Sensorless control temperature identification method for UAV power system

Abstract

A sensorless control temperature identification method for a UAV power system belongs to the field of motor control technology. The present invention addresses the problem that temperature identification in a UAV permanent magnet synchronous motor drive system depends on rotor position information, which is in conflict with the closed-loop coupling of the sensorless control itself, which requires estimation of the rotor position. The method comprises: establishing a complex vector model of a permanent magnet synchronous motor based on extended back electromotive force in an estimated synchronous rotating γ-δ reference frame; establishing a full-order state sliding film observer based on a complex vector with stator current and extended back electromotive force as state variables; then performing forward Euler discretization to obtain a full-order state sliding film observer in a discrete domain, observing the stator current estimate and the extended back electromotive force estimate; and then obtaining the rotor observation position and the rotor observation speed; designing an affine projection algorithm, combining the segmented regularization factor strategy to identify the stator resistance in real time, and then obtaining the motor temperature estimate based on the stator resistance estimate obtained by identification. The present invention is used for motor temperature identification.

Description

Technical Field

[0001] The invention relates to a sensorless control temperature identification method for a UAV power system, and belongs to the technical field of motor control. Background Art

[0002] Permanent magnet synchronous motors (PMSMs), with their high efficiency, high power density, and excellent dynamic performance, have become the core drive unit in drone propulsion systems. With the widespread application of drones in logistics, inspection, and other fields, however, the stringent lightweight requirements and complex operating conditions (such as high altitude and extreme temperature fluctuations) of drones pose greater challenges to the reliability of PMSM control systems.

[0003] The key to precise control of permanent magnet synchronous motors (PMSMs) lies in real-time and efficient acquisition of rotor position and speed information. However, traditional PMSM control relies on mechanical position sensors, which are bulky, costly, and susceptible to environmental interference, making them inadequate for lightweight and highly reliable drones. Consequently, "plug-and-play" sensorless control technologies are gaining popularity. Research on position sensorless control can be categorized into low-speed and medium- and high-speed approaches based on the motor's operating speed. Low-speed PMSM sensorless control typically employs signal injection, processing the response to the injected high-frequency signal to obtain the rotor position error. The rotor position is then determined using a phase-locked loop (PLL) or other similar techniques. Medium- and high-speed PMSM sensorless control typically employs model-based approaches, estimating rotor pole position information using back-electromotive force (EMF) or flux linkage models derived from fundamental frequency excitation. Sliding mode observers have attracted considerable attention due to their high accuracy and robustness.

[0004] However, permanent magnet synchronous motors (PMSMs) often experience significant temperature increases during actual drone operation, especially during heavy-load transportation or high-temperature inspections. Excessive temperatures can cause permanent magnet demagnetization and stator winding insulation damage, potentially leading to motor system failure and severely impacting reliability and service life. Therefore, for sensorless PMSM drive systems in drones, online temperature monitoring is crucial for safe system operation. Traditional temperature monitoring relies on built-in temperature sensors (such as thermocouples and thermistors), but these have limitations such as high hardware cost, space constraints, and slow dynamic response. Consequently, sensorless online temperature identification methods have become a research trend. Existing technologies fall into two main categories: thermal model-based methods and stator resistance-based methods.

[0005] Although sensorless control and temperature monitoring technologies have made progress, the following key issues still exist in the permanent magnet synchronous motor drive system of UAVs: the traditional reduced-order sliding mode observer ignores the dynamic characteristics of the current, resulting in increased position estimation errors at high speeds or sudden load changes; the temperature identification method relies on rotor position information, while sensorless control itself requires estimation of the rotor position, which forms a closed-loop coupling contradiction with sensorless control, and the fixed-parameter resistance identification algorithm has a poor convergence speed and large steady-state error when the speed suddenly changes.

[0006] Therefore, in order to improve the operational reliability and thermal safety of the permanent magnet synchronous motor sensorless control system for UAVs, the study of the full-order state sliding mode observer sensorless control technology with online temperature identification has important theoretical significance and application value. Summary of the Invention

[0007] Aiming at the problem that temperature identification in the permanent magnet synchronous motor drive system of a UAV depends on the rotor position information, which is inconsistent with the closed-loop coupling of the sensorless control itself, which requires the estimation of the rotor position, the present invention provides a temperature identification method for the sensorless control of the UAV power system.

