Machine Learning HVAC Analysis to Reduce False Maintenance Alarms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional HVAC system data analysis systems require manual modification of criteria and often fail to accurately predict maintenance needs, leading to unnecessary site visits and equipment downtime.
Innovation Solution
A system comprising sensors and a computing device that uses machine learning models to analyze data from building equipment, requesting expert feedback when data significantly deviates from prior patterns, and automatically updating models for improved prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional data analysis systems use a specific set of manual criteria to analyze HVAC system data, then the system structure is simple and easy to implement, but the system cannot accurately predict maintenance needs and requires periodic technician site visits
Solution Approach 1:
The system enables self-service through automated machine learning models that continuously learn from HVAC system data without requiring manual criterion modification. The model automatically identifies maintenance needs and predicts system performance, eliminating the need for periodic technician site visits while improving prediction accuracy beyond traditional manual criteria
Solution Approach 2:
The patent replaces the mechanical system of manual criterion-based analysis with an intelligent machine learning system. The ML model processes sensor data automatically, substituting human expert judgment with algorithmic prediction that continuously improves through learning from historical data, thereby enhancing reliability without proportional increases in operational complexity
2Measurement precision
If machine learning algorithms are incorporated into data analysis systems to predict system performance, then prediction accuracy improves, but the system requires more computational resources and complex infrastructure
Solution Approach 1:
The system performs preliminary action by pre-processing and storing historical HVAC data in structured formats before actual prediction is needed. This preparation work includes data cleaning, feature extraction, and model training in advance, allowing the ML system to make rapid accurate predictions during operation without requiring complex real-time computational infrastructure
Solution Approach 2:
The patent introduces an intermediary layer between raw sensor data and prediction outputs - a trained machine learning model that has been pre-configured with domain knowledge. This intermediary translates complex sensor readings into interpretable maintenance predictions, reducing the computational complexity required at the point of use while maintaining high prediction accuracy
3Ease of operation
If traditional analysis systems notify maintenance needs based on fixed criteria, then the system is easy to operate, but it generates false alarms when data that normally indicates an issue is not actually a concern
Solution Approach 1:
The system applies dynamics by transitioning from static fixed thresholds to dynamic adaptive criteria. The machine learning model continuously adjusts maintenance notification thresholds based on learned patterns from historical data, allowing the system to distinguish between normal variations and actual maintenance needs. This dynamic approach maintains ease of operation while significantly improving notification accuracy by reducing false alarms
Data Source
AI summary
One embodiment of the present disclosure is a system for predicting performance of building equipment. The system comprises one or more sensors in communication with the building equipment, and the sensors are operable to detect characteristics from the building equipment. The system further comprises a computing device in communication with the sensors and in the same geographic location as the sensors. The computing device comprises one or more memory devices configured to store instructions that, when executed on one or more processors, cause the one or more processors to receive data from the sensors, the data based on the detected characteristics. The one or more processors also generate, based on a machine learning model and the data, a predicted performance of the building equipment when the machine learning model comprises a prior data substantially similar to the data.