Method for detecting short-circuits in a coil

1. A method for detecting winding short-circuits in an electric machine comprising:

a) arranging a coil in an air gap arranged between the rotor and the stator of the electric machine;

b) calculating a minimum duration (t min )of two immediately sequential zero crossings of a signal curve U(t) generated by means of the coil, taking the rotational frequency and the number of pole pairs of the electric machine into account;

c) recording the signal curve U(t) generated by means of the coil during operation of the electric machine, having at least the duration of one revolution of the rotor;

d) determining the zero crossings of the signal curve U(t) and storing the times of said zero crossings;

e) determining the zero crossings of the signal curve U(t)-c corrected by an offset c, and identifying at least one pair of immediately sequential zero crossings, the time interval of which is longer than the minimum duration (t min ), where c is not equal to zero;

f) in the event that a pair is not identified in step e), repeating step e) until a pair is identified, wherein the offset c is varied in the direction from the zero point to a global extreme value of the signal curve U(t);

g) identifying at least one of the two stored times which lie between and closest in time to the pair, and;

h) extracting two half-waves from the signal curve U(t) using the times identified in step g), wherein each half-wave corresponds to half a revolution of the rotor;

i) comparing the two half-waves.

2. The method as claimed in claim 1 , further comprising:

e 1 ) determining the zero crossings of the signal curve U(t)-d corrected with an offset d, and identifying at least one pair of immediately sequential zero crossings, whose time interval is longer than the minimum duration (t min ), where d is not equal to 0 and has the opposite arithmetic sign to c;

f 1 ) in the event that a pair is not identified in step e 1 ), repeating step e 1 ) until a pair is identified, wherein the offset d is varied in the direction from the zero point to the other global extreme value of the signal curve U(t); and wherein in step h one of the two half-waves is extracted making use of the time points identified in step e 1 ).

3. The method as claimed in claim 1 ,

wherein the number of pole pairs of the electric machine is one, and

each of the two half-waves is delimited respectively by two of the time points identified in step g) and/or

g 1 ) that are immediately sequential.

4. The method as claimed in claim 3 ,

wherein the number of pole pairs of the electric machine is larger than one, and each of the half-waves is formed of a number of partial waves corresponding to the number of pole pairs,

wherein each partial wave is delimited in each case by two of the immediately sequential times identified in step g) and/or g 1 ).

5. The method as claimed in claim 2 ,

wherein in steps d), e) and/or e 1 ), the zero crossings are determined by formation of y 0 =Ut=α*Ut=α+1 for all the points of the signal curve U(t) and of the corrected signal curve U(t)-c, where Ut=α is a signal value in U(t) or U(t)-c, and Ut=α+1 is the immediately sequential signal value.

6. The method as claimed in claim 5 ,

wherein, in the case where y 0 is negative, the two points associated with Ut=α and Ut=α+1 are linearly interpolated for determination of the zero crossing.

7. The method as claimed in claim 1 ,

wherein the signal curve U(t) exhibits the electrical voltage generated in the coil or the current magnitude generated in the coil.

8. The method as claimed in claim 1 ,

wherein the signal curve U(t) recorded in step c) is smoothed by means of a filter, a Bezier filter, a median filter and/or a gradient filter.

9. The method as claimed in claim 1 ,

wherein the electric machine is a generator, a synchronous machine, and/or an electric motor.