[0008] A sensorless control temperature identification method for a UAV power system of the present invention comprises:

[0009] Based on the stator voltage equation and flux equation of the permanent magnet synchronous motor, a permanent magnet synchronous motor model based on extended back electromotive force in the dq coordinate system is obtained; an estimated synchronous rotating γ-δ reference system is established, and the permanent magnet synchronous motor model based on extended back electromotive force in the dq coordinate system is converted to the estimated synchronous rotating γ-δ reference system, thereby obtaining a permanent magnet synchronous motor complex vector model based on extended back electromotive force in the estimated synchronous rotating γ-δ reference system;

[0010] Based on the permanent magnet synchronous motor complex vector model, an estimated state equation with stator current and extended back electromotive force as state variables in a synchronous rotating γ-δ reference frame is obtained, and a full-order state sliding film observer based on a complex vector is established based on the state equation;

[0011] Performing forward Euler discretization on the complex vector-based full-order state sliding film observer to obtain a full-order state sliding film observer in a discrete domain, which is used to observe the stator current estimate and the extended back electromotive force estimate; then, based on the stator current estimate and the extended back electromotive force estimate, using a normalized orthogonal phase-locked loop to obtain the rotor observed position and the rotor observed speed;

[0012] An affine projection algorithm is designed based on the stator current estimate and the extended back electromotive force estimate; the affine projection algorithm is combined with a piecewise regularization factor strategy to identify the stator resistance in real time, and then the motor temperature estimate is obtained based on the identified stator resistance.

[0013] The beneficial effects of the present invention are as follows: the method of the present invention improves the dynamic observation accuracy of the rotor position by integrating the dynamic characteristics of the full-order state sliding mode observer and the real-time temperature identification technology based on the affine projection algorithm, realizes the decoupling estimation of the temperature in the full speed range, improves the convergence and accuracy of the identification, and can effectively avoid the risk of permanent magnet demagnetization caused by model mismatch and parameter drift in traditional solutions, while reducing hardware costs and improving the lightweight level of the system, providing technical guarantee for the long-term and reliable operation of UAVs in extreme environments.

[0014] The present invention establishes a full-order sliding mode observer with stator current and extended back electromotive force as state variables based on the extended back electromotive force model in the estimated synchronous rotating γ-δ reference frame. Then, an affine projection algorithm is designed by using the motor model based on the extended back electromotive force in the estimated synchronous rotating γ-δ reference frame to avoid dependence on the rotor position; the resistance is identified in real time in combination with the piecewise regularization factor strategy to balance the convergence speed and steady-state accuracy. The method of the present invention is used for the sensorless control system of permanent magnet synchronous motor for unmanned aerial vehicles. It can achieve good steady-state tracking performance and high estimation accuracy in both steady-state operation and acceleration and deceleration operation, and shows good convergence and small steady-state error in the acceleration and deceleration process, effectively improving the convergence speed and accuracy of identification, and realizing high-precision, safe and stable operation of permanent magnet synchronous motor for unmanned aerial vehicles under sensorless conditions.

[0015] While ensuring the lightweight and reliability of the UAV drive system, the method of the present invention significantly improves the position estimation accuracy and the real-time performance of temperature monitoring through a full-order state sliding mode observer and online temperature decoupling identification, thereby improving the practicality of the method of the present invention and providing an innovative solution for highly reliable and lightweight motor control. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is an overall block diagram of the sensorless control temperature identification method for the UAV power system of the present invention;

[0017] Figure 2 is the estimated synchronously rotating γ-δ reference frame with Schematic diagram of the relative positions of the axis system and the dq coordinate system;

[0018] Figure 3 It is the flow chart of online temperature recognition using affine projection algorithm;

[0019] Figure 4 is a comparison diagram of the encoder speed and the observer speed in Example 1;

[0020] Figure 5 is a schematic diagram of the angle error in Example 1;

[0021] Figure 61 is a schematic diagram of experimental results of stator temperature identification and error analysis when the motor speed is accelerated in a step form of 500 rpm in Example 2;

[0022] Figure 7 1 is a schematic diagram of experimental results of stator temperature identification and error analysis when the motor speed is decelerated in a step form of 500 rpm in Example 2. DETAILED DESCRIPTION

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0024] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0025] The present invention will be further described below with reference to the accompanying drawings, but is not intended to limit the present invention.

