The Start of Combustion Prediction for Methane-Fueled HCCI Engines: Traditional vs. Machine Learning Methods

Table 4

The coefficient vector of ML linear regression methods.

Model	N (rpm)	T (K)	P (kPa)		EGR (%)	RH (%)	Real CR	Intercept
OLS	4.2962e − 03	−2.493e − 01	−4.0262e − 02	2.1140e + 01	3.0853e − 02	−5.5125e − 03	−2.2814	499.1991
Lasso	0.0037	−0.2292	−0.0416	0.00	0.03259	−0.0061	−1.9549	491.9509
Ridge	4.2417e − 03	−2.4664e − 01	−4.0385e − 02	1.7200e + 01	3.2108e − 02	−5.7550e − 03	−2.2502	498.7426
SVR	5.2693e − 01	1.3572e + 01	1.7384	1.3029e + 03	−3.1742	8.9697e − 01	6.8226	−6227.37
BR	4.2954e − 03	−2.4926e − 01	−4.0264e − 02	2.1079e + 01	3.0872e − 02	−5.5163e − 03	−2.2809	499.1922
Huber	−7.5840e − 03	6.0737e − 01	−2.3513e − 01	1.7234e − 02	−1.2149e − 01	−1.2149e − 01	8.0966	0.6408