Academia.eduAcademia.edu

Figure 7 – uploaded by Mejosh TJ

See full PDFdownloadDownload figure
Figure 5: Feed Forward NN for sector selection IET-UK International Conference on Information and Communication Technology in Electrical Sciences

Figure 5 Feed Forward NN for sector selection IET-UK International Conference on Information and Communication Technology in Electrical Sciences

Related Figures (24)

Where Vd» Vass lds, Igs Rs Ls R, Ly Lins Ws, Ws, War, Wo and 0, are the d-q axes voltages and currents, stator resistance, stator inductance, rotor resistance, rotor inductance, mutual inductance between the stator and rotor windings, stator flux linkages, rotor flux linkages and the rotor position respectively. transforming the three phase quantities into two phase direct and quadrature axes quantities. The mathematical model in compact form can be given in the stationary reference frame as fallows [8].
Where Vd» Vass lds, Igs Rs Ls R, Ly Lins Ws, Ws, War, Wo and 0, are the d-q axes voltages and currents, stator resistance, stator inductance, rotor resistance, rotor inductance, mutual inductance between the stator and rotor windings, stator flux linkages, rotor flux linkages and the rotor position respectively. transforming the three phase quantities into two phase direct and quadrature axes quantities. The mathematical model in compact form can be given in the stationary reference frame as fallows [8].
Figure 1: Stator and rotor flux-linkage space vectors Where Ge is the angle between the stator and rotor flux linkage space vectors, as shown in Fig.1.
Figure 1: Stator and rotor flux-linkage space vectors Where Ge is the angle between the stator and rotor flux linkage space vectors, as shown in Fig.1.
Figure 2: Proposed ANN based DTC Scheme From equation (5) it is clear that the motor torque can be varied by changing the rotor or stator flux vectors. The rotor time constant of a standard squirrel-cage induction machine is very large, thus the rotor flux linkage changes slowly compared to the stator flux linkage. However, during a short transient, the rotor flux is almost unchanged. Thus rapid changes of the electromagnetic torque can be produced by rotating the stator flux in the required direction, which is determined by the torque command. On the other hand the stator flux can instantaneously be accelerated or decelerated by applying proper stator voltage phasors. Depending on the position of the stator flux, it is possible to switch on the suitable voltage vectors to control both flux and torque.
Figure 2: Proposed ANN based DTC Scheme From equation (5) it is clear that the motor torque can be varied by changing the rotor or stator flux vectors. The rotor time constant of a standard squirrel-cage induction machine is very large, thus the rotor flux linkage changes slowly compared to the stator flux linkage. However, during a short transient, the rotor flux is almost unchanged. Thus rapid changes of the electromagnetic torque can be produced by rotating the stator flux in the required direction, which is determined by the torque command. On the other hand the stator flux can instantaneously be accelerated or decelerated by applying proper stator voltage phasors. Depending on the position of the stator flux, it is possible to switch on the suitable voltage vectors to control both flux and torque.
Table 2: The Voltage Switching Vector look-up table 3 Artificial Neural Network Based Voltage Vector Selection
Table 2: The Voltage Switching Vector look-up table 3 Artificial Neural Network Based Voltage Vector Selection
Table 1: Switching Logic The optimal voltage vector table is given in table 2, where k (k= 1, 2..., 6) indicates the sector position of stator flux.
Table 1: Switching Logic The optimal voltage vector table is given in table 2, where k (k= 1, 2..., 6) indicates the sector position of stator flux.
Artificial Neural Network Based Direct Torque Control of Induction Motor Drives
Artificial Neural Network Based Direct Torque Control of Induction Motor Drives
Figure 4: Feed forward NN for Theta estimation There are three neural networks. First is to estimate the value of stator flux position, 9e as shown in the Fig.4. This is the angle between the stator flux and the rotor flux. It is a two input-one output feed-forward network with three layers. The input layer has 6 neurons of hyperbolic tangent sigmoid transfer function, first hidden layer has 4 neurons of log sigmoid transfer function and the output layer has 1 neuron of linear function.The necessary steps to adjust these weights associated with the hidden neurons can be found in [13]. The training method used was the Levenberg-Marquardt back- propagation method [14]. The input given are the‘d-axis stator flux’ and ‘q-axis stator flux’.
