Big data analysis of industrial manufacturing

Big data analysis of industrial manufacturing

Big data is not just the accumulation of a large amount of data. An important attribute of big data is that people try to collect and understand the ever-changing data types. If only a large amount of data of the same type is collected, no amount of data can be called big data.

How to realize intelligent manufacturing is a problem that everyone cares about. From Michael Porter of Harvard Business School to Wharton Business School, there is a general understanding that digital transformation is the way to realize intelligent manufacturing. Importantly, this knowledge also comes from many world-class manufacturing enterprises and entrepreneurs.

This knowledge is based on the integration of many technical trends, such as Internet of Things, Cyber System (CPS), Industrial Internet of Things, Mobile Technology, Artificial Intelligence, Cloud Computing, Virtual/Virtual Augmented Reality (VR/AR) and Big Data Analysis. We must stay awake and don't simply think that with these technologies, the next five years will be the golden age of manufacturing. There is a simple reason. The transformation process of this new manufacturing culture is quite complicated and difficult. Without the integration of industries, enterprises and users, this transformation cannot be realized. Digital transformation not only means simple enterprise digitalization, but also takes numbers as the core driving force of intelligent manufacturing and integrates industrial chain and value chain with data.

Since the industrial revolution, in order to improve operations, manufacturers have been interested in collecting and storing data. With the passage of time, the demand for data analysis in manufacturing industry will increase. However, in the past many years, the fundamental motivation of using data has not changed, the complexity of data has increased, and the ability of transforming data into intelligence has become greater and greater.

20 12, Gartner gave the definition of big data, especially emphasizing that big data is a diversified information asset, which not only pays attention to actual data, but also pays attention to big data processing methods. The size of data itself is not the core index to judge the value of big data, but the real-time and diversity of data have a more direct impact on the definition and value of big data.

When discussing industrial big data analysis, I noticed two different views:

The first view is that manufacturing has always had big data. For decades, our enterprise has been collecting data through historical records, MES, ERP, EAM and other application systems. Big data is a new hot word in some industrial chains, especially in marketing.

The second view is that from the perspective of industrial big data, manufacturing is an unopened market, or a newly opened market. There are many different types of data, but now they have not been applied to the analysis.

Considering these viewpoints, we should always maintain an appropriate skepticism when facing any new market formulation, including noun explanation, definition or analysis framework. I prefer the second view here. Our manufacturing industry does have a lot of data, but this is not the meaning of "big data" that most of us understand from the market. How to define the big data of manufacturing industry before figuring out the analysis of industrial big data? Here, we can further understand the characteristics of big data through its three characteristics.

data source

There are two main sources of industrial big data. The first is smart devices. Pervasive computing has a lot of space, and modern workers can take a pervasive sensor and other equipment to participate in production and management. Therefore, the industrial data source is the association between a large number of devices of about 28 billion, which is one of the data sources we need to collect in the future.

The second data comes from the data generated by human trajectory, including the internal processes of procurement, production, logistics and sales in modern industrial manufacturing chain and external Internet information. Through the combination of behavior trajectory data and equipment data, big data can help us analyze and mine customers, and its application scenarios include real-time core transactions, services, back-office services and so on.

Data relation

Data must be analyzed in the corresponding environment to understand the relationship between data. For example, every new model will go through a series of cruel flight tests before being delivered to airlines. Extreme weather test is one of the tests. The purpose of this test is to ensure that the engine, materials and control system of the aircraft can work normally under extreme weather conditions.

The key to deal with the problem is to find the root of the problem, eliminate the known mistakes and ensure the reliability and effectiveness of the solution. Once the root cause is found and determined, and acceptable emergency measures are taken, the problem can be regarded as a known error. The problem investigation process must collect all available information related to the incident to determine and eliminate the root cause of the incident and the problem. Data collection and analysis must be combined with environmental data of events/problems.

Data value

For digital transformation, big data should not only pay attention to the actual amount of data, but also pay attention to the application of big data processing methods in specific occasions, so that data can generate great innovation value. If you leave the design of revenue consideration or return on investment (ROI) and blindly seek big data, big data analysis can neither land nor create value for enterprises.

Definition of industrial big data analysis

Engine is the heart of an airplane, and it is also the most important thing related to aviation safety and life safety. In order to monitor the engine status in real time, most modern civil aviation companies have installed the aircraft engine health management system. Data collected by sensors, transmitting systems, signal receiving systems, signal analysis systems, etc. It will be transmitted by VHF or satellite communication through the aircraft communication addressing and reporting system, which is why GE's engine monitoring system obtains more than 1PB data every day.

The production execution system (MES) is exactly the same as the aircraft engine health management system. We can collect a large number of process variables, measurement results and other data in real time from the production of the factory. Reports generated based on a large number of data sets, or analysis of basic statistics, are not enough to be called big data analysis in manufacturing.

