Active stabilization of DC link in motor drive systems

Abstract

The present disclosure provides a system and method that provides active damping at the input of a motor drive system without the need for the hardware used in conventional and RC damping circuits. According to the disclosure a virtual damping network is realised at the input of the motor drive system based on modification of field-oriented control, FOC. More specifically, a control algorithm creates a virtual damping impedance at the motor drive input by applying a damping algorithm to both q and d components of the motor current.

Claims

1. A method of providing active damping to a motor drive system, the method comprising: determining a flux component and a torque component of a motor drive current; comparing the flux component and torque component with a respective desired flux component and desired torque component; using a controller to provide a motor control output from each comparison; adding, to each motor control output, a result of the active damping implemented using a transfer function with constant coefficients Ad_d(s) and Ad_q(s) performed on a voltage measured at an input to the motor drive system, wherein the coefficients of the transfer function are adapted depending on an operating point; pulse width modulating the damped motor drive output to provide a three phase motor drive output; and tuning the transfer function, wherein an input admittance of the motor drive system is derived from: ${Z_{\mathrm{damp}}^{-1}\left(s\right).Math.}_{\mathrm{Surface}\mathrm{PM}\mathrm{motor}}=\left(\frac{A_{d\_q}\left(s\right).Math.\left(I_{\mathrm{Lq}}.Math.Z_L\left(s\right)+V_i.Math.D_q\right)+A_{d\_d}\left(s\right).Math.\left(I_{\mathrm{Ld}}.Math.Z_L\left(s\right)+V_i.Math.D_d\right)}{Z_L\left(s\right)+{\mathrm{PI}}_i\left(s\right).Math.F_i\left(s\right).Math.K_{\mathrm{PWM}}.Math.e^{-{\mathrm{sT}}_d}}\right).Math.F_v\left(s\right).Math.K_{\mathrm{PWM}}.Math.e^{-{\mathrm{sT}}_d},$ where K.sub.PWM is a gain of a PWM modulator, Fi(s) is the transfer function of an AC current sensor, Fv(s) is the transfer function of a DC-link voltage sensor, ZL(s) relates to a series inductive impedance of an electrical motor, Pli(s) relates to a current controller Pli(s), ILq, ILd represent operating point of the machine currents in d-q frame, Dd and Dq represent operating point of the duty cycle in d-q frame, and Vi is operating point of the DC-link voltage and Td represents a total loop delay, and wherein the transfer function is tuned to achieve a predetermined value for input admittance.

2. The method of claim 1, further comprising: adding a function of the voltage measured at the input, to the result of the active damping network functions.

3. The method of claim 1, wherein the motor drive system includes a plurality of motor drives or other three phase voltage source converters.

4. The method of claim 1, further comprising: activating or deactivating the damping according to predetermined criteria related to the motor drive system.

5. A method of providing active damping to a motor drive system, the method comprising: determining a flux component and a torque component of a motor drive current; comparing the flux component and torque component with a respective desired flux component and desired torque component; using a controller to provide a motor control output from each comparison; adding, to each motor control output, a result of the active damping implemented using a transfer function with constant coefficients Ad_d(s) and Ad_q(s) performed on a voltage measured at an input to the motor drive system, wherein the coefficients of the transfer function are adapted depending on an operating point; pulse width modulating the damped motor drive output to provide a three phase motor drive output; wherein: $A_{d\_q}\left(s\right)=\frac{Z_L\left(s\right)+{\mathrm{PI}}_i\left(s\right).Math.F_i\left(s\right).Math.K_{\mathrm{PWM}}}{I_{\mathrm{Lq}}.Math.Z_L\left(s\right)+V_i.Math.D_q}.Math.\frac{1}{F_v\left(s\right).Math.K_{\mathrm{PWM}}}.Math.\frac{1}{Z_{\mathrm{emulated}\_q}\left(s\right)}$ $A_{d\_d}\left(s\right)=\frac{Z_L\left(s\right)+{\mathrm{PI}}_i\left(s\right).Math.F_i\left(s\right).Math.K_{\mathrm{PWM}}}{I_{\mathrm{Ld}}.Math.Z_L\left(s\right)+V_i.Math.D_q}.Math.\frac{1}{F_v\left(s\right).Math.K_{\mathrm{PWM}}}.Math.\frac{1}{Z_{\mathrm{emulated}\_d}\left(s\right)}$ where K.sub.PWM is a gain of a PWM modulator, Fi(s) is the transfer function of an AC current sensor, Fv(s) is the transfer function of a DC-link voltage sensor, Zemulated is an emulated damping impedance, Pli(s) relates to a current controller Pli(s), ILq, ILd represent operating point of the machine currents in d-q frame, Dd and Dq represent operating point of the duty cycle in d-q frame, and Vi is operating point of a DC-link voltage and Td represents a total loop delay.

