Library for applying a non-uniform DFT over simple time series data from sensors

Question

I have real-time data from sensors monitoring pumps. I need to analyze the duty-cycle of the pumps in the frequency domain. I am pretty rusty in FFT, so I was hoping to find a C++ or Python library that would have a simple API taking in time-series data (i.e. a series of {timestamp, value} real-number pairs), and returning a series of frequencies, amplitudes and phases.

I found several packages taking in complex number series. Moreover, their example code typically uses random data as input, which makes it hard to integrate with production code. A Matlab-based solution would also be hard for me to integrate into my code.

One of the packages I've found:

https://github.com/flatironinstitute/finufft

I know that the problem is my lack of knowledge, but given that sensor-data analysis is such a hot topic today, I was surprised to realize that such a library is so hard to find.

Thanks in advance for your help.

Answer 1

With Python technical computing, the first place you should look for functions is the Numpy and Scipy libraries. It is these libraries that allow you to replace MATLAB with Python in many instances. There are are a set of Numpy functions for discrete fourier transforms. Also for Scipy there are fft functions. If the data is not sampled a regular intervals, you might want to resample it.