Hasan Ocak

Automatic detection of epileptic seizures in EEG using discrete wavelet transform and approximate entropy

In this study, a new scheme was presented for detecting epileptic seizures from electro-encephalo-gram (EEG) data recorded from normal subjects and epileptic patients. The new scheme was based on approximate entropy (ApEn) and discrete wavelet transform (DWT) analysis of EEG signals. Seizure detection was accomplished in two stages. In the first stage, EEG signals were decomposed into approximation and detail coefficients using DWT. In the second stage, ApEn values of the approximation and detail coefficients were computed. Significant differences were found between the ApEn values of the epileptic and the normal EEG allowing us to detect seizures with over 96% accuracy. Without DWT as preprocessing step, it was shown that the detection rate was reduced to 73%. The analysis results depicted that during seizure activity EEG had lower ApEn values compared to normal EEG. This suggested that epileptic EEG was more predictable or less complex than the normal EEG. The data was further analyzed with surrogate data analysis methods to test for evidence of nonlinearities. It was shown that epileptic EEG had significant nonlinearity whereas normal EEG behaved similar to Gaussian linear stochastic process.

Tool wear condition monitoring in drilling operations using hidden Markov models (HMMs)

Monitoring of tool wear condition for drilling is a very important economical consideration in automated manufacturing. Two techniques are proposed in this paper for the on-line identification of tool wear based on the measurement of cutting forces and power signals. These techniques use hidden Markov models (HMMs), commonly used in speech recognition. In the first method, bargraph monitoring of the HMM probabilities is used to track the progress of tool wear during the drilling operation. In the second method, sensor signals that correspond to various types of wear status, e.g., sharp, workable and dull, are classified using a multiple modeling method. Experimental results demonstrate the effectiveness of the proposed methods. Although this work focuses on on-line tool wear condition monitoring for drilling operations, the HMM monitoring techniques introduced in this paper can be applied to other cutting processes.

Optimal classification of epileptic seizures in EEG using wavelet analysis and genetic algorithm

In this study, a new scheme was presented for the optimal classification of epileptic seizures in EEG using wavelet analysis and the genetic algorithm (GA). In the proposed scheme, normal and epileptic EEG epochs (windows) were decomposed into various frequency bands through a fourth-level wavelet packet decomposition. Approximate entropy (ApEn) values of the wavelet coefficients at all nodes of the decomposition tree were used as a feature set to characterize the predictability of the EEG data within the corresponding frequency bands. Then, the GA was used to find the optimal feature subset that maximizes the classification performance of a learning vector quantization (LVQ)-based normal and epileptic EEG classifier. Clinical EEG data recorded from normal subjects and epileptic patients were used to test the performance of the new scheme. It was demonstrated that the new scheme was able to classify the normal and epileptic EEG epochs with 94.3% and 98% accuracy, respectively. It was also shown that, if the GA was not used for the optimal feature selection, the classification accuracies dropped noticeably.

Estimation of the running speed and bearing defect frequencies of an induction motor from vibration data

Abstract This paper presents two separate algorithms for estimating the running speed and the bearing key frequencies of an induction motor using vibration data. Bearing key frequencies are frequencies at which roller elements pass over a defect point. Most frequency domain-based bearing fault detection and diagnosis techniques (e.g. envelope analysis) rely on vibration measurements and the bearing key frequencies. Thus, estimation of the running speed and the bearing key frequencies are required for failure detection and diagnosis. The paper also incorporates the estimation algorithms with the most commonly used bearing fault detection technique, high-frequency demodulation, to detect bearing faults. Experimental data were used to verify the validity of the algorithms. Data were collected through an accelerometer measuring the vibration from the drive-end ball bearing of an induction motor (Reliance Electric 2HP IQPreAlert)-driven mechanical system. Both inner and outer race defects were artificially introduced to the bearing using electrical discharge machining. A linear vibration model was also developed for generating simulated vibration data. The simulated data were also used to validate the performance of the algorithms. The test results proved the algorithms to be very reliable.

A new bearing fault detection and diagnosis scheme based on hidden Markov modeling of vibration signals

This paper introduces a new bearing fault detection and diagnosis scheme based on hidden Markov modeling (HMM) of vibration signals. First features are extracted from amplitude demodulated vibration signals obtained from both normal and faulty bearings. The features are based on the reflection coefficients of the polynomial transfer function of the autoregressive model of the vibration signal. These features are then used to train HMMs to represent various bearing conditions. The technique allows for online detection of faults by monitoring the probabilities of the pre-trained HMM for the normal case. It also allows for the diagnosis of the fault by the HMM that gives the highest probability. The new scheme was tested with experimental data collected from drive end ball bearing of an induction motor (Reliance Electric 2HP IQPreAlert) driven mechanical system.