6. The method of claim 5, further comprising: adding a function of the voltage measured at the input, to the result of the active damping network functions.

7. The method of claim 5 wherein the motor drive system includes a plurality of motor drives or other three phase voltage source converters.

8. The method of claim 5, further comprising: activating or deactivating the damping according to predetermined criteria related to the motor drive system.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a simplified circuit diagram of a DC-fed drive system incorporating the active damping technique of this disclosure.

(2) FIG. 2 shows a schematic of the implementation of the active damping control

(3) FIG. 3 is a model representing the virtual damping impedance resulting from the active damping technique of the disclosure.

(4) FIG. 4 illustrates the behaviour of the input voltage and input current of a motor drive system connected to a DC bus, with and without active damping.

(5) FIG. 5 illustrates a possible implementation of d-q current set points for the proposed active damping method.

DETAILED DESCRIPTION

(6) Referring first to FIG. 1, a typical DC-fed motor drive system can be seen for driving a motor 1 to drive e.g. an actuator 2. The motor is a three-phase motor. The motor drive is a DC-fed system. Therefore, a DC link 3 is provided between the source 4 and the motor drive. Conventionally, as described above, a filter 5 is provided in the DC link, in the form of capacitors and inductors. A damping circuit 6 is also provided—here in parallel with the filter capacitor (other damping realizations are possible). The technique of the present disclosure involves the incorporation of an active damper control to create a virtual impedance at the input of the motor drive and this avoids the need for the damper circuit (shown in FIG. 1 crossed out). A smaller filter capacitor can also be used because of the active control technique of this invention. The power source, in the example shown is a DC source, but this can also be an AC source or DC source of another voltage value.

(7) The control technique of this disclosure takes the voltage across the filter capacitor at the input of the motor drive as an input and also reference d-q motor current, and motor speed, as will be discussed further below.

(8) The technique of this disclosure is a modification of the known field-oriented control (FOC), also known as vector control. In FOC, the stator currents of the three-phase motor are identified as two orthogonal components represented by a vector. The FOC aims to optimise magnetic field of the motor by adjusting the phase of the voltage applied to the motor so as to ensure the current components are at the desired value.

(9) The modified technique of the disclosure will now be explained in more detail with specific reference to FIG. 2.

(10) The top part of FIG. 2 shows a conventional power switching system of a motor drive including the filter capacitor and the switches 7 that provide controlled output to drive one or more motors 1.

(11) The control for active damping is shown in detail in the bottom part of FIG. 2. With reference to FIG. 1, it will be seen that this control will be incorporated across the power switching system.

(12) In the bottom part of FIG. 2, the elements to the left of the drawing, surrounded by dotted lines correspond to elements that are known from conventional FCO. The components towards the middle of the drawings, within full lines, are optional elements as described below. The elements within dashed lines are features required to perform the modified technique of this disclosure. This is clearly just one realisation of the system structure. Note that the elements inside the FOC are represented by transfer functions in continuous time domain, but implementations in discrete domain may be used in practice.

(13) To control the drive of the motor, the motor stator current is separated into two components, a flux or direct component d and a torque component q. A desired motor speed ω.sub.mref is provided as a control input and is compared, in a comparator, with an actual motor speed ω.sub.m. From this, using a proportional integral (PI) controller 8 PI.sub.w, a reference or target value for the torque component of the stator current i.sub.Lqref is determined. The target flux current component i.sub.Ldref is provided as another input to the control system. Other controllers different from PI may be used as well.

(14) The actual d and q components of motor current are then compared, respectively, with the target or reference values and the results of the comparison are provided to respective controller 9, 10 (e.g. PI controllers). The PI outputs are used as duty cycle commands, in d-q frame, (i.e. dd, dq) for a pulse-width modulator (PWM). Scaling factors as a function of input voltage may be used (see division blocks).

