فصلنامه روشهای هوشمند در صنعت برق
پیاپی 50 (تابستان 1401)

Compressors are of particular importance due to their wide application in various industries for compression and gas transmission. Due to the increasing use of compressors in the industry, it is imperative to determine a mathematical model for the compressor to design a control system, analyze it and its simulating in corresponding computer software like MATLAB. In recent years, intelligent modeling such as neural and fuzzy networks due to the more realistic performance of these models has been considered by researchers. Its types have been used for modeling. Intelligent methods have a high ability to stablish a relation between input and output data. On the other hand, in compressors surge phenomenon is a challenging case which occurs very quickly and will cause damage to the compressor and the production process. This phenomenon is defined as an instability in the operation of the compressor. Surge control expands the operating range of the compressor and exhibit the occurrence of this phenomenon. This article reviews modeling, surge phenomenon, and types of active and inactive surge control. For modeling, from the Moore–Greitzer’s model to intelligent models, it has been studied that according to the obtained results, it can be said that the Moore–Greitzer’s model has more errors over time than intelligent models and is a suitable model. Not from the compressor. Next, the control of the surge phenomenon in the compressor is checked by active and passive methods. The compressor's control in the active mode does not require a compressor performance curve. Still, in the passive method, it controls the compressor by using the performance curve and creating a safety margin from the surge line. So as not to enter the surge phenomenon. This makes active control performance more efficient than passive control.

In this paper, a ZVS interleaved flyback converter with two transformers is presented, which consists of two active clamp flyback converters and the main switch of one converter acts as an auxiliary switch of another converter. This converter has less auxiliary elements and less voltage and current stress compared to similar soft switching interleaved flyback converters. The introduction of a new auxiliary circuit for soft switching, in addition to increasing efficiency, minimizes the number of added semiconductors. Also, another advantage of this structure is the applicability of the provided auxiliary circuit to other isolated converters. The soft switching conditions in this converter are created by the auxiliary circuit in such a way that the converter switches turn on and off under ZVS conditions and the converter diodes turn on and off under ZCS conditions. The efficiency of the proposed ZVS interleaved flyback converter at full load is increased by 5%. Another advantage of the proposed converter is that the Q2 switch, in addition to providing zero voltage switching conditions for the Q1 switch, it also transmits energy and increases the density of the converter power and reduces the current stress. The converter is thoroughly analyzed and a 300W laboratory prototype is made to confirm its correct operation and practical results are presented.

In the age of health information analysis, the disease diagnostic code is considered as the patient's privacy. Achieving this code is the most important need for the analysts while anonymizing the code is necessary for people when publishing health information. Disease diagnostic codes, usually presented based on international classifications, are displayed in the form of a taxonomy. In real life, patients only allow the category of the disease diagnostic code to be disclosed, not the original disease diagnostic code. Conventional privacy-preserving models often distort the category of the disease diagnostic code. Preserving privacy accompanying the data utility has always been a critical issue in the dissemination of health information. In this study, a new anonymization method is presented in a way that all attributes of health information can be published without distortion to maintain the utility of the data. So, the published information protects the privacy of patients, so that the experts' expectations and the utility of analysts are desired as expected. The innovative method disseminates health information in a way that the maximum probability of disclosing the disease diagnostic code is always less than or equal to the threat threshold defined by the expert, and on the other hand, the membership analysis error is reduced. The new method is scalable under certain conditions. The results of the practical evaluation of patient data obtained from one of the hospitals in Isfahan are evidence of the effectiveness of the proposed method.

With the increasing interest in Joint Photographic Experts Group (JPEG) image compression, one of the most important issues in digital image manipulation is finding a proper method to detect double JPEG compression. This paper introduces a trained adaptive filter based on spatial-domain convolutional autoencoder (CAE). This filter can remove interference information caused by image content to have a more accurate detection. The convolutional neural network (CNN) has been widely employed for accurate image classification; therefore, a CNN is used in the classification part of the proposed algorithm. The proposed model is based on consecutive CAE with CNN, which is able to provide acceptable detection accuracy and sensitivity to quality factors (QFs) in two scenarios, i.e. aligned and non-aligned forgeries. This model improves the sensitivity to quality factors by up to 86% in the relative error reduction (RER) rate in some cases. Other experiments such as manipulation localization on the RAISE dataset have been performed to evaluate the proposed method. These results show the superior performance of this method compared to similar algorithms in the situations that the quality factor of the second compression is greater the quality factor of the first compression.

