In this article, we present ARPM (Augmented Reality-based Preventive Maintenance) system, which aims at offering to the technician of a Cement Factory a virtual assistance in real time using various forms of 3D models, images or even texts, in order to help him to maintain and inspect his equipment. This system not only offers the possibility of reducing costs of the preventive maintenance, but also allows monitoring and even anticipating failures. In order to ensure a reliable maintenance operation, we adapted our system to be compatible with the CMMS system (Computerized Maintenance Management System) already exists in the Cement factory

In this paper, we propose an algorithm for classification based on extreme learning machine, named TS-ELM (Two feature Spaces Extreme learning Machine). It is a hybridization of two distinct feature mapping of SLFNs (Single hidden layer feedforward neural network). Both are realized based on ELM theories. The first feature space resulted from a random kernel distributions of a randomly connected input samples, that inspired initially from conventional random connected neural network and it is also included in ELM theories. The second feature space that gives the training or testing samples is the result of a random probability distribution of fully connected samples taken from the first feature space, and adjusted with their own weight and threshold parameters. This algorithm was applied on three datasets of multiclass and binary classification, and the results shows that it has strong classification capabilities and universal approximation as well. The mechanisms of the algorithm and universal approximation are discussed in this paper.

Automatic Speech Recognition can be considered as a transcription of spoken utterances into text which can be used to monitor/command a specific system. In this paper, we propose a general end-to-end approach to sequence learning that uses Long Short-Term Memory (LSTM) to deal with the non-uniform sequence length of the speech utterances. The neural architecture can recognize the Arabic spoken digit spelling of an isolated Arabic word using a classification methodology, with the aim to enable natural human-machine interaction. The proposed system consists to, first, extract the relevant features from the input speech signal using Mel Frequency Cepstral Coefficients (MFCC) and then these features are processed by a deep neural network able to deal with the non uniformity of the sequences length. A recurrent LSTM or GRU architecture is used to encode sequences of MFCC features as a fixed size vector that will feed a multilayer perceptron network to perform the classification. The whole neural network classifier is trained in an end-to-end manner. The proposed system outperforms by a large gap the previous published results on the same database.

The automation of production systems has been an
answer to the changing and competitive industrial context and
works by extracting data and experiences of experts. This
automation is a double-edged sword; on one hand, it increases
the productivity of the technical system (cost reduction,
reliability, availability, quality), but, on the other hand, it
increases the complexity of the system. This has led to the need
of efficient technologies, such as Supervisory Control and Data
Acquisition (SCADA) systems and techniques that could
absorb this complexity such as artificial intelligence and fuzzy
logic. In this context, we develop an application that controls
the pretreatment and pasteurization station of milk localized in
Batna (Algeria) by adopting a control approach based on
expert knowledge and fuzzy logic.