HMM-Based Fault Detection and Diagnosis Scheme for Rolling Element Bearings

In this paper, we introduce a new bearing fault detection and diagnosis scheme based on hidden Markov modeling (HMM) of vibration signals. Features extracted from amplitude demodulated vibration signals from both normal and faulty bearings were used to train HMMs to represent various bearing conditions. The features were based on the reflection coefficients of the polynomial transfer function of an autoregressive model of the vibration signals. Faults can be detected online by monitoring the probabilities of the pretrained HMM for the normal case given the features extracted from the vibration signals. The new technique also allows for diagnosis of the type of bearing fault by selecting the HMM with the highest probability. The new scheme was also adapted to diagnose multiple bearing faults. In this adapted scheme, features were based on the selected node energies of a wavelet packet decomposition of the vibration signal. For each fault, a different set of nodes, which correlates with the fault, is chosen. Both schemes were tested with experimental data collected from an accelerometer measuring the vibration from the drive-end ball bearing of an induction motor (Reliance Electric 2 HP IQPreAlert) driven mechanical system and have proven to be very accurate.

Post-sigh breathing behavior and spontaneous pauses in the C57BL/6J (B6) mouse.

The purpose was to examine sighs and spontaneous pauses in regard to the stability of resting breathing in the B6 strain, compared to the A/J strain. A 5-HT1A receptor agonist (buspirone) and a chromosomal substitution strain (B6a1) were used to further alter breathing patterning. Ten-minute recordings of room air breathing were collected from unanaesthetized B6, A/J, and B6a1 mice. Despite no differences between strains in the magnitude and incidence of sighs, post-sigh apneas, the variation for duration of expiration (Te) after sighs, and the number of spontaneous pauses were greater in the B6, while Shannon Entropy (nonlinear metrics) for Te after sighs was lower in B6, compared to the other strains. Buspirone and chromosomal substitution eliminated post-sigh apneas and decreased spontaneous pauses. A greater irregularity and the lower complexity of post-sigh breathing in B6 are reversed by elements on A/J chromosome 1 and by increased 5-HT1A serotonergic tone.

Real time monitoring of tool wear using multiple modeling method

Real time monitoring of tool wear in machining operations is very crucial in order to prevent tool failures, increase machine utilization and decrease production cost in an automated manufacturing environment. In general the price of the tool is relatively low, but the failure can cause incomparably higher production cost. Over the years, a wide variety of tool condition monitoring (TCM) techniques has been developed based on sensor signals such as cutting forces, acoustic emission and vibration. In this paper, a real time monitoring technique based on multiple modeling method is presented for drilling operations which is one of the most widely used manufacturing operations. The multiple modeling method utilizes cutting forces (thrust and torque) collected during the drilling operation and classifies these forces using hidden Markov models (HMM) to determine wear status of the drill bit. Experimental results have been presented in order to demonstrate the effectiveness of the proposed method.

A support vector machine-based online tool condition monitoring for milling using sensor fusion and a genetic algorithm

In machining systems, the quality of the manufactured part is directly related to the condition of the tool used. Sharp tools are mostly used on the final machining pass to obtain enhanced dimensional accuracy and surface smoothness. Worn tools on the other hand are typically used for coarse machining. The operator usually makes tool assignments based on his experience, the wear levels of the tools and the type of machining task. However, this kind of operator judgment is bound to errors and may not be reliable in processes requiring high precision. Therefore, a tool condition monitoring system is highly desirable to achieve the best results in machining quality. In this study, three-axis cutting forces, torque, three-axis accelerometer and acoustic emission signals were analyzed and used for the development of an online tool condition monitoring system. Various time domain and statistical features extracted from these signals were used to train support vector machine models in a binary decision tree, which was used to predict the condition of the cutting tool. The genetic algorithm was employed for reducing the dimensionality of the feature set by selecting the features that correlates best with the tool condition. Nine experiments were carried out at different cutting conditions. Experimental results demonstrated the efficacy of the proposed scheme. The classification rates for the tool condition monitoring system before and after inclusion of the genetic algorithm step were determined as 89% and 100%, respectively.

Design and Implementation of a Multi Sensor Based Brain Computer Interface for a Robotic Wheelchair

In this study, design and implementation of a multi sensor based brain computer interface for disabled and/or elderly people is proposed. Developed system consists of a wheelchair, a high-power motor controller card, a Kinect camera, electromyogram (EMG) and electroencephalogram (EEG) sensors and a computer. The Kinect sensor is installed on the system to provide safe navigation for the system. Depth frames, captured by the Kinect’s infra-red (IR) camera, are processed with a custom image processing algorithm in order to detect obstacles around the wheelchair. A Consumer grade EMG device (Thalmic Labs) was used to obtain eight channels of EMG data. Four different hand movements: Fist, release, waving hand left and right are used for EMG based control of the robotic wheelchair. EMG data is first classified using artificial neural network (ANN), support vector machines and random forest schemes. The class is then decided by a rule-based scheme constructed on the individual outputs of the three classifiers. EEG based control is adopted as an alternative controller for the developed robotic wheelchair. A wireless 14-channels EEG sensor (Emotiv Epoch) is used to acquire real time EEG data. Three different cognitive tasks: Relaxing, math problem solving, text reading are defined for the EEG based control of the system. Subjects were asked to accomplish the relative cognitive task in order to control the wheelchair. During experiments, all subjects were able to control the robotic wheelchair by hand movements and track a pre-determined route with a reasonable accuracy. The results for the EEG based control of the robotic wheelchair are promising though vary depending on user experience.