Data-based modeling for dynamic behavior

The design of experiments (DoE) element enables you to achieve maximum model accuracy with a minimum measurement effort. The DoE experimental design module (ExpeDesDynamic) makes it easy to plan the underlying measurements, and to propose a time curve of the adjustment variables in which different final values with different gradients are approached methodically and in a space-filling way. It allows you to adjust the range of parameter variation, to add stationary points and excerpts from real driving cycles, and to consider specific aspects and peculiarities of dynamic measurements carried out on engine test benches. Graphical representations of the curves (trajectories) in all dimensions facilitate straightforward validation of your DoE plan.

ASCMO-DYNAMIC uses time-based scope views to display the system behavior based on the model. It also offers several other options for visualizing dependencies and influences, including the scatter plot. Extensive analysis options include statistical information such as mean, extreme values and root mean square model error (RMSE). In addition, model predictions can be compared with real measurements to evaluate curve accuracy over time. To achieve a suitable sampling rate during measurement, ASCMO-DYNAMIC allows you to reduce the amount of data by downsampling. This increases the quality of the model and reduces the time required for model training.

ASCMO-DYNAMIC makes it easy for less experienced users to apply the advanced modelling methods required for accurate, robust prediction of the dynamic/transient system behavior. At the same time, it gives experts the freedom to adjust model parameters with a high degree of flexibility. ASCMO-DYNAMIC offers a wide variety of modelling approaches such as Gaussian process models, recurrent or convolutional neural networks, ensemble models, as well as autoencoder for anomaly detection. Using ASCMO-DYNAMIC automated machine-learning, the tool automatically determines the specific mathematical function that best represents the real system behavior. All models can be exported in formats such as Simulink, C code. and FMU/FMI for use outside ASCMO.

ASCMO-DYNAMIC allows you to use the models as a virtual measuring instrument. An example of this is the prediction of emissions over a complete driving cycle based on time series of input variables. For all desired parameter combinations of input variables, the model then calculates the corresponding output variables. Just like real measurements (for example from the engine test bench), this artificial data can be used, analyzed graphically, and exported as a CSV file for further processing in Excel.