Motor cogging torque self-adaptive compensation method

Abstract

The invention relates to the technical field of motor cogging torque compensation, and discloses a motor cogging torque self-adaptive compensation method, which comprises the following steps: step 1, multi-temperature-point initialization modeling: a system acquires compensation current data of a plurality of position points at a plurality of stable temperature points, and then generates an initial compensation table LUTi at each temperature point Ti, forming an initial compensation model; 2, online real-time compensation is carried out, the position information theta and the temperature information Tcur of the motor are collected, compensation current data LUTcur are directly obtained or calculated according to the initial compensation model, and online real-time compensation is completed; and step 3, automatically updating the model, performing micro-amplitude online updating on LUTcur data under Tcur, forming an optimization compensation table by using the updated data, and storing and forming an optimization compensation model. According to the invention, the problems of high cost and low accuracy when a control algorithm is adopted to compensate the cogging torque of a single specific motor in the prior art are solved.

Description

Technical Field

[0001] This invention relates to the field of servo motor control, and specifically to a method for adaptive compensation of motor cogging torque. Background Technology

[0002] When a permanent magnet motor with a cogging structure rotates, it causes changes in magnetic field energy, resulting in cogging torque. The periodic torque generated by cogging torque can cause various problems for the motor, including mechanical vibration, noise, decreased positioning accuracy, limited starting performance, and reduced overall efficiency. To reduce the damage caused by cogging torque, common methods in existing technologies to avoid or suppress it include structural optimizations such as optimizing slot-pole matching, tilting stator slots or rotor poles at a certain angle, using fractional slots or fractional poles, and improving the uniformity of air gap size. Algorithmic compensation schemes, such as incorporating rotor position-based prediction compensation into the control algorithm, are also used. However, structural optimizations cannot completely eliminate cogging torque, making algorithmic compensation schemes, once the motor structure is fixed, more effective in further reducing the impact of cogging torque pulsation.

[0003] Existing control algorithm compensation schemes mainly include loop compensation, suppression, and point calibration schemes. Loop compensation schemes pre-store a "position-compensation current" lookup table (LUT) in the controller. This table is typically generated at room temperature after measuring samples of the same model of motor before shipment. However, different motors have different characteristics, making it impossible to achieve precise current compensation for a specific motor. Suppression schemes rely entirely on feedback mechanisms such as PID controllers, generating suppressive torque by detecting speed or position deviations—a passive and lagging compensation method. Point calibration schemes involve self-calibration in some advanced systems, but the calibration process is usually completed under a single, uncontrolled ambient temperature, resulting in a static compensation table that does not consider temperature variables. Additionally, there are mathematical model-based schemes, which require accurate motor mathematical models and extensive calculations of cogging torque data, resulting in a heavy computational burden. Therefore, existing cogging torque compensation schemes cannot provide cost-effective and accurate current compensation for individual motors. Summary of the Invention

[0004] The present invention aims to provide an adaptive compensation method for motor cogging torque, so as to solve the problem of low accuracy when using control algorithms to compensate for the cogging torque of a single specific motor in the prior art.

[0005] To solve the above problems, the present invention adopts the following technical solution: an adaptive compensation method for motor cogging torque, the first step of which is to initialize and model the motor at multiple temperature points, control the motor to rotate at a constant speed, perform online cogging calibration at multiple stable temperature points, collect compensation current data at multiple position points, and then generate an initial compensation table LUT_i at each temperature point T_i to form an initial compensation model. Step 2: Online Real-Time Compensation. During system operation, the system collects motor position information θ and temperature information T_cur in real time. Using the current temperature T_cur and the current position θ as input, the system queries the initial compensation model. If the current temperature T_cur is the temperature point detected in the initial compensation model, the corresponding compensation current is directly used. If the current temperature T_cur is not the temperature point detected in the initial compensation model, the system calculates the compensation current data LUT_cur at the current temperature T_cur based on the compensation current data of the two adjacent temperature points before and after T_cur in the initial compensation model, and then compensates the motor with the calculated compensation current data LUT_cur in real time. Step 3: Automatic Model Update. When the motor reaches a stable state, the compensation current required to maintain a stable speed at a certain position of the motor within a certain period is collected. Then, the collected compensation current data is compared with the compensation current value in the initial compensation model or the compensation current value calculated by the initial compensation model. The LUT_cur data under T_cur is updated slightly online. The updated data forms an optimized compensation table, which is stored and forms an optimized compensation model.

