Modern methods of engine development use complex mathematical models. Adding advanced components such as variable valve trains or direct injection systems to the model increases the degrees of freedom resulting in a high number of measurements for validation. Steadily rising costs for development, time and staff make it crucial for industry to improve the quality of measurements with advanced analysis techniques. Often, such models consider the simulated system as stationary, implying that system variables no longer change with time.This paper presents an internal combustion engine measurement system utilizing algorithms for the real-time evaluation of the state of the engine or its components. Several approaches have been reviewed and tested regarding their applicability. The most straightforward algorithms compare the gradient of a sensor signal to a pre-defined threshold. These techniques are known to be very robust and suitable for motor applications but must be carefully adjusted. The signal quality often varies in engine environments, and typically slow sample rates (1-10 Hz) render frequency-based filtering impractical. A combination of Savitzky-Golay filtering and Total-Variation filtering is used in implementing the system described in this paper. Calibration of the system is required and conducted with the support of a training method. A global weighting function considers the significance of particular signals on the state of the general system.Measurements on an engine test rig validate the assumption that the introduced system can reduce the measurement efforts as test time decreases while reliability of measurements increases.