Reducer Wear Prediction Using Grease Iron Powder Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing maintenance systems for reducers in multi-axis robots lack efficient methods to predict the optimal time for maintenance based on the concentration of iron powder in grease, leading to potential mechanical failures and downtime due to unpredictable wear and tear.
Innovation Solution
A control system that utilizes machine learning to build a concentration estimation model correlating rotational speed, torque, and iron powder concentration in the reducer's grease, allowing for the derivation of a recommended maintenance time by analyzing the time transition of these parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional maintenance methods are used without machine learning prediction, then maintenance timing is determined by fixed schedules, but this leads to unpredictable wear and tear and potential mechanical failures
Solution Approach 1:
The system performs preliminary analysis by continuously monitoring grease condition parameters (iron powder concentration, viscosity, temperature) and using machine learning models to predict future degradation trends. This allows maintenance to be scheduled before actual failure occurs, transforming reactive maintenance into proactive maintenance and preventing unexpected downtime.
Solution Approach 2:
The system implements continuous feedback loops where sensor data from the reducer (temperature, vibration, grease conditions) is constantly fed into machine learning models. The models update predictions based on this feedback, allowing dynamic adjustment of maintenance timing. This closed-loop system improves reliability by adapting to actual wear patterns rather than following fixed schedules.
2Measurement precision
If machine learning models are trained with comprehensive data including iron powder concentration, then prediction accuracy improves, but data collection and processing complexity increases
Solution Approach 1:
The system uses a multi-functional sensor platform that simultaneously measures multiple parameters (temperature, vibration, iron powder concentration, viscosity) using a single integrated system. This universal approach improves prediction accuracy through comprehensive data collection while avoiding the complexity of multiple separate measurement systems. The machine learning model processes all these parameters together to generate unified maintenance predictions.
3Duration of action of stationary object
If maintenance is performed based on predicted iron powder concentration thresholds, then component lifespan is extended, but frequent monitoring and analysis are required
Solution Approach 1:
The machine learning system operates autonomously to monitor grease conditions, predict wear trends, and generate maintenance recommendations without requiring constant human intervention. The system self-manages the complex data analysis and prediction tasks, extending component lifespan through intelligent monitoring while minimizing the productivity impact by automating rather than manualizing the monitoring process.
Data Source
AI summary
A control system for an actuator includes a non-transitory computer readable medium storing a database in which first information, second information and third information are stored and correlated with each other, and processing circuitry that performs machine learning based on the first information, the second information and the third information stored in the database. The first information is associated with a rotational speed of a reducer in the actuator, the second information is associated with a torque acting on the reducer, the third information indicates a concentration of iron powder in grease in the reducer, and the machine learning builds a concentration estimation model indicating a relationship between the first information, the second information and the third information.