[0006] The principle of this scheme is as follows: In this application, online cogging calibration is first performed at multiple stable temperature points, and compensation current data of multiple position points on the motor at the corresponding stable temperature points are collected. Based on the compensation current data corresponding to each temperature point T_i, an initial compensation table is generated and an initial compensation model is formed. Then, during the actual use of the motor, the current position information θ and the current temperature information T_cur of the motor are collected in real time. Using the current temperature T_cur and the current position θ as input, the sparse temperature-position compensation model is queried. During the query, there are two cases: the current temperature T_cur is the same as or different from the multiple stable temperature points collected in step one. When T_cur is the same as a stable temperature point collected in step one, the current compensation value at different rotor positions in the initial compensation table in step one is directly superimposed on the current loop as a feedforward to compensate for the cogging torque. If the current temperature T_cur is not one of the multiple stable temperature points in step one, for example, the current T_cur is 40℃, while the temperature points during online cogging calibration in step one are 30℃ and 50℃, the compensation current when the current T_cur is 40℃ is calculated based on the compensation current of the two temperature points before and after 40℃ (i.e., 30℃ and 50℃). The calculated compensation current value is superimposed on the current loop as a feedforward to achieve compensation for the cogging torque.

[0007] Furthermore, in step three, the model can be further automatically updated. Specifically, when the system detects that the motor has reached a stable state, at which point the motor temperature is stable, the system collects data on a specific position of the motor within a certain period and the magnitude of the compensation current required to maintain a stable speed at that position. Then, the collected compensation current data is compared with the compensation current value predicted by the current model, thereby performing a small-scale online update to the LUT_cur data under T_cur. The updated data is then used to form an optimized compensation table, which is stored and forms the optimized compensation model. This step further corrects and improves the data, enabling the model to continuously learn and optimize itself during motor operation, allowing it to continuously approach the true optimal compensation value and reduce the negative impact of cogging torque during motor operation.

[0008] The beneficial effects of this application are as follows: 1. By using temperature-position parameters as a key dimension for motor cogging torque compensation, the compensation current can be dynamically adjusted at the corresponding position of the motor according to temperature changes in practical applications, effectively reducing the impact of cogging torque and overcoming the performance degradation caused by temperature drift in principle.

[0009] 2. Different motors can autonomously learn and obtain their corresponding compensation current: Compared to the existing technology of directly using the same lookup table for a certain type and model of motor, due to the differences between motors, even motors of the same type and model may have differences in manufacturing processes, resulting in different actual compensation current requirements for each motor in actual use. Therefore, using the same lookup table for the same type and model of motor has the problem of poor compensation accuracy. In this application, an initial compensation model is formed in the initial stage, and then continuously optimized and adjusted in the actual use of the motor to form an optimized compensation model. Based on the actual operating conditions of each motor, a "tailor-made" compensation model is formed, which can better adapt to the unique characteristics of each motor's random time and temperature changes, achieving true individualization and self-adaptation.

[0010] 3. Continuous Learning and Improvement: In this application, based on the intelligent triggering learning mechanism of the motor's operating conditions, when the motor reaches a stable state, the system determines that the motor's operating condition is in a "learning window" period. Then, under stable conditions, it collects the compensation current required to maintain a stable speed at a certain position of the motor within a certain period. The compensation current is calculated and converted into the compensation current for motor cogging torque compensation. The compensation current calculated based on actual detection is compared with the compensation current value in the initial compensation model or the compensation current value calculated through the initial compensation model. Based on the comparison results, the current compensation data is updated slightly online. Finally, the updated data forms an optimized compensation table, which is stored and forms an optimized compensation model. This allows the compensation parameters to be continuously optimized during the stable phase of the motor's operation. Subsequently, during the operation of the motor, the rotor angle can be read in real time, and the compensation current value can be superimposed on the current command through the controller by looking up the optimized compensation table, enabling the motor to run stably for a long time.

