Building Equipment Reliability Modeling for Chiller Fault Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Predicting chiller faults in HVAC systems is challenging due to various influencing factors, and existing machine learning models are prone to inaccuracies from incorrect training data, leading to overestimation of failure probabilities and unnecessary maintenance costs.
Innovation Solution
A method that involves generating reliability models by calibrating runtime data using climate information and combining warranty claim data with censored data to train Weibull or Cox models, while trimming statistically insignificant data and accounting for idle times, to accurately predict chiller component failures and initiate automated maintenance actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional machine learning models are trained with available data to predict chiller failures, then failure prediction capability is provided, but accuracy deteriorates due to overestimation of failure probabilities and false alarms
Solution Approach 1:
The patent applies preliminary action by calibrating runtime data with climate information before training the reliability model. This preprocessing step adjusts the training data to account for environmental factors that affect chiller operation, thereby improving the accuracy of failure probability predictions before the actual prediction process begins
Solution Approach 2:
The patent changes parameters by transforming raw runtime data into calibrated runtime data using climate-based adjustments. This parameter transformation modifies the input features to better reflect actual operational conditions, leading to more accurate failure probability estimates and reduced false alarms
2Reliability
If maintenance is performed frequently to ensure equipment reliability, then equipment reliability is improved, but maintenance costs increase due to unnecessary maintenance activities
Solution Approach 1:
The system performs preliminary prediction of failure probabilities before maintenance is scheduled. By using the calibrated reliability model to forecast actual failure risks, the system enables maintenance to be performed only when genuinely needed, avoiding unnecessary maintenance activities and associated costs while still ensuring equipment reliability
Solution Approach 2:
The patent implements feedback by using predicted failure probabilities to dynamically adjust maintenance scheduling. The system continuously monitors calibrated failure risks and provides feedback that guides when maintenance should be performed, creating an adaptive maintenance strategy that balances reliability with cost efficiency
Data Source
AI summary
A method for affecting operation of building equipment includes providing a plurality of reliability models that model failure probabilities of components of the building equipment as functions of equipment runtime, providing associations of the components with a plurality of subsystems of the building equipment, calculating, for the plurality of subsystems of the building equipment, probabilities of subsystem failure based on the reliability models for the components and the associations, and initiating an automated action to affect operation of the building equipment based on the probabilities of subsystem failure.