• Ei tuloksia

3 EXCITATION CONTROL OF DTC SYNCHRONOUS MOTOR DRIVES

3.3 Combined open loop and feedback control

The excitation control in DTC synchronous machine drives has formally the same features as the excitation control in network connected synchronous generators. There is no stator current control, but stator currents and load angle are formed according to the load and the controlled field current.

Additionally the frequency and the modulus of the stator flux linkage vector are controlled in DTC.

The excitation control has the same goals in both cases. The dynamic stability is one of the key issues from the excitation control point of view. Further the power factor is controlled by adjusting the field current.

The main goal in synchronous generator excitation control is to prevent electromechanical oscillations in the generator and to secure the generator and the power system stability during the load transients. The main difficulties in the feedback control are the nonlinear character of a synchronous machine and the parameter variations due to saturation and temperature variation. Hsu and Wu (1988) have proposed an adaptive method, where the process is linearised piece wise and the controller variables are updated according to the working point. Also the feedback can be linearised, as e.g. in the excitation control method represented by Mielczarski and Zajaczkowski (1991). Nonlinear methods have been applied for the synchronous generator excitation control as well, e.g. by Xianrong et al (1993). Adaptive state feedback control has been proposed by Fork and Schreurs (1988). Even fuzzy sets (Handschin et al 1993) and neural networks (Zhang and El-Hawary 1994) have been proposed for the excitation control of synchronous generators.

Despite of the formal similarities the same methods are not necessarily applicable for the excitation control of synchronous generators and DTC synchronous motor drives. The operating range of DTC controlled synchronous motor, including the field weakening range, is large compared to that of a synchronous generator and thus the parameter variation is larger. The linearisation in the case of DTC is difficult due to the wide working area. On the other hand, the stator voltage is a controlled variable in the DTC system unlike in the case of synchronous generators. It is thus obvious, that the excitation control has different roles in the DTC synchronous motor drives and in synchronous generator control.

The dynamic behaviour of the machine is an important issue in both cases. In the case of DTC the main control variables, which are the stator flux linkage modulus and the torque, can be controlled accurately to a certain extent even without any field current adjustment. However, since the maximal static and dynamic performance of the DTC synchronous motor drive was found to be dependent on the field current adjustment, the excitation control should be able to follow the main control loop of DTC in load transients.

The open loop control of a synchronous motor magnetic state has been widely used in current vector controlled synchronous motor drives. In the transvector control of a smooth air gap synchronous motor (Bayer et al 1971) the magnetising current reference im,ref is calculated according to the air gap flux linkage modulus |ψm|. It is further divided into the rotor magnetising current reference imf,ref and the stator magnetising current reference ims,ref. The stator field current ims,ref occurs only in the dynamic changes

i l

ms,ref m ref mf,act m

,

The synchronous motor control method proposed by Mård et al (1990-I) calculates the field current reference value as follows

i

( )

l i

which keeps the load angle δ constant. Niiranen (1993) proposes an excitation control method according to the reaction excitation control combined with the power factor correction term if,corr

if,ref lmd i i

md

d,ref f,corr

=ψ − +

. (3.8)

The reaction control part of the field current reference compensates the d-axis damper winding current and requires fast dynamics for the excitation unit. The power factor correction works with a slow time constant.

Good dynamics for the excitation unit is required also in the control method proposed by Yamamoto et al (1993), where the inverse dynamic model of a synchronous motor is used. The field current reference is then

( ) ( ( ) ( ) ) ( ( ) ( ) ) ( )

Eq. (3.9) is presented in the Laplace domain. The given examples show, that in current vector controlled synchronous motor drives the excitation control is normally related to the stator current control and no fast feedback is utilised as in the excitation control methods for synchronous generators. Due to the missing feedback of the machine magnetic state, the drive performance is dependent on the accuracy of the saturation model. DTC is less dependent on the saturation model, since the voltage model is not inductance parameter dependent. For this reason, the open loop excitation control method can very well be used for the electrically excited DTC synchronous motor drive.

Eq. (2.41) makes it possible to relate the excitation control to the DTC main control and it can be implemented as an open loop control. This is useful, since the magnetic energy balance between the stator and the rotor will be found rapidly after a load change without oscillations, which might occur in the case of an unoptimally tuned feedback control. Furthermore, this method allows the excitation control with the same speed as the torque is controlled. That is important for both the stability and the dynamic performance of a drive, especially in the field weakening range.

The drawback of the open loop excitation control is, that the control method is not able to compensate erroneous inductance values. Excitation curve calculation can be improved by using inductance parameters given by a saturation model. The saturation model is required also for the current model, which is needed specifically in the low speed operations, where the voltage model is no more reliable.

The minimum for the stator current and unity power factor is obtained, when the stator current vector is and the stator flux linkage vector ψs are orthogonal to each other. The orthogonality of two vectors can be investigated with the scalar product. If unity power factor is desired, the scalar product can be used to calculate an error signal ε for the excitation control

ε =ψs⋅ =is ψsx sxisy syi , (3.10)

where the stator reference frame oriented current and stator flux linkage components are used. The DTC modulation principle causes modulation noise in the stator current. The magnitude of the ripple is both switching frequency and transient inductance dependent. For that reason the error signal in Eq. (3.10) is also disturbed. A further source of error in Eq. (3.10) occurs to be the stator flux linkage estimate. Even a small error in the stator flux linkage estimate causes eccentricity in the real stator flux linkage. As a result, the stator current will be eccentric too, and the error signal in Eq. (3.10) will oscillate with the basic electrical frequency.

Because of inaccuracies in the feedback signal of Eq. (3.10) the feedback signal should be filtered with a low pass filter before using it for control purposes. The power factor control is then slow compared to the torque control dynamic requirements and the control method is not sufficient to handle fast load transients.

By combining the fast open loop control implemented by the excitation curve calculation, Eq.

(2.41), and the slow feedback control for power factor correction with error signal from Eq. (3.10), a good excitation control method for DTC synchronous machine drive can be achieved. This control method fulfills two main features, which are; a quick response in load transients in order to ensure the drive stability and a stator current minimisation with the slow power factor control. The block diagram of the excitation system is represented in Fig. (3.3).

i

Figure 3.3 DTC synchronous motor drive excitation system with a fast open loop control and a feedback power factor control, which uses filtered values ψs,f and is,f. The feedback uses P-controller with a gain Pex. The output of the feedback controller has been limited.