Torque / suspension pulsation suppression method in commutation process of magnetic suspension motor
By adopting a comprehensive control strategy of PI controller, allocation function, inverse model, model prediction control and online optimization in the magnetic levitation motor, the problem of torque and levitation force pulsation during phase commutation is solved, and better suspension performance and higher speed applications are achieved.
Patent Information
- Application Number
- CN202510242461.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-03
- Publication Date
- 2025-05-27
AI Technical Summary
During the phase exchange process, there are problems of torque and buoyancy force pulsation, which affects its suspension performance and operating stability.
The control strategy including PI controller, torque and buoyancy force distribution function, torque current and buoyancy current inverse model, model prediction control module, power converter and overlapping interval online optimization is adopted. The torque/suspended force pulsation during phase commutation is suppressed by defining the torque distribution function and buoyancy force distribution function, the angle within the online optimization distribution function, and the current tracking algorithm.
It effectively weakens the torque and levitation force pulsation during phase commutation, improves the suspension performance of the magnetic levitation motor, and broadens its application at higher speeds.
Smart Images
Figure CN120049787A_ABST
Abstract
Description
Technical Field
[0001] The present invention discloses a method for simultaneously suppressing torque / suspension pulsation during the commutation process of a magnetic levitation motor, which relates to motor technology, control technology, automation and intelligent control, and energy and power systems. Background Art
[0002] A magnetic levitation motor is a motor that does not require mechanical contact. It uses magnetic force (usually electromagnetic force) to levitate the rotor part and make it operate without friction. This can significantly reduce wear, improve efficiency, and allow for higher operating speeds. The research on magnetic levitation technology can be traced back to the 1950s and 1960s, initially mainly applied to maglev trains. In these systems, magnetic force is used to levitate the train on the track, thereby reducing friction. With the development of material science and control theory, magnetic levitation motors have gradually moved from the laboratory to practical applications. The core of a magnetic levitation motor lies in the suspension system, which usually uses electromagnets or superconductors to generate sufficient magnetic force to keep the rotor levitating in the air. The suspension system can be active (using feedback control) or passive (using permanent magnets). The drive system involves converting current into rotational motion. Usually, the principles of synchronous motors or asynchronous motors are used, and the rotation of the rotor is achieved by controlling the interaction of current and magnetic field. The control system of a magnetic levitation motor is very complex and requires real-time monitoring and adjustment of the electromagnetic force to keep the rotor in a predetermined position. Common methods include PID control, fuzzy control, and adaptive control, etc. To control the speed and position of the rotor, advanced algorithms are usually used to precisely adjust the current and magnetic field. For example, vector control technology and direct torque control are commonly used drive control technologies. Magnetic levitation motors have significant advantages in high-speed, precision, and high-efficiency applications. For example, in high-precision machining equipment, laboratory equipment, spacecraft, and high-speed transportation systems, magnetic levitation motors can bring extremely high performance improvements. Maglev trains are one of the most successful applications of magnetic levitation technology, such as the superconducting maglev train (Maglev) in Japan and the Transrapid system in Germany. Summary of the Invention
[0003] The present invention discloses a method for simultaneously suppressing torque / suspension pulsation during the commutation process of a magnetic levitation motor. This control strategy includes three key technologies: defining a torque distribution function and a suspension force distribution function, online optimizing the internal angle of the distribution function, and a current tracking algorithm.
[0004] The technical solution of the present invention is as follows: A method for suppressing both torque / suspension pulsations during the commutation process of a magnetic levitation motor. The control method includes a PI controller, a torque and suspension force distribution function, an inverse model of torque current and suspension current, a model predictive control module, a power converter, and online optimization of the overlap interval connected in sequence. The PI controller outputs the given torque and suspension force to the torque and suspension force distribution function to generate the torque and suspension force that each phase should generate. Secondly, the inverse model of torque current and suspension current solves the given current according to the given torque and suspension force. Subsequently, the model predictive control module generates the optimal switching state based on the given current and the feedback current and outputs it to the power converter, thereby generating the voltage signal for driving the motor. Finally, the overlap angle is adjusted online based on the feedback speed and current.
