Multimodal data reduction agent for high density data in IIoT applications

A method, comprising:

• receiving, by a system comprising a processor, a data batch comprising time-series values of a data tag of an industrial device;
• selecting, by the system, a data reduction algorithm from multiple predefined data reduction algorithms based on a determination of whether a probability distribution of the time-series values is contains a single mode, multiple modes, or no modes;
• and
• applying, by the system, the data reduction algorithm to the data batch to yield a reduced data set,
• wherein the multiple predefined data reduction algorithms comprise a first data reduction algorithm for probability distributions containing a single mode, a second data reduction algorithm for probability distributions containing multiple modes, and a third data reduction algorithm for probability distributions containing no modes.

A non-transitory computer-readable medium having stored thereon instructions that, in response to execution, cause a system comprising a processor to perform operations, the operations comprising:

• receiving a data batch comprising time-series values of a data tag of an industrial device;
• determining whether a probability distribution of the time-series values is a unimodal distribution, is a multimodal distribution, or has no mode;
• selecting, based on a result of the determining, a data reduction algorithm from multiple predefined data reduction algorithms, wherein the multiple predefined data reduction algorithms comprise a first data reduction algorithm for unimodal probability distributions, a second data reduction algorithm for multimodal probability distributions, and a third data reduction algorithm for probability distributions containing no modes;
• and
• applying the data reduction algorithm to the data batch to yield a reduced data set.