[0026] Combine Figures 1 to 3 As shown, the present invention provides a sensorless control temperature identification method for a UAV power system, comprising:

[0027] Based on the stator voltage equation and flux equation of the permanent magnet synchronous motor, a permanent magnet synchronous motor model based on extended back electromotive force in the dq coordinate system is obtained; an estimated synchronous rotating γ-δ reference system is established, and the permanent magnet synchronous motor model based on extended back electromotive force in the dq coordinate system is converted to the estimated synchronous rotating γ-δ reference system, thereby obtaining a permanent magnet synchronous motor complex vector model based on extended back electromotive force in the estimated synchronous rotating γ-δ reference system;

[0028] Based on the permanent magnet synchronous motor complex vector model, an estimated state equation with stator current and extended back electromotive force as state variables in a synchronous rotating γ-δ reference frame is obtained, and a full-order state sliding film observer based on a complex vector is established based on the state equation;

[0029] Performing forward Euler discretization on the complex vector-based full-order state sliding film observer to obtain a full-order state sliding film observer in a discrete domain, which is used to observe the stator current estimate and the extended back electromotive force estimate; then, based on the stator current estimate and the extended back electromotive force estimate, using a normalized orthogonal phase-locked loop to obtain the rotor observed position and the rotor observed speed;

[0030] An affine projection algorithm is designed based on the stator current estimate and the extended back electromotive force estimate; the affine projection algorithm is combined with a piecewise regularization factor strategy to identify the stator resistance in real time, and then the motor temperature estimate is obtained based on the identified stator resistance.

[0031] Furthermore, in the dq coordinate system, the stator voltage equation and flux equation of the permanent magnet synchronous motor are:

[0032] (1),

[0033] (2),

[0034] In the formula is the d-axis stator voltage, is the q-axis stator voltage, is the stator resistance, is the d-axis stator current, is the q-axis stator current, p is the differential operator, is the d-axis stator flux, is the q-axis stator flux, is the motor electrical angular velocity; is the d-axis stator inductance, is the q-axis stator inductance, is the permanent magnet flux;

[0035] By transforming formula (1) and formula (2), we can obtain the permanent magnet synchronous motor model based on extended back electromotive force in the dq coordinate system:

[0036] (3),

[0037] (4),

[0038] In the formula To expand the back electromotive force.

[0039] The method for obtaining the estimated complex vector model of the permanent magnet synchronous motor based on the extended back electromotive force in the synchronous rotating γ-δ reference frame is:

[0040] Write formula (3) and formula (4) in complex vector form:

[0041] (5),

[0042] (6),

[0043] In the formula is the dq axis stator voltage in complex vector form, is the dq axis stator current in complex vector form, is the expanded back EMF in complex vector form;

[0044] Formulas (5) and (6) show the complex vector structure of the permanent magnet synchronous motor. The dq axis components are converted into the real and imaginary parts of the complex vector model respectively. The two-dimensional structure of the original model can be converted into a one-dimensional structure to simplify subsequent analysis.

[0045] Combine Figure 2 As shown, an estimated synchronous rotating γ-δ reference frame is established, and formulas (5) and (6) are converted to the estimated synchronous rotating γ-δ reference frame to obtain the complex vector model of the permanent magnet synchronous motor based on the extended back electromotive force in the estimated synchronous rotating γ-δ reference frame:

[0046] (7),

[0047] (8),

[0048] (9),

[0049] In the formula is the estimated stator voltage in the synchronously rotating γ-δ reference frame, is the observed rotor speed, is the estimated stator current in the synchronously rotating γ-δ reference frame, is the estimated expanded back EMF in the synchronously rotating γ-δ reference frame; is the rotor position error, is the rotor speed error; is the rotor position, is the rotor observation position.

