Virtual Cylinder Pressure Sensor (VCPS) with Individual Variable-Oriented Independent Estimators

Tremendous amount of useful information can be extracted from the cylinder pressure signal for engine combustion control. However, the physical cylinder pressure sensors are undesirably expensive and their health need to be monitored for fault diagnostic purpose as well. This paper presents the results of the development of a virtual cylinder pressure sensor (VCPS) with individual variable-oriented independent estimators. Two neural network-based independent cylinder pressure related variable estimators were developed and verified at steady state. The results show that these models can predict the variables correctly compared with the extracted variables from the measured physical cylinder pressure sensor signal. Good generalization capabilities of the developed models are observed in the sense that the models work well not only for the training data set but also for the new inputs that they have never been exposed to before. Different neural network structures and input variable combinations were explored and compared in terms of performance. The results of this paper clearly show that the VCPS with individual variable-oriented independent estimators can greatly reduce the modeling task as well as improve the robustness and accuracy of the models. The VCPS for transient conditions are suggested.