China
Africa
Explore chapters and articles related to this topic
Modulation Systems and Characteristics
Published in Jerry C. Whitaker, Power Vacuum Tubes, 2017
A number of modulating schemes have been developed to take advantage of the noise immunity afforded by a constant amplitude modulating system. Pulse time modulation (PTM) is one of those systems. In a PTM system, instantaneous samples of the intelligence are used to vary the time of occurrence of some parameter of the pulsed carrier. Subsets of the PTM process include the following: Pulse duration modulation (PDM), where the time of occurrence of either the leading or trailing edge of each pulse (or both pulses) is varied from its unmodulated position by samples of the input modulating waveform. PDM also may be described as pulse length or pulse width modulation (PWM).Pulse position modulation (PPM), where samples of the modulating input signal are used to vary the position in time of pulses, relative to the unmodulated waveform. Several types of PTM waveforms are shown in Figure 2.28.Pulse frequency modulation (PFM), where samples of the input signal are used to modulate the frequency of a series of carrier pulses. The PFM process is illustrated in Figure 2.29.
Methodology for data-driven predictive maintenance models design, development and implementation on manufacturing guided by domain knowledge
Published in International Journal of Computer Integrated Manufacturing, 2022
Oscar Serradilla, Ekhi Zugasti, Julian Ramirez de Okariz, Jon Rodriguez, Urko Zurutuza
According to the Nyquist-Shannon sampling theorem, a signal of unknown frequency locations has to be sampled at least at 2 times its frequency in order to enable signal reconstruction, thus maintaining enough information to avoid nonreversible information loss by the aliasing effect, presented by Mishali and Eldar (2009). Anyway, collecting more data than needed is preferable to collecting less than that stated, given that in oversampled data, downscaling is possible, but under-sampled data cannot reconstruct original data correctly. However, big data collection and storage result in higher costs, so the collection strategy should be correctly designed to fit use-cases requirements to reduce costs and computational time. The use of signal processing techniques is encouraged to design a suitable data collection strategy that addresses the use-case’s PdM characteristics. Signal processing techniques can help to determine a suitable sampling frequency. Moreover, signal processing techniques include filters such as IIR Filters, Chebysev, Butterworth or Bessel as stated by Almaged and Hale (2019), which can be used to reduce the bandwidth of a signal that has a higher sampling frequency than required. When the sampling rate of the variables is different, in order to enable data analysis in any timestep for all available variables, timestep by imputation such as repeating last value or interpolation can be useful.