Direct access to Huawei Cloud IoT Intelligent Manufacturing behind the HUAWEI MATE 40 production line

Abstract:Digital Twin? Find physical world devices in the digital world!

This article is shared from the HUAWEI MATE 40 production line directly hit Huawei Cloud IoT Intelligent Manufacturing Factories Digital Transformation, the original author is Qiming.

Part 1: Digital Twin in the Era of Intelligent Industry 4.0

I. Industry 4.0, the intelligent era has come

Looking back at human history, we have gone through three industrial revolutions together.

The first was the era of steam engine, which ushered in the replacement of manual labor with machines. The second was the electrical age, when the development of natural science and industry were closely combined, and science played a more important role in promoting productivity. The third time is the information age, the transformation of science and technology into direct productivity speed up rapidly.

Now we are in the midst of the fourth revolution, Industry 4.0: The Age of Intelligence. “The essence of Industry 4.0 is to transform from economies of scale to economies of scope through data flow automation technology, and to build heterogeneous and customized industries at the cost of homogenized scale. This is a crucial role for the reform of the industrial structure.”

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 achieve the “C2B2C” business model.

II. The pains and difficulties in the current digital transformation of factories

At present, we are still in the stage of “Industry 4.0”. A large number of factories have started their own road of intelligent transformation, such as building applications to collect data visualization, to maximize the value of data. However, in the process of this practice, problems continue to emerge, such as:

2.1 Data/information islands with stacks

A factory, in different stages, because of the different projects, may find different suppliers to undertake. The segmented project providers lead to different system applications. Visually speaking, the multi-system is not interconnected, just like independent “chimneys”, each “chimney” has “smoke”, but they are not interconnected. In the stage of Industry 4.0, no intercommunication means information island, which means the scattered distribution of digital assets of enterprises, high maintenance cost and low use efficiency.

2.2 Slow application on-line, time consuming and energy consuming

As mentioned in the first point, the lack of interconnection between different systems leads to “reinventing the wheel” for new applications: there is a lot of rework for each application, which wastes manpower and material resources, and takes a long time. More important is the problem of data processing with new applications: due to the lack of unified modeling, each application needs to process the original data repeatedly. Two “duplicates” make the already high cost even more “worse”;

2.3 High threshold for data analysis

Factories or enterprises want to reduce costs and increase efficiency. For example, they want to find rules through analyzing existing data so as to optimize the process, but they are discouraged by the high threshold of data analysis. The most critical reason for this is that its business scenario is not clear and it has not found a good data platform.

Third, finding the right platform is half the battle

The above pain points and difficulties are encountered by most manufacturers in the industrial field in the process of “Industry 4.0” exploration, and through which is the “application”. That is to say, software developers do not do enough hierarchical decoupling is one of the important reasons for the above problems. Based on the “application”, the plant went through three periods of three modes:

3.1 Mode 1: “Chimney” application

Prior to Industry 4.0, due to the lack of applications and practices, most factory applications, as described above, were “chimney-type” :

This leads to, first, lack of overall planning, independent deployment of each application, data collected and used separately based on business needs; Second, low efficiency, such as repeated data collection, has a great impact on production.

3.2 Mode 2: Platform decoupling — unified digital acquisition platform

After the concept of “platform” was proposed, factory managers gradually realized that perhaps, there needs to be a “platform” between the production line and the application. Such decoupling can make the application and the production line and the application and the application interconnect with each other. This is a fundamental model of Industry 4.0.

The generation of mode two enables professional data collection teams to complete as much data collection as possible, and centralized, unified and open the data collection, thus improving the overall efficiency. But we can see that even in this case, the use of the data is still independent, and there is no real convergence. The data obtained in the production line or production equipment is still metadata. After the application obtains the data, it still needs to conduct secondary processing and use of the data separately, which leads to a lot of repeated work in the data processing among the applications.

3.3 Mode 3: Data processing — unified twin model

Huawei IoT has its own method to solve the problem of “application decoupling” and “unified data processing” simultaneously.

