Predictive Maintenance System Using Sensory Data Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current maintenance strategies for devices, such as printers, often result in either unexpected costs or unnecessary expenses due to reactive or preventive maintenance approaches, which can lead to increased downtime and inefficient resource allocation.
Innovation Solution
Implementing predictive maintenance driven by data analytics and machine learning techniques to identify when devices are likely to fail, allowing for proactive scheduling and reducing maintenance costs by monitoring sensory inputs and storing data in a centralized database to train supervised machine learning models for accurate intervention predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If reactive maintenance is used, then maintenance costs are incurred only when needed, but device downtime increases and productivity decreases
Solution Approach 1:
The system performs preliminary actions by continuously monitoring device sensory inputs and using machine learning models to predict failures before they occur. This allows maintenance to be scheduled in advance, preventing unexpected downtime while avoiding unnecessary maintenance interventions.
Solution Approach 2:
The system implements feedback by continuously collecting sensory input data from devices, analyzing it through machine learning models, and using the predictions to optimize maintenance scheduling. This closed-loop feedback enables dynamic adjustment of maintenance strategies based on actual device conditions.
2Reliability
If preventive maintenance is used, then device reliability is improved, but unnecessary maintenance costs increase and resource allocation becomes inefficient
Solution Approach 1:
The system transitions from static preventive maintenance schedules to dynamic maintenance planning based on real-time device conditions. Machine learning models continuously analyze sensory inputs to adjust maintenance timing, ensuring interventions are performed only when actually needed rather than on fixed schedules.
Solution Approach 2:
The system changes the parameter of maintenance timing from fixed calendar-based intervals to condition-based intervals determined by machine learning predictions. This allows maintenance to be aligned with actual device degradation patterns rather than arbitrary time schedules.
3Measurement precision
If more sensory monitoring is implemented, then prediction accuracy improves, but device complexity and implementation costs increase
Solution Approach 1:
The system uses universal machine learning models that can process multiple types of sensory inputs from various devices through a centralized platform. This multi-functional approach allows the same prediction engine to handle different device types and sensor configurations, reducing overall system complexity.
Solution Approach 2:
The patent introduces a centralized database and machine learning platform as intermediaries between device sensors and maintenance decision-making. This intermediary layer consolidates data processing and analysis, preventing complexity from propagating to individual devices while maintaining high prediction accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Examples include a non-transitory machine-readable storage medium having stored thereon machine-readable instructions executable to cause a processing resource to monitor sensory inputs related to a device, monitor a first maintenance intervention related to the device, store data relating to the monitored sensory inputs and the first maintenance intervention in a centralized database, and predict a second maintenance intervention based on the data stored in the centralized database.