Industrial Big Data: Workshop IoT Data Management

In the article "Re-cognition of Manufacturing Data Management", we divide industrial big data into public resource data, engineering data, management data and IoT data. This article mainly talks about IoT data, and it is also the last article in the data management section. Traditional management systems use people as data collection terminals, use processes to solidify organizational behavior, and use indicators to measure and evaluate the efficiency of processes and organizations. The Internet of Things for industrial enterprises is to connect people and things, systems and things, and use things as data collection terminals, and people or systems to analyze and make decisions. Data analysis and optimization is one of the key technologies of the Internet of Things, and it is also the key point for the Internet of Things to play its value in the future. The Internet of Things has many applications in industry, such as logistics warehousing, production and manufacturing, product operation and maintenance, etc. Here we focus on production and manufacturing and product operation and maintenance.

Section 1 How IoT Data is Organized

The production and manufacturing IoT application of industrial enterprises is generally called workshop IoT or manufacturing IoT. By using RFID sensors, wireless network communication, GPS positioning, voice and video systems and other technologies to link manufacturing plans with information such as "people, machines, materials, methods, and environment" of manufacturing resources, the five major manufacturing resources can be intelligently identified, located, tracked, monitored and managed to meet the management requirements of enterprise command and dispatch, environmental monitoring and other aspects. The five major manufacturing resources are divided into static attributes and dynamic attributes. For example, the static attributes of a machine tool can be divided into management information (equipment code, equipment name, equipment classification, etc.), static parameters (working environment, feed speed, cutting parameters, etc.), and dynamic parameters (machine tool status, lathe completion rate, lathe load rate, maintenance records, etc.). Static attributes are not affected by the production process and are determined before the production process begins. They are constant data in workshop site management, but these data are not always fixed. They can be adjusted by users after the production process ends; dynamic data is data that is always changing, and most workshop IoT data are dynamic data.

Section 2 Management Technology of IoT Data

The workshop Internet of Things is a typical complex information system that involves various aspects of data management, mainly including: data quality control, data fusion and integration, complex event processing, data storage and processing, and security access control.

Data quality control: The data quality of IoT can be measured by three indicators: accuracy, confidence, and integrity. In terms of improving the quality control of RFID and sensor network data, the main methods are to remove over-read and misread data and fill in missed data. Data cleaning usually uses probability statistics and spatiotemporal correlation methods.

Data fusion and integration: The polymorphism of data objects in the IoT data space is manifested in multiple types, heterogeneity, and no unified model. Therefore, on the one hand, it is necessary to build a unified data model for the workshop to express data in a unified way; on the other hand, based on the unified data model, research is conducted on how to map and transform heterogeneous data into a unified data framework; on the other hand, the data sources in the IoT are distributed, autonomous, and independent. In the process of data integration, it is sometimes necessary to automatically discover relevant data sources; on the other hand, the source of the data must be recorded to achieve data traceability; on the other hand, the manufacturing resources in the workshop are constantly changing, and this change will have an impact on the consistency, version, and model update of the data, so the process of data evolution must be recorded.

Complex event processing: In typical IoT applications, the upper-level system is responsible for monitoring the status and behavior of each object and controlling it to respond intelligently and complete corresponding behaviors according to established procedures. The behavior of an object is usually expressed in the form of events.

Security access control: Due to the openness of the Internet of Things, how to protect the privacy of sensor data has become a thorny issue. Because these massive data are easy to obtain, if the information is retrieved from the Internet, and complex reasoning techniques are used, private information can be deduced. The heterogeneity and mobility of objects in the Internet of Things increase the complexity of privacy protection.

Section 3 Application Model of IoT Data

1. Production control of IoT workshops: As the "brain" of the entire production, the workshop production command and dispatch center needs to coordinate and dispatch various resources and production capabilities of the workshop. The center integrates ERP, MES, MDC and other system data, and uses large electronic screens as carriers to display the operation status of each production site, emergencies, event tracking, and improvement status. Combined with the lean production concept, the production task execution of each production unit is managed throughout the process. After statistics, analysis, induction, and prediction, the production data is visualized and managed, which provides a panoramic display of the company's production task execution status, provides enterprise decision-making support analysis, and ensures that production tasks are completed on time in terms of quality and quantity.

2. Quality control of IoT workshops: A steel company is the largest special steel manufacturer in China. On one of its silicon steel production lines, due to the influence of multiple complex factors, a corrugated defect called longitudinal stripes sometimes forms on the surface of the finished product. The longitudinal stripe defect not only affects the appearance of the product, but also has a direct impact on the physical properties of the product such as interlayer resistance, electromagnetic properties and lamination performance. The longitudinal stripe defective steel accounts for about 30% of the production volume, causing huge losses to the company every year. Through data mining technology, the production process data is deeply analyzed and mined, and a product quality optimization model is established to reduce the defective rate and improve product quality.