Unmanned aerial vehicle motor rotating speed control method based on active disturbance rejection

Abstract

The invention provides an unmanned aerial vehicle motor rotating speed control method based on active disturbance rejection. The method comprises the following steps: 1) measuring the rotating speed of a motor by adopting a high-precision rotating speed sensor based on a quantum effect; 2) designing a tracking differentiator TD to smooth an expected rotating speed, and outputting a rotating speed tracking signal and a rotating speed tracking signal change rate; 3) introducing a dynamic weight matrix to form a dynamic weight expansion state observer DWESO, and taking an output state variable estimation value and a total disturbance estimation value as the input of the NLSEF; 4) automatically searching the optimal combination of the gain parameters of the DWESO by adopting an ALA (artificial Louver Mouse Algorithm), so that the optimal state variable estimation value and the optimal total disturbance estimation value are output to the NLSEF; compared with the prior art, the method for controlling the rotating speed of the unmanned aerial vehicle has the advantages that the dynamic response speed, the anti-jamming capability and the parameter adaptability of the system are remarkably improved, and the flight performance and the robustness of the unmanned aerial vehicle are improved.

Description

Technical Field

[0001] The present invention relates to the field of UAV motor speed control, and a method for controlling the speed of a UAV motor based on active disturbance rejection. Background Art

[0002] Due to its advantages such as low cost, high mobility, and zero personnel risk, UAVs are widely used in military, logistics, rescue and other fields. However, its strong non-linear and multi-variable coupled dynamic characteristics and complex external disturbances, such as sudden air flow changes and extreme environments, pose severe challenges to motor speed control. Traditional control systems rely on linearized models and empirical parameter tuning, facing problems such as insufficient measurement accuracy (traditional sensors are vulnerable to electromagnetic interference and temperature drift) and weak anti-disturbance ability (PID controllers are difficult to compensate for composite disturbances). Existing improvement schemes such as sliding mode control or fuzzy algorithms still have limitations such as limited parameter adaptability, complex calculations, or dependence on accurate models.

[0003] Therefore, there is an urgent need for a comprehensive control method that integrates high-precision sensing, dynamic disturbance compensation, and intelligent optimization to improve the control accuracy and robustness in complex scenarios. Summary of the Invention

[0004] Object of the Invention: Aiming at the problems in the background art, the present invention provides a method for controlling the speed of a UAV motor based on active disturbance rejection, aiming to solve the deficiencies of traditional UAV motor speed control systems in measurement accuracy, anti-interference ability, and controller parameter tuning. Through multi-technology collaborative optimization, the present invention significantly improves the system's dynamic response speed, anti-interference ability, and parameter self-adaptability, thereby improving the flight performance and robustness of the UAV.

[0005] Technical Solution: The present invention discloses a method for controlling the speed of a UAV motor based on active disturbance rejection, including the following steps:

[0006] A high-precision speed sensor based on the quantum effect is used to detect the actual speed of the UAV motor, and a tracking differentiator TD is used to smooth the desired speed. The output speed tracking signal and the change rate of the speed tracking signal are used as the inputs of a dynamic weight extended state observer DWESO;

[0007] The dynamic weight extended state observer DWESO introduces a dynamic weight matrix on the basis of the traditional ESO; an artificial lemming algorithm ALA is used to automatically search for the optimal combination of the gain parameters of the dynamic weight extended state observer DWESO, and the estimated values of the state variables and the estimated total disturbance output by the optimized dynamic weight extended state observer DWESO are used as the inputs of a non-linear state error feedback NLSEF;

[0008] The nonlinear state error feedback NLSEF calculates the ideal control quantity without disturbance and feedforward compensation according to the input of tracking differentiator TD and dynamic weighted extended state observer DWESO;

[0009] Finally, the final control quantity is calculated based on the ideal control quantity output by the nonlinear state error feedback NLSEF and the total disturbance estimate output by the dynamic weighted extended state observer DWESO, so as to achieve precise control of the motor speed.

