IFAC-PapersOnLine最新文献

A new fault tolerant estimation method of unmanned aerial vehicle (UAV) dynamics in the presence of sensor/actuator faults with both adaptivity and robustness is proposed. Choosing between robust and adaptive approaches in the event of a sensor/actuator fault is the topic of this study. We describe two methods: a robust technique with R-adaptation and an adaptive method with Q-adaptation. Fault detection in the Kalman filter is based on the chi-square distribution of the normalized quadratic innovation function (NQI). After detection of fault it is proposed to run simultaneously both, R-adaptive and Q-adaptive Kalman filters and compare their estimation performances to distinguish the sensor and actuator faults. As a performance criterion the mean of the quadratic differences between estimation and extrapolation values of robust and adaptive filters is proposed to use.

In the study, the noncentral Wishart matrix trace-based fault detection statistics is proposed for sensor/actuator fault detection in the presence of measurement bias. As the difference from the most of existing innovation-based fault detection methods, this approach allows to detect sensor/actuator faults in the presence of additive measurement errors. The trace of the noncentral Wishart matrix is used in this method for the fault detection statistics. The proposed innovation approach-based sensor/actuator fault detection using trace of the noncentral Wishart matrix is applied to a dynamic model of an unmanned aerial vehicle (UAV). The sensor bias and actuator loss of control effectiveness type faults are considered. The proposed and traditional methods for detecting faults in the presence of slowly developing gyroscope drift are considered and compared.

This paper aims at detecting soft faults, i.e. degradations, in Y-shaped communication networks and locating the faulty branch. The proposed method is based on the transmission coefficient (TC) between each source and the receivers. The estimation of the TC is performed online by power line communication technology through orthogonal frequency division multiplexing scheme. Then, a health indicator per receiver is computed and sent back to the source, which computes a set of structured residual signals for fault detection and localization. This method is validated by data collected on a laboratory test bench. To enhance the robustness of the residuals to noise, modified residuals are proposed. These new residuals are obtained by weighting the centered residuals by the Frequency Response Assurance Criterion. The experimental results confrm the robustness of the weighted residuals while maintaining their sensitivity to soft faults.

This paper deals with the modeling of a photovoltaic system connected to a grid for the simulation of normal and faulty operations and the generation of a data-set for learning a fault detection algorithm based on a Stacked Autoencoder. To evaluate the effectiveness of the proposed approach, a Mean Squared Error is used. This method enables early fault detection, enhancing system relability and efficiency while addressing the need for proactive fault management in the system under normal conditions. Obtained results under different radiation and temperature conditions highlight the relevance of the proposed model and the effectiveness of the fault detection algorithm.

This paper proposes a bidirectional rapidly-exploring random trees (RRT) algorithm to solve the motion planning problem for hybrid systems. The proposed algorithm, called HyRRT-Connect, propagates in both forward and backward directions until an overlap between the forward and backward propagation results is detected. Then, HyRRT-Connect constructs a motion plan through the reversal and concatenation of functions defined on hybrid time domains, ensuring that the motion plan satisfies the given hybrid dynamics. To address the potential discontinuity along the flow caused by tolerating some distance between the forward and backward partial motion plans, we reconstruct the backward partial motion plan by a forward-in-hybrid-time simulation, effectively eliminating the discontinuity. The proposed algorithm is applied to an actuated bouncing ball system to highlight its computational improvement.

In this paper, we provide an automata-based framework for verifying diagnosability property of Cyber-Physical Systems leveraging a notion of so-called hybrid barrier certificates. Concretely, we first construct a so-called (δ,K)-deterministic finite automata ((δ,K)-DFA) associated with the desired diagnosability property, which captures the occurrence of the fault to be diagnosed. Having a (δ,K)-DFA, we show that the verification of diagnosability properties is equivalent to a safety verification problem over a product system between this DFA and the dynamical system of interest. We further show that such a verification problem can be solved via computing hybrid barrier certificates for the product system. To compute the hybrid barrier certificates, we provide a systematic technique leveraging a counter-example guided inductive synthesis framework. Finally, we showcase the effectiveness of our results through a case study.

This paper focuses on distributed controller synthesis for ensuring the safety of large-scale systems with unknown dynamics and known interconnection structures, employing control barrier certificates. Our approach centers on a distributed framework tailored to create control barrier certificates for safety, utilizing the interconnections within large-scale systems. We introduce a novel application of graph neural networks to synthesize these certificates and distributed controllers in a data-driven fashion. The formal correctness of trained networks is subsequently validated through data-driven techniques involving the solution of a sampling-based optimization problem. To illustrate the effectiveness of our methodology, we conduct experiments on a complex interconnected system comprising as many as 1000 components.

Most control synthesis methods under temporal logic properties require a model of the system, however, identifying such a model can be a challenging task. In this work, we develop a direct data-driven control synthesis method for temporal logic specifications, which does not require this explicit modeling step, capable of providing certificates for the general class of linear systems. After collecting a single sequence of input-output data from the system, we synthesize a controller, such that the controlled system satisfies a (possibly unbounded) temporal logic specification. The underlying optimization problem is solved by mixed-integer linear programming. We demonstrate the applicability of the results through simulation examples.

This study compares methodologies for fault classification in reciprocating compressors, focusing on traditional Machine Learning (ML) with classical feature extraction processes and one-dimensional Convolutional Neural Networks (1D-CNN) in Deep Learning (DL). Both techniques demonstrated viability by employing a dataset of compressor vibration signals encompassing ten fault classes. While ML achieved a classification accuracy of 86%, DL reached 90.709%, highlighting its superior learning and generalization abilities, although with longer training times. These findings suggest that, despite ML being effective when relevant prior knowledge is available, DL, particularly with 1D-CNN, offers enhanced fault classification performance for this study case at the expense of additional processing resources.

A two-degree-of-freedom Twin Rotor MIMO (Multiple-Input-Multiple-Output) System (TRMS) is an aerodynamic laboratory equipment at the School of Engineering, University of Leicester for control theory experimentation in the undergraduate (UG) curriculum, Open-Days (ODs), Offer-Holder-Days (OHDs) and research. It is crucial in demonstrating system modelling, simulation, real-time testing, open/closed loop control, and controller design (Proportional-Integral-Derivative, Linear Quadratic Regulator, Model Predictive Control). Feedback received indicates that TRMS experiments have successfully attracted many candidates at ODs/OHDs to the UG aerospace engineering degree programme while giving current students a sense of real-world applicability. Opportunities to further enrich the UG curriculum are explored.