Machine Learning Power Failure Prediction for Safe Machine Retraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing production systems in unstable power conditions face issues such as sudden shutdowns, machine damage, and defective workpieces due to power instability, with previous solutions requiring uninterruptible power supply systems that are costly and not applicable to all environments.
Innovation Solution
A predicting device using machine learning to forecast power instability by correlating measurement data with failure notifications, which includes a state observing section, judgment data acquiring section, and learning section, and a control device that receives failure predictions to retract working machines to a safe state, deployable in cloud, fog, and edge computing environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If an uninterruptible power supply system is used to protect against power instability, then reliability is improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent introduces a predicting device as an intermediary component that monitors power supply states and predicts failures before they occur. This mediator enables proactive protection without requiring a full uninterruptible power supply system, thus improving reliability while avoiding the complexity and cost of complete UPS infrastructure.
Solution Approach 2:
The system performs preliminary action by predicting power supply failures before they actually occur. The predicting device analyzes current power supply states and forecasts potential failures, allowing the production system to take preventive measures (such as saving data or stopping operations) before the power failure happens, thereby avoiding the need for reactive UPS systems.
2Reliability
If an uninterruptible power supply system is deployed to all machines, then reliability is improved, but cost increases enormously
Solution Approach 1:
The predicting device serves as a cost-effective intermediary that provides failure prediction capabilities to multiple machines simultaneously. Instead of equipping each machine with expensive UPS systems, a single predicting device can monitor and predict failures across the entire production system, dramatically reducing overall costs while maintaining reliability.
Solution Approach 2:
The predicting device is designed with universal applicability, serving multiple machines and production systems with a single unit. This multi-functional approach allows one predicting device to protect numerous machines from power-related failures, eliminating the need for individual UPS systems at each machine and significantly reducing the total quantity of protective equipment needed.
3Reliability
If traditional power protection methods are used, then reliability is improved, but adaptability to different environments deteriorates
Solution Approach 1:
The predicting device acts as an adaptable intermediary that can be integrated into various production environments without requiring complete UPS infrastructure. It provides failure prediction capabilities that work across different machine types and power supply configurations, making the solution versatile and environmentally adaptable while maintaining protective reliability.
Solution Approach 2:
The system adapts to different environments by monitoring and analyzing various power supply parameters specific to each environment. The predicting device adjusts its prediction algorithms based on local power conditions, machine types, and operational patterns, enabling reliable failure prediction across diverse production environments without requiring environment-specific UPS systems.
Data Source
AI summary
A predicting device of a production system includes a machine learning device that learns the relationship between a change in measurement data indicating the state of a power supply and a failure which occurs in the power supply. The machine learning device learns the measurement data including at least a measurement value of electric power consumption in a factory by correlating a state variable indicating the current state of an environment with judgment data indicating a failure notification indicating the occurrence of a failure. A control device of the production system includes a receiving section that receives a prediction notification of a failure which occurs in the power supply, the failure being predicted based on a change in the measurement data indicating the state of the power supply, and a retracting operation control section that makes a working machine transition to a safely retracted state when receiving the prediction notification.