Filter design using radial basis function neural network and genetic algorithm for improved operational health monitoring

The problem of denoising damage indicator signals for improved operational health monitoring of systems is addressed by applying soft computing methods to design filters. Since measured data in operational settings is contaminated with noise and outliers, pattern recognition algorithms for fault detection and isolation can give false alarms. A direct approach to improving the fault detection and isolation is to remove noise and outliers from time series of measured data or damage indicators before performing fault detection and isolation. Many popular signal-processing approaches do not work well with damage indicator signals, which can contain sudden changes due to abrupt faults and non-Gaussian outliers. Signal-processing algorithms based on radial basis function (RBF) neural network and weighted recursive median (WRM) filters are explored for denoising simulated time series. The RBF neural network filter is developed using a K-means clustering algorithm and is much less computationally expensive to develop than feedforward neural networks trained using backpropagation. The nonlinear multimodal integer-programming problem of selecting optimal integer weights of the WRM filter is solved using genetic algorithm. Numerical results are obtained for helicopter rotor structural damage indicators based on simulated frequencies. Test signals consider low order polynomial growth of damage indicators with time to simulate gradual or incipient faults and step changes in the signal to simulate abrupt faults. Noise and outliers are added to the test signals. The WRM and RBF filters result in a noise reduction of 54 - 71 and 59 - 73% for the test signals considered in this study, respectively. Their performance is much better than the moving average FIR filter, which causes significant feature distortion and has poor outlier removal capabilities and shows the potential of soft computing methods for specific signal-processing applications.

A New Structure Preserving Energy Function Incorporating Transmission Line Resistances

A novel method to account for the transmission line resistances in structure preserving energy functions (SPEF) is presented in this paper. The method exploits the equivalence of a lossy network having the same conductance to susceptance ratio for all its elements to a lossless network with a new set of power injections. The system equations and the energy function are developed using centre of inertia (COI) variables and the loads are modelled as arbitrary functions of respective bus voltages. The application of SPEF to direct transient stability evaluation is presented considering a realistic power system example.

Distributed nonlinear optimization algorithms for general network problems

There are a number of large networks which occur in many problems dealing with the flow of power, communication signals, water, gas, transportable goods, etc. Both design and planning of these networks involve optimization problems. The first part of this paper introduces the common characteristics of a nonlinear network (the network may be linear, the objective function may be non linear, or both may be nonlinear). The second part develops a mathematical model trying to put together some important constraints based on the abstraction for a general network. The third part deals with solution procedures; it converts the network to a matrix based system of equations, gives the characteristics of the matrix and suggests two solution procedures, one of them being a new one. The fourth part handles spatially distributed networks and evolves a number of decomposition techniques so that we can solve the problem with the help of a distributed computer system. Algorithms for parallel processors and spatially distributed systems have been described.There are a number of common features that pertain to networks. A network consists of a set of nodes and arcs. In addition at every node, there is a possibility of an input (like power, water, message, goods etc) or an output or none. Normally, the network equations describe the flows amoungst nodes through the arcs. These network equations couple variables associated with nodes. Invariably, variables pertaining to arcs are constants; the result required will be flows through the arcs. To solve the normal base problem, we are given input flows at nodes, output flows at nodes and certain physical constraints on other variables at nodes and we should find out the flows through the network (variables at nodes will be referred to as across variables).The optimization problem involves in selecting inputs at nodes so as to optimise an objective function; the objective may be a cost function based on the inputs to be minimised or a loss function or an efficiency function. The above mathematical model can be solved using Lagrange Multiplier technique since the equalities are strong compared to inequalities. The Lagrange multiplier technique divides the solution procedure into two stages per iteration. Stage one calculates the problem variables % and stage two the multipliers lambda. It is shown that the Jacobian matrix used in stage one (for solving a nonlinear system of necessary conditions) occurs in the stage two also.A second solution procedure has also been imbedded into the first one. This is called total residue approach. It changes the equality constraints so that we can get faster convergence of the iterations.Both solution procedures are found to coverge in 3 to 7 iterations for a sample network.The availability of distributed computer systems — both LAN and WAN — suggest the need for algorithms to solve the optimization problems. Two types of algorithms have been proposed — one based on the physics of the network and the other on the property of the Jacobian matrix. Three algorithms have been deviced, one of them for the local area case. These algorithms are called as regional distributed algorithm, hierarchical regional distributed algorithm (both using the physics properties of the network), and locally distributed algorithm (a multiprocessor based approach with a local area network configuration). The approach used was to define an algorithm that is faster and uses minimum communications. These algorithms are found to converge at the same rate as the non distributed (unitary) case.

Experimental investigation of the dependence of barrier height on metal work function for metal---SiO2---p---Si (MIS) Schottky-barrier diodes in the presence of inversion

The dependence of barrier height on the metal work function of metal-SiO2-p-Si Schottky barrier diodes was investigated and nonlinearity was found. This is explained by the theoretical model proposed recently by Chattopadhyay and Daw. The values of interface trap density and fixed charge density of the insulating layer of the diodes were calculated using this model and found to be appreciably different from those estimated by the usual method.

