| 摘要 |
<p>PURPOSE:To shorten a time required for retrieval by dividing time series data into comparatively small sections, Fourier-transforming the divided sections, finding out the correlation factor of the Fourier factor, comparing the correlation factor with a previously set correlation factor reference value, and detecting a part continuing the sections decided at its existence of similarity as similar data. CONSTITUTION:The time series data are Fourier-transformed at every fixed period and Fourier series having the number of orders sufficient for expressing the action of original time series data is stored. Time series data to be retrieved are also Fourier-transformed at every fixed period and a correlation factor between the Fourier factor obtained by said Fourier transformation and the Fourier factor of the already stored Fourier series is calculated. A part continuing sections having large correlation factors is detected as similar time series data. Consequently, parts similar to a specific operation state can be completely retrieved from the past process operation record within a short period based on objective evaluation.</p> |
