Kingshook Bhattacharyya

Artificial neural network based prediction of drill flank wear from motor current signals

In this work, a multilayer neural network with back propagation algorithm (BPNN) has been applied to predict the average flank wear of a high speed steel (HSS) drill bit for drilling on a mild steel work piece. Root mean square (RMS) value of the spindle motor current, drill diameter, spindle speed and feed-rate are inputs to the network, and drill wear is the output. Drilling experiments have been carried out over a wide range of cutting conditions and the effects of drill wear, cutting conditions (speed, drill diameter, feed-rate) on the spindle motor current have been investigated. The performance of the trained neural network has been tested for new cutting conditions, and found to be in very good agreement to the experimentally determined drill wear values. The accuracy of the prediction of drill wear using neural network is found to be better than that using regression model.

Robust Unknown Input Observer for Nonlinear Systems and Its Application to Fault Detection and Isolation

A robust unknown input observer for a nonlinear system whose nonlinear function satisfies the Lipschitz condition is designed based on linear matrix inequality approach. Both noise and uncertainties are taken into account in deriving the observer. A component fault detection and isolation scheme based on these observers is proposed. The effectiveness of the observer and the fault diagnosis scheme is shown by applying them for component fault diagnosis of an electrohydraulic actuator.

Application of wavelet packet analysis in drill wear monitoring

In this work, an attempt has been made to develop a drill wear monitoring system which is independent to cutting conditions of the drilling process. A cost effective Hall-effect current sensor, which does not interfere with the process, has been used for acquiring motor current signature during drilling under different cutting conditions. An advanced signal processing technique, the wavelet packet transform has been used on the acquired current signature to extract features for indirect representation to the amount of drill wear. Experimental sensitivity analysis reveals that in comparison to time domain features, wavelet packet features are more sensitive to flank wear and less sensitive to the cutting conditions. A multilayer neural network model has then been developed to correlate the extracted wavelet packet features with drill flank wear. Experimental results show that the proposed drill wear monitoring system can successfully predict the flank wear with acceptable accuracy.

FUZZY RADIAL BASIS FUNCTION (FRBF) NETWORK BASED TOOL CONDITION MONITORING SYSTEM USING VIBRATION SIGNALS

Thriving automation in industries leads to more research on the tool condition monitoring systems for better accuracy and fast recognition/evaluation of tool wear. Research on the applicability of the new advances in the soft-computing as well as in the signal processing fields is the inevitable consequence. In this work, a new soft-computing modeling technique, fuzzy radial basis function (FRBF) network has been applied to the prediction of drill wear using the vibration signal features. This work presents the wear prediction performance comparison of this new model with three other already tried and established soft-computing models, such as back propagation neural network (BPNN), radial basis function network (RBF) and normalized radial basis function network (NRBF), for both time-domain as well as wavelet packet approaches of feature extraction. Experimental results show that FRBF model with wavelet packet approach produces the best performance of predicting flank wear.

Modeling and simulation of centerless grinding of ball bearings

Abstract A dynamic analysis of the process of modification of surface topography of a ball bearing during the grinding process using bond graph techniques has been attempted. Such modification results from material removal by abrasion. The position of any point of the surface profile, with respect to a nominal subcutaneous surface, has been expressed as a bivariate, orthogonal series of spherical harmonics. The angular coordinates of the aforesaid point in a spherical coordinate system are taken as the two variables. The net material removal rate has been expressed as a sum of the rate of change of the individual harmonic coefficients. This approach has thereby broken down the process of modification of surface topography into a number of orthogonal modes. While the net rate of deformation is obtained from an elastoplastic analysis of the contact forces, the wear rate is easily obtainable as an empirical function of the contact forces, velocities and material properties of the surfaces in contact. An orthonormal expansion of the net rate of deformation or net material removal rate gives the rate of change of coefficients of the spherical harmonics, which are considered as slowly varying functions of time. The modeling technique captures microscopic, features of surface profiles at a point, in terms of cyclic functions of the coordinates of the point in a body fixed frame. These ideas have been used in modeling the change in the surface profile of a ball bearing during the process of grinding. The effect of various process parameters, such as grinding speed, preload, and entry point orientation have been studied using this paradigm. The present work also studies the effect of the random entry configuration in the grinder and shows that entry configurations have a very significant effect on the final shape of the ball, based on which observation the authors make an argument in favor of studying grinding processes in terms of the entire population in order to decide the suitability of operating parameters, rather than simulating a single sample in isolation.

Drill Wear Monitoring through Current Signature Analysis using Wavelet Packet Transform and Artificial Neural Network

In automated manufacturing systems, tool wear monitoring plays an important role in ensuring the dimensional accuracy of the workpiece and automatic cutting process without any failure. It is, therefore, very essential to develop simple, reliable, process condition independent and cost effective online tool wear monitoring system. In this work, an attempt has been made to develop a drill tool wear monitoring system which fulfills most of the above essential requirements. A cost effective Hall-effect current sensor, which does not interfere with the process, has been used for acquiring motor current signature during drilling under different cutting conditions. A more advanced signal processing technique, wavelet packet transform has been implemented on the acquired current signature to extract features which are more sensitive to drill wear and less sensitive to the process conditions. A multilayer neural network model has been developed to correlate the extracted features with drill wear. Experimental results show that the proposed drill wear monitoring system can successfully predict the drill wear with acceptable accuracy.

An unknown input Kalman filter based component FDI algorithm and its application in automobiles

An Unknown Input Kalman Filter (UIKF) based Component Fault Detection and Isolation (CFDI) technique for a dynamical system, affected by both plant and measurement noise, is presented. The Fault Detection and Isolation (FDI) algorithm, which consists of two steps, is developed with the assumption that the fault occurs in a single component of the system. In step 1, the detection of the fault and the isolation of the faulty region are achieved. In the next step, the faulty parameter is isolated from the faulty region. The method is applied on a road vehicle model to show the effectiveness of the algorithm.

Application of wavelet packet transform based Normalised Radial Basis Function Network in a machining process

In this work, an attempt has been made to develop a drill wear prediction system. A Hall-effect current sensor has been used for acquiring motor current signals during drilling under different cutting conditions. Wavelet packet transform has been used on the acquired current signals to extract features. A normalised Radial Basis Function (RBF) neural network model has then been developed to correlate the extracted features with drill wear. The proposed network outperforms the standard RBF neural network in terms of training error and also in terms of prediction error.

Strategies for intelligent drill wear prediction using multiple sensor signals

In this work, strategies for artificial intelligence (AI)-based drill flank wear prediction systems have been presented. Four AI-based models, namely, back propagation neural network (BPNN), radial basis function network (RBF), normalised radial basis function network (NRBF) and fuzzy radial basis function network (FRBF) have been used. Signals have been acquired from dynamometer, current sensor and accelerometer during drilling under different cutting conditions. Wear sensitive features in time-domain and frequency domain have been extracted. Different strategies, i.e., various combinations of the selected features and process parameters used as inputs to the AI models, are formulated. For most of the strategies, BPNN model gives better wear prediction results followed by NRBF model. But for noisy features such as features from vibration signal can be better related to drill wear using FRBF model.

Fault detection and isolation of vehicle suspension system using robust unknown input observers

A method of Fault Detection and Isolation (FDI) for vehicle suspension systems in the presence of exogenous noise and parametric uncertainties is presented. First, a new kind of reduced-order Robust Unknown Input Observer (RUIO) is designed based on Linear Matrix Inequality (LMI) approach. Then, using a bank of such observers, a component FDI algorithm is derived. Simulation results with the help of half-car model show the efficacies of the FDI algorithm as well as the observer.