CHITRA: Cognitive handprinted input-trained recursively analyzing system for recognition of alphanumeric characters

A novel system for recognition of handprinted alphanumeric characters has been developed and tested. The system can be employed for recognition of either the alphabet or the numeral by contextually switching on to the corresponding branch of the recognition algorithm. The two major components of the system are the multistage feature extractor and the decision logic tree-type catagorizer. The importance of ldquogoodrdquo features over sophistication in the classification procedures was recognized, and the feature extractor is designed to extract features based on a variety of topological, morphological and similar properties. An information feedback path is provided between the decision logic and the feature extractor units to facilitate an interleaved or recursive mode of operation. This ensures that only those features essential to the recognition of a particular sample are extracted each time. Test implementation has demonstrated the reliability of the system in recognizing a variety of handprinted alphanumeric characters with close to 100% accuracy.

Hot gaseous coronae of early-type galaxies and their radio luminosity function

Recent X-ray observations have revealed that early-type galaxies (which usually produce extended double radio sources) generally have hot gaseous haloes extending up to approx102kpc1,2. Moreover, much of the cosmic X-ray background radiation is probably due to a hotter, but extremely tenuous, intergalactic medium (IGM)3. We have presented4–7 an analytical model for the propagation of relativistic beams from galactic nuclei, in which the beams' crossing of the pressure-matched interface between the IGM and the gaseous halo, plays an important role. The hotspots at the ends of the beams fade quickly when their advance becomes subsonic with respect to the IGM. This model has successfully predicted (for typical double radio sources) the observed8 current mean linear-size (approx2Dsime350 kpc)4,5, the observed8–11 decrease in linear-size with cosmological redshift4–6 and the slope of the linear-size versus radio luminosity10,12–14 relation6. We have also been able to predict the redshift-dependence of observed numbers and radio luminosities of giant radio galaxies7,15. Here, we extend this model to include the propagation of somewhat weaker beams. We show that the observed flattening of the local radio luminosity function (LRLF)16–20 for radio luminosity Papproximately 1024 W Hz-1 at 1 GHz can be explained without invoking ad hoc a corresponding break in the beam power function Phi(Lb), because the heads of the beams with Lb < 1025 W Hz-1 are decelerated to sonic velocity within the halo itself, which leads to a rapid decay of radio luminosity and a reduced contribution of these intrinsically weaker sources to the observed LRLF.

Medium access control schemes for local area networks with multiple priority function

This paper deals with the development and performance evaluation of three modified versions of a scheme proposed for medium access control in local area networks. The original scheme implements a collision-free and fair medium arbitration by using a control wire in conjunction with a data bus. The modifications suggested in this paper are intended to realize the multiple priority function in local area networks.

Study of the random vibration of nonlinear systems by the Gaussian closure technique

A technique is developed to study random vibration of nonlinear systems. The method is based on the assumption that the joint probability density function of the response variables and input variables is Gaussian. It is shown that this method is more general than the statistical linearization technique in that it can handle non-Gaussian excitations and amplitude-limited responses. As an example a bilinear hysteretic system under white noise excitation is analyzed. The prediction of various response statistics by this technique is in good agreement with other available results.

The non-linear response of a gyrostabilized platform: Effects of correction torque time delay and of random inputs

First, the non-linear response of a gyrostabilized platform to a small constant input torque is analyzed in respect to the effect of the time delay (inherent or deliberately introduced) in the correction torque supplied by the servomotor, which itself may be non-linear to a certain extent. The equation of motion of the platform system is a third order nonlinear non-homogeneous differential equation. An approximate analytical method of solution of this equation is utilized. The value of the delay at which the platform response becomes unstable has been calculated by using this approximate analytical method. The procedure is illustrated by means of a numerical example. Second, the non-linear response of the platform to a random input has been obtained. The effects of several types of non-linearity on reducing the level of the mean square response have been investigated, by applying the technique of equivalent linearization and solving the resulting integral equations by using laguerre or Gaussian integration techniques. The mean square responses to white noise and band limited white noise, for various values of the non-linear parameter and for different types of non-linearity function, have been obtained. For positive values of the non-linear parameter the levels of the non-linear mean square responses to both white noise and band-limited white noise are low as compared to the linear mean square response. For negative values of the non-linear parameter the level of the non-linear mean square response at first increases slowly with increasing values of the non-linear parameter and then suddenly jumps to a high level, at a certain value of the non-linearity parameter.

