Development of the nearest neighbour multivariate localized regression modelling technique for steady state engine calibration and comparison with neural networks and global regression

Chi

2011

Self Cite

This is the first part of a study investigating a model-based transient calibration process for diesel engines. The motivation is to populate hundreds of parameters (which can be calibrated) in a methodical and optimum manner by using model-based optimization in conjunction with the manual process so that, relative to the manual process used by itself, a significant improvement in transient emissions and fuel consumption and a sizable reduction in calibration time and test cell requirements is achieved. Empirical transient modelling and optimization has been addressed in the second part of this work, while the required data for model training and generalization are the focus of the current work. Transient and steady-state data from a turbocharged multicylinder diesel engine have been examined from a model training perspective. A single-cylinder engine with external air-handling has been used to expand the steady-state data to encompass transient parameter space. Based on comparative model performance and differences in the non-parametric space, primarily driven by a high engine difference between exhaust and intake manifold pressures (DP) during transients, it has been recommended that transient emission models should be trained with transient training data. It has been shown that electronic control module (ECM) estimates of transient charge flow and the exhaust gas recirculation (EGR) fraction cannot be accurate at the high engine DP frequently encountered during transient operation, and that such estimates do not account for cylinder-to-cylinder variation. The effects of high engine DP must therefore be incorporated empirically by using transient data generated from a spectrum of transient calibrations. Specific recommendations on how to choose such calibrations, how many data to acquire, and how to specify transient segments for data acquisition have been made. Methods to process transient data to account for transport delays and sensor lags have been developed. The processed data have then been visualized using statistical means to understand transient emission formation. Two modes of transient opacity formation have been observed and described. The first mode is driven by high engine DP and low fresh air flowrates, while the second mode is driven by high engine DP and high EGR flowrates. The EGR fraction is inaccurately estimated at both modes, while EGR distribution has been shown to be present but unaccounted for by the ECM. The two modes and associated phenomena are essential to understanding why transient emission models are calibration dependent and furthermore how to choose training data that will result in good model generalization.

Section: Introductionmentioning

confidence: 99%

Development of a model-based transient calibration process for diesel engine electronic control module tables – Part 1: data requirements, processing, and analysis

Chi

2011

Self Cite

“…Similarly, the neural network architecture of one hidden layer with 10 neurons using the sigmoid transfer function and output layer with a linear transfer function was used based on previous performance and recommendations. 4,27 The number of hidden neurons has been shown to be of little significance, as long as there are enough neurons to approximate the function, and early stopping (or other method to prevent overfitting) is employed. 62 Three techniques for utilizing the best combination matrix Three techniques for utilizing the best combination matrix for each modeling method were investigated.…”

Section: Methodsmentioning

confidence: 99%

“…The author has previously investigated purely empirical methods for steady-state and dynamic engine calibration. [4][5][6][7] Since emission compliance technologies have matured and manufacturers of these technologies have consolidated, engine manufacturers have discovered that engine calibration can be a competitive advantage, see, for example, work by Kerekes et al 8 The traditional ''art'' of engine calibration practiced by a few experts with limited knowledge sharing has been transformed into an area of systematic scientific enquiry with its own conferences, for example, the design of experiments (DoEs) conference organized by IAV GmbH 9 and the dedicated calibration/control sessions at the Society of Automotive Engineers (SAE) World Congress. 10 Several commercial products such as the AVL ''Cameo,'' the MathWorks ''Model-Based Calibration (MBC)'' toolbox, the IAV ''EasyDoE'' and the Ricardo ''Global DoE Toolkit'' are available for model-based calibration, in addition to other related products that enable automated data acquisition for engine calibration; see work by Koegeler et al 11 for an example of calibration performed with a commercial product.…”

Section: Introductionmentioning

confidence: 99%

Advantages and applications of transforming empirical model input space with dimensional models

2014

Model-based calibration of steady-state engine operation is commonly performed with highly parameterized empirical models that are accurate but not very robust, particularly when predicting highly nonlinear responses such as diesel smoke emissions. To address this problem, and to boost the accuracy of more robust non-parametric methods to the same level, GT-Power was used to transform the empirical model input space into multiple input spaces that simplified the input-output relationship and improved the accuracy and robustness of smoke predictions made by three commonly used empirical modeling methods: Multivariate Regression, Neural Networks and the k-Nearest Neighbor method. The availability of multiple input spaces allowed the development of two committee techniques: a 'Simple Committee' technique that used averaged predictions from a set of 10 pre-selected input spaces chosen by the training data and the ''Minimum Variance Committee'' technique where the input spaces for each prediction were chosen on the basis of disagreement between the three modeling methods. This latter technique equalized the performance of the three modeling methods. The successively increasing improvements resulting from the use of a single best transformed input space (Best Combination Technique), Simple Committee Technique and Minimum Variance Committee Technique were verified with hypothesis testing. The transformed input spaces were also shown to improve outlier detection and to improve k-Nearest Neighbor performance when predicting dynamic emissions with steady-state training data. An unexpected finding was that the benefits of input space transformation were unaffected by changes in the hardware or the calibration of the underlying GT-Power model.

“…The full quadratic form was found superior to linear or pure quadratic models and included 28 regressor variables corresponding to the six parameters that could be calibrated. A BoxCox [15] transform was used for the emission models based on previous work by Brahma et al [16,17], which suggests that an exponential dependence of emissions on engine parameters increases prediction accuracy because it represents the physical phenomena better than a linear quadratic expression. However, the Box-Cox transform, commonly used by statisticians, imposes an exponential dependence between emissions and all independent variables, as illustrated by equation (1).…”

Section: Modelling Methods and Model Formmentioning

confidence: 99%

Development of a model-based transient calibration process for diesel engine electronic control module tables – Part 2: modelling and optimization

Chi

2011

Self Cite

This is the second part of a study investigating a model-based transient calibration process for diesel engines. The first part addressed the data requirements and data processing required for empirical transient emission and torque models. The current work focuses on modelling and optimization. The unexpected result of this investigation is that when trained on transient data, simple regression models perform better than more powerful methods such as neural networks or localized regression. This result has been attributed to extrapolation over data that have estimated rather than measured transient air-handling parameters. The challenges of detecting and preventing extrapolation using statistical methods that work well with steady-state data have been explained. The concept of constraining the distribution of statistical leverage relative to the distribution of the starting solution to prevent extrapolation during the optimization process has been proposed and demonstrated. Separate from the issue of extrapolation is preventing the search from being quasi-static. Second-order linear dynamic constraint models have been proposed to prevent the search from returning solutions that are feasible if each point were run at steady state, but which are unrealistic in a transient sense. Dynamic constraint models translate commanded parameters to actually achieved parameters that then feed into the transient emission and torque models. Combined model inaccuracies have been used to adjust the optimized solutions. To frame the optimization problem within reasonable dimensionality, the coefficients of commanded surfaces that approximate engine tables are adjusted during search iterations, each of which involves simulating the entire transient cycle. The resulting strategy, different from the corresponding manual calibration strategy and resulting in lower emissions and efficiency, is intended to improve rather than replace the manual calibration process.