Linear Regression and its use in predicting the link between ionization current and the pressure signal in a hybrid mode engine

Homogenous Charge Compression Ignition (HCCI) is an alternative to Spark Ignited (SI) combustion, which can provide part-load efficiencies as high as compression ignition engines and energy densities as high as SI engines, without high levels of NOx or Particulate Matter (PM). The principle of operation involves reaching the thermal oxidization barrier of a homogeneous air-fuel mixture. This combustion practice is enabled by diluting then compressing the mixture with the Trapped Residual Gases (TRG) to dilute the initial charge thus keeping combustion temperatures down. Introduction of exhaust gasses in the mixture can be achieved by the use of early exhaust valve closure and late inlet valve opening. The charge is well mixed avoiding particulate emissions, and by using exhaust gasses for load regulation the need for throttled operation is removed allowing the realization of high efficiencies, low pumping losses and a resulting 15 - 20 % improvement in fuel economy. One of the major disadvantages of HCCI is that it can only be used at part load conditions, and so it has been proposed that a 'hybrid mode' engine operation strategy could be used, where the engine uses SI operation at cold start, high load and idle. This demands a closed loop feedback control signal to give a seamless transition between SI (where Start of Combustion (SOC) is dictated by the spark plug) and HCCI (where SOC is unknown). This paper investigated the use of an ion current sensor to accurately predict the Peak Pressure Position (PPP) and Peak Pressure Magnitude (PPM). This is achieved with the mathematically simple and computationally efficient approach of linear regression. This technique can be used between engine cycles so adjustments can be made to engine strategies to compensate for difficult conditions. The results showed that there is a definite correlation between the two signals and that the ion current can be used to control the engine during HCCI mode. Finally a good balance between computation time and accuracy was obtained using 20 cycles