Insulation failure between winding turns is one of the main causes of incipient winding fault in a transformer. During the operation of a transformer, strong electric fields are applied to the dielectric material of its windings. Dielectric deterioration and aging is one of the major cause of short circuit faults in transformer windings. Due to the probable occurrence of this type of defect and its extension, its early detection is a very important task in power systems. In this paper, it is shown that by measuring the phase difference between winding voltage and winding current of a loaded transformer, the existence of internal winding fault can be detected. For online fault detection, an intelligent system (neural-fuzzy system) has also been proposed.Both simulation results and laboratory tests confirm the ability of the proposed method for the detection of internal winding faults especially at light loads. With this method, there is no need to de-energize the power transformer.

Wideband linear frequency modulation (LFM) signals are widely used in systems such as radar, sonar, and mobile. 2D-DOA algorithms for LFM signals are relying on a large number of snapshots. For this reason, they are not suitable for low-power applications. In this paper, we present an algorithm-centered estimation method with low estimation of signal parameters via rotational invariance technique (ESPRIT) calculations based a 2D circular array using a fractal Fourier transform (FrFT). Furthermore, the utilization of a circular array facilitates the two-dimensional DOA calculation. Therefore, the procedure is that firstly, we develop the Dechirping process for LFM signals using the FrFT; secondly, we extend the ESPRIT algorithm- as used for linear arrays (ULA) - for 2D circular arrays (UCA). Finally, DOA calculations are made for a low number of snapshots with low computational volume. The simulation results of the proposed MESPRIT (i.e. modified ESPRIT) algorithm show that this algorithm outperforms compared to other algorithms like MUSIC and TSFDOA. We also have shown that the proposed method has an acceptable accuracy in low SNRs and creates less error in high SNRs. It was also demonstrated that for all algorithms, accuracy of azimuth angle is better than the elevation angle’s.

In this paper, a new temperature sensor based on the performance of carbon nanotube transistors in the subthreshold region is designed and simulated which leads to reduction of power consumption. Also, a differential amplifier is used in the output of the sensor. in order to keep the values of gain and common mode level voltage due to temperature changes, a method has been proposed that can compensate for these parameters variation due to temperature variation in the range of -30 ºC to +125 ºC. The proposed temperature sensor with its amplifier can be used as a system on the chip surface for temperature monitoring and control. The proposed temperature sensor in the CNTFET carbon nanotube field effect transistor technology with a supply voltage of 0.5 V in the sub-threshold area is simulated by HSPICE software with a 32nm CNT model. The simulation results show that at proposed circuits can measure the temperature in range of -30 ºC to +125 ºC linearly with a sensitivity of 1 mV/ ºC and consumes only 123 nW at room temperature. Also, the error measured at 125 ºC is about 2.5 mV, which means an error of 1.25 ºC at this temperature.

One of the most important advances in Raman spectroscopy in recent years has been its integration with scanning probe microscopes (SPM), especially atomic force microscopes (AFM). Currently, AFM is recognized as one of the best imaging methods for studying the distribution of heterogeneous surface in nanoscale dimensions. Scientists are now focused on obtaining more enhancement factor of electric field, to the extent that detection and mapping of only one molecule has become possible with this method. Therefore, spatial resolution is being improved in detecting sub-molecule levels. In this paper, using the finite difference time domain (FDTD) calculation method, the effect of changing the parameters of the probe such as cone angle, tip radius and its material on the electric field intensity near the apex of the probe is investigated. In addition, the effect of polarization of light on the increase of electric field has been analyzed. The simulation results obtained for different cone angles show that the cone angle of 30 degree creates the highest amount of electric field enhancement factor at the tip apex. Furthermore, the use of laser source with radially polarized light and the use of substrate are very effective factors on improving the electric field enhancement factor. Finally, the maximum value of electric field enhancement factor of the proposed configuration is 3.2×104, where this value has been improved significantly comparing to the results reported in the previous papers published in this field.