[0011] 4. Achieve high-precision compensation across the entire temperature range: In this application, multiple temperature points and corresponding location points of multiple compensation current data are first collected in the initial compensation table. Then, during the subsequent use of the motor, the compensation current values ​​of different temperature points and corresponding locations are continuously increased through detection and interpolation compensation. Finally, the compensation current value is continuously optimized when the motor reaches a stable state, so that the motor can obtain consistent and excellent smooth performance in cold start, hot operation and even any operating environment temperature.

[0012] 5. Simplified application process and reduced cost: In this application, only the tooth groove needs to be calibrated in step one to form an initial compensation table. There is no need for manual calibration of the motor compensation amount under different temperature environments. After the motor is put into use, steps two and three can be completed automatically to finally form an optimized compensation table, so that the model can be automatically updated and optimized, reducing costs and the threshold for use. At the same time, by online learning to compensate for the impact of slow-changing factors such as component aging and mechanical wear, the system maintains the peak performance of the system throughout its entire life cycle and reduces maintenance requirements.

[0013] Preferably, as an improvement, the condition for determining the stable state of the motor in step three is that the current fluctuation range is less than or equal to 0.02A.

[0014] Preferably, as an improvement, the condition for determining the stable state of the motor in step three is that the speed fluctuation range of the motor is less than or equal to 2 RPM.

[0015] Preferably, as an improvement, in step three, when the motor reaches a stable state, the period for collecting the motor position and the compensation current required to maintain a stable speed is 5-10 complete mechanical cycles of the motor rotating at a constant speed.

[0016] Preferably, as an improvement, a speed loop integrator is used in step three to collect the compensation current of the motor when it is in a stable state in step three.

[0017] Preferably, as an improvement, in step three, the recursive least squares method or an adaptive filtering algorithm is used to perform a small-scale online update of the LUT_cur data under T_cur.

[0018] Preferably, as an improvement, in step three, the updated data is formed into an optimization compensation table and written into non-volatile memory.

[0019] Preferably, as an improvement, when the system performs online tooth groove calibration at multiple stable temperature points in step one, the temperature points include -20℃, -10℃, 0℃, 25℃, 50℃, and 75℃.

[0020] Preferably, as an improvement, when collecting compensation current data at multiple locations in step one, the sampling interval between the multiple locations is less than or equal to 1°.

[0021] Preferably, as an improvement, in step one, when determining the stable temperature point and in step two, when collecting temperature information T_cur, the detection signals of the shaft voltage, shaft current, and electrical angular velocity of the permanent magnet synchronous motor are acquired. A resistance observer model is established based on the detection signals, and the resistance model temperature T̂ is calculated. Simultaneously, a motor thermal balance temperature model is established based on the detection signals to evaluate the winding temperature T. The two models are then fused, and a small gain k_f is used to allow the winding temperature T to slowly converge towards the resistance model temperature T̂. And obtain the temperature information T_cur. Detailed Implementation

[0022] The following detailed description illustrates the specific implementation method: Example

[0023] This first embodiment describes an adaptive compensation method for motor cogging torque, comprising the following steps: Step 1: Initial modeling at multiple temperature points. The system controls the motor to rotate at a constant speed, performs online cogging calibration at multiple stable temperature points, collects compensation current data at multiple locations, and then generates an initial compensation table LUT_i for each temperature point T_i, storing it to form the initial compensation model. In practical applications, multiple stable temperature points include -20℃, -10℃, 0℃, 25℃, 50℃, and 75℃. When collecting compensation current data at multiple locations, the sampling interval between the multiple locations is less than or equal to 1°.

[0024] Step 2: Online Real-Time Compensation. During system operation, the system collects motor position information θ and temperature information T_cur in real time. Using the current temperature T_cur and current position θ as input, the system queries the initial compensation model. If the current temperature T_cur is the temperature point detected in the initial compensation model, the corresponding temperature point's compensation current is directly used to compensate for the cogging torque. If the current temperature T_cur is not the temperature point detected in the initial compensation model, the system calculates the compensation current data LUT_cur at the current temperature T_cur based on the compensation current data of the two adjacent temperature points before and after T_cur in the initial compensation model. Then, the calculated compensation current data LUT_cur is applied to the motor in real time to improve the stability of the motor during use.

