller; S5.4, when the monitoring module detects that a certain motor is stalled, the force and torque distribution matrix in step S5.1 is set to zero in the column corresponding to the motor, the matrix F is updated, and then steps S5.2 and S5.3 are continued to obtain the motor signal control quantity after the control distribution is re-distributed. S5.4, when the monitoring module detects that a certain motor is stalled, the force and torque distribution matrix in step S5.1 is set to zero in the column corresponding to the motor, the matrix F is updated, and then steps S5.2 and S5.3 are continued to obtain the motor signal control quantity after the control distribution is re-distributed.

3. The method of claim 2, wherein, When the 2nd motor is stopped, the updated force and torque distribution matrix is: 。 4. The method of claim 1, wherein, The step S6 comprises the following sub-steps: S6.1, in combination with the remaining motor quantity, the maximum allowed tilt angle of the UAV after the propeller is broken is calculated according to the attitude envelope linear function ; S6.2, calculating the vertical motion dynamics under the stability constraint to ensure the attitude stability in the low speed working condition; S6.3, converting the return path generation into a constrained optimization, and ensuring the autonomous safe return ability of the unmanned aerial vehicle under the propulsion system failure condition through a nonlinear constraint optimization and stability algorithm, so as to realize the unification of fault tolerance and navigation safety.

5. The method of claim 4, wherein, the maximum allowed tilt angle The calculation process is as follows: ， wherein, is the nominal maximum attitude angle; Nr denotes the number of fault-free motors, N denotes the total number of motors; is the power exponent, which characterizes the influence of the system redundancy; is the fault attenuation coefficient, which reflects the multiple fault effect.

6. The method of claim 4, wherein, The vertical motion dynamics equation is: ， ， ， wherein, is an angular rate dependent adaptive convergence gain function; is a height rate of change, maximum not exceeding 0.5 m / s; is a base convergence rate coefficient; is an angular velocity regulation gain coefficient; is a sigmoid saturation function, ensuring smooth bounded output; is the body angular rate vector; is the Euclidean norm; is a minimum angular velocity threshold to maintain attitude stability; is an angular velocity normalized reference value; is a hyperbolic tangent function, providing a smooth saturation characteristic; h is the current flight height; is the home point target height; is a height convergence region scale parameter.

7. The method of claim 4, wherein, The constrained optimization is: ， wherein, is a three-dimensional trajectory of the UAV; is an instantaneous yaw angle; is a yaw angle at the time of failure occurrence; is a yaw keeping weight coefficient; The sufficient condition for the safe return of the unmanned aerial vehicle is ensured by the following simultaneous inequalities: ， ， ， wherein, maximum flight speed of the drone, minimum torque required to maintain stability; maximum torque provided to the remaining motors; air density related parameters; moment of inertia tensor of the drone; , design tolerance parameters.

Images