[0050] Based on formula (7) and formula (8), the estimated state equation in the synchronous rotating γ-δ reference frame with stator current and extended back electromotive force as state variables is obtained:

[0051] (10),

[0052] In the formula Express Find the derivative, Express Find derivatives;

[0053] According to formula (10), a full-order state sliding film observer based on complex vector is established:

[0054] (11),

[0055] Where:

[0056] (12),

[0057] In the formula for The estimated value of For Find the derivative, for The estimated value of is the feedback gain matrix of the full-order state synovial observer based on complex vectors; is the damping ratio of the second-order system based on the full-order state sliding film observer of the complex vector, is the second-order system angular frequency of the full-order state sliding film observer based on complex vector.

[0058] The full-order state synovial observer adds a feedback correction channel in the design. The actual status of the system When they are not equal, it is reflected in their output and They are also not equal, so an error signal is generated, which is fed to the input of the observer through the feedback gain matrix K, and participates in adjusting the state of the observer so that it approaches the true state of the system with a certain accuracy and speed.

[0059] By performing forward Euler discretization on formula (11), we can obtain the full-order state sliding film observer in the discrete domain:

[0060] (13),

[0061] In the formula is the discrete sampling moment of the driving system, is the sampling period, is the estimated value of stator current, For extended back EMF estimation.

[0062] The full-order state synovial observer in the discrete domain can be used to obtain the observed axis and ; Estimation of extended back EMF Using a normalized orthogonal phase-locked loop to obtain the rotor observation position and rotor observed speed .

[0063] Furthermore, to achieve online identification of motor temperature, the preferred method is resistance-based online identification. This is because there is a clear linear relationship between stator resistance and motor temperature. By identifying the resistance value in real time, the temperature change of the stator winding can be accurately calculated.

[0064] The affine projection algorithm is used to estimate the motor resistance parameters. The discrete system time model of the affine projection algorithm is designed as follows:

[0065] (14),

[0066] In the formula is the discrete system output matrix, Input matrix for discrete system, is the true value of the vector to be identified, is the identification value of the vector to be identified, is the iteration step length, The smaller the value, the smaller the steady-state error and the slower the convergence speed, and vice versa; is the regularization factor, To avoid numerical problems in the process of matrix inversion and overcome possible singular situations of the algorithm, adjustment parameters are introduced; is the identity matrix.

[0067] According to formula (13) and formula (14), the stator resistance is identified in real time:

[0068] (15),

[0069] In the formula is the estimated value of stator current The γ-axis component, is the estimated value of stator current The δ-axis component of is the stator voltage The γ-axis component, To extend the back EMF estimate The γ-axis component, is the stator current The γ-axis component, is the stator current The δ-axis component of is the stator resistance identification value.

[0070] The affine projection algorithm uses an extended back-EMF-based complex vector model of the permanent magnet synchronous motor in an estimated synchronously rotating γ-δ reference frame for real-time identification. This is because the γ-δ reference frame includes position information estimated by the sensorless observer, while the extended back-EMF model includes the EMF generated by the permanent magnets and stator inductance, meaning that the model accounts for the effects of rotor salient poles. Therefore, the affine projection algorithm using the extended back-EMF-based complex vector model of the permanent magnet synchronous motor provides more accurate real-time estimation. Furthermore, the entire system does not require any position sensors for variable transformation, making it directly applicable to a wide range of fields, such as sensorless motor control, condition monitoring, and fault detection.

[0071] Meanwhile, the affine projection algorithm is derived by using the observer current equation, and thus has stronger robustness under speed variation conditions.

[0072] To improve the convergence speed and steady-state error of the algorithm identification, this implementation adopts a segmented approach to select the regularization factor of the affine projection algorithm. By selecting an appropriate regularization factor and iteration step size, online temperature identification can be relatively fast and accurate, improving algorithm performance.