(15) In a conventional FOC, the voltage outputs would be pulse width modulated to provide drive outputs to drive the motor. If the system is controlled in this manner, the motor drive system behaves as a constant power load at frequencies within the bandwidth of the control system, i.e. the controller compensates input voltage disturbances so that, if the input voltage increases, the input current decreases so that the power flow into the system is constant. From a dynamic standpoint, a constant power load behaves as a negative resistance, and this may lead to an unstable system unless a relatively large dc-link capacitor or passive damping network is incorporated into the system.

(16) As indicated in the introduction, some methods have been proposed in the literature to mitigate this issue, e.g. see [3] and [4]. The idea is to measure the dc-link voltage, perform convolution through a certain differential equation, and apply a compensation signal into the FOC control so that the negative resistance effect is compensated at the resonance frequency of the LC filter. This way, closed-loop operation of the system can be achieved while the instability problem is addressed. A problem with this approach is that the active damping effect depends on the operating point of the d-q currents and the duty cycle. Depending on the system parameters, this may lead to operating conditions where no active damping effect is achieved, in particular at light loads. This may be problematic in motor drive systems that make use of state-of-the art capacitor and inductor technology to realize an input filter with high Q factors, as this may lead to amplification of oscillations in the DC-bus that could impact the system operation.

(17) The aim of the present technique, as mentioned above, is to achieve active damping at all operating points.

(18) The present technique modifies the above-described FOC by adding an additional input-output dynamic network 12 as shown in the dashed lines in FIG. 2. The DC link voltage measured at the terminals of the DC link filter capacitor is added, via respective linear time invariant transfer functions Ad_d(s) and Ad_q(s) to each of the d and q outputs of the FOC (prior to the PWM). The exponential factor e.sup.(−sT) here represents the inherent loop time-delay present in synchronous digital implementations, where T is the sampling period of the loop.

(19) As mentioned above, the dynamic networks for active damping in FIG. 2 are implemented using linear time invariant transfer functions, which can be realized in the discrete domain using a discrete difference equation. However, other realizations may be also possible (e.g. realizations using adaptive coefficients depending on operating point).

(20) This modification to FOC introduces a virtual impedance into the control (shown in the model in FIG. 3).

(21) Using the modification of addition of the linear time invariant transfer functions, the input impedance of the motor drive is modified so that constant power load behaviour is not exhibited at the input port of the motor drive, at the resonant frequency of the input filter. Instead, the motor drive exhibits the behaviour of a damping network in parallel with the normal input impedance of the drive using conventional FOC, i.e. emulating a classical passive damping means described above. A model illustrating the concept is shown in FIG. 3.

(22) Furthermore, as innovation over the state of the art, by modulating the “d” current reference set point of the drive as proposed in this invention at low torque levels and speed levels, the system can actively damp the DC-link at all operating points. The adaptation of set points is illustrated in FIG. 5. When torque is low (i.e. the q component is low), the “d” component is increased, in this modified technique to ensure that the motor drive has damped input impedance at all operating points, and not only at rated power. It is important, particularly in aerospace applications, to handle susceptibility requirements even at low and medium power levels. It should be noted that other means to increase damping performance at light loads may also be applied, e.g. using linear time-variant networks to implement Ad_d and Ad_q and adapting them as a function of the operating point. In FIG. 5, linear evolution of the set points is illustrated with respect to the torque demand, but other evolutions are possible.

(23) Thus, when perturbations are applied in the input voltage at this resonant frequency, the (modified) motor impedance is used to perform damping.

(24) This invention may also be applied to introduce active damping into a plurality of motor drive systems connected in a common DC-bus. In such a scenario, either some of them, or all of them, may be fitted with the proposed active damping method.

(25) Furthermore, the technique could also be fitted with means to detect existence of oscillations in order to enable or disable the active damping.

(26) According to a particular aspect of this disclosure, a new tuning process for the linear time invariant transfer functions Ad_d(s) and Ad_q(s) has also been developed so that the virtual impedance can resemble a desired passive network (e.g. like a conventional parallel RC damping circuit). This tuning process simplifies the design process for engineers as they can use well-established equations for passive damping design, and they can readily predict the expected performance of the drive at the input port, and achieve similar behaviour as obtained by conventional passive dampers.

(27) The tuning process also facilitates prediction of the behaviour of the active damping technique by analytical means. The tuning process developed by the inventors is simple and is described below.

