Parameter scheme for abnormal noise analysis of electric components (Part 2)

Machine speed fluctuation

Even when the motor is in a stable operating state, the instantaneous speed of the motor will still fluctuate to a certain extent. When the frequency of this fluctuation phenomenon is relatively low, it often gives people a very poor subjective feeling. Therefore, it is necessary to test the motor speed in the experiment. When the motor under test is small or it is inconvenient to directly test the speed for other reasons, the vibration noise signal can also be used to extract the speed. Both PULSE Labshop and BK Connect have the function of automatic speed extraction, and PULSE Labshop supports online real-time speed extraction.

Take the left figure below as an example. Due to the fluctuation of motor speed, the frequency of motor vibration shows obvious periodic changes. This periodic change of frequency is linearly proportional to the periodic change of speed. Therefore, these vibration spectra can be used to extract speed data. The result of the right figure below is the speed data extracted from the left data.

Relative Pitch

When the motor is running stably, if there is a deviation and fluctuation in the speed, the harmonic components of the motor noise will be deviated and fluctuated. In order to quantify this phenomenon, the relative pitch parameter can be used. Referring to the SAE paper (SAE 2019-01-1521), the definition of relative pitch is Relative Pitch = 19.9317 * log10(f/fmax)

Where f is the harmonic frequency at each moment during operation, and fmax is the maximum value of the harmonic frequency during operation.

If the noise harmonic frequency is not easy to obtain, the harmonic components in the speed signal or vibration signal can also be used for calculation. The following figure is an example of the relative pitch analysis result. The upper and lower limits of the relative pitch change are set according to the average result of the offset and the tolerance range.

Order Analysis and Order Tracking

For rotating machinery, order analysis and order tracking are two common analysis methods. In the order analysis results, the horizontal axis is frequency. As shown in the left figure below, there are some bright diagonal lines. The slope of the diagonal line corresponds to the ratio of the X-axis frequency to the Y-axis rotation frequency, which is the corresponding order. There is a bright diagonal line of the 69th order in the figure, which means that the 69th order has a large noise. In the order tracking result on the right, taking the same original time domain data as an example, the horizontal axis is the order, and the corresponding 69th order is a vertical bright line.

The two methods can provide similar information, but there are also obvious differences. In the order analysis on the left, it is convenient to view the relationship between the natural frequency and the order, and in the order tracking on the right, it is convenient to view the data of different orders. Another difference is the order resolution. At low speeds, the bright lines of each order in the order analysis are very dense (the closer to the origin, the denser the diagonal lines), which is not conducive to distinguishing different orders. In order tracking, no matter what range the speed is in, the same order resolution can be maintained (each order is a vertical line with a fixed interval). Both methods have their own advantages. In practical applications, it is necessary to choose the appropriate method according to the focus of the analysis.

(non-zero origin) positive and negative order

In the noise and vibration signal of the motor, there are positive and negative orders around the controller switching frequency. BK Connect can display positive and negative orders in the spectrum cloud map. As shown in the left figure below, there is ±2 Order noise (two bright oblique lines) around 1720Hz. The color of +2 Order is brighter at high speed, indicating that the noise is greater. In order to compare the ±2Order noise, the spectrum cloud map is sliced ​​by order (also called order extraction), as shown in the right figure below. Above 2000RPM, +2 Order is significantly higher than -2 Order, and is very close to the full-band Total value. Therefore, +2 Order is the main component of this motor noise at high speed.

Kurtosis and Peak Factor

Motors often produce impact noise during the start-up and stop phases and during operation. When evaluating the amplitude of impact noise, in addition to common parameters such as sound pressure level and loudness, kurtosis and crest factor may also be required. Kurtosis (or kurtosis) is often used in statistics to evaluate the dispersion characteristics of data. The more dispersed the data is, the greater the kurtosis is.

Take the following figure as an example. When impact noise appears in the sound pressure data of the original time domain signal of the noise (in the blue box), the absolute value of the sound pressure value becomes larger, away from 0Pa, and more divergent than when there is no impact noise. Therefore, in the Kurtosis vs. time result on the left, the kurtosis at the moment of impact noise will be significantly greater than at other moments. The crest factor is the ratio of the peak value to the effective value in the signal. Taking the same motor impact noise data as an example, the peak value of the impact noise can also be reflected in the CrestFactor vs. time.

Summarize

Electric components usually include drive motors and actuators, which may produce abnormal noises of different characteristics during operation. When testing and analyzing such abnormal noises, some special parameters are needed to quantify the abnormal noises. HBK's BK Connect software includes a variety of objective parameter calculation functions. Users can use these parameters directly, or derive other parameters based on actual problems with the help of other tools such as MS Excel and MATLAB.

This article combines some measured data and analysis results to introduce various parameters, including:

• Sound Pressure Level (SPL)

• Psychoacoustic parameters: Loudness, Sharpness, Fluctuation Strength, Roughness

• AM parameters: Modulation, Envelope

• Pure tone parameters: Prominence Ratio, Tone-to-noise Ratio, Tonality

• Spectral parameters: FFT, 1/3 Octave, Critical Band

• Statistical parameters: percentile, percentile frequency

• Deviation and fluctuation parameters: Warble, Speed ​​fluctuation, Relative Pitch

• Order analysis and order tracking

• (non-zero origin) positive and negative order

• Kurtosis and Crest Factor

In HBK's previous practical application cases and consulting service projects, these parameters can effectively deal with most abnormal noise problems, and conduct quantitative research on abnormal noise through appropriate objective parameters. In addition to the electric parts industry, the objective parameters mentioned in this article are also applicable to abnormal noise problems in other similar industries.