ACOEM (France)

Anomaly detection is a key issue in many fields of application. In condition-based maintenance, we are navigating between two key issues: the health status assessment and the identification of the default (diagnosis itself). The first can be answered by an "unsupervised" approach, which seeks to detect changes in the behavior of complex industrial equipment, in order to detect faults or measurement problems as early as possible. The latter, on the other hand, implies supervised methods that are rarely applicable to complex equipment due to a lack of labeled data. The aim of this paper is to present a methodology for setting up a more suitable monitoring system than the simple detection of threshold exceedances usually used in maintenance. Due to the multiplicity of equipment to be monitored, the automatic anomaly detection process needs to prioritize the equipment to be analyzed and provide key indications to help experts for diagnosis. The main constraint is that the number of false alarms must not overload the control room, while at the same time not tolerating any missed detections. The method we have developed consists in learning and running a behavioral model based on an analysis of the distances between a new measurement and measurements from a reference period, which makes it close to k-NN type methodologies. This methodology named Smart Exception (SE), which meets the needs defined above, incorporates a process for automatically determining the end of the reference period, a definition of the adaptive distance metric, and the ability to reduce false alarms over time by taking continuous human analysts feedback into account.

This article presents the work carried out as part of a project aiming to regulate the noise exposure of workers of PETROBRAS refining units. This is a corporate project carried out over a three years’ period (2012 to 2015) by the Supply business unit of the Brazilian company. The uniqueness of this project is due to its scale. Indeed, this encompassed twelve refineries located throughout Brazil. The corporate approach to deal with noise exposure risk leads to an overall data consolidation allowed by a standardized methodology. The methodology used is split into four phases. First, a long but necessary planning stage, then an exhaustive assessment of the current noise situation of production areas was carried out during the first twelve months of the project. This stage allowed the involvement of several important tools used to identify and rank noise exposure issues. This lead to noise maps of every industrial area and an acoustic database of equipment exceeding regulatory limits (sound power level and directivity). Third, every industrial area was modelled into noise prediction software compliant with ISO 9613-2 standard. Finally, acoustic damping solutions were designed and proportioned by size of equipment. This mixed approach of measurement and numerical simulation mobilized a dedicated ACOEM project team comprising of eight acoustic experts. Principal achievement of this project, concluded in 2015, is a clear vision of the noise control priorities allowed by standardized methodology on all refineries in a short time. This achievement leads to a coherent set of controls for the future investments to be done to mitigate noise exposure risks. In addition to highlighting the benefits of integrated Health Safety and Environment (HSE) management, this work provides an understanding on the use of virtual computing doses from propagating acoustic model.

Unattended noise measurements are more and more common for noise assessment in the environment. Multiple sources are usually measured with a random position with respect to the measurement point. Noise generated by ground transportation, leisure activities, construction sites is coming from all directions, although mainly the horizontal direction. Placed vertically and configured for a reference direction of 90° from its axis, the goal is to meet the requirements of the IEC 61672 standard on sound level meters taking into account noise incidence from the horizontal direction. The main technical difficulty is the criterion for the maximum level difference allowed between two random incidence angles (directivity). The objective can be fulfilled using a cone-shaped device on top of the microphone. When measuring attended noise with the instrument in hand, the sound level meter must be pointed at the source according to standard IEC 60651. The purpose of the paper is to describe the different research & development phases to fulfill IEC 61672 sound level meter standard for 0° and 90° reference directions with the same device.