Air conditioning load learning apparatus and air conditioning load prediction apparatus
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing air conditioning load prediction methods struggle to accurately account for the impact of building operations on internal heat generation due to the difficulty in obtaining and completeness of operation-related information.
Innovation Solution
An air conditioning load learning apparatus that utilizes an actual load acquisition unit, first information acquisition unit, and a learning unit to generate a learning model associating building operation information with actual air conditioning load, enabling the effect of operations on internal heat generation to be grasped.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical model-based air conditioning load prediction is used, then prediction framework is established, but accuracy is insufficient due to inability to capture building operation effects
Solution Approach 1:
The patent merges the physical model-based prediction method with a data-driven learning model. The physical model provides a baseline prediction framework, while the learning model captures the effects of building operations that the physical model cannot represent. By combining these two approaches, the system achieves both the structural rigor of physical models and the flexibility to capture operational nuances, thereby improving overall prediction accuracy.
Solution Approach 2:
The patent introduces a learning model as an intermediary between the physical model and the final prediction result. This learning model acts as a mediator that processes building operation information and adjusts the physical model's predictions to reflect actual operational effects. The intermediary learning model bridges the gap between the simplified physical model and the complex reality of building operations.
2Measurement precision
If building operation information is collected to improve prediction accuracy, then effect of operations can be grasped, but information acquisition becomes difficult and incomplete
Solution Approach 1:
The learning model is designed to work with the available building operation information without requiring a complete and sophisticated information acquisition system. It automatically learns from the data that is provided, adapting to the nuances of building operations even when the information is incomplete. This self-service approach allows the system to improve prediction accuracy without imposing heavy requirements on information collection infrastructure.
Solution Approach 2:
The patent transforms building operation information into appropriate input parameters for the learning model. By changing the form and representation of the operation information into suitable parameters, the system can effectively utilize available data without requiring complex acquisition systems. The parameter transformation process makes the system robust to incomplete or varying quality of input information.
Data Source
AI summary
An air conditioning load learning apparatus includes an actual load acquisition unit, a first information acquisition unit, and a learning unit. The actual load acquisition unit acquires an actual air conditioning load in a target space inside a target building. The first information acquisition unit acquires first information about an operation of the target building. The learning unit generates a learning model using at least the first information as an explanatory variable and a value regarding the actual air conditioning load as an objective variable. An air conditioning load prediction apparatus includes the air conditioning load learning apparatus.