Abstract: Systems, techniques, and computer-program products that, individually and in combination, permit machine condition monitoring are provided. In some aspects, state estimation and anomaly localization can be determined jointly. To that end, in some embodiments, systems can be configured using at least a synthetic training dataset. The synthetic training dataset includes sensor output data that incorporates synthetic a random amount of noise to each one of multiple sensor devices that probe an industrial machine. The training dataset also includes synthetic information indicative of location of anomalous sensor device(s) of the multiple sensor devices. Therefore, the systems can learn to determine state estimation and anomalous localization concurrently, in a single operation. Accordingly, the training of the systems is consistent with the operation of the systems during machine condition monitoring.

Abstract: A machine fault diagnosis system is provided. The system combines a rule-based predictive maintenance strategy with a machine learning system. A simple set of rules defined manually by human experts is used to generate artificial training feature vectors to portray machine fault conditions for which only a few real data points are available. Those artificial training feature vectors are combined with real training feature vectors and the combined set is used to train a supervised pattern recognition algorithm such as support vector machines. The resulting decision boundary closely approximates the underlying real separation boundary between the fault and normal conditions.

Abstract: In a framework for acquiring and analyzing data from a network of sensors, plug-in software interfaces are used to provide scalability and flexibility. Data collection set-up data is exchanged through one or more first plug-in software interfaces with data collection devices, to configure the processor to collect measurement data from the data collection devices. Analysis set-up data is exchanged through one or more second plug-in software interfaces with one or more data analysis software packages, to configure the processor to provide a predefined subset of the measurement data to the data analysis software packages and to accept analysis results from the data analysis software packages. Measurement data and analysis results are subsequently exchanged through the plug-in interfaces.

Abstract: A machine fault diagnosis system is provided. The system combines a rule-based predictive maintenance strategy with a machine learning system. A simple set of rules defined manually by human experts is used to generate artificial training feature vectors to portray machine fault conditions for which only a few real data points are available. Those artificial training feature vectors are combined with real training feature vectors and the combined set is used to train a supervised pattern recognition algorithm such as support vector machines. The resulting decision boundary closely approximates the underlying real separation boundary between the fault and normal conditions.

Abstract: The present invention provides methods and apparatus for determining and utilizing detection models, such as models for machine condition monitoring. Specifically, the present invention provides a method for identifying and prioritizing labeled data. The model allows a monitored system to be associated with a calibrated and ordered set of states. Further, in machine condition monitoring, the machine condition is associated with the entire set of states in a particular order with a relevance zero-point. That is, a ranked set of calibrated data describing machine conditions is augmented with an annotation indicating a cut-off between relevant and non-relevant data.

Abstract: Machine condition monitoring on a system utilizes a wireless sensor network to gather data from a large number of sensors. The data is processed using a multivariate statistical model to determine whether the system has deviated from a normal condition. The wireless sensor network permits the acquisition of a large number of distributed data points from plural system modalities, which, in turn, yields enhanced prediction accuracy and a reduction in false alarms.

Abstract: A method and framework are described for detecting changes in a multivariate data stream. A training set is formed by sampling time windows in a data stream containing data reflecting normal conditions. A histogram is created to summarize each window of data, and data within the histograms are clustered to form test distribution representatives to minimize the bulk of training data. Test data is then summarized using histograms representing time windows of data and data within the test histograms are clustered. The test histograms are compared to the training histograms using nearest neighbor techniques on the clustered data. Distances from the test histograms to the test distribution representatives are compared to a threshold to identify anomalies.

Abstract: A method and apparatus for detecting faults in power plant equipment is discloses using sensor confidence and an improved method of identifying the normal operating range of the power generation equipment as measured by those sensors. A confidence is assigned to a sensor in proportion to the residue associated with that sensor. If the sensor has high residue, a small confidence is assigned to the sensor. If a sensor has a low residue, a high confidence is assigned to that sensor, and appropriate weighting of that sensor with other sensors is provided. A feature space trajectory (FST) method is used to model the normal operating range curve distribution of power generation equipment characteristics. Such an FST method is illustratively used in conjunction with a minimum spanning tree (MST) method to identify a plurality of nodes and to then connect those with line segments that approximate a curve.

Abstract: Condition signals of machines are observed and one or more discontinuities are detected in the condition signals. The discontinuities in the condition signals are compensated for (e.g., by applying a shifting factor to models of the signals) and trends of the compensated condition signals are determined. The trends are used to predict future fault conditions in machines. Kalman filters comprising observation models and evolution models are used to determine the trends. Discontinuity in observed signals is detected using hypothesis testing.

Abstract: A method for monitoring machine conditions provides additional information using a one-class classifier in which an evaluation function is learned. In the method, a distance is determined from an anomaly measurement x to a boundary of a region R1 containing all acceptable measurements. The distance is used as a measure of the extent of the anomaly. The distance is found by searching along a line from the anomaly to a closest acceptable measurement within the region R1.

