Numerical Controller Abnormality Detection via Neighborhood Method
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional numerical controllers face challenges in detecting and quantifying abnormalities, such as spindle collisions, due to difficulties in setting appropriate threshold values and distinguishing between different operating conditions, which hinders effective countermeasure implementation.
Innovation Solution
A numerical controller that uses a neighborhood method to detect abnormalities by collecting sampling values during normal machining and operation, computing abnormality degrees based on the distance between these values, and presenting countermeasures associated with the calculated degree.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional threshold determination methods are used to detect abnormalities, then abnormality detection capability is provided, but it is difficult to distinguish between different operating conditions (heavy-duty cutting vs. spindle collision, high-speed fast-forward vs. spindle collision) and appropriate threshold values are hard to set
Solution Approach 1:
The system performs preliminary data collection during normal operation to build a database of normal sampling values before actual abnormality detection begins. This preliminary action creates a reference dataset that enables the neighborhood method to function without requiring manual threshold setting, thereby resolving the contradiction between reliable detection and ease of operation.
Solution Approach 2:
The system uses its own normal operation data to automatically establish detection criteria through the neighborhood method. By serving itself with internally collected data rather than requiring external threshold configuration, the system eliminates the operational burden of threshold setting while maintaining reliable abnormality detection.
2Ease of operation
If traditional neighborhood method is used to detect abnormalities, then occurrence of abnormality can be detected without pre-defined thresholds, but the degree of abnormality cannot be quantified
Solution Approach 1:
The invention transitions from binary abnormality detection (abnormal/normal) to a multi-level assessment by introducing the concept of abnormality degree. This dimensional extension allows the system to not only detect whether an abnormality occurs but also to quantify its severity, thereby resolving the contradiction between operational simplicity and information completeness.
3Measurement precision
If traditional methods require accumulation of abnormality data through trial runs, then sufficient data for threshold setting can be obtained, but large amount of trial run time is required
Solution Approach 1:
The system collects normal operation data in advance during regular machining operations, storing this data for later use in the neighborhood method. This preliminary data collection eliminates the need for separate trial runs dedicated to abnormality data accumulation, thereby resolving the contradiction between measurement precision and time loss.
Data Source
AI summary
A numerical controller that detects occurrence of an abnormality according to a neighborhood method includes a sampling value acquisition unit configured to collect sampling values indicative of a state of a machine or environment, wherein the sampling values are collected during normal machining and during operation; a learning unit configured to generate a set of the sampling values during the normal machining; and an abnormality degree determination unit configured to compute an abnormality degree on the basis of a distance between the sampling value during the operation and the set of the sampling values during the normal machining.