[0010] Furthermore, a high-precision speed sensor based on quantum effects detects the actual speed of the drone motor. Using the magnetic sensitivity of diamond nitrogen vacancy (NV) color centers, the speed measurement is achieved by detecting the magnetic field changes of a magnetic marker on the rotating shaft of the drone motor. The steps are as follows:

[0011] Step 1.1: N pairs of magnetic poles are evenly arranged on the surface of the UAV motor shaft. Each rotation generates N periodic changes in the magnetic field. The peak value of the magnetic field generated by a single magnetic pole is B. peak , must satisfy B peak >1μT, 1μT=10 -6 Tesla;

[0012] Step 1.2: Irradiate the NV color center with a 532 nm laser to excite fluorescence at a frequency of f MW The microwave is used to measure the magnetic field by optical detection magnetic resonance technology. The magnetic field change ΔB(t) causes the fluorescence intensity to be periodically modulated with a modulation frequency f sensor The formula for the energy level splitting frequency difference caused by magnetic field changes is related to the rotation speed and is as follows:

[0013]

[0014] Among them, Δf is the energy level splitting frequency difference, B is the magnetic field strength, γ e is the electron gyromagnetic ratio, which determines the sensitivity of the magnetic field to the energy level splitting and is taken as 28 GHz / T;

[0015] Step 1.3: Convert the fluorescence signal into a digital frequency f using a lock-in amplifier and a frequency counter sensor , according to the detected frequency f sensor And the number of magnetic poles N, calculate the UAV motor speed ω, the formula is as follows:

[0016]

[0017] By leveraging the high sensitivity of quantum effects to tiny changes, more accurate and stable UAV motor speed measurements can be achieved.

[0018] Furthermore, the discrete form formula of the tracking differentiator TD is as follows:

[0019]

[0020] Among them, fhan(·) is the fastest control law function, v is the motor's desired speed signal, v1 is the desired speed tracking signal, v2 is the change rate of the desired speed tracking signal, r is the speed tracking factor that controls the tracking speed of the tracking differentiator, h is the filtering factor that determines the quality of the filtering effect, and T is the sampling step of the system, which affects the discretization accuracy of the tracking differentiator.

[0021] Furthermore, the dynamic weight matrix adjusts the weights of state estimation and disturbance estimation in control in real time according to the state of the system. When the disturbance is large, the weight of disturbance estimation is increased to quickly suppress the disturbance; when the system is stable, the weight of state estimation is increased to improve the control accuracy. The formula of the dynamic weight matrix is as follows:

[0022]

[0023] Among them, α is the weight coefficient of state z2, which is dynamically adjusted according to the speed error e, and φ is the weight coefficient of the total disturbance estimation value z3, which is dynamically adjusted according to the change rate of the speed error dynamically adjusted, α0, φ0 ∈ (0, 1], are the initial weight coefficients, and γ, σ > 0, are the attenuation factors;

[0024] The formula of the discrete form of DWESO is as follows:

[0025]

[0026] Among them, fal(·) is a non-linear function used for the state update of ESO. z1 is the estimated speed value, z2 is the estimated change rate of speed, representing the dynamic characteristics of the system, z3 is the total disturbance estimation value, representing the uncertainty of the system, e is the speed error, ω is the actual speed of the motor measured by a high-precision sensor, u is the control quantity, representing the control command, b is the amplification factor of the control quantity, α1 is the feedback strength of the main control state error, affecting the dynamic response speed, α2 is the smoothness of the main control error differentiation, affecting the anti-noise ability, β1, β2, β3 are the gain parameters of ESO, β1 affects the convergence speed of the motor speed estimation, β2 affects the convergence speed of the motor speed change rate estimation, β3 affects the convergence speed of the total disturbance estimation value, and h is the sampling step;

[0027] Furthermore, the artificial lemming algorithm ALA is used to automatically search for the optimal combination of the gain parameters of the dynamic weight extended state observer DWESO. Its steps include:

[0028] Step 2.1: Population initialization, randomly generate initial candidate solutions for combinations of multiple DWESO gain parameters β1, β2, β3;

[0029] Step 2.2: Iteratively update the gain parameters β1, β2, and β3 of DWESO by simulating four strategies of lemmings' long-distance migration (i.e., global exploration), burrowing (i.e., local exploration), foraging (i.e., spiral search), and escaping from natural enemies (i.e., flight escape).

