Abdelaziz Lakehal

Fault diagnosis of electrical power transformer based on water content analysis using Bayesian network

Water content and breakdown voltage of dielectric oil are generally unstable parameters. Exceeding limit permissible threshold of one of parameters implies corrective actions because they are directly related to the oil ability to isolate. In this paper a model based on a Bayesian network (BN) is used to diagnose the causes of transformer failures. The proposed model is used to diagnose the water content in the oil, and to predict the breakdown voltage. A case study of a main transformer (MT) of a power plant is presented to show the effectiveness of our model.

Optimisation of an emergency plan in gas distribution network operations with Bayesian networks

The focus of this paper is on the development of a predictive emergency plan for operating gas networks in crisis situations (major incident or disaster). The first contribution of this paper is summarised in the first part, which defines the essential elements of an emergency plan and the centre position of reliability in such plan. A Bayesian model is implemented; it allows the estimation of probabilities of valve closure based on system-level performance for isolating gas distribution network. Also it allows the prioritisation of revamp work and capacity-upgrade actions related to existing gas pipe networks for the sake of safe operations. Finally a case study of a distribution network supplying a city is presented. The paper demonstrates that Bayesian networks allow for the management and the predictive control of gas networks and the simulation of effect of maintenance and investment actions on the performance of isolating system.

Suggestion for a new design of the piezoresistive accelerometer

ABSTRACT The piezoresistive accelerometer is an electronic device used to measure the vibration level of rotating machines; we can say that an accelerometer more precise and more sensitive is a more reliable sensor. In our paper, based on the improvement of some parameters of piezoresistive accelerometer for develop their accuracy and their sensitivity. This improvement comes from extracting a suitable mathematical model of the accelerometer; this model allows proposing a new design of piezoresistive accelerometer.

New model of piezoelectric accelerometer relative movement modulus

A piezoelectric accelerometer is the first element of a vibration measurement chain, and its improvement can enhance measurement quality. In this work, a new model of relative movement modulus as a function of measurement error is developed, thus minimizing the measurement error and increasing the measurement precision of the accelerometer. Therefore, a precise relationship between the movement relative frequency and the piezoelectric accelerometer natural frequency is extracted in order to determine a good frequency range relative to the piezoelectric accelerometer. Thus, the failure risk of the resonance phenomenon is minimized and the accelerometer piezoelectric reliability is optimized. The developed model is confirmed and validated by experimental tests. The purpose and the objective of this work is to improve the performance and the design of the accelerometer.

Vibration sensor mechanical sensitivity improvement

In this paper, we research the mechanical sensitivity of the vibration sensor; The mechanical sensitivity mathematical model is developed in function of the relative movement modulus in firstly and in function of the measurement error in secondly. The developed model allowed improving the mechanical sensitivity by enhancing the performance of vibration sensor (the best choice of damping rate and the frequency ratio). The right choice of the damping rate and the frequency range allowed keep the mechanical sensitivity constant. This model is validated by computer simulation and experimental tests.