[0025] Meanwhile, in this embodiment, when collecting motor position information θ and temperature information T_cur in real time, the motor position information can be collected by the encoder inside the motor, while when collecting motor temperature information T_cur, temperature sensors such as thermistors and thermocouples can be used to collect and detect the operating temperature of the motor.

[0026] Step 3: Automatic Model Update. When the motor reaches a stable state, the compensation current required to maintain a stable speed at a certain position of the motor within a certain period is collected. In this embodiment, the motor's stable state determination condition is that the motor speed fluctuation range is less than 2 RPM and the current fluctuation range is less than or equal to 0.02A. The time period for collecting the compensation current required to maintain a stable speed at a certain position of the motor is 5-10 complete mechanical cycles of the motor rotating at a constant speed, preferably 5 complete mechanical cycles. During the collection process, a speed loop integrator is used to collect the compensation current of the motor in the stable state in this step. In this embodiment, after collecting the compensation current required to maintain a stable speed within the corresponding period of the motor, the collected compensation current data is compared with the compensation current value predicted by the initial compensation model. The initial compensation model is updated according to the comparison result. Specifically, the recursive least squares method or adaptive filtering algorithm is used to perform a small-amplitude online update of the LUT_cur data under T_cur. The updated data forms an optimized compensation table, which is stored and forms an optimized compensation model.

[0027] In this embodiment, a motor with model number PD42-2-1670-TMCL is used as an example. Its power is 52W, rated voltage is 24V, torque is 0.125N•M, rated speed is 4000RPM, and dimensions are 42.00mm×42.00mm. The steps for motor cogging torque compensation are as follows: Step 1: Initial modeling at multiple temperature points. The system controls the motor to rotate at a constant speed, performs online cogging calibration at multiple stable temperature points, collects compensation current data at multiple locations, and then generates an initial compensation table LUT_i for each temperature point T_i, storing it to form an initial sparse temperature-position compensation model. In practical applications, multiple stable temperature points include -20℃, -10℃, 0℃, 25℃, 50℃, and 75℃. When collecting compensation current data at multiple locations, the sampling interval for multiple locations is 1°. Table 1 shows the initial compensation table for a certain temperature point.

[0028]

[0029] Step 2: Online Real-Time Compensation. During system operation, the system collects motor position information θ and temperature information T_cur in real time. Using the current temperature T_cur and current position θ as input, it queries the initial compensation model. If the current temperature T_cur is a temperature point detected in the initial compensation model, the corresponding compensation current at that temperature point is directly used to compensate the motor's cogging torque. If the current temperature T_cur is not a temperature point detected in the initial compensation model, the system calculates the compensation current data LUT_cur at the current temperature T_cur based on the compensation current data of the two adjacent temperature points before and after T_cur in the initial compensation model. The calculated compensation current data LUT_cur is then superimposed into the current command through the controller to compensate the motor's cogging torque. For example, if the detected current temperature is 30℃, the system obtains the compensation current values ​​at T=25℃ and T=50℃ from Step 1, and then calculates the compensation current value at T=30℃ using linear interpolation. During actual motor operation, the compensation current value at T=30℃ will be calculated.