[0030] Step 2.3: Dynamically adjust the weights of the algorithm in the global exploration and local development stages to avoid premature convergence and ensure the optimization quality of the UAV motor speed control parameters.

[0031] Step 2.4: Select the optimal DWESO gain parameter combination with the integral time absolute error (ITAE) as the performance index to optimize the UAV motor speed control performance.

[0032] Step 2.5: After reaching the maximum number of iterations, output the global optimal solution.

[0033] Furthermore, the nonlinear state error feedback (NLSEF) is used to dynamically adjust the control quantity and directly compensate for the disturbance effect. The control law formula is as follows:

[0034]

[0035] where k1 and k2 are feedback gains that determine the weight of the error term, u0 is the control input of the uncompensated disturbance, which is the ideal control quantity calculated by the controller based on the error between the desired trajectory and the actual state, and u 终 is the final control input after compensating for the disturbance, and precise control of the UAV motor speed is achieved by calculating u 终 .

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] (1) The present invention introduces a high-precision rotation speed sensor based on the quantum effect, significantly improving the rotation speed measurement accuracy and stability, and laying a foundation for high-precision rotation speed control; (2) introducing a dynamic weight matrix to solve the inherent contradiction between the fast response and anti-interference ability of the traditional ESO, and real-time adjusting the priority of state and disturbance estimation, improving the tracking accuracy and anti-disturbance ability of the motor speed; (3) using ALA to optimize the gain parameters β1, β2, and β3 of DWESO, improving the disturbance estimation accuracy of DWESO and enhancing the parameter adaptive ability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is the system block diagram of the present invention;

[0039] Figure 2 is the system flowchart of the present invention;

[0040] Figure 3 is the ALA flowchart;

[0041] Figure 4 It is a performance comparison chart of DWESO and traditional ESO;

[0042] Figure 5 It is a performance comparison chart of ALA and GA. Specific implementation manner

[0043] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0044] This embodiment provides a method for controlling the rotational speed of a drone motor based on active disturbance rejection, including the following steps:

[0045] Step 1: Use a high-precision rotational speed sensor based on the quantum effect to take the actually measured rotational speed of the drone motor as the input of the ESO.

[0046] Adopt the magnetic sensitivity characteristic of the diamond NV color center to realize rotational speed measurement by detecting the magnetic field change of the magnetic mark on the rotating shaft of the drone motor. The steps are as follows:

[0047] Step 1.1, Magnetic mark installation: Uniformly arrange N pairs of magnetic poles on the surface of the rotating shaft of the drone motor. Each rotation generates N periodic magnetic field changes, and the magnetic field peak value generated by a single magnetic pole is B peak , which needs to satisfy B peak > 1 μT, 1 μT = 10 -6 Tesla;

[0048] Step 1.2, Quantum sensor magnetic field detection: Irradiate the NV color center with a 532 nm laser to excite fluorescence, apply microwaves with a frequency of f MW , and measure the magnetic field through optically detected magnetic resonance technology. The magnetic field change ΔB(t) causes periodic modulation of the fluorescence intensity. The modulation frequency f sensor is related to the rotational speed. The formula for the energy level splitting frequency difference caused by the magnetic field change is as follows:

[0049]

[0050] where, Δf is the energy level splitting frequency difference, B is the magnetic field intensity, and γ e is the electron gyromagnetic ratio, which determines the sensitivity of the magnetic field to energy level splitting, and is usually taken as 28 GHz / T;

[0051] Step 1.3, Frequency signal extraction and rotational speed calculation: Convert the fluorescence signal into a digital frequency f sensor through a lock-in amplifier and a frequency counter. According to the detected frequency f sensor and the number of magnetic pole pairs N, calculate the rotational speed ω of the drone motor. The formula is as follows:

[0052]

[0053] Utilize the high sensitivity of the quantum effect to minute changes to achieve higher-precision and more stable measurement of the rotational speed of the UAV motor.

