Composite signals are defined as data generated from existing data or from models. Composite signals can be used to add modeled data values or relationships between data columns to quality control tests.
Python facilitates a wide range of analysis options that can be incorporated into Pecos using composite signals. For example, composite signals can be created using the following methods available in open source Python packages (e.g., numpy, scipy, pandas, scikit-learn, tensorflow):
Rolling window statistics
Pecos can also interface with analysis run outside Python using the Python package subprocess.
Once a composite signal is created, it can be used directly within a quality control test, or compared to existing data and the residual can be used in a quality control test.
In the Simple example, a very simple ‘Wave Model’ composite signal is added to the PerformanceMonitoring object.
>>> clocktime = pecos.utils.datetime_to_clocktime(pm.df.index) >>> wave_model = pd.DataFrame(np.sin(10*(clocktime/86400)), ... index=pm.df.index, columns=['Wave Model']) >>> pm.add_dataframe(wave_model)