Satyabroto Sinha

Invariance of stereo images via the theory of complex moments

Abstract For an approximate epipolar registration of stereo images the transformation between the images can be approximated from an affine one to a rotational and translational one. This property can be used to find the point of correspondences of stereo images in conjunction with complex moments which are themselves rotational invariant. Corners are chosen as features and around them an intensity kernel is defined and complex moments are calculated. Then the correspondences are found from similar corners by finding the L2 norm of the invariances between the two images. From the matched points the respective affine models of the images are also created.

An expert system approach to control system design and analysis

The architecture of CONEX-expert system for control system design and analysis-is presented. The important features of CONEX are discussed and they include: a structured organization of the short-term database, a clustering criteria for partitioning the rule-base into connected rule-spaces, a language for representing the domain-specific rules, and an inference engine that can adopt different search strategies for different rule-spaces. Additionally, an explanation generation module gives CONEX the attribute of user friendliness. >

Artificial neural-network model-based observers

Describes a pseudorandom testing scheme for fault diagnosis of analog integrated circuits. The goal is to implement a BIST technique with both a built-in pattern generator and a response analyzer for fault diagnosis. We have chosen a diagnostic framework for the analog ICs using a pseudorandom noise generator as the test-pattern generator and a model-based observer to detect and diagnose faults. The observer is implemented through a multilayer feedforward ANN trained with a back-error propagation (BEP) algorithm. Both the test-pattern generator and the model-based observer proposed in this article can be implemented either on- or offline depending on the need of the application and silicon area overhead.

A pseudo-random testing scheme for analog integrated circuits using artificial neural network model-based observers

This paper presents a simple and very efficient testing strategy for fault diagnosis of analog integrated circuits. The methodology is based on a technique of using a pseudorandom noise as the test pattern and a model-based observer for fast and robust testing and fault diagnosis. By incorporating device-level faults the efficiency is illustrated for stand-alone as well as embedded operational amplifiers as examples. The simulation results obtained are very encouraging. The technique can be viewed as a built-in self-test along with a design for testability scheme that dramatically improves the fault coverage and can be implemented for both on-line and off-line depending on the need of the application and silicon area overhead. Its main advantages are: a universal input stimulus (white noise) is used and thus test generation can be avoided, good and faulty signatures for high quality testing can be easily constructed and testing cost can be minimized, the technique is very efficient and robust, and the scheme can be well suited for built-in self-test implementation.

Microcomputer as a Programmable Controller for State Feedback Control of a DC Motor Employing Thyristor Amplifier

A microprocessor-based state feedback control of a single-phase thyristor bridge/dc motor drive system is presented. Considering the drive system as a regulator problem in multivariable system theory, suitable control laws are developed using a state space model and applying a pole placement technique for fast regulation coupled with system stability by state feedback. To improve the response further, feedforward control is included in the control law so that the output follows the reference with minimum delay. A minimal order observer is also incorporated to make a quick estimate of the armature current (assumed to be inaccessible) from the knowledge of output speed and inputs of the system. Finally, the controller including the feedback control, feedforward control, and minimal order observer has been implemented in real time using an Intel 8085 microprocessor kit with suitable interfacing and standard software package. The experimental setup is fabricated and tested, and the test results are compared with digital simulation results for step changes in reference and load torque.

A method to diagnose faults in analog integrated circuits using artificial neural networks with pseudorandom noise as stimulus

This paper describes a new, fast and economical strategy for fault diagnosis of analog integrated circuits. The methodology is based on a technique of using a pseudo random noise generator as the test pattern generator and a model-based observer, which is implemented through a feed forward artificial neural network in the form of a single hidden-layer perceptron. The proposed methodology can be implemented in any personal computer with a data acquisition card for on-line operation. Its main advantages are the low time requirement for learning and diagnosing. The method is quite robust and is able to detect small component variations without problems. This technique has been successfully applied to diagnose both hard and soft faults in a bipolar junction transistor based operational amplifier and a MOS operational amplifier.

TRANSELEX: A knowledge-based approach for the selection of transducers using a dynamic database

Abstract TRANSELEX (TRANsducer SELection EXpert) is a single expert system for selection of transducers in the areas of temperature, flow and pressure measurement to meet the requirements of the user. It allows numerical ranges to be inputted for greater accuracy of the selection process. At present the knowledge base of TRANSELEX has 44 sensors and 55 rules. This paper provides an overview of the expert program.

DEXT: an integrated knowledge-engineering environment for control-engineering applications

Abstract The paper presents dext , an integrated knowledge-engineering tool for control-engineering applications. First, the domain of control engineering has been analysed so that a few of the key requirements to be satisfied by the tool may be identified. A large number of software packages developed over the years for dealing with different aspects of control engineering constitute the primary knowledge base of an expert system for such applications. Unfortunately, these packages have been developed in different languages, and they differ in their performance with regard to computational efficiency and numerical stability. Hence, transparent access to a bank of such packages must be allowed, and such accesses must be independent of the language in which the packages are written. Apart from providing such an interface, the proposed tool must also support a variety of data types. A structured organization of the design information is also required, as the expert system may refer to such facts at different levels of abstraction. The structure of the rules and the facts is described in detail. The salient features of dext include the feature of transparent interfacing with the external packages, and enhancement of inferencing efficiency through the techniques of rule partitioning and intelligent backtracking.

Design of industrial regulators using expert systems approach

The applicability of the expert system approach to the design of industrial regulators is demonstrated. A brief overview of an expert system, CONEX, has been presented first. The result of applying CONEX to solve two industrial regulation problems is then discussed. The designs obtained by CONEX have been compared with those obtained through ‘non-expert-system’ approach.

Medium range forecasting of power system load (energy) demand

The most popular Box-Jenkins method, generally used for short-term forecasting, is modified to make it suitable for medium and long-range forecasting. The non-stationarity and seasonality have been identified and, after removing trends and/or seasonality, the series are tested for stationarity by various methods. The series have been fitted for different auto-regressive moving average (ARMA) models in the multiplicative modes. The parameter values have been determined from autocorrelation function (a.c.f.) and partial auto-correlation function (p.a.c.f.) cor-relograms and the whiteness of the residue has been checked. A forecast has been made for energy demand for one year with the help of this model and the result has been compared with actual demand.