Filter design using radial basis function neural network and genetic algorithm for improved operational health monitoring

The problem of denoising damage indicator signals for improved operational health monitoring of systems is addressed by applying soft computing methods to design filters. Since measured data in operational settings is contaminated with noise and outliers, pattern recognition algorithms for fault detection and isolation can give false alarms. A direct approach to improving the fault detection and isolation is to remove noise and outliers from time series of measured data or damage indicators before performing fault detection and isolation. Many popular signal-processing approaches do not work well with damage indicator signals, which can contain sudden changes due to abrupt faults and non-Gaussian outliers. Signal-processing algorithms based on radial basis function (RBF) neural network and weighted recursive median (WRM) filters are explored for denoising simulated time series. The RBF neural network filter is developed using a K-means clustering algorithm and is much less computationally expensive to develop than feedforward neural networks trained using backpropagation. The nonlinear multimodal integer-programming problem of selecting optimal integer weights of the WRM filter is solved using genetic algorithm. Numerical results are obtained for helicopter rotor structural damage indicators based on simulated frequencies. Test signals consider low order polynomial growth of damage indicators with time to simulate gradual or incipient faults and step changes in the signal to simulate abrupt faults. Noise and outliers are added to the test signals. The WRM and RBF filters result in a noise reduction of 54 - 71 and 59 - 73% for the test signals considered in this study, respectively. Their performance is much better than the moving average FIR filter, which causes significant feature distortion and has poor outlier removal capabilities and shows the potential of soft computing methods for specific signal-processing applications. (C) 2005 Elsevier B. V. All rights reserved.

Design of Single-Input-Single-Output Compliant Mechanisms for Practical Applications Using Selection Maps

We present an interactive map-based technique for designing single-input-single-output compliant mechanisms that meet the requirements of practical applications. Our map juxtaposes user-specifications with the attributes of real compliant mechanisms stored in a database so that not only the practical feasibility of the specifications can be discerned quickly but also modifications can be done interactively to the existing compliant mechanisms. The practical utility of the method presented here exceeds that of shape and size optimizations because it accounts for manufacturing considerations, stress limits, and material selection. The premise for the method is the spring-leverage (SL) model, which characterizes the kinematic and elastostatic behavior of compliant mechanisms with only three SL constants. The user-specifications are met interactively using the beam-based 2D models of compliant mechanisms by changing their attributes such as: (i) overall size in two planar orthogonal directions, separately and together, (ii) uniform resizing of the in-plane widths of all the beam elements, (iii) uniform resizing of the out-of-plane thick-nesses of the beam elements, and (iv) the material. We present a design software program with a graphical user interface for interactive design. A case-study that describes the design procedure in detail is also presented while additional case-studies are posted on a website. DOI:10.1115/1.4001877].

Transport properties of phosphomolybdate and phosphotungstate glasses

The d.c. conductivity of phosphomolybdate and phosphotungstate glasses is discussed. The conductivity of these glasses is due to the hopping of electrons between two valence states (Mo5+ to Mo6+ or W5+ W6+). In some of the glasses, the activation energy itself is found to be a function of temperature. This appears to be due to thermally activated and variable-range hopping mechanisms operating in different temperature regimes. The relation between conductivity and the [M5+]/[Mtotal](M ≡ Mo, W) ratio does not show any systematic variation. This anomaly can be understood using the structural models of these glasses. In contrast, Mott's theory and the Triberis and Friedman model have been used to obtain conductivity parameters such as the percolation distance Rij and 2agrRij (agr is the tunnelling probability). The conductivity parameter 2agrRij is quite useful to resolve the controversy regarding the tunnelling term exp(2agrRij) existing in the literature. For low values of 2agrRij, it is shown that the exp (2agrRij) term is very significant.

Polarization Caging in Diffusion-Controlled Electron Transfer Reactions in Solution

In some bimolecular diffusion-controlled electron transfer (ET) reactions such as ion recombination (IR), both solvent polarization relaxation and the mutual diffusion of the reacting ion pair may determine the rate and even the yield of the reaction. However, a full treatment with these two reaction coordinates is a challenging task and has been left mostly unsolved. In this work, we address this problem by developing a dynamic theory by combining the ideas from ET reaction literature and barrierless chemical reactions. Two-dimensional coupled Smoluchowski equations are employed to compute the time evolution of joint probability distribution for the reactant (P-(1)(X,R,t)) and the product (p((2))(X,R,t)), where X, as is usual in ET reactions, describes the solvent polarization coordinate and R is the distance between the reacting ion pair. The reaction is described by a reaction line (sink) which is a function of X and R obtained by imposing a condition of equal energy on the initial and final states of a reacting ion pair. The resulting two-dimensional coupled equations of motion have been solved numerically using an alternate direction implicit (ADI) scheme (Peaceman and Rachford, J. Soc. Ind. Appl. Math. 1955, 3, 28). The results reveal interesting interplay between polarization relaxation and translational dynamics. The following new results have been obtained. (i) For solvents with slow longitudinal polarization relaxation, the escape probability decreases drastically as the polarization relaxation time increases. We attribute this to caging by polarization of the surrounding solvent, As expected, for the solvents having fast polarization relaxation, the escape probability is independent of the polarization relaxation time. (ii) In the slow relaxation limit, there is a significant dependence of escape probability and average rate on the initial solvent polarization, again displaying the effects of polarization caging. Escape probability increases, and the average rate decreases on increasing the initial polarization. Again, in the fast polarization relaxation limit, there is no effect of initial polarization on the escape probability and the average rate of IR. (iii) For normal and barrierless regions the dependence of escape probability and the rate of IR on initial polarization is stronger than in the inverted region. (iv) Because of the involvement of dynamics along R coordinate, the asymmetrical parabolic (that is, non-Marcus) energy gap dependence of the rate is observed.