Abstract: Explain in detail how Huawei cloud IoTA service helps IoT data developers quickly realize the value of IoT data, and bring you hands-on practice to complete a set of industrial Internet of things solution development.

This article is shared from huawei cloud community “[Cloud Resident Co-create] Twin? Digital Twin? Welcome to the geek world”, author: Qiming.

Before we begin, let’s make an explanation of the noun in the title:

  • Huawei Cloud IoTA data Analysis: Also known as Huawei Cloud IoTA (Analyst) Huawei cloud IoTA service is based on the Internet of Things asset model, integrating IoT data integration, cleaning, storage, analysis and visualization, reducing the development threshold, shortening the development cycle, and helping IoT data developers quickly realize the value of IoT data. Through hands-on practice, developers in the field of intelligent manufacturing experience how to complete the development of a set of industrial Internet of Things solutions based on Huawei cloud IoT edge computing, equipment access, data analysis capabilities, from equipment access to the production line.

  • Digital twinning: Digital twinning is a simulation process that integrates multi-disciplinary, multi-physical quantity, multi-scale and multi-probability by making full use of data such as physical model, sensor update and operation history, and completes mapping in virtual space to reflect the whole life cycle process of corresponding physical equipment. Digital twinning is an out-of-reality concept that can be viewed as a digital mapping system of one or more important, interdependent equipment systems.

Through the above explanation, I believe you have a certain understanding of the title and what we are going to introduce today. Next, let’s get to the topic.

In the era of Industry 4.0, factory digitization is in danger

Looking back at the history of mankind, we have successfully experienced three industrial revolutions:

The first was the age of steam engines, which ushered in the replacement of manual labor by machines. The second is the age of electricity, the development of natural science and industry closely integrated, science played a more important role in promoting productivity; The third is the information age, the transformation of science and technology into direct productivity speed up.

Now, we have ushered in the fourth revolution, namely, Industry 4.0: intelligent age. “The essence of industry 4.0 is to transform economies of scale into economies of scope through data flow automation technology, and build heterogeneous and customized industries at the cost of homogeneous scale. This will play a crucial role in industrial restructuring.

As a new round of industrial revolution, the core feature of Industry 4.0 is interconnection. Industry 4.0 represents the intelligent production of “Internet + manufacturing” and breeds a large number of new business models, which can really help realize the “C2B2C” business model.

At present, everyone is still in the “industry 4.0” groping stage. A large number of factories have started their own intelligent transformation, such as building applications to visualize collected data and maximize the value of data. However, in the process of practice, problems kept emerging, such as:

Islands of data/information with smokestacks

1) Multiple applications are built independently, lacking unified standards, and data sharing and information exchange between applications cannot be effectively carried out, forming data and information islands;

2) The enterprise’s digital assets are distributed in small quantities, with high maintenance costs and low use efficiency;

2. The application takes time and effort to come online

1) Repeated wheel manufacturing, each application will have a lot of repeated work, waste human and material resources, time-consuming;

2) The amount of data processing is complex and there is no unified modeling, which leads to repeated processing of original data in all applications;

3. High threshold of data analysis

1) Low data utilization, large amount of data restriction, underutilization, value mining is only the tip of the iceberg;

2) Key reasons: unclear business scenarios, high data development technology threshold, lack of good data platform;

One of the reasons that may lead to the above key problems, we think, may be related to the lack of hierarchical decoupling of software development.

Software development does not do enough hierarchical decoupling

Why is software development not layered enough to decouple?

Let’s compare the traditional software development process with the huawei Cloud IoTA model. There are two patterns in the traditional software development process:

Pattern 1: Build applications independently.

1) Half of them lack overall planning. Each application is deployed independently, and data is collected and used based on business needs

2) Low efficiency, such as repeated data collection, which has a great impact on production

Mode 2: Build based on data acquisition platform

1) Start to consider a unified data acquisition platform, with professional data acquisition teams completing as much data collection as possible and centralized, unified and open;

2) Overall efficiency has been improved, but the use of data is still independent without real integration, and there are still a lot of repetitive work in data processing between applications.

So how does Huawei cloud solve this problem? This is where the number twin comes in.

