A survey on online learning and optimization for spark advance control of SI engines

Abstract

One of the most important factors affecting fuel efficiency and emissions of automotive engines is combustion quality that is usually controlled by managing spark advance (SA) in spark ignition (SI) engines. With increasing sensing capabilities and enhancements in on-board computation capability, online learning and optimization techniques have been the subject of significant research interest. This article surveys the literature of learning and optimization algorithms with applications to combustion quality optimization and control of SI engines. In particular, this paper reviews extremum seeking control algorithms for iterative solution of online optimization problems, stochastic threshold control algorithms for iterative solution of probability control of stochastic knock event, as well as feedforward learning algorithms for iterative solution of operating-point-dependent feedforward adaptation problems. Finally, two experimental case studies including knock probabilistic constrained optimal combustion control and on-board map learning-based combustion control are carried out on an SI gasoline engine.

Cycle-to-cycle combustion variability modelling in spark ignited engines for control purposes

“…In other works, they included a third layer to control knock under a desired level, either by a knock constraint which is updated in an OL map 20 or with a likelihood-based control. 21 Corti and co-workers 22,23 also proposed the use of extremum seeking control over the spark and the air-to-fuel ratio (AFR) to optimize a cost function composed from the combustion efficiency, the knock intensity, and the exhaust temperature. Popović et al 24 tested different algorithms of extremum seeking to automatically optimize the variable valve timing (VVT) for minimizing the fuel consumption while Hellstrom et al 25 used extremum seeking for finding the optimal SA in a flex-fuel engine.…”

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

“…Online optimization or calibration of SA using extremum seeking (ES) control has been well studied in existing literatures. 16–22 ES is a control system which is used to determine and maintain the extremum value of a function. Several variants of ES have been developed and have been proven to be robust and efficient in the applications of SA optimization.…”

“…Several variants of ES have been developed and have been proven to be robust and efficient in the applications of SA optimization. 16 For example, sinusoid excitation-based ESs that employ periodic sinusoid waves as excitation signals have been applied to SA optimization (calibration) in flex-fuel/alternative fueled engines. 18,19 Moreover, natural excitation-based ESs that utilize the stochasticity of cyclic combustion indicators (for instance, combustion phase and thermal efficiency) have also been developed.…”

“…The thought of making trade-off between knock prevention and efficiency improvement stimulates the formulation of the probability constrained optimization problem. The data driven–based probability constrained optimization framework for engine control was summarized by Zhang and colleagues 1,16 and first applied for knock probabilistic constrained efficiency optimization. The data driven–based probability constrained optimization method iteratively approaches to the optimal SA for maximizing efficiency in real time.…”

Pigeon-inspired optimization and extreme learning machine via wavelet packet analysis for predicting bulk commodity futures prices

“…Traditional statistical approaches, such as Grey-Markov model [1] and autoregressive integrated moving average (ARIMA) [2,3] have been utilized to predict futures prices. In recent years, intelligent approaches [4][5][6][7], such as artificial neural network (ANN) [8][9][10], support vector machine (SVM) [11][12][13], difference algorithm (DE) [14], and pigeon-inspired optimization (PIO) [15], have better predicting precision for big data. Chen et al [16] utilized the ANN to forecast gold futures prices.…”