In the Internet of Things, there is such a thing as a “twin.” Through the “twin”, the perception of the device and the cognition of the device are processed in a unified way. Also take a factory as an example, there are a lot of production equipment, production lines and other kinds of physical equipment in the factory, so can we help the factory process all these physical equipment one by one through unified modeling and abstract them into digital images?

The answer is yes. By digitizing the physical objects one by one, the interaction between the application and the physical device becomes the interaction between the application and the digital twin. There is a big change in the way this pattern is developed compared to the previous two patterns: we can ignore the underlying physical device, or physical interface, and leave the data modeling part to the IoT’s Unified Twin Model Layer.

The concept of “twinning” means that we need to have a clear understanding of the model in the process of modeling, that is, a wide range of number acquisition ability. After all, in a factory, there will be a variety of equipment which also have a variety of protocols. Second, you need to have a very high level of abstraction, you need to abstract devices from the physical world into models of the digital world that can interact with each other.

Number acquisition ability and abstraction ability are two key capabilities in the development of Internet of Things applications.

Based on Huawei Cloud IoT to bring a new development mode, to help users quickly build the digital transformation of the basic platform.

Next, take Huawei’s own factory as an example, to briefly explain how Huawei Cloud IoT uses the new development model to facilitate the digital transformation of the factory.

As we all know, Huawei itself is also a manufacturing factory. Huawei’s cloud IoT capability is first practiced in its own factory. We take the southern factory, which is the production factory of HUAWEI Mate 40, as an example to construct a digital twin of the production line in the digital world through digital acquisition and modeling of the mobile phone patch process of the factory.

Based on the capability of Huawei Cloud IoT, a unified twin body is completed in the south, and a visual and intelligent application is constructed in the upper layer. The specific architecture diagram is as follows:



In the actual digitization process of the South Factory, there are the following challenges:

• Production line equipment manufacturers/types/models are diverse, involving more than 30 different application layer protocols need to be connected, and it is difficult to collect;
• There are more than thousands of data of measurement points in one production line, and lack of data modeling means leads to poor data processing.

So, how do you save time and effort to digitize from a developer’s perspective? Huawei Cloud IoT has officially made its debut.

Fourth, build digital twin with multi-dimensional model as the core

Behind the practical application of a digital twin, there are many models, such as the model of production line, the model of equipment, the model of quality defect and so on. In the modeling process, from the perspective of looking at the physical objects in the physical world of a factory, the twins of a factory can be divided into two categories: the digital twins of manufacturing and the digital twins of products.

Making digital twins:

• Positioning: Digital mirror of the manufacturing process of the factory, which can reflect the manufacturing process of the factory in real time; Through a unified abstraction of the manufacturing process, different applications can interact based on the same semantics.
• Modeling content: production equipment, production line, production process, quality defects, physical structure and so on;

Product digital twinning:

• Positioning: Organize all kinds of data generated in the production process from the dimension of products in the factory, and reserve the data in the product design stage and product maintenance stage through the docking ability with the digital main line;
• Modeling content: various attributes of the product, production process data, quality data, etc.

These are two very important dimensions of digital abstraction for factory digital twinning. Through the transparent production process of the production line, the production is orderly and controllable, reducing the application on-line time from the original 6-9 months to 3 months; At the same time, twin modeling + intelligent analysis, using data to drive intelligent production, so that the efficiency of data development can be increased by 70%. Through Huawei Cloud IoT, we canThe total factor connection of the factory can be realized quickly, and the data utilization efficiency can be greatly improved by constructing the digital twin model of the factory.

Part 2: Digital Twin Practice Based on South Factory

Back to our topic. The South Plant is the production line for HUAWEI MATE 40. The explosion of mobile phone production, so that the production line digital demand is imminent. Through the digitization of the entire production line, the production process can be improved, the management of manufacturing engineering companies can be optimized, and the management of quality control can be optimized, so that the efficiency of the production line can be greatly improved, and the operation cost can be reduced at the same time.

Above is a multidimensional model of a factory twin. We can see that in the product model, there are the equipment model and the production line model, and then there are the process capacity model, the quality defect model, the equipment physical/structural model and the equipment fault prediction model.