[0030] Step 3: Automatic Model Update. When the motor reaches a stable state, the position of the motor and the compensation current required to maintain a stable speed are collected within a certain period. In this embodiment, the conditions for determining the stable state of the motor are that the motor speed fluctuation range is less than or equal to 2 RPM and the current fluctuation range is less than or equal to 0.02A. The period for collecting the motor position and the compensation current required to maintain a stable speed is 5 complete mechanical cycles of the motor rotating at a constant speed. When collecting the compensation current of the motor in the stable state in this step using the speed loop integrator, since the integral value of the compensation current output by the speed loop integrator is the total compensation current required to overcome all disturbances to maintain a constant speed of the motor, it includes not only the main cogging torque but also other constant load torques such as friction and external loads. Since other constant load torques are constant values, it is necessary to remove other constant load torques when determining the motor cogging torque. In the actual detection process, the speed loop integrator accumulates the compensation current at N points per revolution of the motor, obtaining a total of M*N samples; then, the average value X of point i in M ​​revolutions is calculated. iFor example, if the test involves 3 revolutions, and 360 points are tested per revolution, then the average value of each of the 360 ​​points over the 3 revolutions is calculated. Next, the average value of all points across all revolutions is calculated to obtain the constant component S. Finally, the magnitude of the cogging torque at point i is determined as X. i -S.

[0031] In this embodiment, after collecting the compensation current that maintains a stable speed within the corresponding cycle of the motor, the collected compensation current data is compared with the compensation current value predicted by the interpolated temperature-position compensation model. Specifically, the recursive least squares method is used to perform a small-scale online update on the LUT_cur data under T_cur. The updated data is then used to form an optimized compensation table, which is stored and used to form an optimized temperature-position compensation model. Example

[0032] The difference between Example 2 and Example 1 is that in this example, when the optimized compensation table is formed in step 3, the updated data is written into a non-volatile memory to avoid data loss due to motor power failure. Example

[0033] The difference between Example 3 and Example 1 is that in Step 1, the temperature signal of the motor is directly detected by a temperature sensor. Although this can complete the detection, in actual testing, the hardware detection method using a temperature sensor is not only costly but also prone to failure, affecting the accuracy of the detection. In this example, to further improve the accuracy of the motor temperature signal detection, the following method is used: First, the detection signals of the shaft voltage, shaft current, and electrical angular velocity of the permanent magnet synchronous motor are acquired. Simultaneously, the winding thermal capacitance C_th, phase resistance Rs, and thermal resistance R of the permanent magnet synchronous motor are also acquired. th , excitation flux ψ f A resistance temperature observer model is established based on the detected signal to calculate the resistance model temperature T̂. Simultaneously, a thermal equilibrium temperature model for the heat conduction motor is established based on the detected signal to evaluate and calculate the winding temperature T. The two models are fused, and a small gain k_f is used to allow the winding temperature T to slowly converge towards the resistance model temperature T̂ to obtain the temperature information T_cur. , where k f The value range is 0.0001 to 0.01.

[0034] When detecting specific temperature signals, the formula for calculating the temperature T̂ in the resistance model is: ; , where v q This refers to the shaft voltage, measured in V; i q The quadrature-axis current is expressed in amperes (A); ω e ψ is the electrical angular velocity, measured in rad / s.f The excitation flux linkage is expressed in Wb. A thermal balance temperature model of the motor is established based on the operating signals to estimate the winding temperature T. The calculation formula is as follows: , , Where DT is the sampling period, which is 1–10 kHz in this embodiment; C_th is the winding heat capacity, in J / K, and in this embodiment, the value ranges from 30 to 50 J / K, corresponding to a small-power permanent magnet synchronous motor; i d For direct-axis current, i q Rs is the quadrature-axis current, both in A; Rs is the phase resistance, in Ω; α is the temperature coefficient of copper resistance, with a value of 0.00393 / °C; Rs th Thermal resistance, measured in K / W, is taken from 1.0 to 10 K / W in this embodiment, corresponding to a low-power permanent magnet synchronous motor; T tem The ambient temperature is expressed in °C.

[0035] This embodiment uses a 5KW low-power permanent magnet synchronous motor as an example for illustration. The fixed parameters of the permanent magnet synchronous motor are as follows: phase resistance Rs is 0.15Ω, thermal resistance R... th 1.25K / W, excitation flux ψ f It is 0.05Wb, k f The value is 0.001. In a certain operating state, assuming the temperature T at the previous moment... k At 40℃, the shaft voltage v q 15V, quadrature axis current i q For 10A, i d 0A, electrical angular velocity ω e The electrical angular velocity is 1256.64 rad / s (electrical angular velocity at 3000 rpm), and the winding heat capacity C_th is 32 J / K. First, following the method in step two above, a resistance observer model is established based on the operating signal, and the resistance observer temperature T̂ is calculated. The calculation result is: The current is 0.1588Ω, T̂=39.93℃; then, according to step three, a motor thermal balance temperature model is established based on the operating signal to estimate the winding temperature T. The calculation results are Qin=27.54W, Qout=12.00W, Tk+1=40.000086℃; finally, the two models are fused, and the calculation process and results are as follows: Tnew=T k+1 -k f (T) k+1 -T̂), T fused =39.9999786℃.

