Gautam Sarkar

A Reinforcement-Learning-Based Fuzzy Compensator for a Microcontroller-Based Frequency Synthesizer/Vector Voltmeter

This paper presents the development of an intelligent fuzzy-based compensation scheme, utilizing reinforcement learning methodology, which can be used to compensate the reading of an unknown voltage, in vector form. This compensator is implemented online, in real time, with an indigenously developed microcontroller-based scheme that can be used both as a frequency synthesizer and as a vector voltmeter. This frequency synthesizer/vector voltmeter is developed using a direct digital synthesis method for the frequency synthesizer and a synchronous detection technique for the vector voltmeter. The developed fuzzy compensator has been tested in both offline and online modes, and in both cases, it has been found to significantly improve the accuracy of the measurement compared to those obtained with an uncompensated vector voltmeter. It has been shown that the final compensated measurements are in close agreement with the true unknown voltages under measurement.

Development of a Microcontroller-Based Frequency Synthesizer cum Vector Voltmeter

This paper proposes the development of a microcontroller-based scheme that can be used both as a frequency synthesizer and a vector voltmeter. The scheme employs direct digital synthesis for development of the frequency synthesizer and the synchronous detection technique for developing the vector voltmeter. The system also employs correction polynomials to compensate for errors in practical voltage readings. The final corrected readings of the system have been found to closely match with the true readings.

Neural Compensation for a Microcontroller Based Frequency Synthesizer-Vector Voltmeter

An automated neural network compensation scheme is proposed for an indigenously developed microcontroller based frequency synthesizer-vector voltmeter system, developed using direct digital synthesis and the synchronous detection technique. This compensator, when implemented online, can significantly improve the reading of an unknown voltage (both in magnitude and phase), in real-time. The neural compensator developed is trained offline on the basis of real data acquired from the system, and when this compensator is implemented online, it could outperform polynomial and fuzzy based compensators for a variety of different unknown voltages under measurement.

Hierarchical Extreme Learning Machine-Polynomial Based Low Valued Capacitance Measurement Using Frequency Synthesizer–Vector Voltmeter

This present paper describes the development of a capacitance measurement system in the picofarad region. The system uses an universal serial bus port-based arrangement in conjunction with an indigenously developed Programmable Intelligent Computer microcontroller-based frequency synthesizer-vector voltmeter that can be used to measure the voltage in vector form and the capacitance can be determined using circuit solution technique. An intelligent two-layered, hierarchical reinforcement-based instrumentation scheme is proposed that can be integrated along with the original measurements to significantly improve the system performance. In layer 1, an extreme learning machine-based supervised phase reinforcement scheme is employed to improve the accuracy of the voltage measurement. Subsequently, in layer 2, local polynomial-based reinforcements are employed to improve both the resistive and reactive part measurements in the unknown capacitance. Three variants of ELM-based reinforcements are implemented for capacitance measurements in the range 100-10 000 pF and the utility of the hybrid ELM-polynomial-based reinforcements for such measurements is aptly demonstrated.

Reference input tracking of inversion-based non-minimum phase system using adaptive two-degree-of-freedom control

Reference input tracking plays a major role in control system Engineering. Inverse model technique is very useful for exact tracking of minimum phase system but for non-minimum phase system, it gives unbounded output response. This paper proposes an effective method for achieving the approximate desired trajectory tracking of unstable inversion-based non-minimum phase (NMP) system. Here two-degree-of-freedom control theory has been applied where feedback control is provided by arbitrary pole placement method to create the bounded response of unstable NMP system. Lyapunov based Direct Model Reference Adaptive Control (MRAC) of inverse transfer function model of NMP system act as feed-forward compensator in 2DOF framework to track the reference input trajectory. Unstable non-minimum phase plant can successfully track step, ramp and parabolic input signal with a minimum steady state error shown in the simulation results. Initial undershoot which is obvious in NMP system has been completely removed by this control strategy.

