Homogeneous Charge Compression Ignition (HCCI) offers a significant potential to reduce CO2 as well as NOx and particulate emissions. However ensuring stable and efficient HCCI combustion in practical use is a challenge and requires sophisticated control concepts. In the present paper a closed loop control concept is investigated for this purpose. In order to optimize the closed loop control, adaptations with neural networks are introduced. Different sensor concepts for HCCI combustion control are presented and the benefit of the analysis of crankshaft movements is analyzed.