Machining Step Anomaly Detection Using Adaptive Average Patterns

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional machining load monitoring methods struggle to accurately detect anomalies in machining processes due to variations in tool and work piece states, especially when the same machining step is repeated on multiple work pieces, and fail to account for tool wear and other conditions.

Innovation Solution

An anomaly detection device and method that collect and record machining execution information at regular intervals, select a subset of data suitable for calculating an average pattern, and compare real-time data with this pattern to detect anomalies, taking into account cumulative tool usage and other conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If reference data is determined by one round of trial cutting, then the monitoring method is simple, but the accuracy of anomaly detection deteriorates due to variations in tool and work piece

Engineering Contradiction:

Improvesimplicity of monitoring methodVSAvoidaccuracy of anomaly detection

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The system performs preliminary trial cutting multiple times before actual machining to collect sampling data. This preliminary action allows the system to establish a baseline of normal machining variations before monitoring begins, enabling accurate anomaly detection during actual machining operations.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The reference data is dynamically updated by combining sampling data from multiple trial cutting rounds. Instead of using static reference data from a single trial, the system continuously refines the reference data to adapt to variations in tool wear, work piece material, and other machining conditions.

Inventive Principle:

Principle #15Dynamics

2Measurement precision

If reference data is determined from multiple trial cuttings, then the accuracy of reference data improves, but the complexity of the monitoring method increases

Engineering Contradiction:

Improveaccuracy of reference dataVSAvoidcomplexity of monitoring method

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs a predetermined number of trial cutting rounds (excessive action) to ensure sufficient sampling data is collected. This approach guarantees that the reference data is statistically reliable without requiring complex adaptive algorithms to determine when enough data has been collected.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The system creates a simplified mathematical model (average value and dispersion) that copies the essential characteristics of multiple trial cutting results. This allows complex multi-round trial data to be represented by simple statistical parameters that are easy to store and compare during monitoring.

Inventive Principle:

Principle #26Copying

3Ease of operation

If the same reference data is used for all work pieces, then the monitoring method is consistent, but the reliability of anomaly detection deteriorates when tool wear occurs

Engineering Contradiction:

Improveconsistency of monitoring methodVSAvoidreliability of anomaly detection

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The system continuously monitors machining load during actual cutting and compares it against the reference data. When anomalies are detected, the system can trigger feedback mechanisms such as alerting operators to tool wear conditions or automatically adjusting machining parameters, thereby maintaining reliable detection despite tool degradation.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

By establishing reference data from multiple trial cuttings before production begins, the system creates a baseline that accounts for initial tool conditions. This preliminary characterization of normal variations enables the system to reliably detect deviations caused by tool wear during extended machining operations.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11226613B2Anomaly detection device, anomaly detection server and anomaly detection method

Publication Date: 2022.01.18 FANUC LTD

AI summary

An anomaly detection device includes: a machining state collection unit which collects machining execution information at a predetermined time interval; a machining execution information recording unit which records the collected machining execution information in a storage unit; a selection unit which selects, from a set of a plurality of pieces of machining execution information recorded by executing a machining command a plurality of times, a subset of the machining execution information in order to calculate an average pattern according to a machining step which is an analysis target; an average pattern calculation unit which calculates the average pattern corresponding to the machining step of the analysis target based on the subset; and an anomaly detection unit which compares the machining execution information in the machining step of the analysis target with the average pattern to detect whether or not an anomaly occurs in the machining step of the analysis target.