[0073] Based on stator resistance identification value The estimated motor temperature is:

[0074] (16),

[0075] In the formula is the estimated motor temperature, is the ambient temperature The resistance of the lower stator winding, is the temperature coefficient of resistance.

[0076] Figure 1 middle is the speed given value, is the given value of the γ-axis component of the stator current, is the given value of the δ-axis component of the stator current, is the stator voltage Axis component, is the stator voltage Axis component, is the three-phase current, is the stator current Axis component, is the stator current Axis component;

[0077] Figure 3 middle is the regularization factor value in the low-speed segment, is the regularization factor value in the medium speed segment, is the regularization factor value of the high-speed segment;

[0078] Figure 4 The observer speed is the rotor observation speed ;

[0079] Figure 5 The angle error is the rotor position error ;

[0080] Figure 6 The identified temperature is the estimated value of the motor temperature , in the figure, Y represents the temperature coordinate axis.

[0081] The following specific examples are used to verify the beneficial effects of the present invention:

[0082] Embodiment one:

[0083] In order to verify the effectiveness of the method of the present invention, verification was carried out on a constructed UAV experimental platform.

[0084] The main parameters of the permanent magnet assisted synchronous reluctance motor used are: rated voltage 44.4V, rated current 11A, d-axis inductance L d = 27.5e-6H, q-axis inductance L q = 42.2e-6H, pole pair number p = 21, stator resistance Rs = 0.09Ω, permanent magnet flux = 0.002425Wb.

[0085] like Figure 4 and Figure 5 As shown in Figure 1, the experimental results of the full-order state sliding film observer based on the extended back electromotive force are shown, where the motor speed increases from 0 to the maximum speed in a step of 500 rpm and 100% rated load is applied. Figure 4 The waveforms of the encoder speed (actual speed) and the observer speed are shown. It can be seen that the speed error of the motor does not exceed ±10rpm during rated steady-state operation. Figure 5 Shown is the rotor angle error, given by Figure 5 It can be seen that the maximum steady-state electrical angle deviation is at 3350 rpm, and the steady-state electrical angle deviation does not exceed 0.35 rad. Experimental results show that the sensorless control method of the present invention can achieve good steady-state tracking performance and high estimation accuracy in both steady-state operation and acceleration and deceleration operation, and the system has good load-bearing performance.

[0086] Example 2:

[0087] The key to online parameter identification of sensorless control systems of UAVs is whether the algorithm identification value can track the actual value when the operating speed changes.

[0088] In order to verify the performance of the online temperature recognition algorithm based on the affine projection algorithm proposed by the method of the present invention, Figure 6 The experimental results of stator temperature identification and error analysis are shown when the motor speed is accelerated in a step form of 500 rpm. The online temperature estimation error is less than 3.02°C, and the estimated stator temperature can quickly and accurately track the actual measured temperature. Figure 7 The experimental results of stator temperature identification and error analysis when the motor speed is decelerated in a 500 rpm step form. The online temperature estimation error is less than 2.06°C, which shows that the stator temperature information can be accurately tracked in real time across the entire speed range.

[0089] Figure 6 and Figure 7The experimental results show that since the affine projection algorithm uses a segmented method to select the regularization factor, it can still quickly and effectively estimate the motor temperature in the case of a sudden speed step change. It has good convergence and high identification accuracy during acceleration or deceleration, thereby realizing online monitoring of the drone motor temperature within the full speed range.

[0090] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely illustrative of the principles and applications of the invention. It should be understood that many modifications may be made to the illustrative embodiments, and that other arrangements may be devised, without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in ways other than those described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in conjunction with other described embodiments.

