From magnetic signal to wavelet and wirelet

The principle of magneto-inductive testing was discovered by the English in the years 1830 to 1850 with an application dedicated to the inspection of gun barrels. A variant of magnetic rope inspection was developed by MacCann & Colson in the mines in South Africa around 1906. In 1931, Prof. Woernle of the IFT Institute, Stuttgart developed the first magneto-inductive coil technology for intercepting and measuring the voltage induced by the leakage magnetic flux emitted by a wire break in a steel wire rope. Starting in 1931, MRT technology entered the golden age of the development with for instance the Swiss (Signum) and Polish (Krakow Institute) inspection machines.

The source of magnetization of electromagnetic origin (Integra) was replaced in 1970 by permanent magnets (Kündig). In 1981, Polish Stachurski introduced Hall sensors to measure the loss of metallic cross section of a rope. The ferrite magnets are in turn supplanted by the rare earth magnets of the Neodymium type in the early 1990s. Until then, research focused on machine technology. A shift took place from 1985 onwards to the era of software, which took a closer look at signal analysis with an acceleration from 1999 onwards, born of the pioneering work of Mrs Weischedel, Tytko, Chaplin, Dohm and Nussbaum. Dohm is one of the first to point out the weaknesses of MRT analysis by organizing a robin test round demonstrating erroneous predictions in the context of mine cables. A theoretical vacuum related to the prediction of wire breaks is partially filled by the work of Nussbaum which is the cornerstone of future research.

The first softwares dedicated to the analysis of rope magneto-inductive signals are designed by S. Winter, followed by S. Messmer, G. Hinterndorfer and others. Although intensive efforts have been made to improve the design of the auscultation machines, it should be noted that the results have been much worse in the analysis and interpretation of magnetic signals. It is important to remember that steel wire rope technology has evolved considerably with the use of increasingly high-grade wire and a more hostile environment: increasing rope operating speed and a constant trend more cycles of fatigue in a shortened time.

Pr. Pistòra who invented the defectoscope in 1999 is the first to project the idea of ​​using wavelets as a tool for processing magneto-inductive signals. Faced with the theoretical and algorithmic difficulties of wavelet analyzes but also facing a deaf silence of the scientific community, he had to resign himself to giving up his idea which was later taken up by other contemporaries from 2014. In parallel, Mr Weischedel from NDtech was a pioneer who began early in 1985 to use signal processing techniques to isolate the signatures of defects hidden in rope ground signal. In 2013, he developed an ingenious technique called "wire rope roughness" to create indicators of the evolution of defects built on raw signals. Unknowingly, he had constructed a cousin filtering technique rediscovering the precepts of wavelet analysis applied to magnetic signals.

Magnetic rope testing is like a chain comprising four links: the machine, the in-situ test, the analysis and the interpretation of the magnetic signals and finally the reporting step. A fragile or broken link has important implications for the quality of the evaluation. The magneto-inductive method has limitations related to the physics of the process but also to a "faulty" analysis of the signals

Nussbaum was the first to establish the theoretical basis for the induced signals of wire breaks: a broken wire signature is constructed as the superposition of two signals - a signal U1 of loss of cross section of the wire and a signal U2 returning from the section of the wire and has an evanescent waveform which has two negative amplitude nipples separated from a central peak of positive amplitude.
It is striking to note the similarity of the magnetic signal of a wire break with for example the Mexican hat wavelet or the Cauchy wavelet. The first idea is therefore to use a wavelet transform to analyze the content of magnetic signals to identify the presence of these broken wire signatures. After the resemblance traits, it is easy to see nuances between the candidate wavelet patterns. The second idea that follows naturally is the following: the signature of a wire break is itself a wavelet. This is the starting point of wirelets, a family of wavelets whose construction is dictated by the laws of physics.