[0036] The above descriptions are merely embodiments of the present invention, and common knowledge such as specific technical solutions and / or characteristics are not described in detail here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the technical solutions of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application shall be determined by the content of its claims, and the specific embodiments described in the specification may be used.

Claims

1. A method for adaptive compensation of motor cogging torque, characterized in that: Includes the following steps, Step 1: Initialize modeling at multiple temperature points. Control the motor to rotate at a constant speed. The system performs online cogging calibration at multiple stable temperature points, collects compensation current data at multiple position points, and then generates an initial compensation table LUT_i for each temperature point T_i to form an initial compensation model. Step 2: Online Real-Time Compensation. During system operation, the system collects motor position information θ and temperature information T_cur in real time. Using the current temperature T_cur and the current position θ as input, the system queries the initial compensation model. If the current temperature T_cur is the temperature point detected in the initial compensation model, the corresponding compensation current is directly used. If the current temperature T_cur is not the temperature point detected in the initial compensation model, the system calculates the compensation current data LUT_cur at the current temperature T_cur based on the compensation current data of the two adjacent temperature points before and after T_cur in the initial compensation model, and then compensates the motor with the calculated compensation current data LUT_cur in real time. Step 3: Automatic Model Update. When the motor reaches a stable state, the compensation current required to maintain a stable speed at a certain position of the motor within a certain period is collected. Then, the collected compensation current data is compared with the compensation current value in the initial compensation model or the compensation current value calculated by the initial compensation model. The LUT_cur data under T_cur is updated slightly online. The updated data forms an optimized compensation table, which is stored and forms an optimized compensation model.

2. The method of claim 1, wherein: In step three, the condition for determining the stable state of the motor is that the current fluctuation range is less than or equal to 0.02A.

3. The method of claim 1, wherein: In step three, the condition for determining the stable state of the motor is that the speed fluctuation range of the motor is less than or equal to 2 RPM.

4. The method of claim 1, wherein: In step three, when the motor reaches a stable state, the period for collecting the motor position and the compensation current required to maintain a stable speed is 5-10 complete mechanical cycles of the motor rotating at a constant speed.

5. The method of claim 1, wherein: In step three, a speed loop integrator is used to collect the compensation current of the motor when it is in a stable state in step three.

6. The method of claim 1, wherein: In step three, the recursive least squares method or adaptive filtering algorithm is used to perform micro-updates of the LUT_cur data under T_cur online.

7. The method of claim 1, wherein: In step three, the updated data is used to form an optimization compensation table and written into non-volatile memory.

8. The method of claim 1, wherein: In step one, the system performs online tooth cog calibration at multiple stable temperature points, including -20℃, -10℃, 0℃, 25℃, 50℃, and 75℃.

9. A method of adaptive cogging torque compensation for an electric machine as recited in claim 8, characterized in that: When collecting compensation current data at multiple locations in step one, the sampling interval between the multiple locations is less than or equal to 1°.

10. The adaptive compensation method for motor cogging torque according to claim 1, characterized in that: In step one, when determining the stable temperature point, and in step two, when collecting temperature information T_cur, the detection signals of the shaft voltage, shaft current, and electrical angular velocity of the permanent magnet synchronous motor are acquired. A resistance observer model is established based on these detection signals, and the resistance model temperature T̂ is calculated. Simultaneously, a motor thermal equilibrium temperature model is established based on the detection signals to evaluate the winding temperature T. The two models are then fused, and a small gain k_f is used to allow the winding temperature T to slowly converge towards the resistance model temperature T̂. And obtain the temperature information T_cur.

Images