Modular Building Equipment Fault Prediction With Sparse Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fault detection methods for building equipment in building management systems often rely on robust historical data, which is not always available in practice, making it difficult to predict and mitigate faults effectively.
Innovation Solution
A method involving the creation of a fault prediction model, which includes preprocessing training data by calculating synthetic features, resampling, and interpolating to achieve a constant sample rate, and modularly constructing the model using combinations of autoencoders and classifiers to predict faults in building equipment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If robust historical data with multiple instances of different types of fault events is used for fault detection, then fault detection accuracy is improved, but data availability deteriorates
Solution Approach 1:
The system performs preliminary actions by pre-processing training data before fault detection is needed. This includes calculating synthetic features based on chiller fault rules, resampling data to achieve constant sample rates, and interpolating missing values. These preliminary data preparation steps enable the model to achieve high fault detection accuracy even when actual historical fault data is limited or unavailable.
2Adaptability or versatility
If modular model construction with multiple autoencoder and classifier combinations is used, then model adaptability is improved, but device complexity deteriorates
Solution Approach 1:
The fault prediction model is segmented into modular components: multiple autoencoder options (principle component analysis, long short-term memory network, feedforward network, convolutional network) and multiple classifier options (Gaussian classifier, feedforward model, XGBoost model). This segmentation allows the system to adapt to different data characteristics and fault types by selecting appropriate combinations, while maintaining manageable complexity through standardized module interfaces.
Solution Approach 2:
The modular architecture provides universality by enabling the same framework to handle various fault detection scenarios through different autoencoder-classifier combinations. The system can universally apply the selected model组合 to different building equipment types and fault conditions, making the solution versatile across multiple applications without requiring completely separate systems for each scenario.
Data Source
AI summary
A method includes obtaining a fault prediction model for building equipment, predicting, with the fault prediction model, both (i) whether a fault will occur during a first prediction bin and (ii) whether a fault will occur during a second prediction bin, performing a first mitigating action for the building equipment if the fault is predicted to occur during the first prediction bin, and performing a second mitigating action for the building equipment if the fault is predicted to occur during the second prediction bin.