Non-Bayesian social learning theory provides a framework that models distributed inference for a group of agents interacting over a network. Agents iteratively form and communicate beliefs about an unknown state of the world with their neighbors using a learning rule. Existing approaches assume agents have access to precise statistical models (in the form of likelihoods) for the state of the world. However in many situations, such models must be learned from finite data. We propose a social learning rule that takes into account uncertainty in the statistical models using second-order probabilities. Therefore, beliefs derived from uncertain models are sensitive to the amount of past evidence collected for each hypothesis. These beliefs characterize whether or not the hypotheses are consistent with the true state of the world. We explicitly show the dependency of the generated beliefs with respect to the amount of prior evidence. Furthermore, as the amount of prior evidence goes to infinity, learning occurs and is consistent with traditional social learning theory.

Heat exchangers are critical components of the environmental control system (ECS) of an aircraft. The ECS regulates temperature, pressure, and humidity of the cabin air. Fouling of the heat exchangers in an ECS may occur due to the deposition of external substances (e.g., debris) on the fins that obstruct the air flow, which increases the pressure drop across the heat exchanger and degrades its efficiency. Fouling is a critical issue, because it necessitates time consuming, periodic, and expensive maintenance. In this regard, this paper presents a two step process for fouling diagnosis of the heat exchanger: optimal sensor set selection that contains the most relevant information for fault classification and robust data analysis and sensor fusion in the presence of various uncertainties for the inference of fouling severity via different machine learning tools. This process of heat exchanger fouling diagnosis is implemented and tested on the data generated from an experimentally validated high-fidelity Simulink model of the ECS provided by an industry partner.

This paper presents a distributed supervisory control algorithm that enables opportunistic sensing for energy-efficient target tracking in a sensor network. The algorithm called Prediction-based Opportunistic Sensing (POSE), is a distributed node-level energy management approach for minimizing energy usage. Distributed sensor nodes in the POSE network self-adapt to target trajectories by enabling high power consuming devices when they predict that a target is arriving in their coverage area, while enabling low power consuming devices when the target is absent. Each node has a Probabilistic Finite State Automaton which acts as a supervisor to dynamically control its various sensing and communication devices based on target's predicted position. The POSE algorithm is validated by extensive Monte Carlo simulations and compared with random scheduling schemes. The results show that the POSE algorithm provides significant energy savings while also improving track estimation via fusion-driven state initialization.

This paper addresses the issue of fault diagnosis in the heat exchanger of an aircraft Air Conditioning System (ACS). The heat exchanger cools the air by transferring the heat to the ram-air. Due to a variety of biological, mechanical and chemical reasons, the heat exchanger may experience fouling conditions that reduces the efficiency and could considerably affect the functionality of the ACS. Since, the access to the heat exchanger is limited and time consuming, it is preferable to implement an early fault diagnosis technique that would facilitate Condition Based Maintenance (CBM). The main contribution of the paper is pre-flight fault assessment of the heat exchanger using a combined model-based and data-driven approach of fault diagnosis. A Simulink model of the ACS, that has been designed and validated by an industry partner, has been used for generation of sensor data for various fouling conditions. A total of nine different fouling levels are simulated including the nominal condition. Subsequently, the output temperature data of the heat exchanger is analyzed using the Principal Component Analysis (PCA) method for feature extraction. The Support Vector Machine (SVM) and the k-Nearest Neighbor (k-NN) methods are used for data classification into different faulty conditions. The results are evaluated by cross-validation and presented in terms of the confusion matrix, Correct Classification Rate (CCR), sensitivity, and specificity.

This paper addresses the issues of Fault Detection and Isolation (FDI) in complex networked systems such as the Environmental Control System (ECS) of an aircraft. The ECS controls and supplies pressurized air to the aircraft and consists of multiple subsystems that in turn consist of interconnected components, heterogeneous sensing devices, and feedback controllers. These complex interconnections and feedback control loops make fault detection and isolation a very challenging task in the ECS. For example, a faulty component yields off-nominal outputs which are inputs to the other coupled components. This coupling leads to off-nominal outputs from otherwise healthy components, thus causing unwanted false-alarms. Secondly, due to off-nominal inputs, the healthy components are driven beyond their normal operating conditions, leading to cascading failures. In this regard, this paper presents a hierarchical fault detection and isolation method that facilitates FDI in complex networked systems while reducing the computational complexity and false alarms. The underlying algorithm is called System-Level Isolation and Detection (SLIDE), and it relies on a diagnostic tree that systematically isolates faulty subsystems and their components. The proposed method is tested and validated on the data generated from an experimentally validated simulation model of the ECS provided by our industry partner. The results show that the application of the SLIDE algorithm reduces the computational complexity and the false alarm rate, while maintaining a high Correct Classification Rate (CCR).