| 摘要 |
<p>A time-series signal predicting apparatus which predicts future time-series data on the basis of past time-series data and present time-series data. Vector generation means generates n-dimensional vector data (n: natural number) on the basis of n time-series data inputted, and the data are stored in vector memory means. Another vector generation means generates (n + 1)-dimensional vector data on the basis of the past time-series data inputted, classifies similar vectors into L vector groups by using the vector data of up to the n-dimensions and stores them in vector group memory means. Next, the vector classification means classifies the inputted vector data into the vector groups, and vector selection means selects k vector data, in the order of similarity to the time-series data inputted later, from the vector group into which the vector data are classified. Past data extraction means extracts k data of the (n + 1)th dimension from (n + 1)-dimensional vectors belonging to the selected vector group, and outputs them with their similarities. Data predicting means predicts and outputs time-series data at the next time after the inputted time-series data on the basis of the k data so extracted, the inputted vector data and the similarities.</p> |