With a model such as the one above, it is possible to:

More comprehensive contextualized cognition:

1) Unified digital Modeling language, semantic layer intercommunication

2) Graphical/template-based modeling development environment, improving efficiency by 40%

3) Powerful model computing engine, supporting a maximum of 100 million model nodes, seamless AI, intelligent expansion

Broader physical perception:

1) 10+ mainstream native protocols, 60+ standard protocols

2) Cloud side collaboration, custom extension of protocol plug-in

3) The module is pre-integrated with SDK, and the cloud is connected immediately after power-on

By digitizing the physical objects one by one, the interaction between application and physical device is transformed into the interaction between application and digital twin. This is a big change from the first two: we can ignore the underlying physical devices, or interfaces, and hand off data modeling to IoT’s “unified twin model layer.”

IoT data analysis twin modeling concept analysis

Next, we will introduce some concepts related to “twin modeling.”

Digital twin: in the broad sense, it is a system composed of physical objects, digital mirror and interactive system, and in the narrow sense, it is the digital mirror of physical objects.

Digital twin maturity model can be divided into five stages: digitalization, interaction, foreknowledge, foresight, and collective wisdom. It has surpassed any concept of existing digitalization, Internet of Things, simulation, big data, artificial intelligence, cloud computing and so on.

1) Digitalization: objects in the physical world can be digitally modeled to obtain a digital model;

2) Interaction: information and data can be transmitted between digital objects and physical objects in real time to produce interactive effects. For example, some control commands can be issued to the equipment (turning on the street lamp);

  • Prophet: Can predict the future based on complete information and clear mechanism. For example, the weather forecast in daily life can predict the weather in the future;

  • 1. To speculate about the future based on incomplete information and unclear mechanisms;

3) Co-governance: multiple digital twins can share wisdom and co-evolve, which is a higher performance.

Digital twinning in IoT: Real-time and accurate mapping of objects in the physical world to the digital world by defining models and aggregating relevant data to build digital twinning carriers.

IoT digital twin application scenario

Based on the above concepts, we introduce the digital twin application scenario. Digital twin is a universal technology, which is becoming a new starting point for national digital transformation and enabling technology to support the digital transformation of all sectors of society. The following table shows the application scenarios and values of digital twin core:

In the “smart factory” industry, for example, leading companies are already developing digital twins. In the process of developing equipment, the digital twin is developed and delivered simultaneously, thus realizing the ability of digital management of the whole equipment in its life cycle. At the same time, relying on the on-site data collection, combined with its twin analysis, can provide product fault analysis, fault prediction, remote management and other value-added services, ultimately can effectively improve user experience, reduce the overall operation and maintenance costs, and can strengthen the competitiveness of the enterprise.

The factory digital twin consists of manufacturing digital twin and product digital twin, and the multi-model behind it is the core

Factory digital twin consists of two parts: manufacturing digital twin and product digital twin.

Create digital twins:

Positioning: the factory’s manufacturing process is digitally mirrored, which can reflect the factory’s manufacturing process in real time; Through the unified abstraction of manufacturing process, different applications can interact with each other based on unified semantics.

Modeling content: production equipment, production line, production process, quality defects, physical structure, etc.

Product Digital twin:

Positioning: organize all kinds of data generated in the production process from the dimension of product manufacturing in the factory, and reserve the ability to connect with the digital main line to get through the data of product design stage and product maintenance stage;

Modeling content: product attributes, production process data, quality data, etc.

There are multiple models behind the digital twinning of the entire factory:

From bottom to top, we can see: equipment model, production line model, process capacity model, quality defect model, etc., and then equipment failure prediction model and equipment physical/structure model, etc. A variety of models, corresponding to different use scenarios.

So, how is such a model designed and landed?

We need to remember that when building digital asset models for things in the physical world, we must define the asset model before we create the asset.

The digital twin of IoT consists of two parts, one is the model and the other is the asset. How do you understand these two concepts? A model is the object oriented equivalent of a class, a class. Classes are static, defining something abstract.

And the asset instance is equivalent to the object in the object oriented instance, that is, the class instantiation creates an object, and uses it.

Open the model, which is divided into attributes and analysis tasks. Properties can be analogous to member variables in our class; Analysis tasks, analogous to class member methods, define the logic of some analysis techniques for attributes.

In terms of attributes, it can be divided into static attributes, measurement data attributes and analysis task attributes. At the same time, the analysis task can also be divided into transformation computing, aggregation computing and flow computing.

Introduction to modeling effect of factory digital twin production line and equipment

So, what is the effect of Huawei cloud IoTA service?

