Time-Series Prediction Device Using Dynamic Data Selection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for predicting future data using time-series data are compromised when outlier values are present, leading to decreased prediction precision, especially when the number of data points is limited, as they often require replacing actual values with prediction values, thereby reducing the usable data for subsequent predictions.

Innovation Solution

A prediction device that acquires and processes time-series data, determining whether to include present time data based on prediction variation, using only past time data when the variation is significant, and both past and present data when the variation is within acceptable limits, to enhance prediction accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If actual values are replaced with prediction values when large prediction errors occur, then prediction reliability is improved, but the number of usable actual values for prediction decreases

Engineering Contradiction:

Improveprediction reliabilityVSAvoidnumber of usable data points

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent changes the parameter of data usage by introducing a dynamic selection mechanism based on prediction error magnitude. Instead of always replacing or always using actual values, the system adapts the data usage strategy by comparing prediction errors against thresholds, thereby optimizing both reliability and data quantity utilization through parameter-based decision making

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If present time data is used in prediction, then prediction precision is improved, but prediction stability deteriorates when outlier values are present

Engineering Contradiction:

Improveprediction precisionVSAvoidprediction stability

Core Design Contradiction:

Measurement precisionVSStability of the object's composition

Solution Approach 1:

The patent implements a feedback mechanism where prediction errors are calculated and used to control the data selection process. The system continuously monitors prediction errors, compares them against thresholds, and adjusts data usage accordingly - using present time data when errors are small (high precision) and excluding it when errors are large (maintaining stability), thereby resolving the contradiction between precision and stability

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent introduces dynamics into the data selection process by making the inclusion of present time data conditional rather than static. The system dynamically adjusts whether to use present time data based on real-time prediction error assessment, allowing the prediction model to adapt its data composition according to current data quality conditions

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS10540609B2Prediction device, prediction method, and recording medium

Publication Date: 2020.01.21 CANON KK

AI summary

To precisely predict future data even when the number of pieces of time-series data is small, in predicting the future data, using the time-series data. When the future data is predicted using the time-series data, whether present time data is used is determined based on prediction variation or a data transition, and then the prediction of the future data is performed.