The diversity of data types is an important attribute of industrial big data analysis.

Big data is not just the accumulation of a large amount of data. An important attribute of big data is that people try to collect and understand the ever-changing data types. If only a large amount of data of the same type is collected, no amount of data can be called big data.

For example, the time series collected in the production environment simulates process variables, and the data type is single, so it is easy to establish indexes. Even with Qian Qian, it is not enough to become big data.

Data must include high variability and species diversity. There are numerous applications of big data in manufacturing plants, but it does not include simply classifying and displaying a series of process measurement results. For these tasks, you can complete the basic statistical display. Some big data databases or data lakes are also composed of text information, image data, geographical or geological information and unstructured information, such as data types obtained through social media or other collaborative platforms.

Manufacturing information structure is generally divided into two layers, one is management layer and the other is automation layer. Decision support, management, production execution, process control and equipment connection and perception are realized from three dimensions: management, production execution and control. Big data analysis in manufacturing industry refers to combining structured system data and unstructured data in management and automation layers with a common data model, and then discovering new insights through advanced analysis tools.

The Significance of Big Data Analysis to Enterprise Production Intelligence

The core of manufacturing innovation is to rely on a large number of cutting-edge technologies. Advanced technology is a means of innovation. With the support of new technology, enterprise management application systems, such as ERP and EAM, can be integrated with related systems of industrial automation through integrated manufacturing operation management system MOM. On the basis of integrated manufacturing operation management, an integrated manufacturing enterprise information system solution integrating IT+MOM+MES+BI is realized.

From the perspective of the integration of industrialization and modernization, information system providers should unify planning, standards, functional design and implementation strategies from the perspective of the main information system providers (MIV) of enterprises. Assist enterprises in risk control, reduce investment, reduce operation and maintenance costs, and realize the complete integration of enterprise information systems.

In particular, the enterprise management information platform is usually regarded as an integration and dashboard tool for manufacturing enterprise management. Many suppliers not only invest heavily in their proprietary integration with ERP and automation systems, but also invest in open integration, dashboard and mobile technology, hoping to provide measurement standards for decision makers who need correct information anytime and anywhere.

Three ways of big data analysis in manufacturing industry

The first way is to use open technology and platform to move the data of any system to any other place.

The construction project of manufacturing operation management system is a systematic project, which is not only a traditional software system we understand, but also a platform for project implementation and service. This needs to reflect the comprehensive management ability and soft power of manufacturing enterprises in project management and strategic "customer service".

The whole platform should be structured from three stages: early stage, project implementation and after-sales service. In the early planning, we should pay attention to standards, design and implementation, especially to form a unified docking with the integrated management information system. With the formulation of unified planning in the early stage, the link of project implementation can integrate industry experience, integration ability, implementation ability and software development ability. In particular, it is necessary to establish and form a super team system within the organization. Continuous service, long-term operation, and integration of Internet of Things application with internet plus strategy of "software+cloud service" are the key considerations for follow-up services.

In the analysis of big data in manufacturing industry, it is necessary to strengthen the support for subsequent continuous services through the application of Internet of Things technology. Through the industrial Internet of Things, we can strengthen and lock the long-term cooperation between supply chain enterprises by responding to customers in time, regularly checking the software and hardware systems of the Internet of Things, providing emergency spare parts, providing consumables and improving applications. By managing the platform and Internet of Things data, we can continue to provide valuable services to our customers.

The second method is to invest in a data model that can handle structured and unstructured data in the system architecture stack inside and outside the factory.

New technology is the core of innovation revolution, and one of the most important features is integration, that is, the integration of manufacturing operation management system MOM with ERP, EAM, OA and business analysis, including one-click login, interface integration, message push, workflow integration, master data and application integration bus and platform.

Because the master data between these systems are unified, the data interaction between all systems depends on the application system bus for data interaction, and seamless integration and analysis are realized after integrating cross-system business processes, workflows and service processes. For enterprise managers, after one-click login, the necessary information most related to management can be personalized according to different positions. This is the sharing concept brought by the Internet.

The third way is to combine advanced analysis tools such as time series, images, videos, machine learning, geospatial, prediction models, optimization, simulation and statistical process control with the big data platform in manufacturing enterprises, so as to gain insight into the situation that has not yet appeared. Combining IOT data such as sensors, sensors, transmission networks and application software with management application software will be a main direction of big data analysis in manufacturing industry in the future.

Cultivate big data analysis experts within the enterprise.

As an industry, we need to organically develop industry-specific big data analysis tool sets, so that current industry experts can realize digital transformation from enough data science. In order to promote the transformation, we need a large number of excellent enterprises to use this method and prove its value to others or peers.