The highest frequency that is currently allowed for the analysis service is 120Hz. This is a limiting when you want to use it to do analysis on higher frequency data. A higher limit, such as 1kHz would be an appreciable improvement.
ADMIN RESPONSE
Aug 20, 2022
Any calculations done at this data rate should be done at the hardware/control level and not at a software level like the PI System.
In response to Jerome Lefebvre, "Currently working on the case where a cu..."
Currently working on the case where a customer has 1 MHz vibration sensor (AE sensor). Data is stored in the PI Data Archive at a slower rate, in large part due to performance consideration.
While, the data is stored at a high frequency, there isn't much of a need to display information that this data rate, thus we were looking into seeing if the analysis service could produce aggregations of those data.
For example, taking averages every 1 milliseconds.
I am curious - at what rate are they able to store the data in the PI Data Archive? How many of such high frequency streams would they typically have?
Can you give an example of the data that requires analysis at 1kHz? If so, you are talking hardware level millisec responses...
We do have examples of data collection at 1kHz, 10kHz, 20kHz etc. for a single stream but the analysis is done on samples every 5 mins or so.
In response to Gopal Gopalkrishnan, "Can you give an example of the data that..."
Currently working on the case where a customer has 1 MHz vibration sensor (AE sensor). Data is stored in the PI Data Archive at a slower rate, in large part due to performance consideration.
While, the data is stored at a high frequency, there isn't much of a need to display information that this data rate, thus we were looking into seeing if the analysis service could produce aggregations of those data.
For example, taking averages every 1 milliseconds.
The goal is for manufacturing, and the events they are trying to understand happen in the range of 100ms or so. With the current limit on analysis system, this would mean only around a dozen or so aggregates would be displayed for each of those events.
Yes, I am fully aware that this use case is pushing the PI System quite a lot.
In response to Jerome Lefebvre, "Currently working on the case where a cu..."
For the vibration data coming from NI devices - at say, 30kHz, 50kHz etc. the FFT analysis is run on the device - and you bring data related to the spectrum - see the CBM 2017 Lab examples.
In response to Gopal Gopalkrishnan, "For the vibration data coming from NI de..."
Yes, that is what is done in this case as well.
The customer also wants to be able store and visualize the raw data as well as they can. They are quite aware that this means at a lower rate, but the question is always, by how much of a lower rate.
In response to Jerome Lefebvre, "Yes, that is what is done in this case a..."
In order to store and visualize raw data at high resolution, they will need to either create (i.e. time-consuming) or purchase the capability. A potential solution is the Setpoint MPS / CMS . It integrates with PI AF to store high-resolution waveforms and provides a GUI to visualize the data (e.g. spectrum plots, waveforms, polar plots, etc.).
In response to Jerome Lefebvre, "Yes, that is what is done in this case a..."
In order to store and visualize raw data at high resolution, they will need to either create (i.e. time-consuming) or purchase the capability. A potential solution is the Setpoint MPS / CMS . It integrates with PI AF to store high-resolution waveforms and provides a GUI to visualize the data (e.g. spectrum plots, waveforms, polar plots, etc.).
In response to Jerome Lefebvre, "Yes, that is what is done in this case a..."
In order to store and visualize raw data at high resolution, they will need to either create (i.e. time-consuming) or purchase the capability. A potential solution is the Setpoint MPS / CMS . It integrates with PI AF to store high-resolution waveforms and provides a GUI to visualize the data (e.g. spectrum plots, waveforms, polar plots, etc.).
In response to Jerome Lefebvre, "Yes, that is what is done in this case a..."
In order to store and visualize raw data at high resolution, they will need to either create (i.e. time-consuming) or purchase the capability. A potential solution is the Setpoint MPS / CMS . It integrates with PI AF to store high-resolution waveforms and provides a GUI to visualize the data (e.g. spectrum plots, waveforms, polar plots, etc.).
In response to Jerome Lefebvre, "Yes, that is what is done in this case a..."
In order to store and visualize raw data at high resolution, they will need to either create (i.e. time-consuming) or purchase the capability. A potential solution is the Setpoint MPS / CMS . It integrates with PI AF to store high-resolution waveforms and provides a GUI to visualize the data (e.g. spectrum plots, waveforms, polar plots, etc.).