[0054] Step 2: Use TD to smooth the desired rotational speed, and take the output rotational speed tracking signal and the change rate of the rotational speed tracking signal as the inputs of the NLSEF.

[0055] Introduce the TD-smoothed desired rotational speed signal, which effectively reduces the influence of noise on the control system. The discrete form formula of TD is as follows:

[0056]

[0057] where fhan(·) is the fastest control law function, v is the desired rotational speed signal of the motor, v1 is the desired rotational speed tracking signal, v2 is the change rate of the desired rotational speed tracking signal, r is the rotational speed tracking factor, which controls the tracking speed of the tracking differentiator, h is the filtering factor, which determines the filtering effect, and T is the sampling step of the system, which affects the discretization accuracy of the tracking differentiator.

[0058] Step 3: Introduce a dynamic weight matrix on the basis of the traditional ESO, and take the estimated values of the state variables and the estimated value of the total disturbance output by the DWESO as the inputs of the NLSEF. Use ALA to automatically search for the optimal combination of the gain parameters of the DWESO, so that it outputs the optimal estimated values of the state variables and the total disturbance to the NLSEF.

[0059] The formula of the discrete form of the traditional ESO is as follows:

[0060]

[0061] where fal(·) is a nonlinear function used for the state update of the ESO, z1 is the estimated value of the rotational speed, z2 is the estimated value of the change rate of the rotational speed, representing the dynamic characteristics of the system, z3 is the estimated value of the total disturbance, representing the uncertainty of the system, e is the rotational speed error, ω is the actual rotational speed of the motor, measured by the high-precision sensor in Claim 1, u is the control quantity, representing the control instruction, b is the amplification factor of the control quantity, α1 is the feedback strength of the main control state error, which affects the dynamic response speed, α2 is the smoothness of the differentiation of the main control error, which affects the anti-noise ability, β1, β2, and β3 are the gain parameters of the ESO, β1 affects the convergence speed of the estimation of the motor rotational speed, β2 affects the convergence speed of the estimation of the change rate of the motor rotational speed, β3 affects the convergence speed of the estimation of the total disturbance value, and h is the sampling step;

[0062] Traditional ESOs cannot dynamically adjust the priority of state estimation and disturbance estimation and have difficulty striking a balance between rapid response and anti-interference. However, the dynamic weight matrix can adjust the weights of state estimation and disturbance estimation in control in real time according to the system state. When the disturbance is large, the weight of disturbance estimation is increased to quickly suppress the disturbance; when the system is stable, the weight of state estimation is increased to improve control accuracy. This dynamic adjustment can improve the flexibility and adaptability of control. The formula of the dynamic weight matrix is as follows:

[0063]

[0064] Among them, α is the weight coefficient of state z2, which is dynamically adjusted according to the speed error e, and φ is the weight coefficient of the total disturbance estimate z3, which is adjusted according to the speed error change rate. Dynamic adjustment, α0, φ0∈(0,1], are the initial weight coefficients, γ, σ>0, are the attenuation factors;

[0065] The discrete form of DWESO is as follows:

[0066]

