Bayesian Multivariate Time-Series Prediction for Missing Manufacturing Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manufacturing systems face challenges in predicting and adjusting multivariate time series data, which are essential for optimizing equipment performance and quality control, due to missing values and asynchronous data acquisition, leading to inefficiencies and potential equipment failures.
Innovation Solution
A method and system utilizing a Bayesian model to interpolate missing values and predict multivariate time series data, including uncertainty, to adjust manufacturing parameters, such as scheduling maintenance and operational settings, by recording, storing, and optimizing data within specific time windows.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional prediction methods are used for multivariate time series data, then the system can handle complete data sets, but the prediction accuracy deteriorates when missing values or asynchronous data acquisition occur
Solution Approach 1:
The patent introduces a Bayesian model as an intermediary mechanism that handles missing data and asynchronous data acquisition. The model uses probabilistic inference to estimate missing values and reconcile timing differences between data sources, allowing the system to maintain prediction reliability even when data is incomplete or arrives at different times. The Bayesian framework acts as a mediator between the imperfect real-world data and the prediction requirements.
2Measurement precision
If complex Bayesian models are used to handle missing data and predict multivariate time series, then prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-defining the Bayesian model structure, priors, and computational frameworks before actual prediction tasks. The system prepares the probabilistic models, likelihood functions, and inference algorithms in advance, so that when data arrives (even if incomplete or asynchronous), the computational heavy lifting has already been done. This reduces real-time computational complexity while maintaining high prediction accuracy.
3Productivity
If real-time prediction and adjustment of manufacturing parameters is implemented, then equipment performance and quality control improve, but data processing time and computational resources increase
Solution Approach 1:
The patent implements continuous prediction and adjustment by making the Bayesian model operate in a continuous or near-continuous manner. Instead of batch processing, the system continuously updates predictions as new data arrives, allowing for real-time equipment performance optimization. The continuous operation of the predictive model ensures that manufacturing parameters are adjusted promptly based on the latest available information, maintaining productivity while managing processing time through efficient incremental updates.
Data Source
AI summary
The present disclosure describes a method of controlling a manufacturing system using multivariate time series and includes storing recording data as a plurality of time series, each time series having a first recorded value and a final recorded value, interpolating, within a first time window, missing values in the plurality of time series using a Bayesian model, the missing values falling between a first and an end time of the respective time series, storing the interpolated values as prediction data, each interpolated value including an uncertainty, loading recorded data of a second time window, loading prediction data of the second time window, predicting, using the Bayesian model, values for each time series that is absent recorded data and prediction data, storing the predicted values, each prediction value including an uncertainty, and adjusting a device that generates the recorded data based on the prediction values within the second time window.