Production Process State Recognition Using Power Consumption Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current scalable condition monitoring systems in automated manufacturing face challenges in real-time fault recognition and process state monitoring, particularly in complex machining lines where minor abnormalities can lead to significant production disruptions and increased costs due to the inability to detect faults promptly and efficiently.
Innovation Solution
A scalable real-time state recognition method using online machine learning for pattern recognition in streaming sensor data, which generates power, energy, and time-related indicators for intelligent monitoring and visualization, allowing for efficient fault detection and predictive maintenance with minimal computational power.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional condition monitoring systems are used in complex machining lines, then fault detection capability is limited, but computational resources and system complexity increase
Solution Approach 1:
The patent extracts only the essential features for fault detection by monitoring power consumption patterns and acoustic emissions, rather than implementing comprehensive multi-sensor monitoring systems. This extraction approach achieves reliable fault detection while minimizing system complexity by focusing on the most informative signals.
Solution Approach 2:
The monitoring system is designed to detect multiple types of faults (tool wear, machine anomalies, process deviations) using a unified approach based on power consumption analysis and acoustic emission monitoring. This multi-functional capability achieves comprehensive fault detection without requiring separate specialized systems for each fault type.
2Loss of time
If real-time monitoring of all machining processes is implemented, then fault detection timeliness improves, but computational power requirements increase
Solution Approach 1:
The system implements real-time monitoring selectively by continuously analyzing power consumption patterns and acoustic emissions only when anomalies are detected or during critical machining operations. This partial monitoring approach achieves timely fault detection without requiring continuous full-scale computational analysis of all processes.
Solution Approach 2:
The system performs preliminary analysis of power consumption patterns to establish baseline signatures of normal operation before actual machining begins. This preliminary characterization enables faster real-time detection by comparing live data against pre-established patterns, reducing computational requirements during critical monitoring phases.
3Productivity
If comprehensive process monitoring is implemented in automated machining lines, then production disruption prevention improves, but system cost and complexity increase
Solution Approach 1:
The system uses power consumption patterns as an intermediary indicator that reflects the actual mechanical state of machining processes. By monitoring this intermediate electrical parameter rather than directly measuring mechanical variables, the system achieves comprehensive process monitoring with simpler, more cost-effective sensors and analysis methods.
Solution Approach 2:
The monitoring system leverages existing electrical infrastructure and power measurement capabilities already present in automated machining lines. By utilizing the electrical power system that already serves the machines for their primary function, the system achieves production monitoring without requiring separate dedicated sensing infrastructure, thereby reducing overall system complexity.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The invention relates to a method for the scalable real-time state recognition of processes and/or sub-processes during production with electrically driven production plants and for the generation of corresponding key figures, comprising the generation of a time sequence of measurement data M by determining a power consumption of at least one production device with a predetermined sampling rate; Transmission of the sequence of measured data to a computing unit via streaming protocols, analysis of current properties of M by means of a current measuring point of M and a data set relating to a past measuring point of M and creation of real-time key figures which are representative for the current properties of M. A recognition and/or identification of current states of production processes and/or sub-processes on the basis of real-time key figures and preferably creation of a second set of associated real-time-key figures is implemented.