Kenneth A. Marko

Growing Structure Multiple Model Systems for Anomaly Detection and Fault Diagnosis

A new anomaly detection scheme based on growing structure multiple model system (GSMMS) is proposed in this paper to detect and quantify the effects of anomalies. The GSMMS algorithm combines the advantages of growing self-organizing networks with efficient local model parameter estimation into an integrated framework for modeling and identification of general nonlinear dynamic systems. The identified model then serves as a foundation for building an effective anomaly detection and fault diagnosis system. By utilizing the information about system operation region provided by the GSMMS, the residual errors can be analyzed locally within each operation region. This local decision making scheme can accommodate for unequally distributed residual errors across different operational regions. The performance of the newly proposed method is evaluated through anomaly detection and quantification in an electronically controlled throttle system, which is simulated using a high-fidelity engine simulation software package provided by a major automotive manufacturer for control system development. DOI: 10.1115/1.3155004

Growing structure multiple model system based anomaly detection for crankshaft monitoring

While conventional approaches to diagnostics focus on detecting and identifying situations or behaviors which have previously been known to occur or can be anticipated, anomaly detection focuses on detecting and quantifying deviations away from learned “normal” behavior. A new anomaly detection scheme based on Growing Structure Multiple Model System(GSMMS) is utilized in this paper to detect and quantify the effects of slowly evolving anomalies on the crankshaft dynamics in a internal combustion engine. The Voronoi sets defined by the reference vectors of the growing Self-Organizing Networks(SONs), on which the GSMMS is based, naturally form a partition of the system operation space. Regionalization of system operation space using SONs makes it possible to model the system dynamics locally using simple models. In addition, the residual errors can be analyzed locally to accommodate unequally distributed residual errors in different regions.

Support Vector Machine with Fuzzy Decision-Making for Real-world Data Classification

This paper proposes an improved model for the application of support vector machine (SVM) to achieve the real-world data classification. Being different from traditional SVM classifiers, the new model takes the thought about fuzzy theory into account. And a fuzzy decision-making function is also built to replace the sign function in the prediction stage of classification process. In the prediction part, the method proposed uses the decision value as the independent variable of fuzzy decision-making function to classify test data set into different classes, but not only the sign of which. This flexible design of decision-making model more approaches to the properties of real-world conditions in which interaction and noise influence exist around the boundary between different clusters. So many misclassified cases can be modified when these sets are considered as fuzzy ones. In addition, a boundary offset is also introduced to modify the excursion produced by the imbalance of real-world dataset. Then an improved and more robust performance will be presented by using this adjustable fuzzy decision-making SVM model in simulations.

Monitoring of complex systems of interacting dynamic systems

Increases in functionality, power and intelligence of modern engineered systems led to complex systems with a large number of interconnected dynamic subsystems. In such machines, faults in one subsystem can cascade and affect the behavior of numerous other subsystems. This complicates the traditional fault monitoring procedures because of the need to train models of the faults that the monitoring system needs to detect and recognize. Unavoidable design defects, quality variations and different usage patterns make it infeasible to foresee all possible faults, resulting in limited diagnostic coverage that can only deal with previously anticipated and modeled failures. This leads to missed detections and costly blind swapping of acceptable components because of one’s inability to accurately isolate the source of previously unseen anomalies. To circumvent these difficulties, a new paradigm for diagnostic systems is proposed and discussed in this paper. Its feasibility is demonstrated through application examples in automotive engine diagnostics.

Immune Systems Inspired Approach to Anomaly Detection and Fault Diagnosis for Engines

As more electronic devices are integrated into automobiles to improve the reliability, drivability and maintainability, automotive diagnosis becomes increasingly difficult to deal with. Unavoidable design defects, quality variations in the production process as well as different usage patterns make it is infeasible to foresee all possible faults that may occur to the vehicle. As a result, many systems rely on limited diagnostic coverage provided by a diagnostic strategy which tests only for a priori known or anticipated failures, and presumes the system is operating normally if the full set of tests is passed. To circumvent these difficulties and provide a more complete coverage for detection of any fault, a new paradigm for design of automotive diagnostic systems is needed. An approach inspired by the functionalities and characteristics of natural immune system is presented and discussed in the paper. The feasibility of the newly proposed paradigm is also partially demonstrated through application examples.

Immune Systems Inspired Approach to Anomaly Detection, Fault Localization and Diagnosis in Automotive Engines

As more electronic devices are integrated into automobiles to improve the reliability, drivability and maintainability, automotive diagnosis becomes increasingly difficult. Unavoidable design defects, quality variations in the production process as well as different usage patterns make it is infeasible to foresee all possible faults that may occur to the vehicle. As a result, many systems rely on limited diagnostic coverage provided by a diagnostic strategy which tests only for a priori known or anticipated failures, and presumes the system is operating normally if the full set of tests is passed. To circumvent these difficulties and provide a more complete coverage for detection of any fault, a new paradigm for design of automotive diagnostic systems is needed. An approach inspired by the functionalities and characteristics of natural immune system is presented and discussed in the paper. The feasibility of the newly proposed paradigm is also demonstrated through application examples.

Accelerated testing of on-board diagnostics

Modern products frequently feature monitors designed to detect actual or impending malfunctions. False alarms (Type I errors) or excessive delays in detecting real malfunctions (Type II errors) can seriously reduce monitor utility. Sound engineering practice includes physical evaluation of error rates. Type II error rates are relatively easy to evaluate empirically. However, adequate evaluation of a low Type I error rate is difficult without using accelerated testing concepts, inducing false alarms using artificially low thresholds and then selecting production thresholds by appropriate extrapolation, as outlined here. This acceleration methodology allows for informed determination of detection thresholds and confidence in monitor performance with substantial reductions over current alternatives in time and cost required for monitor development. Copyright

Modeling Automotive Engine Control Module with Neural Network Trained by Iterated Kalman Filter Algorithm

An attempt has been made to model a production engine control module. Both extended Kalman filter (EKF) algorithm and iterated extended Kalman filter (IEKF) algorithm are used in the construction of the model. The results shows the model trained by both algorithms can produce accurate results with RMS errors in a range of 2- 3%, while iterated extended Kalman filter algorithm outperforms the extended Kalman filter algorithm