[0005] Furthermore, the torque and suspension force distribution function is as follows:
[0006]
[0007] where f Tk (θ), f Fk (θ) are the torque distribution function and the suspension force distribution function respectively, θ is the rotor position angle, θ Ton is the conduction angle of the torque current, θ Toff is the turn-off angle of the torque current, θ Tov is the commutation overlap angle of the torque current, θ Fon is the conduction angle of the suspension current, θ Foff is the turn-off angle of the suspension current, θ Fov is the commutation overlap angle of the suspension current, θ ov is the commutation overlap angle.
[0008] Furthermore, the torque and suspension force distribution function can make the total output torque and suspension force remain constant during commutation, thereby effectively suppressing the torque / suspension force pulsations during the commutation period.
[0009] Furthermore, the inverse model of torque current and suspension current is as follows:
[0010]
[0011] where J(θ) is the torque coefficient, k i is the current stiffness coefficient, k x is the displacement stiffness coefficient, i T is the torque current, T is the torque, i F is the suspension current, F * is the given suspension force, and x is the displacement.
[0012] The inverse model of torque and suspension force converts the given torque and suspension force into the given current.
[0013] Furthermore, the inverse models of the torque current and the suspension current need to solve for the given current based on the acquired position signals, displacement signals, and the torque and suspension forces of each phase calculated by the torque and suspension force distribution function.
[0014] Furthermore, the model predictive control module optimizes the current tracking strategy, enabling more accurate current tracking during commutation and improving the suspension performance; to protect the hardware circuit of the bearingless switched reluctance motor (BSRM), a protection factor is added to the model predictive control module. If the output switch state is unreasonable, the model predictive control module will ignore it to avoid damage to the hardware circuit.
[0015] Furthermore, the model predictive control module takes the given current and the current detected by the current sensor as inputs and outputs the optimal switch state of the power converter.
[0016] Furthermore, the power converter is an asymmetric half-bridge, which converts the optimal switch state output by the model predictive control into the voltage signal for driving the motor.
[0017] Furthermore, the calculation formula for the commutation overlap angle of the torque current and the suspension current is as follows:
[0018]
[0019] where ω is the angular velocity of the motor, L min is the minimum value of the phase inductance, U is the bus voltage, J(θ) is the torque coefficient, L max is the maximum value of the inductance, k i is the current stiffness coefficient, k x is the displacement stiffness coefficient, x is the displacement; the commutation overlap angles of the torque current and the suspension current are both adjusted according to the rotational speed, and the overlap interval during the commutation period can be adjusted online, so that the current has sufficient time to commutate; each angle in the distribution function is given by the following formula:
[0020]
[0021] where θ Ton is the conduction angle of the torque current, θ Toff is the turn-off angle of the torque current, θ Tov is the commutation overlap angle of the torque current, θ Fon is the conduction angle of the suspension current, θ Foff is the turn-off angle of the suspension current, θ Fov is the commutation overlap angle of the suspension current, θ ov is the commutation overlap angle, θ ac is the rotor position at the maximum value of the inductance.
[0022] When θ FonWhen the limit value is reached, the angle needs to be adjusted according to the following formula to meet the requirements of high-speed operation:
[0023]
[0024] where θ Fon is the conduction angle of the suspension current, θ Foff is the turn-off angle of the suspension current, θ Fov is the commutation overlap angle of the suspension current, θ ov is the commutation overlap angle, ω is the angular velocity of the motor, U is the bus voltage, L max is the maximum value of the inductance, k i is the current stiffness coefficient, k x is the displacement stiffness coefficient, and x is the displacement.
[0025] Furthermore, the calculation of the overlap angle requires the input of the rotational speed and the given current, and the output torque current commutation overlap angle and the suspension current commutation overlap angle are given to the torque and suspension force distribution functions.