Take a SMT production modeling scenario as a demo:

Above is the SMT patch production line. SMT is actually the production process of a PCB board, all electronic products including iPad, mobile phone and other electronic related products, are inseparable from this process.

As can be seen from the demo, there are three production lines in the middle, and the current real-time equipment and OEE indicators of each production line can be seen on the left.

The middle part is the production line equipment, which can be operated by:

  • You can zoom in and you can open it up;

  • You can click a device and see the real-time OEE of the current device in the current cycle.

  • You can monitor key parameter attributes of the device. For example, you can monitor outliers of key attribute values.

  • You can see that if the device is abnormal: red or yellow dots are displayed throughout the system to indicate that the device is abnormal.

On the right, you can see the device’s column data for the past few days or hours.

Or if you’re wondering about OEE, what is OEE?

OEE stands for Overall Equipment Effectiveness. Generally speaking, each production equipment has its own theoretical capacity, to achieve this theoretical capacity must ensure that there is no interference and quality loss. OEE is used to show the ratio of a device’s production capacity to its theoretical capacity.

When calculating OEE, three dimensions will be involved:

  • Time utilization: Time utilization = σ actual running time / σ planned startup time *100%. Used to evaluate the loss caused by the shutdown, including any event that caused the planned production shutdown, such as equipment failure, raw material shortage, production method change, etc.;

  • Performance utilization: Performance utilization = σ [number of outputs * cycle time of a product processed in the proper state of the equipment]/ σ actual running time *100%. Used to evaluate the loss in production speed. Include any factors that cause production to fail to operate at maximum speed, such as equipment wear, material defects, and operator errors;

  • Pass rate: pass rate =[qualified output quantity]/[output quantity]*100%. It is used to evaluate the loss of quality, which is used to reflect the products that do not meet the quality requirements (including reworked products);

The final calculation formula is OEE=[time utilization]*[performance utilization]*[pass rate]*100%, which is a key indicator to measure the overall operating efficiency of the equipment, and is also a key indicator in many electronic manufacturing plants and other similar plants.

Generally speaking, the value of OEE of domestic manufacturers is not too high, generally only 70%, or 80%, or even only about 40%.

Detailed explanation of plant digital twin production line and equipment modeling

What is the general approach to data processing and analysis for the entire IoT data analytics service?

Refer to the picture above:

The first step is the data pipeline. We put the data in through the data pipeline and back it up locally;

The second step is to model the equipment.

The third step is to establish equipment assets.

The fourth step, the equipment after model instantiation and the data poured in, through the equipment asset analysis of the calculation engine, complete the real-time computation-related analysis tasks;

Step 5: Store the data inside the IoT.

Step 6: Open this data to third parties through an API.

Next, take the actual PRINTING press of SMT production line as an example, combine the previous concepts to talk about how to configure and design the model.

The whole figure can be divided into two parts. On the left side is the whole data to complete the dimensions introduced by the concept outline of the relevant model. On the right side is the actual information of a production printing machine.

Previously, we roughly introduced that attributes can be divided into three categories, but did not specifically explain the relevant information of the three categories. Here, we introduce some supplementary information:

Static configuration attributes, which do not need to be reported by the device and seldom change, such as product model, device type, etc.

Measurement data attributes are reported by the device. Generally speaking, that is, data analysis can not be obtained by itself, need others to give the system data. It includes the attributes reported by the device or read by the third-party business system, which are considered as measurement attributes by the system.

Analyze task attributes, which need to be further calculated after the data is reported.

Meanwhile, there are three types of analysis tasks:

Transformation calculation: For a simple example, suppose that the creation contains two attributes, A and B, and we require that a+b= C in the process, then this is a transformation calculation. The transformation attribute must be real-time, and the data timestamp of ab and AB values is the same;

Aggregation calculation: Aggregation is calculated in a time dimension. Assume that the average temperature is required in the past five minutes. If the device reports data every five seconds, average all data reported in the last five minutes, which is equivalent to aggregation calculation in the time dimension.

Flow calculation: Flow calculation is mainly used in complex scenarios where the logic cannot be expressed with a simple if/else. For example, when an asset reports a number of parameters, the system needs to calculate a result from these parameters and then return the asset, so the flow calculation functions as a calculator. The function of stream computing is very powerful. In the factory digital model, most scenes can be realized, such as sliding window, data filtering, adding attributes, etc., which is a relatively common ability.

