Fault Detection and Isolation for UAVs using Neural Ordinary Differential Equations

Enciso-Salas, Luis; Pérez-Zúñiga, Gustavo; Sotomayor-Moriano, Javier

doi:10.1016/j.ifacol.2022.07.200

Cited by 6 publications

(4 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another potential neural architecture with improved robustness with respect to stability is the NeuralODE [33][34][35], which learns a model while considering many integration steps instead of learning in one-step time increments as ResNet does. Even if stability is significantly enhanced, the NeuralODE's training efforts are greater than those required by the ResNet, mainly due to the fact that the back-propagation in the loss function minimization needs the solution of an adjoint problem, which, in some cases, exhibits poor convergence.…”

Section: Final Remarksmentioning

confidence: 99%

Learning Data-Driven Stable Corrections of Dynamical Systems—Application to the Simulation of the Top-Oil Temperature Evolution of a Power Transformer

Ghnatios

Kestelyn

Denis³

et al. 2023

Energies

View full text Add to dashboard Cite

Many engineering systems can be described by using differential models whose solutions, generally obtained after discretization, can exhibit a noticeable deviation with respect to the response of the physical systems that those models are expected to represent. In those circumstances, one possibility consists of enriching the model in order to reproduce the physical system behavior. The present paper considers a dynamical system and proposes enriching the model solution by learning the dynamical model of the gap between the system response and the model-based prediction while ensuring that the time integration of the learned model remains stable. The proposed methodology was applied in the simulation of the top-oil temperature evolution of a power transformer, for which experimental data provided by the RTE, the French electricity transmission system operator, were used to construct the model enrichment with the hybrid rationale, ensuring more accurate predictions.

show abstract

Section: Final Remarksmentioning

confidence: 99%

Learning Data-Driven Stable Corrections of Dynamical Systems—Application to the Simulation of the Top-Oil Temperature Evolution of a Power Transformer

Ghnatios

Kestelyn

Denis³

et al. 2023

Energies

View full text Add to dashboard Cite

show abstract

“…Another potential neural architecture with improved robustness with respect to stability is the NeuralODE [7,8,12], that learns the model while considering many integration steps, instead of learning for one-step time increments as ResNet performs. Even is stability is significantly enhanced, the NeuralODE training efforts are higher than the ones required by the ResNet, mainly due to the fact that the back-propagation in the loss function minimization needs the solution of an adjoint problem, that, in some cases exhibits poor convergence.…”

Section: Final Remarksmentioning

confidence: 99%

Learning Data-Driven Stable Corrections of Dynamical Systems. Application to the Simulation of the Top-Oil Temperature Evolution of a Power Transformer

Ghnatios¹,

Kestelyn²,

Denis³

et al. 2023

Preprint

View full text Add to dashboard Cite

Many engineering systems are described by using differential models whose solutions, generally obtained after discretization, can exhibit a noticeable deviation with respect to the response of the physical systems that those models are expected to represent. In those circumstances one possibility consists of enriching the model in order to reproduce the physical system behavior. The present paper considers a dynamical system, and proposes enriching the model solution by learning the dynamical model of the gap between the system response and the model-based prediction, while ensuring that the time integration of the learnt model remains stable. The proposed methodology is applied on the simulation of the top-oil temperature evolution of a power transformer, whose data is provided by RTE, the French Electricity Transmission System Operator.

show abstract

“…In recent years, Unmanned Aerial Vehicles (UAVs) and drones are highly utilized in a wide range of services such as construction [1], mapping [2,3], the agriculture industries [4], military [5], detecting defects in oil and gas transmission lines, electricity [6,7], and other matters in different industries. The aforementioned types of equipment are prone to malfunctions due to the extent of their application in environments where human access is regarded as limited or difficult.…”

Section: Introductionmentioning

confidence: 99%

“…Further, [5] proposed a robust adaptive observer combined with a radial neural network (RNN) for UAV fault detection. Furthermore, [6] used neural ordinary differential equations for UAV modeling in order to detect fault. In another study, [7] offered adaptive Thau observer (ATO) for estimating UAV states in order to detect and isolate faults in UAV actuators by the amount of residuals.…”

Section: Introductionmentioning

confidence: 99%

Monitoring UAV Status and Detecting Insulation Defects in Transmission Lines with a New Hybrid Classifier based on the Type-2 Fuzzy and Neural Networks

Amiri,

Pourgholi,

Hashjin

et al. 2023

Preprint

View full text Add to dashboard Cite

Monitoring the condition of the transmission lines maintains the health of the power system, prevents unwanted outages, and reduces the costs of repairs and maintenance. This study aims to present a method for monitoring the UAV status and detecting insulation defects in transmission lines with a new hybrid classifier based on the type-2 fuzzy and neural networks. To this aim, the data are preprocessed and balanced, along with the mapping feature vector. Then, two new classifiers are proposed for the image and numerical datasets. The proposed hybrid structure for numerical data contains a Probabilistic Neural Network (PNN), Support Vector Machine (SVM) optimized with Gray Wolf Optimization (GWO) algorithm, and Deep Neural Network (DNN). For image data, the proposed structure is considered as a combination of Convolutional Neural Network (CNN) for feature extraction and three class perceptron classifiers with a hidden layer, SVM optimized with GWO algorithm, and DNN. Finally, the output of the three classifiers enters the Interval Type-2 Mamdani Fuzzy Logic system (IT2FLS) optimized with the Chaos Game Optimization (CGO) algorithm to determine the final state in order to monitor the UAV status and detect the insulation defects. To verify the validity the above-mentioned structure, it is implemented in the MATLAB environment and a laboratory setup is prepared to receive information from the Internet and display the status on the user's mobile phone. The results of the hybrid classifier are compared with those of the previous study. In addition, the classifier proposed for other applications is assessed with two test datasets from the UCI Machine Learning repository to check the generalizability. Based on the results, the proposed hybrid classifier exhibits more accuracy than the other conventional ones.

show abstract

Fault Detection and Isolation for UAVs using Neural Ordinary Differential Equations

Cited by 6 publications

References 6 publications

Learning Data-Driven Stable Corrections of Dynamical Systems—Application to the Simulation of the Top-Oil Temperature Evolution of a Power Transformer

Learning Data-Driven Stable Corrections of Dynamical Systems—Application to the Simulation of the Top-Oil Temperature Evolution of a Power Transformer

Learning Data-Driven Stable Corrections of Dynamical Systems. Application to the Simulation of the Top-Oil Temperature Evolution of a Power Transformer

Monitoring UAV Status and Detecting Insulation Defects in Transmission Lines with a New Hybrid Classifier based on the Type-2 Fuzzy and Neural Networks

Contact Info

Product

Resources

About