[0026] The beneficial effects of the present invention adopting the above technical solutions are as follows:
[0027] 1. The problem of simultaneous torque and suspension force pulsations during commutation in the magnetic levitation motor is weakened, enabling the magnetic levitation motor to obtain better suspension performance.
[0028] 2. The on-line adjustment of the angle in the distribution function enables the magnetic levitation motor to operate at higher rotational speeds, broadening its operating range. Description of the Drawings
[0029] Figure 1 As shown, it is a cosine-type distribution function.
[0030] Figure 2 As shown, it is a schematic diagram of current inductance.
[0031] Figure 3 As shown, it is the angle optimization rule.
[0032] Figure 4 As shown, it is the overall block diagram of the control method Detailed Embodiment
[0033] Next, the technical solution of the present invention will be clearly and completely described in conjunction with the attached Figures 1-4 First, the torque distribution function and the suspension force distribution function
[0034] The torque distribution function (TSF) defines the torque generated in each stage, making the total torque constant and effectively weakening the commutation torque pulsation. In a rotor cycle angle τ, the cosine-type TSF of the k-th phase is as follows:
[0035] The torque distribution function (TSF) defines the torque generated in each stage, making the total torque constant and effectively weakening the commutation torque pulsation. In a rotor cycle angle τ, the cosine-type TSF of the k-th phase is as follows:
[0036]
[0037] Among them, f Tk (θ) is the torque distribution function, θ is the rotor position angle, θ Ton is the conduction angle of the torque current, θ Toff is the turn-off angle of the torque current, θ Tov is the commutation overlap angle of the torque current, θ ov is the commutation overlap angle.
[0038] The suspension force distribution function is as follows:
[0039]
[0040] In the formula, f Fk (θ) is the suspension force distribution function, θ is the rotor position angle, θ Fon is the conduction angle of the suspension current, θ Foff is the turn-off angle of the suspension current, θ Fov is the commutation overlap angle of the suspension current, θ ov is the commutation overlap angle.
[0041] At any moment, the sum of the torque and suspension force distribution functions of each phase winding must be equal to 1, satisfying:
[0042]
[0043] Second, online optimize the distribution function angle
[0044] The torque distribution function and the suspension force distribution function together contain 6 angles: θ Ton , θ Toff , θ Tov , θ Fon , θ Foff , θ Fov . In order to reduce the influence of the trailing current on the suspension system when the torque current is turned off, the torque current needs to be reduced to 0 in advance when the suspension current is turned on. Therefore, the 6 angles need to satisfy the following formula:
[0045]
[0046] Among them, θ is the rotor position angle, θ Ton is the conduction angle of the torque current, θ Toff is the turn-off angle of the torque current, θ Tov is the commutation overlap angle of the torque current, θ Fon is the conduction angle of the suspension current, θ Foff is the turn-off angle of the suspension current, θ Fov is the commutation overlap angle of the suspension current, θ ov is the commutation overlap angle, θ endis the angle when the torque current drops to 0, θ ac is the angle when the phase inductance reaches the maximum value.
[0047] As can be seen from the above equation, as long as θ Tov , θ Fov is determined, the remaining angles can be determined.
[0048] Neglecting the winding resistance and edge effect, the commutation overlap angle θ Tov in the torque system can be obtained as follows:
[0049]
[0050] where θ Tov is the commutation overlap angle; ω is the angular velocity of the motor; L min is the minimum value of the phase inductance; U is the bus voltage; T is the bus voltage; ψ(θ, i) is the magnetic flux. Since the current has dropped to 0 at θ Fon , the voltage equation can be written as:
[0051]
[0052] where U is the bus voltage, L max is the maximum value of the inductance, I ref is the given value of the current, are the times corresponding to θ ac , θ Fon respectively. According to ω is the angular velocity of the motor. The above equation can be transformed into:
[0053]
[0054] Substituting θ ac - θ Fon = θ Fov into it, we get:
[0055]
[0056] where k i is the current stiffness coefficient, k x is the displacement stiffness coefficient, x is the displacement, and F is the suspension force. However, due to the existence of the commutation overlap angle of the suspension current, the suspension current needs to be turned on in advance, which will affect the torque system. In order to weaken the coupling between the two as much as possible, θ Fov needs to be further optimized.
