Machine Fault Probability Modeling from Sensor and Maintenance Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machinery fault prediction methods rely on local views and assumptions, lacking a comprehensive approach to determine fault probabilities and prioritize preventative maintenance tasks effectively across complex systems.
Innovation Solution
A method utilizing machine learning models, such as random forest, kernel random forest, artificial neural networks, or Bayesian networks, to analyze sensor logs and maintenance data, generating fault probabilities and prioritizing maintenance tasks based on statistical and log metrics to reduce future faults.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If local views and assumptions are used for fault prediction, then the monitoring process is simple, but the fault prediction accuracy and comprehensiveness deteriorate
Solution Approach 1:
The patent segments the fault prediction process into multiple independent components: sensor data collection module, maintenance log analysis module, statistical metric calculation module, and risk model evaluation module. Each module processes specific aspects of machine data independently, then integrates results to provide comprehensive fault predictions, thereby maintaining process simplicity while improving accuracy.
Solution Approach 2:
The patent transitions from traditional single-dimension sensor monitoring to multi-dimensional analysis by incorporating temporal dimensions (time-series sensor data), maintenance history dimensions (maintenance logs), and statistical dimensions (calculated metrics). This dimensional expansion enables more accurate fault prediction without proportionally increasing system complexity.
2Reliability
If comprehensive data analysis is performed to determine fault probabilities, then the accuracy of fault prediction improves, but the computational complexity and data processing requirements increase
Solution Approach 1:
The patent performs preliminary actions by pre-calculating statistical metrics from sensor data and pre-processing maintenance logs before fault prediction is needed. Historical data is aggregated and standardized in advance, creating ready-to-use datasets that reduce computational burden during actual fault prediction events while maintaining high reliability.
Solution Approach 2:
The patent introduces statistical metrics as intermediary variables between raw sensor data and fault probability predictions. These metrics serve as simplified representations of complex sensor patterns, mediating the analysis process by translating raw data into meaningful indicators that feed into the risk model, thereby reducing overall computational complexity.
3Adaptability or versatility
If historical data from multiple machines is utilized, then the predictive capability improves, but the data management and processing complexity increases
Solution Approach 1:
The patent creates a universal data processing framework that handles sensor logs and maintenance logs from multiple machines using the same statistical metrics and risk models. This multi-functional system processes diverse data sources through standardized procedures, improving predictive capability across different machines while avoiding the need for machine-specific processing complexity.
Data Source
AI summary
Systems, methods, non-transitory computer readable media can be configured to access a plurality of sensor logs corresponding to a first machine, each sensor log spanning at least a first period; access first computer readable logs corresponding to the first machine, each computer readable log spanning at least the first period, the computer readable logs comprising a maintenance log comprising a plurality of maintenance task objects, each maintenance task object comprising a time and a maintenance task type; determine a set of statistical metrics derived from the sensor logs; determine a set of log metrics derived from the computer readable logs; and determine, using a risk model that receives the statistical metrics and log metrics as inputs, fault probabilities or risk scores indicative of one or more fault types occurring in the first machine within a second period.