[0067] Set parameter values: β1=3, β2=100, β3=300, w=1000, α0=0.8, φ0=0.6, γ=0.5, σ=0.5, b=2, α1=0.5, α2=0.25, h=0.001, such as Figure 4 As shown in the figure, the error curve of the DWESO decreases faster before the disturbance is introduced, indicating that it can approach the target speed more quickly in the initial tracking stage. After the disturbance is introduced, the error of the traditional ESO fluctuates greatly and is significantly affected by the disturbance, with the error remaining at a high level for a long time. The error of the DWESO is significantly smaller and it can recover from the disturbance more quickly, reflecting its better ability to suppress speed errors under complex working conditions. The error of the traditional ESO deteriorates rapidly after the disturbance is introduced, with a large slope and violent fluctuations in the curve, indicating that its anti-interference response is slow. The error curve of the dynamic weight ESO has a smaller slope and a smoother convergence process, indicating that it can respond to disturbances more promptly. By dynamically adjusting the weights, error convergence is accelerated and fluctuations in the transition process are reduced. The fast response and anti-interference performance of the dynamic weight ESO are significantly better than those of the traditional ESO, and its control accuracy is improved by about 47.7% compared with the traditional ESO.

[0068] Use ALA to automatically search for the optimal combination of DWESO gain parameters:

[0069] 1) Initialize the population

[0070] The artificial lemming algorithm is a swarm-based algorithm that requires initializing the positions of all search agents before entering the iterative process. The set of all initial candidate solutions is a matrix consisting of N (swarm size) rows and Dim (number of parameters) columns between the upper and lower bounds of the given problem. The number of parameters that DWESO needs to optimize is 3. Therefore, the vector has the following formula:

[0071]

[0072] where z i,1 = β1, z i,2 = β2, z i,3 = β3;

[0073] Randomly generate the initial population, and the formula is as follows:

[0074] z i,j = LB j + rand × (UB j - LB j ) (8)

[0075] where rand ∈ [0, 1] is used to initialize the parameters, i = 1, 2,..., N, j = 1, 2, 3, LB j and UB j are the upper and lower bounds of the parameters β1, β2, and β3 respectively;

[0076] The distribution range of the initial parameters needs to cover the dynamic characteristics of the UAV motor under different loads and speeds to ensure that the algorithm can quickly search for the optimal parameter combination in subsequent iterations and improve the speed tracking accuracy.

[0077] 2) Fitness function

[0078] Select ITAE as the performance index of the DWESO gain parameter, and the formula is as follows:

[0079]

[0080] where the smaller the fitness value J, the lower the speed overshoot, the shorter the disturbance response time, the smaller the steady-state error, and the better the control performance.

[0081] 3) Behavior modeling and iterative update

[0082] Step 1, long-distance migration (global exploration):

[0083] Simulate the population diffusion, combining Brownian motion and random direction adjustment, and the formula is as follows:

[0084]

[0085] Among them, F is a random direction vector, whose value is 1 or -1, used to introduce randomness to the search direction and enhance the global exploration ability of the algorithm. is the current DWESO gain parameter of the i-th individual. is the ESO gain parameter of the current optimal individual. is the DWESO gain parameter of a randomly selected individual. is a random weight vector, used to balance the influence of the optimal solution direction and the random individual direction. is the Brownian motion vector. Brownian motion is an irregular and continuous random motion, providing random perturbations for the position update of the search agent to avoid the algorithm falling into local optima.

[0086] The drone motor may face strong nonlinear perturbations. Global exploration ensures that the gain parameters of DWESO can adapt to complex working conditions and enhance the anti-perturbation ability.

[0087] Step 2, Digging holes (local exploration):

[0088] Update the position based on the current optimal solution and random individual perturbations. The formula is as follows:

[0089]

[0090] Among them, is the DWESO gain parameter of a randomly selected individual, and L is the periodic perturbation factor to avoid local stagnation.

[0091] For the slight fluctuations in the motor speed, optimize the parameters through local exploration to ensure more accurate estimation of the state and perturbations by DWESO.

[0092] Step 3, Foraging (spiral local search):

[0093] Spiral local search, the formula is as follows:

[0094]

[0095] Among them, spiral is the path form of the search, achieving efficient search in the local area by imitating the foraging behavior of lemmings. radius is the radius of the spiral search, dynamically adjusted by the distance between the optimal solution and the current position, used to control the scope of local search and balance the exploitation ability of the algorithm.

