Proprietary Technology ＆ Philosophy

AI Smart Monitoring in Rotor Equipment Management

With the rapid development of smart manufacturing, the reliability and stability of equipment operations have become critical pillars of business competitiveness. Particularly for rotor equipment, which is widely used in industries like semiconductors, petrochemicals, and power generation, its operational status directly impacts production efficiency and safety.

However, traditional equipment monitoring methods, such as fixed schedule maintenance or ISO vibration standard monitoring, often suffer from limited data and insufficient early anomaly detection, leading to excessive maintenance or undetected faults. To address these challenges, GoodTech Technology combines Artificial Intelligence (AI), Predictive Maintenance (PHM), and Reliability-Centered Maintenance (RCM) to develop an industry-leading monitoring solution for rotating equipment.

Application Overview

This AI monitoring model captures 25 characteristic values through wireless sensors and evaluates the equipment’s health status based on ISO standards’ four levels: Excellent, Normal, Warning, and Danger. The core functions of the AI include:

1. Health Status Assessment:
Analyzes the probability of the equipment’s operational status based on characteristic value combinations.

2. Trend Monitoring:
Displays each monitoring result with color-coded lights to help identify abnormal or deteriorating trends.

3. Preventive Suggestions:
Provides maintenance or servicing recommendations to reduce downtime risk when warning or danger states are detected.

What are the standards for rotor quality?

Technical Overview

Currently, this AI-based monitoring model has been pre-trained by GoodTech Technology, using data from 25 characteristic values collected by wireless sensors. The AI prediction focuses on the quality status levels of rotating equipment, which are categorized in four levels: Excellent, Normal, Warning, and Danger, following ISO standards.

The trend chart shown in the AI model represents the characteristic values sent by the sensors, which are converted into the “probability” of the equipment’s state. By observing the proportion of colored status blocks, users can determine the current health of the equipment at any given point.

Example of AI Results: As shown in the example above, the AI results suggest that the equipment has a 91.2% probability of being in normal operation.

In the Y-direction vibration values, applying ISO vibration standards showed a warning (orange light), which means maintenance should be scheduled. However, when other vibration characteristic values are added and predicted by the AI model, the equipment is determined to be in a Normal-Excellent state, meaning it is still capable of functioning properly.

Apply ISO vibration regulations and display as warning

ISO Vibration Standard vs. AI Monitoring
The above comparison demonstrates that the ISO vibration standard may be stricter, but it is not necessarily inaccurate. Historically, the equipment monitored by the ISO vibration standard was used mainly in industries like petrochemicals, power generation, and nuclear plants, where equipment is highly valuable, and failure without warning can lead to safety hazards. Therefore, experts and researchers have defined stricter standards based on extensive big data analysis of rotor equipment.

This approach aligns with the goals and principles of predictive maintenance. Comparing both management methods, it’s evident that when ISO vibration limits are exceeded, the AI quality status trend also shows a deterioration, reinforcing the importance of AI optimization.

AI-optimized monitoring

The image above shows another example of AI-optimized monitoring. Comparing the two monitoring models, if damage to a machine cannot be detected via vibration velocity, AI’s optimized monitoring model can identify abnormal states. In this case, the AI detected a Warning-Danger state in the equipment. After a month of maintenance and re-commissioning, even though the vibration values showed no significant changes, the AI model detected the machine in an Excellent state after repairs.

This case clearly demonstrates that as computer capabilities grow, adding more vibration features and calculation mechanisms and appropriately designing AI monitoring models can enhance previously defined vibration standards. This leads to better early anomaly detection and improved predictive capabilities.

PHM and RCM in Rotor Equipment Management

With the rapid development of smart manufacturing, the reliability and stability of equipment operations have become crucial pillars for business competitiveness. Rotor equipment, in particular, is widely used in industries such as semiconductors, petrochemicals, and power generation, where its operational status directly affects production efficiency and safety.

However, traditional equipment monitoring methods such as fixed-schedule maintenance or ISO vibration standard monitoring often suffer from single data sources and limited early anomaly detection, leading to excessive maintenance or undetected faults. To tackle these challenges, GoodTech Technology integrates Prognostics and Health Management (PHM) and Reliability-Centered Maintenance (RCM), creating an industry-leading monitoring solution for rotating equipment.

Integration of PHM (Predictive Health Management) in the AI Model

The core of PHM lies in predicting and managing equipment health. The concept is reflected in the following areas:

1. Health Status Evaluation and Trend Prediction:The AI model uses multi-feature data and trend analysis to predict the equipment’s health status. The trend graph (e.g., color bars) clearly shows the probability of each state, helping to quickly assess health trends. By standardizing the modeling approach, health status models are built for different equipment types to ensure comparability of results. The expansion of data sources (e.g., temperature, pressure, and environmental conditions) improves the accuracy of health status evaluations.

2. Anomaly Detection and Early Warning:The AI model can detect subtle but critical feature changes, providing early warnings for potential failures and avoiding intervention only after complete failure. For example, while ISO vibration standards are strict, AI can further incorporate other data points to detect anomalies that traditional methods miss. The model generates personalized early warnings for each piece of equipment based on historical data, with multi-tiered warning strategies to distinguish between minor and serious anomalies. This helps avoid false alarms and ensures smooth operations. Standard operating procedures (SOPs) for anomaly response, including inspections, diagnostics, and scheduling, are provided.

3. Dynamic Optimization and Learning:PHM emphasizes data accumulation and dynamic model updates. The AI model continues to learn new data based on real-world applications and gradually optimizes prediction accuracy. The platform supports continuous learning, automatically using new data to train and update the model. It allows collaboration with equipment suppliers or users for model validation and optimization.

4. Decision Support:The AI model provides maintenance recommendations based on the health status, such as suggesting scheduled maintenance when the state is in “warning,” rather than immediate shutdown, maximizing equipment usage efficiency and providing decision-making reference for management.

Integration of RCM (Reliability-Centered Maintenance) in the AI Model

RCM focuses on systematic analysis to choose appropriate maintenance strategies that improve equipment reliability. The concept is reflected in the following areas:

1. Function-Oriented Analysis:The AI model monitors rotating equipment based on functional requirements, ensuring mechanical operations meet usage goals (e.g., stable vibration data). It focuses on the health of core equipment functions rather than surface issues of individual components. Key performance indicators (KPIs) are defined for each equipment, such as vibration speed and acceleration, while adapting to different industrial scenarios.

2. Failure Mode and Analysis:The AI model builds failure mode models using characteristic data, identifying the key impacts of failures on production. For example, when vibration anomalies are detected in the Y-direction, AI combines other features for a full analysis, avoiding partial judgments. A shared failure database is created to focus on predicting and analyzing common failure modes.

3. Optimizing Maintenance Strategies:RCM advocates choosing the most appropriate maintenance method (e.g., preventive maintenance, condition-based maintenance, or corrective maintenance). The AI model dynamically determines whether immediate repairs are needed or if regular checks should be scheduled based on health status and trends. Maintenance strategies are established based on health status levels, and the impact of different maintenance modes on production is simulated to find the optimal solution. Maintenance plans are automatically updated based on real-time monitoring data.

4. Cost-Benefit Balance:Through AI’s accurate predictions, excessive maintenance is avoided (lowering maintenance costs), and equipment failures due to lack of monitoring are prevented (reducing production loss), thus lowering overall equipment maintenance costs. Additionally, resource allocation is optimized based on health status and maintenance needs, dynamically distributing repair personnel and parts. Health status monitoring models tailored for different industrial fields (e.g., semiconductors, petrochemicals, and power generation) ensure the best adaptability.