Data processing: measuring – calibrating – or testing – Measurement system – Speed
Reexamination Certificate
2002-09-04
2004-02-24
Bui, Bryan (Department: 2863)
Data processing: measuring, calibrating, or testing
Measurement system
Speed
C702S147000
Reexamination Certificate
active
06697759
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to a method of determining the speed of rotation of a motor and to a computer software product according to the method. The device the speed of rotation of which is to be determined is one that comprises a rotating shaft, such as an electric motor, or a similar device comprising a rotating shaft.
BACKGROUND OF THE INVENTION
The invention relates to monitoring the condition of a motor. Motor condition monitoring based on the vibration produced by the motor and a condition analysis made on the basis of the vibration is widely known in the art. There are several motor faults, such as unbalance, looseness, resonance, bearing failures, and the like, which cause abnormal vibration of the motor.
A commonly applied condition indicator is the root mean square (rms) value of a vibrational velocity signal in the region of 10-1000 Hz. The allowed limits are given in ISO standards, such as ISO-10816. The root mean square value allows motor faults to be detected, but it cannot be used for identifying the type of fault concerned. An experienced serviceman, however, is often able to analyse the spectrum of the vibration signal and thereby evaluate also the type of fault.
It is known art to carry out measurements relating to condition monitoring of motors in the following way, for example. When a fault is suspected in a motor, or in connection with a routine maintenance of the motor, the vibration of the motor is measured using a suitable vibration measuring sensor. The measurement is carried out by coupling a speed sensor or an acceleration sensor connected to a data collection means, such as a PC, to the motor, the measurement being then carried out by taking samples of the motor vibration in axial, horizontal and vertical directions at the driven end of the motor shaft and also in one direction at the opposite shaft end. The measurement data obtained from the measuring sensor is stored in the memory of the data collection means and processed by applying software which uses the vibration data to produce a vibration spectrum which is then analysed visually.
Depending on the speed of rotation of the motor's rotor, the direction and magnitude of the forces causing vibration in the motor vary. For this reason mechanical motor faults often cause motor vibration which has a cycle length which is inversely proportional to the speed of rotation of the motor's rotor. The faults thus cause periodic vibrations in the motor, the frequencies of which can be found out by spreading the vibration signal measured from the motor into frequency components. The form of the vibration signal and the periodic frequencies it comprises depend on the motor type, the speed of rotation and the type of malfunction.
For the above reasons, accurate determining of speed is of primary importance in fault analysis. In prior art measurement solutions, rotation speed of motors is measured using tachometer or stroboscope measurements.
There are, however, major drawbacks in the prior art. When a tachometer is used for measuring speed of rotation, there must be one mounted in the motor, or one must be mounted for the measurement. Correspondingly, when a stroboscope is used, the motor must contain the means for carrying out the measurement. Motors do not usually have built-in speed measurement devices, but the motor must be halted for mounting one. However, motors used in industrial processes cannot usually be halted without causing undue harm for the process in which the motor is one component.
A fault analysis method in which measurement data obtained from a measuring sensor is stored in the memory of a data collection means and processed using software which produces from the vibration data a vibration spectrum for visual analysis requires a discrete Fourier transform (DFT) of the signal. In DFT the measurement time and the frequency resolution (the distinction between two consecutive frequency points) are interrelated in that the better the desired resolution, the longer is the measurement time required. A long measurement time is a problem, because the motor load, and thereby its speed of rotation, should remain constant during the measurement to allow accurate and reliable measurement data to be collected. When the motor is used in an industrial process, this is not, however, usually possible without causing undue harm for the process. Secondly, a set of frequency points where the calculation is to be carried out is determined in advance in the DFT method on the basis of the measurement time and sampling frequency. If the speed of rotation is not exactly the same as the frequency at any of the predetermined frequency points, error will occur in the estimation of fault frequencies which are proportional to the speed of rotation, and, consequently, amplitude estimates will also be erroneous.
As illustrated above, drawbacks that often appear in connection with the prior art include the need to halt the motor for the mounting of the speed measurement device, the need for a plural number of measurement devices and, thereby, the need to carry out various measurements to allow an analysis to be made. Moreover, an accurate analysis requires a long measurement time, during which a constant speed of rotation of the motor is required. This naturally slows down and complicates the measurement, and impairs its accuracy and reliability.
BRIEF DESCRIPTION OF THE INVENTION
It is an object of the invention to alleviate the drawbacks of the prior art, and to provide an improved method of determining speed of rotation and computer software implementing the method.
This is achieved by a method and computer software of the present invention which comprise the characteristics set forth in the claims. More precisely, a method according to a preferred embodiment of the invention is primarily characterized by what is stated in the characterizing part of claim
1
.
An underlying idea of the invention is that speed of rotation of a motor is determined by measuring the mechanical vibration of the motor with a suitable measuring sensor. Measurement data is collected from the vibration at a suitable frequency for a predetermined measurement period. The measurement data is stored in the memory of a data carrier as measurement data of a fixed format. The speed of rotation of the motor is then determined from the measurement data by determining the cycle length of a periodic vibration signal in the time domain, and not in the frequency domain, by using a maximum likelihood estimate (MLE) calculated by maximizing a maximum likelihood function (MLF) of the measurement data. The speed of rotation is thus at the frequency point where the maximum likelihood function (MLF) obtains its highest value.
Before the ML function is calculated, the vibration signal is filtered using a bandpass filter on the passband of which the rotation frequency estimated in advance is located. An advantage of this is that interference in the measurement data, if any, can be attenuated.
The speed of rotation is determined using the maximum likelihood estimate of the time domain. This provides an advantage in that the measurement time needed for determining the frequency is now significantly shorter than in the conventional DFT method. The reason for this is that in the maximum likelihood method, frequency is produced as a continuous variable and not as separate values in which the minimum difference between two frequency values, i.e. their resolution, would be determined by the measurement time and the sampling frequency, as in the commonly used DFT method of the frequency domain. In the maximum likelihood method the only factor having an effect on the accuracy at which the speed of rotation can be determined is the magnitude of interference in the measurement signal.
Practice has shown that for the impact of the interference that is in the measurement signals to be eliminated, the length of the measurement period must be more than 50 times the cycle length of the vibration signal. As a result, the required measurement time is sho
Orkisz Michal
Saarinen Kari
ABB Research Ltd.
Bui Bryan
LandOfFree
Method of determining speed of rotation of a motor and a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of determining speed of rotation of a motor and a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of determining speed of rotation of a motor and a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3340291