[0057] The torque coefficient curve of the three-phase 8 / 4 BSRM is such that in the range of θ ∈ [21.5°, 22.5°], the torque coefficient is relatively low and the rate of decline is relatively fast. At this time, the suspension current is small and still in the rising stage. Even if the suspension current is turned on in advance, it will not have a great impact on the torque system. Therefore, in this paper, the suspension current turn-on angle θ Fon is advanced to 21.5° at most. At this time, it is necessary to re-correct the relational expressions of each angle of the suspension system to adapt to a higher rotational speed:
[0058]
[0059] Third, current tracking algorithm
[0060] Due to the poor performance of the traditional current hysteresis loop tracking, which brings the problems of large output torque and suspension force ripple, this paper adopts model predictive control to improve the current tracking accuracy of the system.
[0061] Establish a state equation with current as the state variable:
[0062]
[0063] where i is the phase current, R is the winding resistance, L is the winding inductance, and U is the bus voltage.
[0064] Discretize the above formula to obtain the prediction model of the bearingless switched reluctance motor (BSRM) drive circuit:
[0065]
[0066] In the formula, T s is the sampling period, i(k + 1) is the predicted current of the motor winding at the (k + 1)-th moment, and i(k) is the winding current at the k-th moment.
[0067] In order to select the optimal control signal, calculate the three different switch state combinations of the asymmetric half-bridge drive circuit respectively, compare the magnitudes of the cost function values, and select the switch state combination with the smallest cost function value as the purpose of minimizing the given reference current error of the magnetic axis. Establish the following cost function, and select the switch signal with the smallest cost function value as the control signal of the magnetic bearing drive circuit, that is, solve the optimal switch control signal.
[0068] J = (i ref (k + 1) - i p (k + 1)) 2
[0069] where, i p (k + 1) is the predicted current of the motor winding at the (k + 1)-th moment; i ref (k + 1) is the given current of the winding at the (k + 1)-th moment.
[0070] In addition to controlling the winding current to track the reference current in real time, it is also necessary to consider the maximum current that the motor winding and the switching power amplifier can withstand. Therefore, it is necessary to limit the amplitude of the winding current. Specifically, an item S(i) can be added to the cost function to ensure that the winding control current does not exceed the maximum withstand current, avoiding burning out the winding and the switching power amplifier and causing unnecessary losses. The expression of S(i) is as follows:
[0071]
[0072] where I max is the maximum winding current that can be withstood, and i is the predicted current.
[0073] When the winding control current is less than the maximum withstand current, the output of this item is 0. Therefore, under normal circumstances, it does not affect the selection of the optimal signal by the cost function. However, when the winding control current is greater than the maximum withstand current, the output of this item is infinite. Therefore, it can effectively avoid outputting the corresponding control signal and avoid damaging the device.
[0074] Taking all factors into consideration, the cost function J should be changed to:
[0075] J = (i ref (k + 1) - i p (k + 1)) 2 + S(i p (k + 1))
[0076] where i ref (k + 1) is the given current at time k + 1; i p (k + 1) is the predicted current at time k + 1; S is the protection factor.
Claims
1. A method for suppressing torque / suspension pulsation during the commutation process of a magnetic levitation motor, characterized in that: The control method includes a PI controller, a torque and suspension force distribution function, a torque current and suspension current inverse model, a model predictive control module, a power converter, and an overlapping interval online optimization that are connected in sequence; the PI controller outputs a given torque and suspension force to the torque and suspension force distribution function to generate the torque and suspension force that each phase should generate; Secondly, the torque current and suspension current inverse model solves the given current according to the given torque and suspension force; then, the model predictive control module generates the optimal switching state according to the given current and feedback current and outputs it to the power converter, thereby generating a voltage signal to drive the motor; finally, the overlap angle is adjusted online by the feedback speed and current.
