×

Fuzzy-learning-based extraction of time-series behavior

  • US 8,001,074 B2
  • Filed: 01/31/2008
  • Issued: 08/16/2011
  • Est. Priority Date: 01/31/2008
  • Status: Active Grant
First Claim
Patent Images

1. A computer-implemented method of extracting time-series behavior based on time-series information, the method comprising:

  • loading into a computer time-series data, wherein the loaded time-series data comprises time-series data points comprising an input-component-part and an output-component-part, the input-component-part comprising one or more input components and the output-component-part comprising one or more output components, the one or more input components collectively representing a value from an input space and the one or more output components collectively representing a value from an output space, wherein loading time-series data further comprises incorporating at least one time-related input component of at least one data point into the loaded time-series data;

    dividing into fuzzy regions a range of possible values for each component of the time-series data;

    assigning with one or more computer processors a fuzzy membership function to each fuzzy region;

    generating with one or more computer processors non-conflicting fuzzy rules that are determined at least in part by the fuzzy membership functions and at least in part by how a plurality of the time-series data points are clustered;

    determining a mapping from the input space to the output space based on defuzzification of the fuzzy rules; and

    displaying the mapping in a manner that allows a user to perform time-series prediction or time-series-trend recognition.

View all claims
  • 25 Assignments
Timeline View
Assignment View
    ×
    ×