Hierarchical Clustering for Segmenting Fused Image Using Discrete Cosine Transform with Artificial Bee Colony Optimization

In this paper, a robust and improved image segmentation technique is proposed for segmentation of a fused image based on Discrete Cosine Transform (DCT) with Artificial Bee Colony Optimization (ABC) termed as DCTopti. It is very challenging task to perceive details information from a visual image due to variation of light, reflection, existence of shadow etc., while in case of infrared images, as the energy is tracked in the photometry, image can be acquired at night. Mainly image fusion is done to solve the object detection problem from the image, where the resultant fused image contains more information than the input images. K-means, an iterative method for separating a data into k number of groups has been opted to segment the fused images using Hierarchical algorithm. Structural Similarity Index Measure (SSIM) is used for comparing the quality of the resulting fused image using the proposed technique with other benchmark methods. The result of this proposed Hierarchical K-means Clustering (HKmC) method shows its robustness for segmenting the fused images.

Two-Degree-of-Freedom Control of Non-minimum Phase Mechanical System

This paper proposes Two-degree-of-freedom control of Non-minimum phase (NMP) system. One of the essential tasks for controlling the plant is to follow the desired trajectory. Exact matching of set point and actual output can be possible if inverse transfer function model of the system is connected in cascade with the original one. But it is true, only for minimum phase system; whereas inverse model connected NMP system always produce unbounded output response. Here, a novel technique has been proposed, where inverse model NMP is controlled by Model Reference Adaptive Control (MRAC) as feed-forward compensation and arbitrary pole placement technique for original transfer function model has been used as feedback controller in Two-degree-of-freedom (2DOF) framework. The satisfactory result of 2DOF controlled practical NMP system has been studied in a simulation environment and compared it with the Proportional Integral Derivative (PID) and State feedback control method.

Study of Arrhythmia Using Wavelet Transformation Based Statistical Parameter Computation of Electrocardiogram Signal

Electrocardiogram (ECG) signal analysis has been done for diagnosis of Arrhythmia. ECG signals of Arrhythmia patients have been used for study. Signals are first denoised and then discrete wavelet decomposition is performed. Detail and approximate coefficients are determined and then their Kurtosis values are extracted. Significant changes are noticed at different decomposition levels and noted. These changes are found helpful in diagnosis of Arrhythmia.

Electrocardiogram Signal Analysis for Diagnosis of Congestive Heart Failure

Congestive heart failure (CHF) refers to building up of fluid surrounding the heart causing the pump inefficiently. In various ways attempts are taken to diagnosis such disease; electrocardiogram is one of them. Here electrocardiogram (ECG) signal analysis has been done for diagnosis of congestive heart failure. ECG signal have been collected from normal healthy person as well as patients suffering from congestive heart failure (CHF). As the signals are non-stationary in nature and collected in discrete manner, discrete wavelet transformation (DWT) has been applied on those collected signals. Approximate coefficients have been determined at different decomposition levels and have been assessed by their skewness values (SA). Comparative study has been made on SA for healthy person and CHF patient. Significant difference has been observed at certain DWT levels. Also radar and histogram based comparative assessment have been performed to easily distinguish ECG signals having CHF. The outcome of the work may be useful for diagnosis of such disease.

Sleep Apnea Diagnosis by DWT-Based Kurtosis, Radar and Histogram Analysis of Electrocardiogram

ABSTRACT Sleep Apnea is a serious breathing disorder that occurs when a persons breathing is interrupted during sleeping. People with sleep Apnea stop breathing repeatedly during their sleep, which means that the brain and rest of the body may not get enough oxygen. If left untreated, sleep Apnea can result in a number of health problems, including high blood pressure, stroke, heart failure, irregular heartbeats, heart attacks, and depression. This paper deals with sleep Apnea assessment by discrete wavelet-transformation-(DWT)-based kurtosis, radar and histogram analysis of electrocardiogram (ECG) signals. ECG signals are de-noised by passing them through well-known Savitzky–Golay FIR filter and then are decomposed at different DWT levels, and kurtosis of approximate and detailed coefficients at different DWT levels is measured. Kurtosis at different levels is compared for a healthy person and Apnea patients. Then, radars are formed by kurtosis and compared. Histogram analysis is done on both ECG signals and obtained kurtosis. The comparative study shows that up to DWT level-4, kurtosis of approximate coefficients of Apnea patients is lower than that of a healthy person. However, kurtosis of the approximate coefficient for Apnea patients is greater than that of a normal person at DWT level-7. Up to level 6, Kurtosis of detailed coefficients for Apnea patients is less than that of a normal person. Radar shapes and histogram peaks of ECG signal and kurtosis are also different between a normal person and Apnea patients. Probability in terms of a “p” value for Kurtosis at optimized DWT levels for Apnea patients has shown satisfactory outcome.