The following figure is the configuration of OEE related indicators. If you are interested, you can go to the demo to see the actual situation:

Here, we focus on the pass rate. Pass rate =[qualified output quantity]/[Output quantity]*100%. The “TS_Sum” in the table represents a time series sum, which means you can sum production over a time range, for example, production over five minutes. Other indicators of the calculation method and pass rate similar, not a repeat.

With assets and equipment in hand, we can build the production line model.

The production line model has the same idea as the device model, but there are some special points. For example, the production line has the name of the production line, but the production line is not an actual physical device, it is only a virtual concept, so the production line may not have measurement data attributes. But it can define corresponding analysis tasks, such as time utilization, performance utilization, pass rate, OEE, and so on.

These properties can be obtained by analysis and calculation, but data from the production line all equipment under the data, so the analysis of the corresponding task definition process and equipment are the same, when calculating the value of the only, just take the equipment the following child data after sum, do come out of a total calculation results.

The above is a modeling process of a production line. After you have the production line model, you can build a factory model in the same way. The factory model may be relatively simple, just fill in some of the attributes that need to be configured.

With the three models of equipment, production line and factory, the entire digital factory can be built.

Build production line assets: digital SMT plant

As you can see from the image above, building a digital factory is actually quite simple:

Firstly, the parent asset — SMT factory is constructed based on factory model. Secondly, a sub-asset can be added under it, such as the first production line, we can call SMT production line 1, and then form a second production line. The assets of the two production lines are instantiated based on the production line model. Finally, we can continue to add assets (equipment) under sub-assets, such as lightning engraving machine, printing machine, SMT machine and other sub-equipment. After the whole process, we can build a digital model of the whole factory.

Please note: how the asset number is constructed depends on the company’s own business needs, there is no specific principle frame.

Combined with the real-time data reported by the equipment of the actual production line, the digital twin model and the asset data calculation engine behind can be used to build the SMT plant’s middle line and equipment in the digital world twin, which can reflect the dynamic changes of the actual physical equipment in real time.

The entire digital factory is built, and on the console side it looks like this:

This is an editorial process. You can see a workbench layout on the left, where you can see each device. After clicking on it, you can see the definition of some properties in detail and some content of the analysis task. Publish through the publish button in the upper right corner. After the release, you can record the operation of the total number of assets, directly run after the real-time response to calculate the need for some data.

As can be seen from the above figure, when the whole asset runs, the data calculated in real time on the right side of a production line can be presented in real time, that is to say, the warning situation of the whole asset can be monitored in real time.

In addition to the above data display mode, according to the needs of the business, the data can be displayed into broken line graph, thermal diagram, curve diagram, etc., easy to analyze the graphical display way, get the results you want. An example figure is shown below.

Want to experience a process of production line model, can go to huawei cloud IoT data analysis service (www.huaweicloud.com/product/iot)… Deep experience. Follow the instructions on the Overview page step by step.

Let me introduce another course. Renesas and Huawei have created a digital factory course. This course is mainly to help developers to realize the terminal development process from the device layer to the application layer based on the development board, so as to experience the whole digital factory end-to-end.

From the figure above, we can see that the whole course involves multi-layer architecture of device layer, edge layer, network layer and so on:

Intelligent application: real-time data push, seamless flow, analysis and calculation, rapid construction of industry solutions

1) Based on the digital twin modeling you, to achieve the production line equipment asset modeling, try to calculate the production line OEE indicators;

2) 3D modeling of production line based on topology application to achieve data visualization.

Access intelligence: cloud, network, edge and terminal coordination, to create a stable, reliable, safe and lowest cost connection service

1) Safe and stable connection of 100 million machines and equipment, high concurrent and reliable communication of 100,000 TPS;

2) Full-link log analysis and message tracing, real-time monitoring and sensing of device status, and flexible customization of service indicator alarms.

Perceptual intelligence: Based on the twin practice of Huawei digital Factory, to create a real, professional and reliable experience scene

1) Cooperate with partners to withdraw, provide customized development version, simulate production line equipment data acquisition and universal protocol access;

2) Detailed interpretation of end-side Demo, in-depth analysis of industrial equipment data acquisition indicators, and exploration of digital transformation.

Click to follow, the first time to learn about Huawei cloud fresh technology ~