Induction motors are broadly used as drivers of a large variety of industrial equipment. A proper measurement of the motor rotation speed is essential to monitor the performance of most industrial drives. As an example, the measurement of rotor speed is a simple and broadly used industrial method to estimate the motor’s efficiency or mechanical load. In this work, a new low-cost non-intrusive method for in-field motor speed measurement, based on the spectral analysis of the motor audible noise, is proposed. The motor noise is acquired using a smartphone and processed by a MATLAB-based routine, which determines the rotation speed by identifying the rotor shaft mechanical frequency from the harmonic spectrum of the noise signal. This work intends to test the hypothesis that the emitted motor noise, like mechanical vibrations, contains a frequency component due to the rotation speed which, to the authors’ knowledge, has thus far been disregarded for the purpose of speed measurement. The experimental results of a variety of tests, from no load to full load, including the use of a frequency converter, found that relative errors on the speed estimation were always lower than 0.151%. These findings proved the versatility, robustness, and accuracy of the proposed method.

Induction motors are key pieces of equipment in today’s society, powering a variety of industrial drives and home appliances. The induction motor speed is often used to monitor the performance of all kinds of industrial drives. For example, in the industrial field, the motor speed is very often used to determine the efficiency and mechanical load of motors. In this work, a new simple, low-cost, and nonintrusive procedure is proposed for infield measurement of induction motors speed, which is based on the spectral analysis of the vibration signal of the motors. The motor vibration signal is first acquired using the accelerometers integrated into a basic phone. The acquired signal is then treated by a MATLAB-based algorithm, which can determine the motor speed by identifying the mechanical frequency of the rotor shaft from the harmonic content of the vibration signal. In this way, it is shown that the mechanical frequency corresponding to the speed of rotation of the motors can be acquired by means of the embedded accelerometers of a common smartphone, avoiding the acquisition and installation of external accelerometers. To the authors’ knowledge, this could be the first time that a smartphone has been proposed as a practical means of measuring the speed of a motor by analysing its vibration. Experimental results from an extensive set of tests, including the supply of the motor from a frequency converter, show that the speed can always be measured with a relative error of less than 0.15%.

High-voltage direct current (HVDC) using voltage source converter (VSC) in transmission systems applications are currently a competitive alternative to the traditional AC transmission systems, especially for offshore wind power applications. The increases of rated power and distance to the shore have made VSC-HVDC transmission systems economically more efficient than the conventional solution based on an AC lines. Locating a fault in a submarine DC line must be fast and accurate because of the high cost of the submarine repairs as well as the operation cost (not-supplied energy). This paper proposed a fault location methodology based on artificial neural networks (ANN) for VSC-HVDC transmission system. The methodology only uses instantaneous values of electrical quantities (voltage and current) at one of the VSC terminal eliminating the problem of synchronisation. The proposed methodology has been tested and demonstrated using a typical VSC-HVDC test network, and simulation results show the appropriate performance of the methodology. Keywords: Artificial Neural Networks, Fault Location, HVDC Transmission, VSC-HVDC

line 1: 1 st Given Name Surname line 2: dept. name of organisation (of Affiliation) line 3: name of organisation (of Affiliation) line 1: 1 st Given me Surname line 2: dept. name of organisation (of Affiliation) line 3: name of organisation (of Affiliation

The computational burden and the time required to train a deep reinforcement learning (DRL) can be appreciable, especially for the particular case of a DRL control used for frequency control of multi-electrical energy storage (MEESS). This paper presents an assessment of four training configurations of the actor and critic network to determine the configuration training that produces the lower computational time, considering the specific case of frequency control of MEESS. The training configuration cases are defined considering two processing units: CPU and GPU and are evaluated considering serial and parallel computing using MATLAB ® 2020b Parallel Computing Toolbox. The agent used for this assessment is the Deep Deterministic Policy Gradient (DDPG) agent. The environment represents the dynamic model to provide enhanced frequency response to the power system by controlling the state of charge of energy storage systems. Simulation results demonstrated that the best configuration to reduce the computational time is training both actor and critic network on CPU using parallel computing.