Using Big Data Analytics for Equipment Preventive Maintenance

Introduction

Today’s semiconductor chip processing involves hundreds of discrete steps and extremely expensive machines and equipment.Precision is of paramount importance. Any non-precise movement caused by the failure of a component—for example, a waferslipping from robotic grip—can result in significant financial loss, and even worse, unplanned downtime for maintenance.

Therefore, many hi-tech companies have embarked on implementing Industry 4.0 manufacturing by deploying sensors aroundtheir costly machines and equipment to collect, transmit, aggregate and analyze operating data under an Internet of Things (IoT)architecture. They enable alert and alarm functions and implement preventive maintenance through analysis of both real time and historical data.

The preventive maintenance is a cloud service that analyzes aggregated sensor data to discover patterns so as to alert the property owner to repair or replace problematic machine components in advance of actual failures.

By developing and implementing a preventive maintenance model, the hi-tech companies safeguard operations of their facilities, protect their costly components andsemi-finished products, avoid unnecessary downtime and reduce maintenance costs; their ultimate goals are to improve both throughput and quality.

Our customer in this case was one of famous chip manufacturers, who wished to establish an IoT-enabled intelligent monitoring system overseeing all machines and equipment in their facilities, with the intention of implementing preventive maintenance once the aggregation of data was sufficient for big data analysis.

They started by adding sensors to the robotic arms used to move wafers to lithography equipment in the yellow-lighted clean rooms.The robotic arms were prioritized because the mechanics of the robotic arms were beyond the range of the lithography equipment’s built-in monitoring capabilities, and had been most susceptible to inaccuracy and failure; so this was the first priority for remediation.Sensors were deployed to collect motor vibration and temperature data .

The investment on the deployment was easily justified, considering the fact that a preemptive replacement of a robotic arm washer costs only tens of US dollars if the problem is identified early in its development, compared with the hundreds of thousands of USdollars required to replace a failed servo motor.

Application Requirements

The customer was first required to deploy sensors in the robotic equipment to collect data on motor temperatures and vibrations.They also needed to install an edge intelligence server near the site to collect, integrate, and preprocess raw data in order to help balance cloud analytics loads as well as to transform that data into human-readable information.

To implement preventive maintenance, the customer needed to accumulate the acquired data in a data pool together with otherdata sources from the facility. A reliable backend platform was required for integrating cross-equipment data and establishing interconnectivity with the company’s legacy Hadoop cloud database used for big data analysis.

When big data analysis sets rules for preventive maintenance (for instance, under what conditions alerts or alarms should betriggered? or which component should be replaced?), a logic flow edit-and-control platform is needed to return the analysis outcomes back to field-level controllers for automatic implementation.

System Solution

Advantech provided its Edge Intelligence Server (EIS), a solution-ready platform that delivers IoT connectivity, cross-equipment data integration and database interconnection, edge computing, preconfigured cloud service, as well as vertical application software support, allowing customers to develop fast time-to-market solutions.

The Advantech EIS solution has Advantech’s most powerful WISE-PaaS software packages built in, including remote monitoring software WISE-PaaS/RMM and smart SCADA software WebAccess/SCADA, which provides over 200 drivers for easily connecting with almost all types of PLCs, machines and equipment using different industrial communication protocols (mainly OPC DA andModBus in this case). This helped save much development time in establishing asset connectivity.

For equipment using proprietary protocols developed by specific equipment vendors, the customer can utilize WISE-Agent software contained in the EIS solution to develop protocol conversion modules with provided SDK and sample code. The WISE-Agent is deployed at IoT gateways mainly to convert sensor and equipment data into IoT-standard MQTT format and relay converted data tothe cloud databases or the backend server.

To establish interconnectivity among databases and cloud services, the EIS solution offers WISE-PaaS platform with default supportfor No-SQL MongoDB and a rich amount of RESTful APIs which can be used to integrate with the company’s legacy Hadoop databasefor big data analysis. The Advantech EIS solution also includes the powerful Node-RED logic flow control tool, which allows the user toedit preventive maintenance workflows and remotely implant the resulting module into an edge server (controller) via the WISE-PaaSplatform; this enables the EIS to carry out data preprocessing and preventive maintenance tasks.

Benefits

• Helps reduce equipment maintenance cost, with high Return on Investment

• Provides built-in tools for fast integration of sensor data and applications

• Provides network edge intelligence and logic flow control needed for implementing preventive maintenance

• Helps reduce the system development period with well-rounded hard- and soft-ware integrated solutions plus the helpful presence of Field Application Engineers who assist with system modifications, integration, and tests

• Modulized logic flow control can be exported and transplanted to similar equipment elsewhere so as to save workload anddevelopment time for future projects