2. A method according to claim 1, characterized in that: The torque and suspension force distribution function is as follows: where f Tk (θ), f Fk (θ) are torque distribution function and suspension force distribution function, θ is the rotor position angle, θ Ton is the torque current conduction angle, θ Toff is the torque current cut-off angle, θ Tov is the torque current commutation overlap angle, θ Fon is the conduction angle of the suspension current, θ Foff is the suspension current cut-off angle, θ Fov is the floating current commutation overlap angle, θ ov is the commutation overlap angle.
3. A method according to claim 2, characterized in that: The torque and suspension force distribution function can keep the total output torque and suspension force constant during the commutation period, thereby effectively suppressing the torque / suspension force pulsation during the commutation period.
4. A method according to claim 1, characterized in that: The inverse model of torque current and suspension current is as follows: Where J(θ) is the torque coefficient, k i is the current stiffness coefficient, k x is the displacement stiffness coefficient, i T is the torque current, T is the torque, i F is the suspension current, F * is the given suspension force and x is the displacement. The inverse torque and suspension force model converts the given torque and suspension force into a given current.
5. A method according to claim 4, characterized in that: The torque current and suspension current inverse model needs to solve the given current through the collected position signal, displacement signal, and the torque and suspension force of each phase calculated by the torque and suspension force distribution function.
6. A method according to claim 1, characterized in that: The model predictive control module optimizes the current tracking strategy so that the current can be tracked more accurately during commutation, improving the suspension performance. In order to protect the bearingless switched reluctance motor (BSRM) hardware circuit, a protection factor is added to the model predictive control module. If the output switching state is unreasonable, the model predictive control module will ignore it to avoid damage to the hardware circuit.
7. A method according to claim 6, characterized in that: The model predictive control module outputs the optimal switching state of the power converter based on the given current and the current detected by the current sensor.
8. A method according to claim 1, characterized in that: The power converter is an asymmetric half-bridge, which converts the optimal switching state output by the model predictive control into a voltage signal to drive the motor.
9. A method according to claim 1, characterized in that: The calculation formula of the torque current and suspension current commutation overlap angle is as follows: Where, ω is the motor angular velocity, L min is the minimum phase inductance, U is the bus voltage, J(θ) is the torque coefficient, L max Maximum inductance, k i is the current stiffness coefficient, k x is the displacement stiffness coefficient, x is the displacement; the torque and suspension current commutation overlap angles are adjusted according to the speed, and the overlap interval of the commutation period can be adjusted online, so the current has time to fully commutate; the various angles in the distribution function are given by the following formula: where θ Ton is the torque current conduction angle, θ Toff is the torque current cut-off angle, θ Tov is the torque current commutation overlap angle, θ Fon is the conduction angle of the suspension current, θ Foff is the suspension current cut-off angle, θ Fov is the floating current commutation overlap angle, θ ov is the commutation overlap angle, θ ac is the rotor position when the inductance is maximum. When θ Fon When the limit value is reached, the angle needs to be adjusted according to the following formula to meet the needs of high-speed operation: where θ Fon is the conduction angle of the suspension current, θ Foff is the suspension current cut-off angle, θ Fov is the floating current commutation overlap angle, θ ov is the commutation overlap angle, ω is the motor angular velocity, U is the bus voltage, L max Maximum inductance, k i is the current stiffness coefficient, k x is the displacement stiffness coefficient, and x is the displacement.
10. A method according to claim 1, characterized in that: The overlap angle calculation requires the input of the speed and the given current, and the output torque current commutation overlap angle and the suspension current commutation overlap angle are given to the torque and suspension force distribution function.