Dynamic System Model Identification via Regression Error Rates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dynamic systems, such as HVAC systems, face challenges in accurately modeling and controlling due to inadequate documentation, human behavior, and hidden disturbances, leading to increased costs and deployment issues with manual parameter setup.
Innovation Solution
A method for remotely monitoring dynamic systems to identify a best model among candidate models by estimating parameters, predicting output variables, calculating error rates, and selecting models with the lowest error rates, thereby reducing costs and time required for model identification and model predictive control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual parameter setup is used for model identification, then model accuracy can be maintained, but deployment costs and time increase significantly
Solution Approach 1:
The system performs self-identification of dynamic system models by automatically estimating parameters and evaluating candidate models without requiring manual intervention. The processor autonomously executes regression analyses, calculates error rates, and selects optimal models, enabling the system to serve itself in the model identification process.
Solution Approach 2:
The system changes parameters by automatically estimating model parameters through regression analysis rather than using fixed manual values. It dynamically adjusts parameter selection based on error rate calculations and statistical tests, transforming the parameter setup from a static manual process to a dynamic automated one.
2Measurement precision
If multiple candidate models are evaluated manually, then model selection accuracy improves, but the complexity and cost of deployment increase
Solution Approach 1:
The system implements feedback by calculating error rates for each candidate model and using these results to iteratively refine model selection. The processor evaluates predicted values against actual measurements, feeds back the error information, and uses statistical tests to determine the optimal model based on this feedback loop.
Solution Approach 2:
The model evaluation process is segmented into distinct automated steps: parameter estimation, predicted value calculation, error rate computation, and statistical hypothesis testing. This segmentation allows complex model selection to be broken down into manageable automated tasks that reduce overall deployment complexity.
3Productivity
If automated model identification is implemented, then deployment time and costs are reduced, but measurement precision may deteriorate
Solution Approach 1:
The system replaces manual mechanical processes with automated computational methods. Instead of manual parameter setup and model evaluation, the processor automatically executes regression analyses, calculates error metrics, and performs statistical tests, substituting human effort with algorithmic computation to maintain accuracy while improving efficiency.
Solution Approach 2:
The system performs preliminary actions by pre-defining multiple candidate models with different structures and parameters before the actual model selection process. This preliminary preparation enables the automated system to efficiently evaluate and compare models without requiring complex real-time adjustments, thereby maintaining accuracy while improving deployment speed.
Data Source
AI summary
Identifying models of dynamic systems is described herein. One method for identifying a model of a dynamic system includes estimating a number of parameters for each of a number of models of the dynamic system, predicting an output using the estimated number of parameters for each of the number of models, calculating a rate of error of the predicted output for each of the number of models compared to an observed output, and identifying a best model among the number of models of the dynamic system based on the calculated rate of errors.