The aim of this thesis is to obtain a supervisory control strategy for an high-performance Hybrid Electric Vehicle. These strategies are obtained solving the energy management problem in which the quasi-static vehicle model perform the speed profile of a given drive cycle minimising the fuel consumption. This control problem has to deal both with continuous variables, like the torque split between the two energy converter, and discrete variables, like the engine on/off decision or the gear selection, resulting thus in a Mixed Integer Optimal Control Problem (MIOPC). In this thesis work the problem has been solved using a canonical method, Pontryagin's Minimum Principle (PMP) and an innovative method, which uses the potentiality offered by Linear Programming (LP). To take into account constraints on the battery's State of Charge the problem is transformed from a Two Point Boundary Value Problem (TPBVP) into a Multi Point Boundary Value Problem (MPBVP). Given the high performance in terms of computational times of the Linear Programming, this technique is used to select the optimal gear shifting strategy, together with the torque split and engine on/off decision. In order to do so, a new method has been devised which iterates between the minimisation of the Hamiltonian to calculate the optimal gears and the solution of the Linear Program to calculate the optimal torque split and engine on/off decisions. The effectiveness of this method in reaching the global optimum has been assessed comparing the results with those obtained using Forward Dynamic Programming. The vehicle models object of study are Ferrari LaFerrari and a new hybrid powertrain. Pontryagin's Minimum Principle is able to deal with non-linear, map based models of the components of the powertrains, while for the Linear Programming a convex linear model has been used. The second powertrain offers a huge number of gear combinations, thus to conclude the thesis work a gears pre-selection strategy based on engine and electric motor maps has been devised to reduce the computational times and obtain still optimal results.
Lo scopo di questa tesi è quello di ottenere una strategia di controllo di alto livello per un'auto sportiva ibrida. Queste strategie sono ottenute risolvendo un "energy management problem" in cui al modello quasi-statico del veicolo viene dato in input il profilo di velocità di un ciclo di omologazione in cui si vuole minimizzare il consumo di carburante. Questo genere di problemi devono prendere in considerazione variabili continue, come la distribuzione della coppia fra i due motori, e discrete, come l'accensione/spegnimento del motore a combustione interna o la marcia ingranata, creando quindi un "Mixed Integer Optimal Control Problem (MIOPC)". In questa tesi il problema è stato risolto usando metodi ben noti in letteratura, come il metodo del minimo di Pontryagin (PMP) e innovativi, che sfruttano le potenzialità offerte dalla Linear Programming (LP). Per considerare anche i vincoli sullo stato di carica della batteria, il problema deve essere trasformato: il Two Point Boundary Value Problem (TPBVP) viene trasformato in un Multi Point Boundary Value Problem (MPBVP). Considerando le elevate potenzialità in termini di tempi di calcolo offerte dalla Linear Programming, questo metodo è stato utilizzato per selezionare anche le marce ottime, oltre alla suddivisione della coppia e alla decisione di on/off del motore a combustione interna. Per farlo è stato realizzato un metodo innovativo che itera fra la minimizzazione dell'Hamiltoniano per calcolare le marce ottime e la soluzione del problema lineare per ottenere la suddivisione ottima della coppia e la decisione di on/off del motore a combustione interna. I risultati ottenuti sono stati comparati con quelli ricavati tramite Forward Dynamic Programming per dimostrare che questo metodo è in grado di raggiungere un ottimo globale. I modelli presi in esame in questa tesi sono la Ferrari LaFerrari e un nuovo veicolo ibrido. L'algoritmo basato sul metodo di Pontryagin è in grado di utilizzare modelli matematici non lineari e basati sulle mappe di efficienza dei componenti del powertrain, per la Linear Programming, invece, è stato utilizzato un modello lineare convesso. Il secondo powertrain offre molte possibilità nella scelta delle marce, quindi, per concludere questa tesi, è stato elaborato un algoritmo per la preselezione delle marce basato sulle mappe di efficienza dei due motori. Questo algoritmo aiuta a ridurre i tempi di calcolo e ottenere comunque risultati vicini all'ottimo.
Optimal energy management and gearshift strategies for high-performance hybrid electric vehicles
VISCERA, NICOLA
2017/2018
Abstract
The aim of this thesis is to obtain a supervisory control strategy for an high-performance Hybrid Electric Vehicle. These strategies are obtained solving the energy management problem in which the quasi-static vehicle model perform the speed profile of a given drive cycle minimising the fuel consumption. This control problem has to deal both with continuous variables, like the torque split between the two energy converter, and discrete variables, like the engine on/off decision or the gear selection, resulting thus in a Mixed Integer Optimal Control Problem (MIOPC). In this thesis work the problem has been solved using a canonical method, Pontryagin's Minimum Principle (PMP) and an innovative method, which uses the potentiality offered by Linear Programming (LP). To take into account constraints on the battery's State of Charge the problem is transformed from a Two Point Boundary Value Problem (TPBVP) into a Multi Point Boundary Value Problem (MPBVP). Given the high performance in terms of computational times of the Linear Programming, this technique is used to select the optimal gear shifting strategy, together with the torque split and engine on/off decision. In order to do so, a new method has been devised which iterates between the minimisation of the Hamiltonian to calculate the optimal gears and the solution of the Linear Program to calculate the optimal torque split and engine on/off decisions. The effectiveness of this method in reaching the global optimum has been assessed comparing the results with those obtained using Forward Dynamic Programming. The vehicle models object of study are Ferrari LaFerrari and a new hybrid powertrain. Pontryagin's Minimum Principle is able to deal with non-linear, map based models of the components of the powertrains, while for the Linear Programming a convex linear model has been used. The second powertrain offers a huge number of gear combinations, thus to conclude the thesis work a gears pre-selection strategy based on engine and electric motor maps has been devised to reduce the computational times and obtain still optimal results.File | Dimensione | Formato | |
---|---|---|---|
2019_04_Viscera.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
10.17 MB
Formato
Adobe PDF
|
10.17 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/10589/146186