Abstract: In a framework for acquiring and analyzing data from a network of sensors, plug-in software interfaces are used to provide scalability and flexibility. Data collection set-up data is exchanged through one or more first plug-in software interfaces with data collection devices, to configure the processor to collect measurement data from the data collection devices. Analysis set-up data is exchanged through one or more second plug-in software interfaces with one or more data analysis software packages, to configure the processor to provide a predefined subset of the measurement data to the data analysis software packages and to accept analysis results from the data analysis software packages. Measurement data and analysis results are subsequently exchanged through the plug-in interfaces.

Abstract: The present invention provides methods and apparatus for determining and utilizing case-based ranking methods, such as methods for machine condition monitoring. Specifically, the present invention provides a method for identifying and prioritizing labeled data. The method allows a monitored system to be associated with a calibrated and ordered set of states. Further, in machine condition monitoring, the machine condition is associated with the entire set of states in a particular order with one or more relevance zero-points. That is, a ranked set of calibrated data describing machine conditions is augmented with an annotation indicating a cut-off between relevant and non-relevant data.

Abstract: A method for identifying a potential fault in a system includes obtaining a set of training data. A first kernel is selected from a library of two or more kernels and the first kernel is added to a regression network. A next kernel is selected from the library of two or more kernels and the next kernel is added to the regression network. The regression network is refined. A potential fault is identified in the system using the refined regression network.

Abstract: Documents from a data stream are clustered by first generating a feature vector for each document. A set of cluster centroids (e.g., feature vectors of their corresponding clusters) are retrieved from a memory based on the feature vector of the document using a locality sensitive hashing function. The centroids may be retrieved by retrieving a set of cluster identifiers from a cluster table, the cluster identifiers each indicative of a respective cluster centroid, and retrieving the cluster centroids corresponding to the retrieved cluster identifiers from a memory. Documents may then be clustered into one or more of the candidate clusters using distance measures from the feature vector of the document to the cluster centroids.

Abstract: A method for monitoring machine conditions is based on machine learning through the use of a statistical model. A correlation coefficient is calculated using weights assigned to each sample that indicate the likelihood that that sample is an outlier. The resulting correlation coefficient is more robust against outliers. The calculation of the weight is based on the Mahalanobis distance from the sample to the sample mean. Additionally, hierarchical clustering is applied to intuitively reveal group information among sensors. By specifying a similarity threshold, the user can easily obtain desired clustering results.

Abstract: Disclosed is a technique for classifying tissue based on image data. A plurality of tissue parameters are extracted from image data (e.g., magnetic resonance image data) to be classified. The parameters are preprocessed, and the tissue is classified using a classification algorithm and the preprocessed parameters. In one embodiment, the parameters are preprocessed by discretization of the parameters. The classification algorithm may use a decision model for the classification of the tissue, and the decision model may be generated by performing a machine learning algorithm using preprocessed tissue parameters in a training set of data. In one embodiment, the machine learning algorithm generates a Bayesian network. The image data used may be magnetic resonance image data that was obtained before and after the intravenous administration of lymphotropic superparamagnetic nanoparticles.

Abstract: A method is provided for selecting a representative set of training data for training a statistical model in a machine condition monitoring system. The method reduces the time required to choose representative samples from a large data set by using a nearest-neighbor sequential clustering technique in combination with a kd-tree. A distance threshold is used to limit the geometric size the clusters. Each node of the kd-tree is assigned a representative sample from the training data, and similar samples are subsequently discarded.

Abstract: Documents from a data stream are clustered by first generating a feature vector for each document. A set of cluster centroids (e.g., feature vectors of their corresponding clusters) are retrieved from a memory based on the feature vector of the document and a relative age of each of the cluster centroids. The centroids may be retrieved by retrieving a set of cluster identifiers from a cluster table, the cluster identifiers each indicative of a respective cluster centroid, and retrieving the cluster centroids corresponding to the retrieved cluster identifiers from a memory. A list of cluster identifiers in the cluster table may be maintained based on the relative age of cluster centroids corresponding to the cluster identifiers. Cluster identifiers that correspond to cluster centroids with a relative age exceeding a predetermined threshold are periodically removed from the list of cluster identifiers.

Abstract: A method for monitoring machine conditions provides additional information using a one-class classifier in which an evaluation function is learned. In the method, a distance is determined from an anomaly measurement x to a boundary of a region R1 containing all acceptable measurements. The distance is used as a measure of the extent of the anomaly. The distance is found by searching along a line from the anomaly to a closest acceptable measurement within the region R1.

Abstract: A method for monitoring machine conditions provides additional information using a one-class classifier in which an evaluation function is learned. In the method, a distance is determined from an anomaly measurement x to a boundary of a region R1 containing all acceptable measurements. The distance is used as a measure of the extent of the anomaly. The distance is found by searching along a line from the anomaly to a closest acceptable measurement within the region R1.