BAS Alarm Flood Sequence Mining for Root-Cause Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Building automation systems (BAS) generate high volumes of alarms, leading to 'alarm floods' that overwhelm operators and hinder maintenance efforts, with existing solutions struggling to effectively manage multivariate alarms caused by system abnormalities and interactions.
Innovation Solution
An alarm flood management system using machine learning and pattern mining to predict and suppress alarm floods, extracting and aligning alarm log data, clustering sequences, and identifying patterns to provide root-cause analysis and suppress unnecessary alarms, integrated with ANSI/ASHRAE standard 135-compliant systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the BAS generates alarms according to BACnet standard to monitor system performance, then system monitoring capability is improved, but alarm flood occurs overwhelming operators
Solution Approach 1:
The patent introduces an alarm management system as an intermediary between the BAS and operators. This mediator processes, analyzes, and filters alarms using machine learning models before presenting them to operators, thereby maintaining comprehensive monitoring while reducing operator burden from alarm floods
Solution Approach 2:
The system performs preliminary analysis and pattern recognition on alarm sequences before they reach operators. By pre-processing alarm data and identifying potential issues through machine learning models, the system prepares information in advance, allowing operators to receive only relevant, processed alarm information rather than raw alarm floods
2Ease of operation
If traditional alarm reduction techniques (delay timers, deadbands) are applied to univariate nuisance alarms, then chattering alarms are reduced, but multivariate alarm sequences remain unmanaged
Solution Approach 1:
The patent develops a universal alarm management system that handles both univariate and multivariate alarms through a single machine learning-based framework. The system can process simple chattering alarms as well as complex multivariate alarm sequences, providing adaptable handling for different alarm types without requiring separate management approaches
3Reliability
If operators respond to all BAS alarms individually, then complete alarm coverage is achieved, but maintenance efficiency decreases due to alarm flood
Solution Approach 1:
The patent merges multiple related alarms into single consolidated alarm presentations. By identifying alarm sequences and patterns, the system combines information from multiple alarms into unified maintenance tasks, ensuring complete coverage of all alarm events while presenting reduced, consolidated information to operators to improve maintenance efficiency
Data Source
AI summary
The present disclosure provides an alarm flood management system including an extracting device, an aligning device, a clustering device, a pattern mining device, a controller, and a segmenting device. The extracting device extracts alarm log data into alarm flood sequences. The aligning device aligns the alarm flood sequences for obtaining the relationship between the alarm flood sequences. The clustering device clusters the alarm flood sequences according to the relationship between the alarm flood sequences for obtaining groups of alarm flood sequences. The pattern mining device performs pattern mining on the groups of alarm flood sequences for obtaining patterns of the alarm flood sequences. The controller utilizes the patterns of the alarm flood sequences for providing a root-cause/strategy. The controller suppresses the incoming alarm data by the predicted incoming alarm data and updates the root-cause/strategy according to the predicted incoming alarm data.