COMPUTATIONAL INTELLIGENCE IN ELECTRICAL ENGINEERING (INTELLIGENT SYSTEMS IN ELECTRICAL ENGINEERING)
Year:2018 | Volume:9 | Issue:1 |Papers Count:6

Increasing the number of processor cores leads to increasing the density of the computing power processor and also raising the temperature. Temperature management is very important in these processors. Thermal management methods are introduced to reduce the CPU temperature. Reactive and proactive approaches are two sets of these schemes. Unlike the reactive techniques, proactive methods predict the temperature using thermal prediction model before reaching its threshold. In this paper, a hybrid model of several SVR models is proposed for predicting temperature. An appropriate dataset is created for training proposed model that includes a high diversity of processor temperature variations. Some features of dataset are measured using temperature sensors and system performance counters. Other features, with historical and control names are calculated with the proposed processes to increase the accuracy of thermal model. Two SVR models are used in the proposed thermal model to reduce its operational overhead. The proper features for each SVR model are selected by the feature selection algorithm based on mutual information. The proposed model is evaluated for temperature prediction for 2 to 5 time distances. The results show that with a selection of 11 features for thermal prediction model of the next 2 seconds, the mean absolute error is about 0.5oC.

Using an intelligent method to automatically detect sleep patterns in medical applications is one of the most important challenges in recent years to reduce the workload of physicians in analyzing sleep data through visual inspection. In this paper, a single-channel EEG-based algorithm is presented for automatic recognition of sleep stages using complete ensemble empirical mode decomposition and combined model of genetic algorithm and neural network. The signal is decomposed into IMFs using the complete ensemble empirical mode decomposition and statistical properties of each of the inherent state functions are extracted. In order to optimize and reduce the dimension of the feature vectors, a hybrid model of genetic algorithm and multi-layer propagation neural network is used. Then, McNemar's test is used to confirm the accuracy of the selected features. The final classification is performed on these optimized properties by a perceptron neural network with a hidden layer. On the average, classification accuracy of 98.9%, 97.1%, 96.7%, 94.8% and 93.8% are obtained respectively for 2, 3, 4, 5 and 6 classes with corresponding Kappa cohen coefficients of 0.98, 0.95, 0.95, 0.83 and 0.90. The results prove that the proposed sleep stage classification method has better performance compared to the previously existing methods.

In this paper, the effect of connecting the charging station of electric vehicles is investigated considering the optimal charging and discharging scheduling at the station. Accordingly, in the first section, the planning process of the station aimed at maximizing the mutual benefits.In the second part, uncertainty was noted in the program's presence at the station based on the production of different scenarios. The statistical simulation method has been used to model the uncertainties in the problem. Possible scenarios are selected from generated scenarios considering the diversity of vehicle presence plans, station’s load curves, market price, and network load levels. Therefore, the equivalent load is calculated. In the third section, the effect of station loading on the distribution network was investigated based on equivalent loads (extracted based on optimal planning of the first part). Comparison of the station loading effect is carried out for IEEE 33-bus network in the MATLAB based on the loss indexes (loss cost) and the voltage drop of the buses. The results indicate loss reduction due to optimal planning at the station and its application in appropriate buses.

Employment of photovoltaic (PV) distributed generation (DG) in the distribution system, leads to improvement of network voltage profile and help to generate the power required by the network.Beside the mentioned advantages, increment of these sources penetration rate cause miscoordination between fuses and recloser. In this paper a new adaptive intelligent method of fuse– recloser coordination in distribution systems with high PV penetration rates is presented. The proposed method is a two-phase approach which operates proportional to the connected photovoltaic systems penetration rate. The first phase is based on an adaptive modification of fast operation curve of recloser, proportional to current term of the branch fuse maximum fault current to recloser current which is located at the beginning of the feeder. The second phase is according to modification of fast operation characteristic of recloser, proportional to voltage drop in fault inception period at the recloser location. By using this method, not only the avoidance of unnecessary exits of PV resources, but also the operation of fuse-saving is possible with adaptive correction of the fast-performance curve of the recloser.

Power transformers are one the most expensive and important equipment of power systems that play an important role in the continuous supply of electrical energy. Therefore, their protection has a significant impact on network reliability and stability. The differential protection scheme equipped with harmonic restraint is used for detection of transformer internal faults. But, current transformer (CT) saturation would have undesirable effect on its performance. In some cases, the CT saturation during internal fault can lead to even harmonics which can prevent the relay from sending trip signal.Moreover, the CT saturation due to DC offset of inrush current can result in relay maloperation. In this paper, a new feature is extracted from differential current based on VMD analysis. This feature as well as differential and bias magnitudes are used for detection of internal faults from other operating conditions. In the proposed method, different classifier such as ANN, PNN and ELM have been used as classifier core. The obtained results show that the proposed the combination of VMD and ELM can correctly detect all internal faults even with severe CT saturation.

Auto associative neural networks can be used for nonlinear processing and normalization of data.Because, firstly, they are able to learn and simulate complex nonlinear communications, and secondly, the communications can be learned through analyzing and distributing information on the neurons and weights and then combining the results of their processing. In this way, they actually make an interpolation between the input data and their communications. But these neural networks cannot model attractor dynamics that is obviously used in brain function. In this paper, the output of auto associative neural network is connected to its input, and through recursive connections the ability of attractor behavior in these models is provided. This study showed that a recursive neuron with a logistic function forms two attractors, in its training point and its symmetry, but for it with a sigmoid nonlinear function can be formed an attractor in a certain range. In the experiments on face images, it was shown that the absorbance of the images to their attractors was improved from 52.67% to 87.27% by increasing the number of layers and the supervised layer-by-layer pre-training in order to adjust the attractors.