Lane M. D. Owsley

Self-organizing feature maps and hidden Markov models for machine-tool monitoring

Vibrations produced by the use of industrial machine tools can contain valuable information about the state of wear of tool cutting edges. However, extracting this information automatically is quite difficult. It has been observed that certain structures present in the vibration patterns are correlated with dullness. We present an approach to extracting features present in these structures using self-organizing feature maps (SOFMs). We have modified the SOFM algorithm in order to improve its generalization abilities and to allow it to better serve as a preprocessor for a hidden Markov model (HMM) classifier. We also discuss the challenge of determining which classes exist in the machining application and introduce an algorithm for automatic clustering of time-sequence patterns using the HMM. We show the success of this algorithm in finding clusters that are beneficial to the machine-monitoring application.

Automatic clustering of vector time-series for manufacturing machine monitoring

Our research in online monitoring of industrial milling tools has focused on the occurrence of certain wide-band transient events. Time-frequency representations of these events appear to reveal a variety of classes of transients, and a time-structure to these classes which would be well modeled using hidden Markov models. However, the identities of these classes are not known, and obtaining a labeled training set based on a priori information is not possible for reasons both theoretical and practical. Unsupervised clustering algorithms which exist are only appropriate for single vector patterns. We introduce an approach to unsupervised clustering of vector series based around the hidden Markov model. This system is justified as a generalization of a common single-vector approach, and applied to a set of vector patterns from a milling data set. Results presented illustrate the value of this approach in the milling application.

Feature extraction networks for dull tool monitoring

Automatic feature extraction is a need in many current applications, including the monitoring of industrial tools. Currently available approaches suffer from a number of shortcomings. The Kohonen (1989) self-organizing neural network (SONN) has the potential to act as a feature extractor, but we find it benefits from several modifications. The purpose of these modifications is to cause feature variations to be aligned with the SONN indices so that the indices themselves can be used as measures of the features. The modified SONN is applied to the dull tool monitoring problem, and it is shown that the new algorithm extracts and characterizes useful features of the data.

Using speech technology to enhance isotope ID and classification

Scientific advances are often made when researchers identify mathematical or physical commonalities between different fields and are able to apply mature techniques or algorithms developed in one field to another field which shares some of the same challenges. The authors of this paper have identified similarities between the unsolved problems faced in gamma-spectroscopy for automated radioisotope identification and the challenges of the much larger body of research in speech processing. In this paper we describe such commonalities and use them as a motivation for a preliminary investigation of the applicability of speech processing methods to gamma-ray spectra. This approach enables the development of proof-of-concept isotope classifiers, whose performance is presented for both simulated and field-collected gamma-ray spectra.

Use of modulation spectra for representation and classification of acoustic transients from sniper fire

There are many applications for classification of acoustic transients produced by supersonic projectile fire. Analysis of existing models for such transients suggests they have properties which may be well-captured by the transform of a signal into joint acoustic and modulation frequency: a modulation spectral representation. Simple features are extracted from this representation which enables successful use in such an important classification application.

Self-organizing feature maps with perfect organization

The self-organizing feature maps (SOFMs) introduced by Kohonen (1990) have found use in a wide variety of signal processing applications. The goal of SOFMs is to allow encoding of high-dimensional data vectors in such a manner that a vectors relative position in the codebook is related to the information in the vector in as simple a fashion as possible. One measure of the SOFMs performance in achieving this goal has been proposed by Zrehen and Blayo (1992). According to this measure, many feature maps produced by previous algorithms were disorganized. We review the Zrehen disorganization measure and some of its characteristics. We then show how our version of the SOFM algorithms can accept some simple modifications to produce feature maps which achieve perfect organization under the Zrehen measure of feature map performance. We discuss the emergent geometric properties of resulting feature maps, and illustrate the results using industrial drill vibration data. We note that applying the Zrehen-constrained algorithm to a data set implies certain assumptions about the sets distribution. We discuss the implications of these assumptions in the context of a feature extraction system for industrial milling data.

Broadband mine detection and classification—Preliminary results from a set of low‐frequency shallow‐water experiments

Mine counter measures (MCM) sonars have seen accelerated development over the last few years as Navy interests have shifted into shallow‐water operations. To find mines, most MCM sonar technologies use relatively high‐frequency and narrow‐band signals to produce high‐resolution images of the seafloor. Even so, the problem of discriminating between proud mines and similarly sized false targets and of detecting buried mines at ranges of several hundred meters remains very difficult. The Applied Physics Laboratory‐University of Washington in conjunction with Arete Engineering and Technologies Corporation‐San Diego have been conducting shallow‐water experiments with a relatively low‐frequency and very broadband (2–20 kHz) sonar. The use of low frequencies permits greater penetration into bottom sediments and very broadband signals mitigate the lack of spatial resolution expected from a narrow‐band analysis. A description of the sonar and of the experiments conducted in Puget Sound using mines and minelike fal...

Ordered vector quantization for neural network pattern classification

The accurate classification of time sequences of vectors is a common goal in signal processing. Vector quantization (VQ) has commonly been used to help encode vectors for subsequent classification. The authors depart from this past approach proposing the use of VQ codebook indices, as opposed to codebook vectors. It is shown that one-dimensional ordering of these indices markedly improves the neural-network-based classification accuracy of acoustic time-frequency patterns. The needs for and extensions of multidimensional codebook indices are described.<<ETX>>

A technique for adjusting Gaussian mixture model weights that improves speaker identification performance in the presence of phonemic train/test mismatch.

Speaker identification is complicated by cases where training material is phonemically deficient. Misclassifications can result either because subsequent test material from that speaker contains primarily the phonemes missing from the training data or because that test material is phonemically most consistent with another talker’s model. This situation can arise in any dialog where, for reasons of brevity and clarity, conventions must be imposed on phraseology. We present here a technique for detecting phonemic deficiencies in a speaker model, and then correcting that model to partially compensate for the biased training data. This technique relies upon a specially constructed universal background model (UBM) from which speaker models are adapted. This UBM is formed by weighting several dozen phoneme GMMs using EM training. As a result, each Gaussian component of the UBM (and of the resulting speaker models) corresponds to a specific phoneme. Analysis of the speaker model weights reveals whether the train...

Modeling acoustic target response by component

The scattered acoustic response of underwater objects due to active interrogation has been studied for decades for use in detection and classification applications. As a means of detection, fielded applications date back nearly a hundred years. However, use of responses for robust automated classification has lagged behind, particularly when the internal structure of the objects is of key importance and when the objects may be partially or fully buried. Analytic solutions for simple geometries have provided much understanding of certain physical mechanisms, but transfer to complex structures of practical importance has proven difficult. In recent decades, finite element (FE) modeling has provided a method of accurate simulation of many structures previously considered intractable. However, simulation of such complex objects produces equally complex returns, with the result that the models are often simply considered as a “black box” where the physical interpretation of the response components is tenuous at best. Thus the state of the art is still short of a method for development of robust classification systems for complex objects based on the physics of the objects of interest and the varied conditions under which they may be found. This paper introduces an effort to use FE techniques to simulate individual components of a return by “turning off” most aspects of the physics and allowing the researcher to isolate one mechanism at a time. The goal is a true physical understanding of the complete response, a physically justifiable feature set for classification, and a much simpler path to environmental robustness.