Multi-Resolution Variable Prediction for Technical Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting variables in technical systems, such as fuel cells and internal combustion engines, either focus on long-term or short-term effects, failing to consider both simultaneously effectively.
Innovation Solution
A computer-implemented method using multiple models with different temporal resolutions to predict variables, where parameters from one model are mapped to another, allowing for the consideration of both long-term and short-term effects by training models at various resolutions and using multilayer state space models with linear transition models and additive disturbance variables.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single model is used for predicting variables in technical systems, then the model can be simple and easy to implement, but it cannot effectively consider both long-term and short-term effects simultaneously
Solution Approach 1:
The patent divides the prediction task into multiple models operating at different temporal resolutions. A first model processes high-resolution data for short-term effects, while a second model processes low-resolution data for long-term effects. This segmentation allows each model to specialize in specific time scales, effectively resolving the contradiction between comprehensive adaptability and model simplicity.
Solution Approach 2:
The patent introduces a new dimension of temporal resolution by training models at multiple scales (high resolution for short-term, low resolution for long-term). This dimensional approach allows the system to capture effects across different time scales simultaneously, transforming a single-dimensional prediction problem into a multi-dimensional solution that addresses both short-term and long-term dynamics.
2Measurement precision
If multiple models with different temporal resolutions are used to predict variables, then both long-term and short-term effects can be considered, but the device complexity and computational requirements increase
Solution Approach 1:
The patent implements a dynamic multi-resolution modeling approach where models operate at different temporal resolutions and are selectively applied based on the prediction task requirements. The system dynamically integrates high-resolution short-term predictions with low-resolution long-term predictions, allowing adaptive precision while managing complexity through selective model application rather than always using the full multi-model system.
Solution Approach 2:
The patent employs a nested model structure where a coarse-grained long-term prediction model provides a framework that is refined by fine-grained short-term prediction models. The high-resolution model operates within the temporal context established by the low-resolution model, creating a nested architecture where predictions at one scale inform and constrain predictions at another scale, thereby improving accuracy while organizing complexity in a hierarchical manner.
Data Source
AI summary
A device, computer program, and computer-implemented method for determining a variable of a technical system. An input variable is determined for a first model for determining the variable at a first temporal resolution. A first time series is provided, at the first temporal resolution, including values which characterize an operating variable of the technical system. A second time series is provided. at a second temporal resolution, including values which characterize the operating variable of the technical system, the first and second temporal resolutions being different. The second time series is mapped using a second model for determining a first prediction for the variable of the technical system at the second temporal resolution on the first prediction. Parameters of a second model are determined, using the second time series, which are mapped on parameters of a third model at the first temporal resolution.