Towards real-time optimal energy management of HEV powertrains using stochastic dynamic programming

Leroy, Thomas; Malaizé, Jérémy; Corde, Gilles

doi:10.1109/vppc.2012.6422661

Cited by 24 publications

(20 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using SDP in look-ahead control applications has been studied [21,24]. In these papers, the load has the Markov property and the probability distribution of the load is also independent of time.…”

Section: Load Case Creation For the Sdp Methodsmentioning

confidence: 99%

Development of Look-Ahead Controller Concepts for a Wheel Loader Application

Nilsson

Fröberg

Åslund

2014

Oil Gas Sci. Technol. – Rev. IFP Energies nouvelles

View full text Add to dashboard Cite

-This paper presents two conceptual methods, based on dynamic programming, for onestep look-ahead control of a Continuously Variable Transmission (CVT) in a wheel loader. The first method developed, designated Stochastic Dynamic Programming (SDP), uses a statistical load prediction and stochastic dynamic programming for minimizing fuel use. The second method developed, designated Free-Time Dynamic Programming (FTDP), has vehicle speed as a state and introduces a fixed 0:1 s delay in the bucket controls in a combined minimization of fuel and time. The methods are evaluated using a set of 34 measured loading cycles, used in a 'leave one out' manner. The evaluation shows that the SDP method requires about 1=10th of the computational effort of FTDP and has a more transparent impact of differences in the cycle prediction. The FTDP method, on the other hand, shows a 10% lower fuel consumption, which is close to the actual optimum, at the same cycle times, and is able to complete a much larger part of the evaluation cycles.Re´sume´-De´veloppement de concepts d'une commande pre´dictive, destine´e a`une application pour chargeur sur pneus -Ce document pre´sente deux me´thodes de conception, base´es sur la programmation dynamique, pour la commande a`un pas de pre´diction d'une transmission continuˆment variable (Continuously Variable Transmission, CVT) d'un chargeur sur pneus. La premie`re me´thode de´veloppe´e, appele´e programmation dynamique stochastique (Stochastic Dynamic Programming, SDP) utilise une pre´diction statistique de la charge et la programmation dynamique stochastique pour minimiser l'utilisation de carburant. La seconde me´thode de´veloppe´e, appele´e programmation dynamique a`temps libre (Free-Time Dynamic Programming, FTDP), e´tablit la vitesse du ve´hicule en tant qu'e´tat et introduit un retard de 0,1 s dans les commandes du godet pour minimiser a`la fois l'utilisation de carburant et le temps ne´cessaire a`l'ope´ration. Les me´thodes sont e´value´es en s'appuyant sur 34 cycles de chargement mesure´s, utilise´s selon la me´thode de validation croise´e « leave-one-out ». L'e´valuation montre que la me´thode SDP requiert environ 1 dixie`me de l'effort de calcul de la me´thode FTDP, et qu'elle a un impact plus transparent sur les e´carts dans la pre´diction du cycle. D'un autre coˆte´, avec la me´thode FTDP on obtient une re´duction de 10 % de la consommation de carburant, ce qui est proche de l'optimum re´el, pour les meˆmes dure´es de cycle, et elle permet de re´aliser une plus grande partie des cycles d'e´valuation.

show abstract

Section: Load Case Creation For the Sdp Methodsmentioning

confidence: 99%

Development of Look-Ahead Controller Concepts for a Wheel Loader Application

Nilsson

Fröberg

Åslund

2014

Oil Gas Sci. Technol. – Rev. IFP Energies nouvelles

View full text Add to dashboard Cite

show abstract

“…2) Drive cycle transition probability matrix, as already described by (8). 3) SOC transition matrix, describing the probability of progressing to each SOC at the next timestep, given the SOC and control decision at the present.…”

Section: A Information Requiredmentioning

confidence: 99%

“…Although the two transition probability matrices in (8) and (10) could be combined into a single matrix in practice, they are kept separate because P(X k+1 = x k+1 | x k ) would far exceed the RAM capacity of modern desktop computers.…”

Section: A Information Requiredmentioning

confidence: 99%

“…Higher values of λ require more iterations before the solution converges, and so computation times escalate quickly. It is not necessary to place a cost on SOC deviation to ensure charge sustenance because with a high enough λ, the controller can only be charge depleting over an infinite horizon, though it may be desirable to enable the use of a smaller λ or to reduce the likelihood of the SOC being very low when the vehicle is switched OFF [8]. Our ambition is to have a λ sufficiently high to ensure that the control strategy is charge sustaining, but which does not increase computation time unnecessarily.…”

Section: A Effect Of λ On Charge Sustenancementioning

confidence: 99%

“…The present literature on the optimal control of hybrid vehicles advocates stochastic dynamic programming (SDP) [1]- [8], which aims to use Bellman's principle of optimality applied to a statistical model of what the future is likely to entail, rather than deterministic dynamic programming (DDP), which may only be implemented over a drive cycle that is known precisely in advance. While DDP and SDP are both global optimization techniques, DDP produces a control policy that is time variant, whereas SDP enables a stationary (time-invariant) optimal control policy to be found as a function of vehicle state; the result is globally optimal if run for an infinite period of time over a drive cycle with the same statistical distributions as the data on which it was based.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Stochastic Dynamic Programming in the Real-World Control of Hybrid Electric Vehicles

Vagg

Akehurst

Brace

et al. 2016

IEEE Trans. Contr. Syst. Technol.

View full text Add to dashboard Cite

show abstract

Deep generative models for vehicle speed trajectories

Behnia,

Karbowski,

Sokolov

2023

Appl Stoch Models Bus & Ind

View full text Add to dashboard Cite

Generating realistic vehicle speed trajectories is a crucial component in evaluating vehicle fuel economy and in predictive control of self‐driving cars. Traditional generative models rely on Markov chain methods and can produce accurate synthetic trajectories but are subject to the curse of dimensionality. They do not allow to include conditional input variables into the generation process. In this paper, we show how extensions to deep generative models allow accurate and scalable generation. Proposed architectures involve recurrent and feed‐forward layers and are trained using adversarial techniques. Our models are shown to perform well on generating vehicle trajectories using a model trained on GPS data from Chicago metropolitan area.

show abstract

Towards real-time optimal energy management of HEV powertrains using stochastic dynamic programming

Cited by 24 publications

References 11 publications

Development of Look-Ahead Controller Concepts for a Wheel Loader Application

Development of Look-Ahead Controller Concepts for a Wheel Loader Application

Stochastic Dynamic Programming in the Real-World Control of Hybrid Electric Vehicles

Deep generative models for vehicle speed trajectories

Contact Info

Product

Resources

About