Adaptive Preventive Maintenance via Cost Function Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for preventive maintenance in computing systems face challenges in effectively discriminating between healthy and fully degraded states, leading to inefficiencies in scheduling maintenance, especially in the 'gray area' where differences in prognostic parameter distributions are subtle.
Innovation Solution
The implementation of an adaptive preventive maintenance system that uses a system cost function and a Discrete Parameter Markov chain model to continuously monitor and optimize maintenance schedules based on observed data, minimizing expected maintenance costs and accounting for system degradation states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional preventive maintenance methods are used, then maintenance schedules can be established, but they fail to effectively discriminate between healthy and fully degraded states in the 'gray area' where prognostic parameter distributions are subtle
Solution Approach 1:
The patent transforms the maintenance scheduling problem from a time-based approach to a state-based approach by changing the parameter from time to prognostic parameter values. It partitions the continuous range of prognostic parameter values into discrete operating states, enabling precise discrimination between healthy and degraded states even in the gray area where traditional methods fail.
Solution Approach 2:
The patent segments the continuous range of prognostic parameter values into multiple discrete operating states (S1, S2, ..., Sm), where each state represents a specific degradation level. This segmentation allows for fine-grained discrimination between different system conditions, particularly in the gray area between healthy and fully degraded states.
2Reliability
If preventive maintenance is performed too early, then system reliability is maintained, but maintenance costs increase due to unnecessary interventions
Solution Approach 1:
The patent implements a feedback mechanism where the system continuously monitors prognostic parameter values, compares them against threshold values defining operating states, and adjusts maintenance scheduling decisions accordingly. This feedback loop enables dynamic optimization of maintenance timing based on actual system condition, avoiding both premature and delayed maintenance.
Solution Approach 2:
The patent transitions from static, predetermined maintenance schedules to dynamic, adaptive scheduling based on real-time system state. The optimal maintenance schedule is determined by the current operating state and can be continuously adjusted as the system evolves, allowing maintenance to be performed at the precise moment when it becomes necessary.
3Loss of energy
If preventive maintenance is performed too late, then maintenance costs are reduced, but system failure risk increases
Solution Approach 1:
The patent enables preliminary maintenance action by predicting future system states based on current prognostic parameter trends. By identifying when the system is approaching critical degradation thresholds, maintenance can be scheduled in advance before failure occurs, balancing cost reduction with reliability maintenance.
4Productivity
If individualized maintenance schedules are created for each system, then maintenance optimization is improved, but system complexity and data requirements increase
Solution Approach 1:
The patent reduces system complexity by transforming individualized system behavior into a standardized state classification problem. Instead of dealing with complex continuous prognostic parameter values, the system maps all systems onto a common framework of discrete operating states, enabling simplified decision-making while maintaining individualization through state-specific maintenance policies.
Data Source
AI summary
Systems, methods, and other embodiments associated with adaptively determining a preventive maintenance schedule based on historical system operation are described. The prognostic parameter values are continuously partitioned into a number of operating states based on observed maintenance costs associated with the prognostic parameter values. The operating states range from absolutely healthy, one or more degrees of degradation, to fully degraded. A system cost function is used as the discriminant function. The system cost function is an expected maintenance cost when a given preventive maintenance (PM) schedule is adopted. The system cost function calculates the expected cost based on the observed cost of operation in each of the operating states and a probability of the computing system being in each of the operating states as determined by the PM schedule. The PM schedule that minimizes the cost function is determined to be the optimal PM schedule.