The work presented in this paper will highlight selected artificial intelligence approaches as applied within an Integrated Vehicle Health Management (IVHM) system. The selected vehicle subsystem areas to be discussed include electromechanical actuators (EMAs), propulsion system performance, vehicle structural integrity and general signal anomaly detection. Artificial intelligence methods including neural networks, fuzzy logic and trained probabilistic classifiers are described within the context of the selected subsystem applications. In addition, discussion on individual subsystem health condition indicators as applied within an intelligent, model-based reasoning approach is presented that examines health state and functional availability of individual components, subsystems, and the overall vehicle. The AI implementations described herein illustrate the integration of detection, diagnostic, and prognostic reasoning capabilities from across critical subsystems on a vehicle platform. The examples provided illustrate how the selected AI technologies can be implemented throughout an end-to-end application, from data signal quality checks to off-board prognostic assessments.
This paper presents methods and tools which can be used within the framework of diagnostics and prognostics to accommodate imprecision of real systems. We outline the advantages and disadvantages of the different techniques and show how they can be used in a hybrid fashion to complement each other. We conclude the paper with a number of successful real world examples.
Recent trends focusing on Industry 4.0 concept and smart manufacturing arise a data-driven fault diagnosis as key topic in condition-based maintenance. Fault diagnosis is considered as an essential task in rotary machinery since possibility of an early detection and diagnosis of the faulty condition can save both time and money. Traditional data-driven techniques of fault diagnosis require signal processing for feature extraction, as they are unable to work with raw signal data, consequently leading to need for expert knowledge and human work. The emergence of deep learning architectures in condition-based maintenance promises to ensure high performance fault diagnosis while lowering necessity for expert knowledge and human work. This paper presents developed technique for deep learning-based data-driven fault diagnosis of rotary machinery. The proposed technique input raw three axis accelerometer signal as high-definition image into deep learning layers which automatically extract signal features, enabling high classification accuracy.
Large manufacturing companies are considering to deliver to their customer base "guaranteed uptime" instead of the conventional service contracts. Modern industry is concerned about extending the lifetime of its critical processes and maintaining them only when required. Significant aspects of these trends include the ability to diagnose impending failures, prognose the remaining useful lifetime of the process and schedule maintenance operations so that uptime is maximized. Prognosis is probably the most difficult of the three issues leading to conditionbased maintenance. This paper attempts to address this challenging problem with intelligence-oriented techniques, specifically dynamic wavelet neural networks. Dynamic wavelet neural networks incorporate temporal information and storage capacity into their functionality so that they can predict into the future, carrying out fault prognostic tasks. An example is presented in which a trained dynamic wavelet neural network successfully prognoses a defective bearing with a crack in its inner race.
Complex dynamical systems, such as aircraft, chemical processes, power plants, shipboard equipment, etc., are required to maintain an acceptable level of operational integrity and availability. Current research aims to maximize uptime by maintaining such systems only when required. A viable and cost-effective diagnostic/prognostic system architecture must integrate a number of functionalities while exhibiting attributes of flexibility and scalability. It must account for fault modes that are inherent to the current operating state of the system and its usage patterns (Hadden et al. 1999). Furthermore, it must be able to predict accurately the remaining useful lifetime of failing components and manage effectively uncertainty (Hadden et al. 1999). This paper introduces an integrated diagnostic/prognostic architecture that builds upon means to identify the system's operating mode and usage pattern using concepts from hybrid system theory and Petri networks as decision support tools, mechanisms to extract an optimum feature vector based on data-mining and diagnostic/prognostic algorithms that are designed employing a fuzzy logic expert system paradigm and static/dynamic wavelet neural network constructs for fault detection/isolation and for estimation of the remaining useful lifetime of a failing component. Essential elements of the architecture are implemented and validated on a laboratory scale process consisting of multiple tanks, control equipment, sensors and actuators.