The aim of the present work is to obtain a perfect compensation by extracting accurate harmonic currents. The objective is to avoid the consequences due to the presence of disturbances in the power system. A comparative study of harmonic currents extraction by simulation and implementation is carried out for two different techniques. The first technique is based on the instantaneous powers, taking advantage of the relationship between current and the power transformed from the supply source to the loads. The second is based on ADALINE neural network. The neural method can estimate the harmonic terms individually and online, therefore, the APF can realise a selective compensation. The developed architectures are validated by computer simulation and experimental tests. The algorithms are implemented in the dSPACE Board in order to show the effectiveness and capability of each technique. The results have demonstrated that the speed and the accuracy of the ADALINE can improve greatly the performances of active power filters

This paper is mainly focused on piezoresistive accelerometer performance improvement. Thus a new model of this sensor is derived to enhance its various parameters, such as precision, sensitivity, and reliability. Moreover, applying this model, a new design of piezoresistive accelerometer can be achieved. The developed model is validated by simulations and confirmed by experimental tests to verify its effectiveness in industrial applications. The comparison study showed that the best choice of damping rate to reduce the measurement error and increase the precision corresponds to the rate chosen in this paper. The results have also demonstrated that a more reliable sensor can be designed compared with the existing designs.