When machines such as automobiles, production machinery, office equipment, and home appliances are in operation, physical quantities such as sound, vibration, heat, pressure, and light can be detected, allowing us to understand their operating state. In other words, these physical quantities contain information about the machine's operating state. Physical quantities containing such information are generally called "signals." By processing the signals obtained from machines in some way, it is possible to check and diagnose their operating state.
The widespread adoption of computer technology in recent years has made digital signal processing easily possible. The advantages of digital signal processing include not only the ability to process signals with high precision and at high speed with large amounts of data, but also the ease with which processing results can be reported, reused, and communicated.
Physical signals obtained from machines generally change over time. That is, if we record the changes in the signal with time on the horizontal axis, it will be represented as a wave. This is called the "waveform" of the signal, and since the horizontal axis is time, it is specifically called a time-axis waveform or time-axis signal.
Now, there are various ways to classify these signals, but based on how they appear and their properties, they can be broadly divided into three types: periodic signals, transient signals, and irregular (random) signals.
Periodic and transient signals are also called deterministic signals; they are signals whose current value and subsequent changes can be determined at any given point in time (signals that can be expressed by mathematical formulas). In contrast, random signals are signals whose value can be known at a certain point in time, but whose subsequent behavior is completely unpredictable (not deterministic). Simply put, they are "random" or "disorderly" signals.
From the perspective of persistence, transient signals are finite signals resulting from a single, isolated phenomenon, while periodic and random signals persist indefinitely in terms of time. Periodic signals have a waveform that continues infinitely with a constant repetition time (i.e., period), while random signals produce an infinite number of completely random waveforms.
The physical signals detected by machines generally appear as a composite signal of these three types of signals, but in the case of rotating machines, periodic signals synchronized with the rotational speed tend to appear prominently.
(Excerpt from the email newsletter published on October 21, 2002)