[0096] When the motor speed approaches stability, fine-tune the parameters through spiral search to reduce the steady-state error and improve the control accuracy.

[0097] Step 4, Avoiding natural enemies (Levy flight escape):

[0098] In the final stage, the modeling focuses on the avoidance and protection behaviors of lemmings when encountering danger. Caves are the shelters for lemmings. Once they detect an enemy, lemmings will use their extraordinary running ability to escape back to the caves. At the same time, lemmings will also make deceptive movements to evade the pursuit of predators. The formula is as follows:

[0099]

[0100] Among them, G is the escape coefficient, which controls the step size scaling of Levy flight. As the iteration progresses, the escape step size gradually decreases, and the algorithm shifts from global exploration to local exploitation. Levy flight is a random process with the characteristic of long-distance jumps, which is used to generate random step sizes. σ is the scale parameter, and u and v are random numbers obeying the normal distribution, which are used to calculate the step size of Levy flight. T max is the maximum number of iterations allowed by the algorithm. β ∈ (1, 2) is the shape parameter of the Levy distribution, which determines the distribution characteristics. The smaller β is, the higher the probability of long step sizes appearing, and the stronger the global exploration ability of the algorithm. Γ(·) is the gamma function, which is used to handle mathematical calculations involving integrals.

[0101] The long-distance jump characteristic of Levy flight helps lemmings quickly escape from local areas, simulating the sudden long-distance escape behavior when avoiding natural enemies. When the UAV encounters a sudden strong disturbance, Levy flight helps the parameters to be quickly adjusted to avoid system instability.

[0102] 4) Energy regulation mechanism

[0103] In the artificial lemming algorithm, these four search strategies are closely related to the energy level of the lemmings themselves. In the initial stage, lemmings often participate in the exploration process to find promising areas, while in the later stage of the search, they tend to implement local exploitation to obtain excellent global optimal solutions. To maintain a stable balance between exploration and exploitation, we design an energy factor that gradually decreases during the iteration process. When lemmings have enough energy, they will selectively migrate or dig holes; otherwise, they will forage and avoid predators. The calculation formula of the energy factor is as follows:

[0104]

[0105] When E(t) > 1, exploration behavior (migration or digging holes) is executed; otherwise, exploitation behavior (foraging or escaping) is executed.

[0106] The UAV motor needs to respond quickly (global exploration) during the startup stage, while fine control is required during stable operation (local exploitation). The energy regulation highly matches the control requirements.

[0107] 5) Termination and output

[0108] After reaching the maximum number of iterations, the global optimal solution is output to ensure that the parameter optimization process of the UAV does not exceed the limit under the real-time requirements such as emergency rescue and disaster relief, taking into account both rapid response and control accuracy.

[0109] Set parameter values: the population size is 50, the maximum number of iterations is 50, the dimension of the optimized parameters is 3, β1 ∈ (1, 30), β2 ∈ (1, 300), β3 ∈ (1, 3000), triggering the long-distance migration behavior (global exploration) of lemmings with a probability of 60% and triggering the burrowing behavior (local exploration) with a probability of 40%. By adopting the Levy flight mechanism and dynamic energy regulation, ALA significantly improves the global search ability. At the same time, the combination of the exploration and exploitation stage mechanisms of ALA enables it to quickly break through local minima in the strong exploration stage, and use spiral search to further refine the solution space exploration in the exploitation stage, and it performs more excellently in the convergence speed to the global optimal solution. In contrast, although the traditional GA ensures the diversity of the population through roulette wheel selection, arithmetic crossover and Gaussian mutation, due to its relatively single search mechanism, it performs poorly in the speed and accuracy of converging to the global optimal solution. Especially when dealing with complex multi-modal optimization problems, GA is prone to fall into local optima. As Figure 5 shown, ALA shows a relatively stable and continuously improving trend throughout the optimization process. The convergence speed of ALA is significantly better than that of GA, and the fitness value is about 18.8% higher than that of GA, which fully proves the superiority of the artificial lemming algorithm in complex optimization problems. Through simulation, the optimal gain parameters β1 = 20.87, β2 = 98.6, β3 = 189.1 are obtained.