(28) By performing dynamic analysis of the system shown in FIG. 2, expression (1) can be derived for the input admittance of the motor drive system. This expression assumes constant current reference set points, which is a reasonable approximation if the resonance frequency to be damped is well above the bandwidth of the speed loop. The meaning of those parameters is depicted in FIG. 2. K.sub.PWM is the gain of the PWM modulator, Fi(s) is the transfer function of the AC current sensor, Fv(s) is the transfer function of the DC-link voltage sensor, ZLd(s) relates to the series inductive impedance of the electrical motor (Ld=Lq in this case for the sake of simplicity). PIi(s) relates to the current controller PIi(s), ILq, ILd represent the operating point of the machine currents in d-q frame, Dd and Dq represent the operating point of the duty cycle in d-q frame, and Vi is the operating point of the DC-link voltage. Td represents the total loop delay. This equation assumes the duty cycles to be scaled as a function of the input voltage. The dependency of the input admittance with respect to the operating point, given by current, duty cycle and input voltage, is clearly seen in this equation.

(29) $\begin{array}{cc}{Z_{\mathrm{damp}}^{-1}\left(s\right).Math.}_{\mathrm{Surface}\mathrm{PM}\mathrm{motor}}=\left(\frac{A_{d\_q}\left(s\right).Math.\left(I_{\mathrm{Lq}}.Math.Z_L\left(s\right)+V_i.Math.D_q\right)+A_{d\_d}\left(s\right).Math.\left(I_{\mathrm{Ld}}.Math.Z_L\left(s\right)+V_i.Math.D_d\right)}{Z_L\left(s\right)+{\mathrm{PI}}_i\left(s\right).Math.F_i\left(s\right).Math.K_{\mathrm{PWM}}.Math.e^{-{\mathrm{sT}}_d^{}}}\right).Math.F_v\left(s\right).Math.K_{\mathrm{PWM}}.Math.e^{-{\mathrm{sT}}_d}& \left(1\right)\end{array}$

(30) The transfer functions corresponding to the active damping, namely Ad_d(s) and Ad_q(s) can be tuned so that the input admittance, Ydamp(s), resembles a certain admittance defined by the user. If the equivalent circuit illustrated in FIG. 5 is targeted with Zdamp being a parallel RC damper, expressions (2) and (3) results for Ad_d and Ad_q. Those expressions are made up by a term representing the dynamic compensation of the motor drive dynamics, and a second term including the emulated damping impedance.

(31) $\begin{array}{cc}{A}_{d\_q}\left(s\right)\underset{\mathrm{Dynamic}\mathrm{Compensation}}{\underbrace{=\frac{Z_L\left(s\right)+{\mathrm{PI}}_i\left(s\right).Math.F_i\left(s\right).Math.K_{\mathrm{PWM}}}{I_{\mathrm{Lq}}.Math.Z_L\left(s\right)+V_i.Math.D_q}.Math.\frac{1}{F_v\left(s\right).Math.K_{\mathrm{PWM}}}}}.Math.\underset{\mathrm{Emulated}\mathrm{damping}}{\underbrace{\frac{1}{Z_{\mathrm{emulated}\_q}\left(s\right)}}}& \left(2\right)\\ A_{d\_d}\left(s\right)\underset{\mathrm{Dynamic}\mathrm{Compensation}}{\underbrace{=\frac{Z_L\left(s\right)+{\mathrm{PI}}_i\left(s\right).Math.F_i\left(s\right).Math.K_{\mathrm{PWM}}}{I_{\mathrm{Ld}}.Math.Z_L\left(s\right)+V_i.Math.D_q}.Math.\frac{1}{F_v\left(s\right).Math.K_{\mathrm{PWM}}}}.Math.}\underset{\mathrm{Emulated}\mathrm{damping}}{\underbrace{\frac{1}{Z_{\mathrm{emulated}\_d}\left(s\right)}}}& \left(3\right)\end{array}$

(32) The result of expressions for the active dampers, given by (2) and (3), can be subsequently simplified for implementation purposes, or expanded with additional filtering effects at low frequency or high frequency depending on particular application needs.

(33) It should be noticed that the expressions used for the active damper, given in (2) and (3), could be either made adaptive as a function of the operating point, or linear time invariant implementations could be used, depending on particular application needs.

(34) Accordingly, the size of the DC link capacitor can be minimized and/or the damping circuitry can be eliminated. This has a significant impact on reliability and size in the motor drive system.