Hybrid Regression-ANN Modeling for Low-Data Building Predictions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Building management systems face challenges in predicting relationships between physical plant inputs and outputs, as regression models are less accurate initially due to lack of data, while ANN models require extensive data sets for accurate predictions, necessitating a hybrid approach that leverages both techniques effectively.
Innovation Solution
A building management system incorporating a database, trust region identifier, regression model predictor, distance metric calculator, and artificial neural network model predictor, which uses weighted averages or Kalman filters to determine combined predictions based on regression and ANN model outputs, improving prediction accuracy and adaptability as data sufficiency thresholds are met.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If regression models are used for predictions, then predictions can be made immediately upon startup before large data sets are collected, but prediction accuracy deteriorates due to lack of sufficient operational data
Solution Approach 1:
The system dynamically switches between regression models and ANN models based on data availability. Initially, regression models are used when data is scarce, and as operational data accumulates, the system transitions to ANN models for higher accuracy predictions. This dynamic adaptation resolves the contradiction between immediate prediction availability and prediction accuracy.
Solution Approach 2:
The system changes the modeling approach parameter based on the amount of accumulated operational data. When data quantity is below a threshold, regression models are employed; when data quantity exceeds the threshold, ANN models are activated. This parameter change enables the system to optimize between prediction availability and accuracy at different operational stages.
2Measurement precision
If ANN models are used for predictions, then prediction accuracy improves with access to large data sets, but the system cannot make predictions immediately upon startup due to data requirements
Solution Approach 1:
Regression models are used in advance during the startup phase when operational data is not yet available. These preliminary predictions enable the system to function immediately, and as data accumulates, the system transitions to ANN models for improved accuracy, thus resolving the timing conflict.
Solution Approach 2:
The system dynamically adjusts its modeling strategy based on the accumulation of operational data. The transition from regression to ANN models is not static but evolves over time as data becomes available, allowing the system to achieve both immediate prediction capability and high accuracy predictions as conditions permit.
3Measurement precision
If a hybrid model combining regression and ANN is implemented, then both immediate predictions and high accuracy are achieved, but system complexity increases due to multiple modeling components
Solution Approach 1:
The hybrid modeling system is segmented into distinct components: regression model module, ANN model module, data sufficiency evaluation module, and prediction selection module. Each component has a specific function, and they work together in a coordinated manner. This segmentation manages complexity by creating modular, independently manageable units while achieving the benefit of both model types.
Solution Approach 2:
A data sufficiency evaluation module acts as an intermediary between the regression and ANN models. This mediator assesses whether sufficient operational data is available and determines which model should be activated. The intermediary manages the complexity by providing a clear decision-making layer that coordinates the two modeling approaches without requiring direct integration of their internal complexities.
Data Source
AI summary
A method for initiating and automatically improving model-driven operations in a low-data scenario includes creating a regression model using pre-operation data prior to initiating the model-driven operations, using the regression model to initiate and perform the model-driven operations during an operational stage, collecting operational data during the operational stage, creating a first artificial neural network model using the operational data, transitioning from using the regression model to perform the model-driven operations to using the first artificial neural network model to perform the model-driven operations responsive to the operational data satisfying a first sufficiency threshold.