[0110] Step 4: NLSEF calculates the ideal control quantity without perturbation and feedforward compensation according to the inputs of TD and DWESO. The final control quantity of the system is obtained from the operation relationship between the ideal control quantity output by NLSEF and the total disturbance estimation value output by DWESO. The precise control of the motor speed is achieved by calculating the final control quantity.

[0111] NLSEF is used to dynamically adjust the control quantity to directly compensate for the disturbance effect. Its control law formula is as follows:

[0112]

[0113] where k1 and k2 are feedback gains, determining the weight of the error term, u0 is the control input without compensated disturbance, which is the ideal control quantity calculated by the controller based on the error between the desired trajectory and the actual state, and u 终 is the final control input after compensating for the disturbance, and the precise control of the UAV motor speed is achieved by calculating u 终 .

[0114] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and adjustments can be made, and these improvements and adjustments should also be regarded as the protection scope of the present invention.

Claims

1. A method for controlling the rotational speed of a drone motor based on active disturbance rejection, characterized in that It includes the following steps: A high-precision rotational speed sensor based on quantum effects is used to detect the actual rotational speed of the UAV motor. A tracking differentiator TD is used to smooth the desired rotational speed, and the output rotational speed tracking signal and the change rate of the rotational speed tracking signal are used as the inputs of the non-linear state error feedback NLSEF; The dynamic weight extended state observer DWESO introduces a dynamic weight matrix on the basis of the traditional ESO; The artificial lemming algorithm ALA is used to automatically search for the optimal combination of the gain parameters of the dynamic weight extended state observer DWESO, and the estimated values of the state variables and the total disturbance estimated value output by the optimized dynamic weight extended state observer DWESO are used as the inputs of the non-linear state error feedback NLSEF; The non-linear state error feedback NLSEF calculates the ideal control quantity without disturbance and feed-forward compensation according to the inputs of the tracking differentiator TD and the dynamic weight extended state observer DWESO; Finally, the final control quantity is calculated according to the ideal control quantity output by the non-linear state error feedback NLSEF and the total disturbance estimated value output by the dynamic weight extended state observer DWESO, so as to achieve precise control of the motor rotational speed.

2. The method for controlling the rotational speed of a drone motor based on active disturbance rejection according to claim 1, wherein A high-precision rotational speed sensor based on quantum effects is used to detect the actual rotational speed of the UAV motor. Using the magnetic sensitivity of the diamond nitrogen vacancy NV color center, the rotational speed measurement is realized by detecting the magnetic field change of the magnetic mark on the rotating shaft of the UAV motor. The steps are as follows: Step 1.1: Uniformly arrange N pairs of magnetic poles on the surface of the drone motor shaft. Each rotation generates N periodic changes in the magnetic field, and the peak magnetic field generated by a single magnetic pole is B peak , which needs to satisfy B peak > 1 μT, 1 μT = 10 -6 Tesla; Step 1.2: Irradiate the NV color center with a 532 nm laser to excite fluorescence, apply microwaves with a frequency of f MW to measure the magnetic field through optically detected magnetic resonance technology. The magnetic field change ΔB(t) causes periodic modulation of the fluorescence intensity, and the modulation frequency f sensor is related to the rotational speed. The formula for the energy level splitting frequency difference caused by the magnetic field change is as follows: where Δf is the energy level splitting frequency difference, B is the magnetic field strength, and γ e is the electron gyromagnetic ratio, which determines the sensitivity of the magnetic field to energy level splitting, and takes 28 GHz / T; Step 1.3: Convert the fluorescence signal into a digital frequency f through a lock-in amplifier and a frequency counter sensor , according to the detected frequency f sensor and the number of pole pairs N, calculate the rotational speed ω of the UAV motor, and the formula is as follows: Utilize the high sensitivity of quantum effects to minute changes to achieve higher-precision and more stable measurement of the UAV motor rotational speed.