Claims

1. A sensorless control temperature identification method for a UAV power system, characterized in that: include: Based on the stator voltage equation and flux equation of the permanent magnet synchronous motor, a permanent magnet synchronous motor model based on extended back electromotive force in the dq coordinate system is obtained; an estimated synchronous rotating γ-δ reference system is established, and the permanent magnet synchronous motor model based on extended back electromotive force in the dq coordinate system is converted to the estimated synchronous rotating γ-δ reference system, thereby obtaining a permanent magnet synchronous motor complex vector model based on extended back electromotive force in the estimated synchronous rotating γ-δ reference system; Based on the permanent magnet synchronous motor complex vector model, an estimated state equation with stator current and extended back electromotive force as state variables in a synchronous rotating γ-δ reference frame is obtained, and a full-order state sliding film observer based on a complex vector is established based on the state equation; Performing forward Euler discretization on the complex vector-based full-order state sliding film observer to obtain a full-order state sliding film observer in a discrete domain, which is used to observe the stator current estimate and the extended back electromotive force estimate; then, based on the stator current estimate and the extended back electromotive force estimate, using a normalized orthogonal phase-locked loop to obtain the rotor observed position and the rotor observed speed; An affine projection algorithm is designed based on the stator current estimate and the extended back electromotive force estimate; the affine projection algorithm is combined with a piecewise regularization factor strategy to identify the stator resistance in real time, and then an estimated motor temperature is estimated based on the identified stator resistance; The stator voltage equation and flux equation of the permanent magnet synchronous motor are: （1）， （2）， In the formula is the d-axis stator voltage, is the q-axis stator voltage, is the stator resistance, is the d-axis stator current, is the q-axis stator current, p is the differential operator, is the d-axis stator flux, is the q-axis stator flux, is the motor electrical angular velocity; is the d-axis stator inductance, is the q-axis stator inductance, is the permanent magnet flux; By transforming formula (1) and formula (2), we can obtain the permanent magnet synchronous motor model based on extended back electromotive force in the dq coordinate system: （3）， （4）， In the formula To expand the back EMF; The method for obtaining the estimated complex vector model of the permanent magnet synchronous motor based on the extended back electromotive force in the synchronous rotating γ-δ reference frame is: Write formula (3) and formula (4) in complex vector form: （5）， （6）， In the formula is the dq axis stator voltage in complex vector form, is the dq axis stator current in complex vector form, is the expanded back EMF in complex vector form; Converting formula (5) and formula (6) to the estimated synchronous rotating γ-δ reference frame, we can obtain the complex vector model of the permanent magnet synchronous motor based on the extended back electromotive force in the estimated synchronous rotating γ-δ reference frame: （7）， （8）， （9）， In the formula is the estimated stator voltage in the synchronously rotating γ-δ reference frame, is the observed rotor speed, is the estimated stator current in the synchronously rotating γ-δ reference frame, is the estimated expanded back EMF in the synchronously rotating γ-δ reference frame; is the rotor position error, is the rotor speed error; is the rotor position, is the rotor observation position; Based on formula (7) and formula (8), the estimated state equation in the synchronous rotating γ-δ reference frame with stator current and extended back electromotive force as state variables is obtained: （10）， In the formula Express Find the derivative, Express Find derivatives; According to formula (10), a full-order state sliding film observer based on complex vector is established: （11）， Where: （12）， In the formula for The estimated value of For Find the derivative, for The estimated value of is the feedback gain matrix of the full-order state synovial observer based on complex vectors; is the damping ratio of the second-order system based on the full-order state sliding film observer of the complex vector, is the second-order system angular frequency of the full-order state sliding film observer based on complex vector; By performing forward Euler discretization on formula (11), we can obtain the full-order state sliding film observer in the discrete domain: （13）， In the formula is the discrete sampling moment of the driving system, is the sampling period, is the estimated value of stator current, To extend the back EMF estimation; Estimation of extended back EMF Using a normalized orthogonal phase-locked loop to obtain the rotor observation position and rotor observed speed ; The discrete system time model for designing the affine projection algorithm is: （14）， In the formula is the discrete system output matrix, Input matrix for discrete system, is the true value of the vector to be identified, is the identification value of the vector to be identified, is the iteration step length, is the regularization factor, is the identity matrix; According to formula (13) and formula (14), the stator resistance is identified in real time: （15）， In the formula is the estimated value of stator current The γ-axis component, is the estimated value of stator current The δ-axis component of is the stator voltage The γ-axis component, To extend the back EMF estimate The γ-axis component, is the stator current The γ-axis component, is the stator current The δ-axis component of is the stator resistance identification value.

Images