Pump Maintenance Evaluation Using ML-Based Failure Risk Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately assess the wear state and failure risk of industrial machinery, particularly pumps, due to the complexity of various influencing factors and external influences, making it difficult to implement timely maintenance measures.
Innovation Solution
A method involving the determination of relevant influencing factors, use of a damage relevance model with a machine learning algorithm, and a database for training data to estimate failure probability, allowing for risk-based maintenance recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional monitoring methods are used to assess machine wear state, then the system is simple to operate, but the measurement precision and reliability of failure risk assessment deteriorates due to inability to capture complex influencing factors
Solution Approach 1:
The system segments the complex wear assessment problem into multiple independent influencing factors (operational parameters, environmental conditions, maintenance history). Each factor is evaluated separately by the machine learning model, which then integrates them to produce an overall wear state assessment. This segmentation allows the system to handle complexity while maintaining measurement precision.
Solution Approach 2:
The patent introduces an evaluation unit as an intermediary between the machine components and the monitoring system. This evaluation unit contains the machine learning model that processes multiple influencing factors and translates them into meaningful wear state assessments and failure risk predictions, bridging the gap between raw data and actionable insights.
2Reliability
If comprehensive influencing factors are monitored to improve failure risk prediction, then the reliability of assessment improves, but the device complexity and data processing requirements increase
Solution Approach 1:
The system performs preliminary actions by continuously collecting and storing influencing factor data in a database during normal operation. The machine learning model is pre-trained with historical data to establish relationships between factors and wear patterns. This preliminary preparation enables reliable failure risk prediction without requiring complex real-time processing of all factors simultaneously.
Solution Approach 2:
The system implements feedback mechanisms where the machine learning model continuously learns from actual wear measurements and failure data. The model updates its predictions based on feedback from maintenance outcomes and actual wear states, improving reliability over time while adapting to the specific characteristics of the monitored machine without increasing structural complexity.
3Productivity
If manual inspection data is used for training, then the training process is simple to implement, but the productivity and coverage of data collection deteriorates due to limited human capacity
Solution Approach 1:
The system achieves universality by designing a multi-functional data collection mechanism that automatically gathers multiple types of data simultaneously (operational parameters, environmental conditions, maintenance records) through integrated sensors and data interfaces. This eliminates the need for separate manual collection processes while maintaining ease of operation through automated routines.
Solution Approach 2:
The machine learning model performs self-service by automatically selecting relevant features from the collected data, performing feature engineering, and optimizing its own parameters during training. This reduces the manual effort required for data preparation and model development while significantly increasing data collection productivity and coverage.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for evaluating a necessary maintenance measure for a machine, more particularly a pump, comprising the method steps: determining one or more influencing variables relevant to the wear of or damage to a machine component; feeding the influencing variables to an estimation model for determining a failure risk and/or the failure probability of at least one machine component and/or of the machine; generating a recommendation for a maintenance measure on the basis of the one or more determined failure risks / failure probabilities.