In this section the identification and estimation approaches described in the previous section are tested on an induction motor setup, where inter-turn stator faults can be sim-ulated. The electrical circuit of the stator is shown in Fig. 5.3. The motor used in the
Motor ω Pump
r
Tl
Vs is Q,H
Software algorithm Real world
Observer
Ob
Observer
Oc
abc abc
abc bca
abc cab
is
xb
is
xc
-eyb
eyc Observer
Oa
is
xa
- e
ya
Figure 5.2: The structure of the identification algorithm for identification and estimation of inter-turn short circuits. Hereeya, eyb and eyc are used for identification of the affected phase, andxˆa,xˆbandxˆcare the estimates of the states including the estimates of the fault currentif and the fault sizeγ.
tests is a 1.5 [KW] customized Grundfos motor, supplied with a Danfoss frequency con-verter. The speed, the three phase currents, and the three phase voltages are available at the test setup. The voltage to the motor is controlled using a linear voltage to frequency relation, with a voltage boost at low frequencies. All tests are preformed at supply fre-quencies around 30 [Hz] to avoid too large short circuit currents and thereby burnout of the motor during the tests. The tests are performed with the induction motor connected in a∆-connection as it is shown Fig. 5.3. However, similar results can be found for a Y-connected motor in (Kallesøe et al., 2004c).
In the first, of the two following subsections, the identification capabilities of the proposed algorithm are tested. In the second subsection the estimation capabilities of the adaptive observer are tested. In these tests the algorithm is tested against three different operating conditions. These are,
• Constant speed at 25 [Hz] supply frequency and balanced supply voltage.
• Speed changes at every 1 second between 25 and 40 [Hz] and balanced supply voltage.
• Constant speed at 25 [Hz] supply frequency and a 5 % voltage decrease in phase
a
25%5%
25%
5%
b c
Figure 5.3: The electrical circuit of the stator in the test setup. Two points of phasesa andband their end points are available at the terminal box.
a, meaning that the supply voltage is unbalanced.
5.3.1 Test of Identification Capabilities
In this subsection the identification capability of the identification algorithm, presented in Section 5.2.3, is tested. Three tests are performed, each testing one of the three operating conditions described above. In each of the tests a short circuit of 5% of the windings is introduced in phaseaandcrespectively. The results of the test with constant speed and balanced supply are shown in Fig. 5.4, the results of the test with speed changes and balanced supply are shown in Fig. 5.5, and finally the results of the test with constant speed and unbalanced supply are shown in Fig. 5.6.
All the tests show that the phase, in which the fault is introduced, can be recognised by the level of the observer error signal. From all three tests it is seen that this error signal is considerable lower for the observer modelling the particular fault. However, it is also seen that the level of the observer error signal is changing in the case of a fault, even in the observer modelling the particular fault. This is especially a problem in the case of a fault in phasea, see Figs. 5.4 and 5.5. This unexpected behaviour is due to an inherent imbalance between the phases in the costumer-designed motor used in the tests.
The phenomenon is not so dominant in Fig. 5.6, where the supply voltage is unbalanced.
This is because the unbalance in the voltage does account for some of the imbalance of the phases.
From Fig. 5.5, presenting the results of the test with the speed changes, it is seen that the error signal is in average larger and is oscillating compared to the two other tests. This is due to the violation of the constant speed assumption in the design of the adaptive fault observers. However, it is still possible to recognize the phase, in which the fault is introduced, using the observer error signal.
Comparing the results of Figs. 5.4 and 5.6 it is seen that, beside of the problem with
0 5 10 15 0
1 2 3 4
eA
0 5 10 15
0 1 2 3 4
eB
0 5 10 15
0 1 2 3 4
eC
time [sec]
Turn fault in phase a
Turn fault in phase c
Figure 5.4: The mean square error of the observersOa,Ob, andOc respectively. In this test the speed is constant and the supply voltage is balanced, and faults are imposed seperately in phaseaandc.
the inherent imbalance in the phases, the results are comparable. This shows that the algorithm is able to handle unbalanced supply conditions, which also was expected as no assumption were put on the supply voltage in the design. This means that the observer can manage any distortion of the supply voltage as long as it does not introduce too large oscillations in the speed.
5.3.2 Test of Estimation Capabilities
In this subsection the estimation capability of the adaptive observer, derived in the pre-vious section, is tested. The observer is tested under the three different operating con-ditions described in the start of this section. In each of the tests the algorithm is tested with no short circuit, 5% of the windings short circuited, and 25% of the windings short circuited in phasea. The results of the test with constant speed and balanced supply are shown in Fig. 5.7(a) and 5.7(b). The results of the test with speed changes and bal-anced supply are shown in Fig. 5.8(a) and 5.8(b), and finally the results of the test with constant speed and unbalanced supply are shown in Fig. 5.9(a) and 5.9(b).
All the tests have shown that the observer is stable. From the first test, presented in Fig. 5.7(a) and 5.7(b), it is seen that the speed is estimated without any bias. It is also seen that there is a bias on the estimated fraction of turns in the short circuit. This bias is
0 5 10 15 0
1 2 3 4
eA
0 5 10 15
0 1 2 3 4
eB
0 5 10 15
0 1 2 3 4
eC
time [sec]
Turn fault in phase a
Turn fault in phase c
Figure 5.5: The mean square error of the observersOa,Ob, andOc respectively. In this test the speed is variating and the supply voltage is balanced, and faults are imposed seperately in phaseaandc.
partly due to noise on the measurements, partly due to mismatch between the real motor parameters and the motor parameters used in the observer, and partly due to the initial imbalance between the three stator phases. This bias is repeated in each of the three tests.
Results from the second test, presented in 5.8(a) and 5.8(b), show that the observer is capable of estimating the wanted quantities despite of speed changes. Still it is seen that the speed changes affect the estimated amount of turns in the short circuit. This is because of the constant speed assumption used in the design. It is, however, still possible to use the estimate of the fault.
From the results of the last test, presented in 5.9(a) and 5.9(b), it is seen that an unbalanced supply of 5% is not affecting the performance of the observer.