Figure 4: Feed forward NN for Theta estimation There are three neural networks. First is to estimate the value of stator flux position, 9e as shown in the Fig.4. This is the angle between the stator flux and the rotor flux. It is a two input-one output feed-forward network with three layers. The input layer has 6 neurons of hyperbolic tangent sigmoid transfer function, first hidden layer has 4 neurons of log sigmoid transfer function and the output layer has 1 neuron of linear function.The necessary steps to adjust these weights associated with the hidden neurons can be found in [13]. The training method used was the Levenberg-Marquardt back- propagation method [14]. The input given are the‘d-axis stator flux’ and ‘q-axis stator flux’.
Figure 6: Feed Forward NN for sector selection Last neural network is for the selection of voltage vector as given in Fig.6, which is based on three inputs, flux, torque and the sector.
Figure 6: Feed Forward NN for sector selection Last neural network is for the selection of voltage vector as given in Fig.6, which is based on three inputs, flux, torque and the sector.
Figure 7(a): Stator Current response of conventional DTC Scheme for all conditions.
Figure 7(a): Stator Current response of conventional DTC Scheme for all conditions.
Figure 7(b): Stator Current (Zoomed) response of conventional DTC Scheme for no load condition
Figure 7(b): Stator Current (Zoomed) response of conventional DTC Scheme for no load condition
First the conventional DTC scheme is applied to the induction motor to check the performance under no load and then switch to full load (12-Nm) condition at 0.4 sec. The speed reversal has been applied at 0.55 sec and the results of stator currents for all conditions are presented in Fig.7 (a). Fig.7(c)-(e) shows the torque and the speed response under steady state and transient condition for conventional DTC scheme. Fig.7 (b) and Fig.7 (d) shows the results of the stator current and torque, (zoomed for comparison) present more ripples. The estimated stator flux angle, sector position and flux trajectory are presented in Fig.7. (f)- (h) for the conventional DTC scheme.
First the conventional DTC scheme is applied to the induction motor to check the performance under no load and then switch to full load (12-Nm) condition at 0.4 sec. The speed reversal has been applied at 0.55 sec and the results of stator currents for all conditions are presented in Fig.7 (a). Fig.7(c)-(e) shows the torque and the speed response under steady state and transient condition for conventional DTC scheme. Fig.7 (b) and Fig.7 (d) shows the results of the stator current and torque, (zoomed for comparison) present more ripples. The estimated stator flux angle, sector position and flux trajectory are presented in Fig.7. (f)- (h) for the conventional DTC scheme.
Artificial Neural Network Based Direct Torque Control of Induction Motor Drive:
Artificial Neural Network Based Direct Torque Control of Induction Motor Drive:
Figure 7(e): Speed response of conventional DTC Scheme Figure 8(d): Zoomed torque response for time interval 0.14sec to 0.28 sec of Fig.8 (c)
Figure 7(e): Speed response of conventional DTC Scheme Figure 8(d): Zoomed torque response for time interval 0.14sec to 0.28 sec of Fig.8 (c)
Figure 8(c): Torque response of ANN based DTC scheme for all conditions
Figure 8(c): Torque response of ANN based DTC scheme for all conditions
The results for ANN based DTC scheme under the same loading conditions, speed reversal as in case of conventional scheme are presented in Fig.8.(a)- Fig.8.(h). Fig.8 (h) shows that trajectory is smoothened, and therefore torques ripple and stator current harmonic content reduction is expected. Fig. 8.(a) -Fig.8(b) and Fig.8.(c)-(d) shows that the ripples in the stator current and torque is comparatively less which supports accuracy in the ANN based estimator and validation for ANN based DTC scheme.
The results for ANN based DTC scheme under the same loading conditions, speed reversal as in case of conventional scheme are presented in Fig.8.(a)- Fig.8.(h). Fig.8 (h) shows that trajectory is smoothened, and therefore torques ripple and stator current harmonic content reduction is expected. Fig. 8.(a) -Fig.8(b) and Fig.8.(c)-(d) shows that the ripples in the stator current and torque is comparatively less which supports accuracy in the ANN based estimator and validation for ANN based DTC scheme.
Figure 7 (g): Sectors under conventional DTC Scheme
Figure 7 (g): Sectors under conventional DTC Scheme
Figure 8 (h): d-q-axis stator flux plot (NN based DTC Scheme) Figure 8(g): Sectors under ANN based DTC Scheme
Figure 8 (h): d-q-axis stator flux plot (NN based DTC Scheme) Figure 8(g): Sectors under ANN based DTC Scheme
Figure 8(f): Stator flux angle under ANN DTC Scheme
Figure 8(f): Stator flux angle under ANN DTC Scheme
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved