Process State Prediction with Virtual Error Training Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data-driven simulators face challenges in preventing the propagation of prediction errors over time, particularly in multi-input multi-output non-linear systems, leading to reduced prediction accuracy and generalization performance.
Innovation Solution
A method involving a processor that generates an input-output data set with added virtual error data to train a machine learning model, allowing it to accurately predict future states by incorporating potential errors, thereby preventing error propagation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If a dynamic prediction model is trained using operation data without virtual error data, then the model can be trained quickly and with simple data preparation, but prediction accuracy deteriorates over time due to error propagation in recursive predictions
Solution Approach 1:
The patent applies preliminary action by adding virtual error data to the training dataset before model training. This pre-processing step incorporates potential future errors into the training phase, enabling the model to learn error compensation patterns in advance. The virtual error data is generated by adding noise with specific statistical properties (mean=0, variance=σ²) to the original operation data, creating a more robust training foundation that prevents error propagation during recursive predictions.
2Reliability
If ensemble prediction with multiple models is used to reduce prediction error influence, then prediction reliability improves, but device complexity and computational resources increase
Solution Approach 1:
The patent applies parameter changes by modifying the training data parameters rather than changing the model architecture or数量. Specifically, it transforms the original operation data into augmented training data by adding virtual error components with controlled statistical parameters (mean, variance). This approach maintains a single model structure while improving prediction reliability through parameter-level data transformation, avoiding the complexity of ensemble methods.
3Measurement precision
If more training data is collected to improve prediction accuracy, then model performance improves, but data collection time and storage requirements increase
Solution Approach 1:
The patent applies copying by creating virtual replicas of existing operation data with added error components. Instead of collecting additional real-world data over time, the system generates synthetic training samples by copying and transforming existing data points. Each original data point is replicated with added virtual error, creating augmented training datasets that improve model robustness without requiring extended data collection periods or increased storage capacity.
Data Source
AI summary
An information processing apparatus includes a processor configured to acquire time-series data related to a process value that represents an operation amount with respect to a process and a state of the process, generate an input-output data set in which an operation amount and a process value at a first time are adopted as an input sample and a process value at a second time subsequent to the first time is adopted as an output sample, add virtual error data to each process value of the input sample, and train a machine learning model that outputs a process value at the second time while adopting the operation amount and the process value at the first time as input, by using the input-output data set to which the virtual error data is added.