Adaptive and Unconventional Strategies for Engine Knock Control

International Journal of Engine Research

et al. 2019

This article presents an on-board map learning–based spark advance control framework for combustion engines. The proposed control framework addresses the knock probabilistic constrained thermal efficiency optimization problem with three layers. First, in the upper layer, maps of knock event distribution and thermal efficiency are learned with manifold pressure and combustion phase as inputs. Second, the middle layer generates the knock probability constrained optimal combustion phase reference that is subsequently tracked by a hypothesis test-based feedback controller. Third, the lower layer employs a partial likelihood-based knock controller that retards the spark advance in case of the frequent knock events. The key contributions of this work are the three-layer control framework and the knock event distribution map learning in the upper layer. The knock event is supposed to obey binomial distribution, and the distribution is modeled by beta distribution and learned in the perspective of Bayesian learning. Moreover, the normalization algorithm is proposed for online feedfoward map update. The proposed map learning–based spark advance control framework is experimentally validated in a test bench equipped with a spark-ignition engine.

Section: Problem Descriptionmentioning

confidence: 94%

On-board knock probability map learning–based spark advance control for combustion engines

Zhang

International Journal of Engine Research

et al. 2019

“…The SA is adjusted at every engine cycle in this method. A detailed deterministic analysis of the mean and variance of the SA can be found in Selmanaj et al, 21 which shows that a large K adv or K ret for fast SA adjustment leads to a large SA variance.…”

Section: Knock Probability Control Schemementioning

confidence: 99%

“…As a result, the SA approaches the region boundary faster, but the dispersion of the SA also increases. In Selmanaj et al, 21 the conventional approach is combined with an adaptive method using the likelihood for gain adjustment, which efficiently reduces the SA dispersion. Due to that augmentation, knock events can be assumed as independent and identically distributed (i.i.d.)…”

Section: Introductionmentioning

confidence: 99%

Normal-gamma distribution–based stochastic knock probability control scheme for spark-ignition engines

Zhao

Proceedings of the Institution of Mechanical Engineers, Part D:

2019

Spark timing, one of the essential parameters to control combustion in spark-ignition gasoline engines, is often advanced to optimize the power output and fuel economy. An overly advanced spark timing, or equivalently a large spark advance, however, can lead to severe knocking under heavy load engine operating conditions. In a trade-off between engine damage avoidance and power enhancement, the knock probability has to be regulated at a low percentage. Based on the observation that the logarithm of the knock intensity under steady operating conditions follows a normal distribution, in this research, a Bayesian knock probability estimation method is proposed using the normal-gamma distribution and the observed knock intensity. Based on the estimation, a spark advance control algorithm is also developed. The proposed knock probability control algorithm is validated on a full-scale test bench with a production spark-ignition engine. The results show that the proposed method is capable of regulating the knock probability to be close to the target percentage. With different parameter settings, the controller can further be configured to behave more aggressively or conservatively in knock probability estimation and regulation. In comparison with the conventional controller and the maximum likelihood–based controller, and in the tip-in/tip-out test, the proposed method also presents a quick response to transient engine operating conditions and a low spark advance dispersion after the spark advance converges close to the borderline.

“…Since knock events are highly cyclically independent, 11,16 it is hard to determine whether an operating condition is near the boundary through individual knock events; therefore, some stochastic control methods have been developed that uses the statistic properties of the binary knock signals to regulate the SA and to decrease dispersion. In Selmanaj et al, 17 an adaptive SA controller is proposed where a likelihood ratio between the target knock probability and the estimated knock probability is used to adjust the gains of the conventional method. In Peyton Jones et al, 18,19 a likelihood ratio is used to determine the adjustment volume and whether an SA adjustment is necessary.…”

Section: Introductionmentioning

confidence: 99%

Beta distribution-based knock probability map learning and spark timing control for SI engines

Zhao

Proceedings of the Institution of Mechanical Engineers, Part D:

2021

In gasoline engines, the spark timing is often advanced to increase fuel economy under certain heavy load engine operating conditions. As a compromise between the risk of knock and the power output, spark timing is regulated at the boundary where a low knock probability is tolerated. Due to the stochasticity of binary knock events, it is necessary to have a large number of engine cycles for probability estimations, which can slow down the response speed of a controller to operating condition changes. To speed up the spark timing regulation and to reduce the spark timing variance, in this article, a knock probability feedforward map learning method and a spark timing control method are proposed under a unified framework. A learning method that applies the beta distribution is the key contribution of this work. The beta distribution in the map learning part is used to describe knock probabilities with uncertainties and to determine the next engine operating condition for sampling and map learning. In the spark timing method, the beta distribution is applied in the conventional control method to adjust the control gains. The proposed methods are experimentally validated on a test bench equipped with a production Toyota 1.8 L, 4-cylinder SI engine.