Machine Tool Process Estimation from Power Time-Series Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating the process of machine tools that perform complicated work, such as NC lathes, are inadequate as they rely on total power consumption, which is not necessarily proportional to the work content, making it difficult to specify processes accurately.
Innovation Solution
A management device that acquires and analyzes time-series power consumption data to divide it into sub-time series, clusters them using unsupervised learning, and associates these clusters with specific processes through teaching data, enabling accurate process estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If total power consumption is used to estimate the process, then the method is simple to implement, but the estimation accuracy deteriorates for complicated work
Solution Approach 1:
The patent segments the total power consumption time series into multiple sub-time series, each representing different operational characteristics of the machine tool. By dividing the power consumption data into distinct segments corresponding to different processes (e.g., idle, cutting, feeding), the system achieves accurate process identification without requiring complex measurement systems, thus resolving the contradiction between implementation simplicity and estimation accuracy.
2Productivity
If the machine performs complicated cutting work, then the work content increases, but the proportionality between power consumption and work content deteriorates
Solution Approach 1:
The patent applies local quality by analyzing specific segments of power consumption characteristics rather than treating the total power consumption as a uniform measure. Different portions of the power consumption time series are evaluated with different weights and criteria based on their local characteristics (e.g., peak power during cutting, baseline power during idle), enabling accurate process estimation even when the overall proportionality between total power and work content breaks down during complicated operations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A time series acquisition unit acquires a time series related to power consumption of a machine for a certain time. A classification unit classifies the time series into any one of a plurality of clusters. A process estimation unit estimates a process executed by the machine, based on relationship information indicating a relationship between the plurality of clusters and the process of the machine, and the cluster into which the time series is classified.