Load monitoring is one of the key components of structural health monitoring, aiming at the real-time reconstruction of the loads acting on a structure. In this work, the load estimation is obtained through the application of a linear approach based on calibration matrices, called Calibration Matrix Approach. The real time aerodynamic load and the full-strain field reconstruction are achieved by a least-squares minimization of an error functional defined as a comparison between measured strains in discrete positions and a numerical formulation of the same, function of an equivalent load set, however representative of the actual aerodynamic loading condition. It was verified that, once the equivalent loads have been correctly determined, the methodology is able to reconstruct the full strain (and, thus the stress) field in every point of complexly loaded structures, without requiring any training data set to be defined. Furthermore, a sensor (and load) network optimization, both in terms of number and positions, has been applied for two aeronautical structures, namely an unmanned aerial vehicle (UAV) and a co-cured composite wing-box specimen. Optimization allowed to enhance the calibration matrix load monitoring performances and to limit the number of sensors, decreasing the associated installation costs. In particular, the optimizations has been performed by applying multi-objective genetic algorithms. Since the procedure makes extensive use of numerical simulation, a Finite Element model of the UAV structure has been experimentally validated in terms of strain response, enabling its usage for structural response. A sensitivity analysis in terms of different strain field produced by different simulated flight manoeuvres has been conducted in order to define the features of interest during the optimization procedure, as well as in terms of different fiber optic types (FBGs) to choose the best candidate for real deployment. The optimization procedure provided good results both in terms of sensors network and equivalent load set network for the UAV structure during flight operations. The second structure, namely the co-cured wing box specimen, was considered only in a numerical framework, in preparation for future experimental tests. Once again, a preliminary sensitivity analysis was followed by a sensor network optimization. However, since this structure will be loaded by two actuators in laboratory, the load set is simply defined by the actuators positions themselves. Lastly, performance indices have been calculated for both structures, after a statistical analysis of the loading conditions was performed in a virtual environment with an insight on the procedure robustness against normally distributed disturbances, providing promising results both in terms of force and strain field reconstruction.
Uno degli aspetti più importanti del monitoraggio strutturale è la ricostruzione in tempo reale dei carichi agenti su una struttura. In questa tesi, tramite l’applicazione di un approccio lineare basato su matrici di calibrazione, è stato possibile ricostruire sia i carichi che lo stato di deformazione in strutture aeronautiche complesse, partendo da un numero limitato di misure di deformazione. La procedura si basa sulla minimizzazione dell’errore quadratico tra le deformazioni misurate e quelle ricostruite numericamente tramite la matrice di calibrazione. La definizione del tipo e del numero di carichi equivalenti è di fondamentale importanza per il corretto funzionamento del metodo, poiché essi influenzano direttamente la configurazione della matrice di calibrazione. In questo lavoro di tesi, i carichi aerodinamici distribuiti sono stati semplificati tramite la creazione di una serie di pressioni ad andamento triangolare. Il metodo sviluppato è stato poi applicato per definire la posizione ottimale di sensori in due strutture aeronautiche: un aeromobile a pilotaggio remoto ed una porzione di ala in materiale composito facente parte di un drone di classe MALE (Medium Altitude Long Endurance). In particolare, per effettuare l’ottimizzazione a livello numerico (sulla base di un modello agli elementi finiti) sono stati utilizzati algoritmi genetici, in quanto essi permettono, in breve tempo, il raggiungimento di risultati quasi ottimali. In seguito all’ottimizzazione della rete di sensori, le prestazioni del sistema di ricostruzione e monitoraggio dei carichi in diverse condizioni di carico e in presenza di rumore gaussiano sono state analizzate tramite la definizione di indici statistici. I risultati ottenuti confermano le potenzialità del metodo nello stimare sia i carichi agenti, sia lo stato di deformazione.
Design and implementation of a load monitoring system for aeronautical structures based on a calibration matrix approach
Dal BOSCO, LUCA;BORTOLOTTI, DAVIDE
2018/2019
Abstract
Load monitoring is one of the key components of structural health monitoring, aiming at the real-time reconstruction of the loads acting on a structure. In this work, the load estimation is obtained through the application of a linear approach based on calibration matrices, called Calibration Matrix Approach. The real time aerodynamic load and the full-strain field reconstruction are achieved by a least-squares minimization of an error functional defined as a comparison between measured strains in discrete positions and a numerical formulation of the same, function of an equivalent load set, however representative of the actual aerodynamic loading condition. It was verified that, once the equivalent loads have been correctly determined, the methodology is able to reconstruct the full strain (and, thus the stress) field in every point of complexly loaded structures, without requiring any training data set to be defined. Furthermore, a sensor (and load) network optimization, both in terms of number and positions, has been applied for two aeronautical structures, namely an unmanned aerial vehicle (UAV) and a co-cured composite wing-box specimen. Optimization allowed to enhance the calibration matrix load monitoring performances and to limit the number of sensors, decreasing the associated installation costs. In particular, the optimizations has been performed by applying multi-objective genetic algorithms. Since the procedure makes extensive use of numerical simulation, a Finite Element model of the UAV structure has been experimentally validated in terms of strain response, enabling its usage for structural response. A sensitivity analysis in terms of different strain field produced by different simulated flight manoeuvres has been conducted in order to define the features of interest during the optimization procedure, as well as in terms of different fiber optic types (FBGs) to choose the best candidate for real deployment. The optimization procedure provided good results both in terms of sensors network and equivalent load set network for the UAV structure during flight operations. The second structure, namely the co-cured wing box specimen, was considered only in a numerical framework, in preparation for future experimental tests. Once again, a preliminary sensitivity analysis was followed by a sensor network optimization. However, since this structure will be loaded by two actuators in laboratory, the load set is simply defined by the actuators positions themselves. Lastly, performance indices have been calculated for both structures, after a statistical analysis of the loading conditions was performed in a virtual environment with an insight on the procedure robustness against normally distributed disturbances, providing promising results both in terms of force and strain field reconstruction.File | Dimensione | Formato | |
---|---|---|---|
2019_7_Bortolotti_Dal_Bosco.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
47.05 MB
Formato
Adobe PDF
|
47.05 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/148737