Abnormality Prediction Using Time-Aligned Multiseries Neural Input
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
When predicting abnormal events using a neural network, inconsistencies in measurement times and dimensions between multiple data time series obtained from different measurement instruments can lead to reduced prediction accuracy.
Innovation Solution
A system is proposed that reconstructs multidimensional array data from multiple data time series, incorporating measurement parameters and relative time values. This data is then processed using a neural network structure with an input layer, intermediate layers comprising recurrent neural networks, and an output layer to calculate predictive information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple data time series from different measurement instruments are synthesized as input data for neural network, then prediction accuracy can be improved, but inconsistency in measurement times and dimensions reduces prediction accuracy
Solution Approach 1:
The patent applies preliminary action by performing data preprocessing before feeding to the neural network. Specifically, it aligns multiple data time series to a common time base and handles dimensional inconsistencies through interpolation or padding operations, ensuring consistent input format before prediction.
Solution Approach 2:
The patent introduces an intermediary data processing layer between the raw measurement instruments and the neural network. This intermediary layer includes components for time alignment, dimensionality normalization, and data formatting, which mediate the inconsistencies between different measurement instruments.
2Adaptability or versatility
If data from multiple measurement instruments with different measurement intervals are used, then comprehensive system monitoring is achieved, but time inconsistency between data streams occurs
Solution Approach 1:
The patent performs time alignment as a preliminary action by establishing a common time base before processing. It uses interpolation methods to resample data from instruments with different measurement intervals to match the reference time schedule, eliminating time alignment errors.
Solution Approach 2:
The patent applies dynamics by using flexible time alignment methods that can adapt to varying measurement intervals. The system dynamically adjusts data sampling rates and time stamps through interpolation and resampling operations, allowing integration of data from instruments with different temporal characteristics.
3Quantity of substance
If data with different dimensional structures are combined, then multi-parameter analysis is enabled, but dimensional inconsistency prevents accurate processing
Solution Approach 1:
The patent performs dimensional normalization as a preliminary action by transforming data from instruments with different dimensional structures into a unified format. It uses padding, truncation, or interpolation operations to ensure all data streams have consistent dimensionality before neural network processing.
Solution Approach 2:
The patent applies parameter changes by transforming data dimensions through mathematical operations. It changes the dimensional parameters of data from different instruments to match a target dimension structure, enabling consistent processing while preserving the essential information from multi-dimensional measurements.
Data Source
AI summary
A system for abnormality prediction is provided, which is able to precisely predict occurrences of abnormal events by using a neural network even in the case where inconsistency between multiple data time series obtained by a plurality of measurement instruments occurs. The system for abnormality prediction includes: a data reconstruction system configured to reconstruct, from the plurality of data time series, multidimensional array data that includes, as array elements, a plurality of measurement parameters and a plurality of relative time values that are assigned to the measurement parameters, respectively; and an inference calculator configured to calculate predictive information on the abnormal event, by performing calculation based on a neural network structure which includes an input layer for receiving the multidimensional array data, an intermediate layer structure containing one or more recurrent neural networks, and an output layer.