Contact Us
INF-TPC
A real-time tool trace anomaly detection system based on unsupervised learning

Overview

INF-TPC leverages FDC data with deep learning and statistical analysis to enable fully automated anomaly detection. This system proactively identifies tool anomalies during production to prevent wafer loss and enhance operational efficiency.

Advantages

Ease of use and low-maintenance design
The system supports one-click intelligent grouping, model training, and service deployment without manual intervention.
Unsupervised anomaly detection without manual labeling
By learning data characteristics autonomously, unsupervised anomaly detection eliminates the dependency on massive labeled samples and lowers data processing costs.
Multi-sensor modeling for minimal model footprint and resource usage
With centralized management and intelligent analysis of multi-sensor data, the system models diverse tool sensors to cut resource consumption.
Support detection for multiple anomaly types
The advanced anomaly detection method identifies multiple anomaly types in semiconductor manufacturing, including interval anomalies, PM anomalies, and amplitude anomalies.With an accuracy rate exceeding 95% across diverse scenarios, it provides manufacturers with comprehensive fault diagnostics and robust quality assurance.
Diversified optimization strategies with high accuracy
The system incorporates powerful data augmentation methods and built-in advanced deep learning models to achieve high identification accuracy. It also supports online data reflux to enable continuous model optimization.

Follow Us

Copyright © 2026 Semitronix Corporation  All Rights Reserved.     User Agreement Privacy Policy 浙ICP备2021003814号-1
XML 地图