3. A method for controlling the rotational speed of a drone motor based on active disturbance rejection according to claim 1, characterized in that, The discrete form formula of the tracking differentiator TD is as follows: Where, fhan(·) is the fastest control law function, v is the desired rotational speed signal of the motor, v1 is the desired rotational speed tracking signal, v2 is the change rate of the desired rotational speed tracking signal, r is the rotational speed tracking factor, which controls the tracking speed of the tracking differentiator, h is the filtering factor, which determines the filtering effect, and T is the sampling step of the system, which affects the discretization accuracy of the tracking differentiator.

4. A method for controlling the rotational speed of a drone motor based on active disturbance rejection according to claim 1, characterized in that, The dynamic weight matrix adjusts the weights of state estimation and disturbance estimation in control in real time according to the state of the system. When the disturbance is large, the weight of disturbance estimation is increased to quickly suppress the disturbance; when the system is stable, the weight of state estimation is increased to improve the control accuracy. The formula of the dynamic weight matrix is as follows: where α is the weight coefficient of state z2, which is dynamically adjusted according to the rotational speed error e, and φ is the weight coefficient of the total disturbance estimate value z3, which is dynamically adjusted according to the change rate of the rotational speed error The dynamic adjustment is such that α0, φ0 ∈ (0, 1] are the initial weight coefficients, and γ, σ > 0 are the attenuation factors; The formula of the discrete form of DWESO is as follows: Where, fal(·) is a non-linear function used for the state update of the ESO, z1 is the estimated value of the rotational speed, z2 is the estimated value of the rotational speed change rate, representing the dynamic characteristics of the system, z3 is the estimated value of the total disturbance, representing the uncertainty of the system, e is the rotational speed error, ω is the actual rotational speed of the motor, measured by a high-precision sensor, u is the control quantity, representing the control command, b is the amplification factor of the control quantity, α1 is the feedback strength of the main control state error, which affects the dynamic response speed, α2 is the smoothness of the main control error differential, which affects the anti-noise ability, β1, β2, β3 are the gain parameters of the ESO, β1 affects the convergence speed of the motor rotational speed estimation, β2 affects the convergence speed of the motor rotational speed change rate estimation, β3 affects the convergence speed of the total disturbance estimated value, and h is the sampling step.

5. A method for controlling the rotational speed of a drone motor based on active disturbance rejection according to claim 4, characterized in that, The artificial lemming algorithm (ALA) is used to automatically search for the optimal combination of the gain parameters of the dynamic weight expansion state observer (DWESO). The steps are as follows: Step 2.1: Initialize the population and randomly generate initial candidate solutions for combinations of the DWESO gain parameters β1, β2, and β3. Step 2.2: Iteratively update the DWESO gain parameters β1, β2, and β3 by simulating four strategies of the lemming's long-distance migration (i.e., global exploration), digging holes (i.e., local exploration), foraging (i.e., spiral search), and avoiding natural enemies (i.e., flight escape). Step 2.3: Dynamically adjust the weights of the algorithm in the global exploration and local exploitation stages to avoid premature convergence and ensure the optimization quality of the UAV motor speed control parameters. Step 2.4: Select the optimal DWESO gain parameter combination with the integral time absolute error (ITAE) as the performance index to optimize the UAV motor speed control performance. Step 2.5: After reaching the maximum number of iterations, output the global optimal solution.

6. The method for controlling the rotational speed of a drone motor based on active disturbance rejection according to claim 5, wherein The nonlinear state error feedback (NLSEF) is used to dynamically adjust the control quantity to directly compensate for the disturbance effect. The control law formula is as follows: Among them, k1 and k2 are feedback gains that determine the weight of the error term, and u0 is the control input of the uncompensated disturbance, which is the ideal control quantity calculated by the controller based on the error between the desired trajectory and the actual state. u 终 is the final control input after compensating for the disturbance, and the precise control of the drone motor speed is achieved by calculating u 终 ​

Images