Predictive building control system with neural network based constraint generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing building control systems lack an efficient method to optimize energy consumption by HVAC equipment using predictive models that automatically generate constraints.
Innovation Solution
A neural network-based predictive model is used to generate constraints for optimizing HVAC system operations, integrating with a building management system to control energy consumption based on historical data and real-time conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional control systems are used, then implementation is simpler, but they fail to account for complex interdependencies between building systems and adapt to changing conditions
Solution Approach 1:
The patent introduces a neural network as an intermediary component that processes sensor data and generates setpoint adjustments. This intermediary layer enables the control system to adapt to changing conditions and complex interdependencies without requiring direct modification of the entire control architecture, thus improving adaptability while managing complexity through modular integration.
2Measurement precision
If more sensors and data collection are added, then model accuracy improves, but computational burden and processing time increase
Solution Approach 1:
The system performs preliminary actions by pre-processing sensor data and maintaining a trained neural network model that can quickly generate predictions. The neural network is trained offline on historical data, so during operation, it rapidly processes current sensor inputs without requiring intensive real-time computation, thus achieving high model accuracy while minimizing processing time.
3Loss of energy
If predictive algorithms are implemented, then energy efficiency improves, but system reliability decreases due to computational requirements
Solution Approach 1:
The control system serves itself by using the neural network to autonomously predict optimal setpoints and generate control actions without requiring constant human intervention or complex external computational resources. The system uses its own sensor data and historical patterns to make decisions, improving energy efficiency while maintaining reliability through self-contained operation.
4Adaptability or versatility
If existing control systems are used, then integration is easier, but they operate systems in isolation without considering interdependencies
Solution Approach 1:
The neural network control system is designed with multi-functionality to handle various building systems (HVAC, lighting, security) through a unified predictive framework. It processes diverse sensor inputs and generates coordinated control actions across multiple systems, enabling integrated operation while managing complexity through a universal control architecture that can adapt to different system configurations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A predictive building control system includes equipment operable to provide heating or cooling to a building and a predictive controller. The predictive controller includes one or more optimization controllers configured to perform an optimization to generate setpoints for the equipment at each time step of an optimization period subject to one or more constraints, a constraint generator configured to use a neural network model to generate the one or more constraints, and an equipment controller configured to operate the equipment to achieve the setpoints generated by the one or more optimization controllers at each time step of the optimization period.