Soft Defects Diagnosis in wired networks – SODDA

Indeed, even if defect information is important, being able to inform before the defect’s appearance would be very advantageous. This is forecast: to detect the premises of the appearance of a defect to warn maintenance or the supervisor in order to increase the lifespan and the reliability of the system.
The objective of the project SODDA is to study the signatures of the soft defects, by combining theory and experiment, and to design and test innovative methods adapted to these signatures which are very difficult to detect. The project will be run by an academic consortium, in close connection with an industrial board, responsible for keeping the work in realistic and relevant use cases.
The approach followed in the project will lean on the study of direct problems as the digital characterizations of the signatures of not true defects and on the study of inverse problems as the identifiability of signatures according to the available measures. The digital approach will be led by using several methods (3D finite elements code optimized for wired geometries, code «Laplace«, FDTD, etc.) To confront the results and guarantee that the data used afterward are validated, it will be completed by experimental measures. The study of detectability of a defect will privilege the case of the not invasive measures such as reflectometry, to avoid perturbing the operation of the cable. The study of identifiability with the aim of a finer diagnosis after detection of the defect, can envision richer measures. In our knowledge, this precise characterization of soft defects and the study of their detectability / identifiability has never been done before.

The results of the direct (given experimental and simulation values of reflectometry) or reverse problems (methods of detection and identification), will serve during the project: on one hand to choose and to direct the methods of diagnosis, on the other hand to supply data for validation on test cases proposed by the industrial committee.
The main breakthroughs of the project are:
• The development of simulation modules specialized in wired geometries,
• The creation of a database of signatures and characterizations of soft defects,
• The study of specific diagnosis methods of these defects,
• The study of signal processing methods capable of extracting from the reflectograms the signatures of soft defects.

The problem of the detection and diagnosis of soft defects becomes more and more relevant, because hard defects (short circuit or opened circuit), even if they are common and important for our industrial partners, represent «accomplished« defects, i.e. their diagnosis is made after they are detected by their consequences on the concerned system. A soft defect is thus two types: defect localized on a weak portion of the cable, or on an insulating material (abrasion, destruction) or on the metal (corrosion, microcut), and defect distributed on the whole length of the cable (ageing). It can be also seen as the premise of a future hard defect, its detection allows then to bring an additional information to the diagnosis: the forecast of the hard defect.
The stakes are thus important, because they concern maintenance as well as safety of the systems. The defect forecast will allow delivering information in advance to prepare and target the maintenance. We shall then speak about preventive maintenance.

Cables are present in electric or electronic systems, in a discrete way because they are considered not very important: their role is however fundamental, (power or information supply). However maintenance is neglected and no system exists to provide a fast and precise diagnosis.
The example of transport: the race for more safety and comfort, plus the constraints of sustainable development, involves an increase of the share of electronics in cars (25% in value). In thirty years, the cable length in cars was multiplied by ten, now more than 4000 meters. In a modern car, one counts up to 100 sensors and 1800 connections: the interconnection network is regarded by manufacturers as a weak part of modern vehicles. The increase in complexity is accompanied by the increase in the failure rate.
To help with maintenance, embarked calculators are equipped with self-diagnosis function. This made it possible to see that 70% of equipment returned to manufacturers were in operating condition. The breakdown came from another thing… the wired network and the connectors.
The example of automotive industry is not isolated. Buildings and infrastructures have higher orders of magnitude: 1000 km of cables in a nuclear plant, 5000 km in a modern building, and 50000 km along railways in France. A new scope of application is appearing: smart grids, intelligent electricity networks which use information technologies to optimize energy production and distribution. Cables diagnosis is fundamental in this field, both for the principal distribution network and the district distributors. This can give nightmares to the agents in charge of maintenance.
Because it is a question of maintenance: to detect the precursory signs of wires or connection defects, to precisely locate the defect to help repair it in a minimum time, if possible without stopping the service. The need for a diagnosis system - detection, localization and characterization informations – for defects in a cables network is present for many applications.
This need was identified few years ago and led to the launching of several projects, funded by the ANR or EUREKA: SEEDS followed by 0-DEFECT in the automotive domain, INSCAN for cables along railways, VECADIS for buried HTA cables. This co-operative work made it possible to provide the foundations of diagnosis methods for cables – with a proof of feasibility in the case of hard defects (short-circuit, open circuit) - and some theoretical results on the associated inverse problems in the case of soft faults. They also made it possible to identify their limits. One of the principal limits of these methods, based on the principles of reflectometry, is the difficulty of detecting soft defects (such as rupture of bits, localized wear of insulator, broken shield) which are often the premises of more annoying problems. If it were possible to detect and locate precisely these defects, that would help for preventive maintenance or prognosis. Indeed, even if defect information is important, being able to inform before the defect’s appearance would be very advantageous. This is forecast: to detect the premises of the appearance of a defect to warn maintenance or the supervisor in order to increase the lifespan and the reliability of the system.
The objective of the project SODDA is to study the signatures of the soft defects, by combining theory and experiment, and to design and test innovative methods adapted to these signatures which are very difficult to detect. The project will be run by an academic consortium, in close connection with an industrial board, responsible for keeping the work in realistic and relevant use cases.