By applying the modeling and analysis capabilities of Huawei Cloud IoT data analysis services, the electronic engineering production line and equipment twin can be quickly constructed. In this article, let’s look at how to build a data analysis service model.

I. Introduction to basic concepts

(I) Introduction to the concept of OEE

Before we go into modeling, let’s popularize a basic concept. OEE, Overall Equipment Effectiveness. Generally speaking, each production equipment has its own theoretical capacity, in order to achieve this theoretical capacity must be guaranteed without any interference and quality loss. OEE is used to represent the ratio of the production capacity of the equipment to the theoretical capacity.

When calculating OEE, there are three dimensions involved:

1. Time utilization:Time utilization =Σ actual running time/planned startup time *100%. It is used to evaluate the loss caused by the shutdown, including any event that causes the shutdown of planned production, such as equipment failure, shortage of raw materials, changes in production methods, etc.
   
   2.Performance utilization:Performance utilization =Σ[quantity producedThe cycle time of processing of a product in the due state of the equipment]/Σ the actual running time100%. It is used to evaluate losses in production speed. Include any factors that cause the production not to run at the maximum speed, such as equipment wear and tear, material substandard and operator error;
   
   3.The percent of pass is:Qualified rate =[quantity of qualified output]/[quantity of output]*100%. Used to evaluate the loss of quality, it is used to reflect the product does 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 measure of the overall operational efficiency of the equipment, and is also a key measure in many electronics manufacturing plants and similar facilities.

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

(II) Modeling and analysis renderings of the twin production line and equipment of the plant

The modeling and analysis of the twin production line and equipment can be viewed through some visual management background. Below are three renderings of different functions.

Picture 1: There are 3 production lines, which can be dragged and dropped appropriately. The figure shows the OEE value of each device. Through asset modeling and analysis capabilities, OEE of production lines and equipment can be calculated in real time, key indicators of each equipment can be monitored in real time and historical data can be viewed at the same time.

Picture 2: Device modeling diagram. Through the combination of equipment failure message and equipment model, the running state of equipment is monitored in real time.

Picture 3: Asset Analysis Diagram The asset model analysis capability enables real-time analysis and monitoring of reported equipment data for anomalies. For example, humidity is normally to 45%~63%, if the reported data is not within this range, it is considered abnormal data. The screen will display a yellow dot indicating that the device is reporting an abnormal data. It can be seen that data analysis can be calculated and monitored in real time. If there are some serious anomalies, it can even be pushed to operation and maintenance personnel.

(III) Factory Digital Twin DEMO data processing and analysis process

To achieve the above renderings, we need to go through the following steps (since it is not a real factory, it is a simulation device) :

• Device simulator: Based on the standard material model, the simulator automatically reports the device attribute data through the MQTT protocol in 5 seconds at regular intervals, and can simulate manually starting to report messages, such as the message of equipment running out.
• IoT Device Access Service: Configure device attribute data and device messages to IOTA (data analysis) services by configuring device data forwarding rules.
• IoT data analysis service: Receive equipment data based on data pipeline, and generate OEE related data of production line and equipment in real time through asset modeling and calculation and analysis capabilities, and judge whether there is abnormal information in the data.
• 3D application: obtain data by calling IOTA API, display production line and equipment in 3D form, view OEE of production line and equipment, key indicators of equipment, failure information, etc., and find relevant historical data at the same time. This is the renderings of the second part.

(IV) Analytical process within IoT data analysis

Next, let’s focus on the flow process within IoT Data Analysis Services.

Step one, the data pipeline. We bring the data in through the data pipe, and we back it up locally;

The second step is to model the device.

The third step is to establish equipment assets.

The fourth step is to complete the analysis tasks related to real-time calculation by analyzing the equipment instantiated by the model and the input data through the calculation engine of equipment asset analysis.

Fifth, store the data inside the IoT.

Step six is to make this data available to third parties through the API.

See the following figure for details:

In this process, we need to explain in detail how step 2 and step 3 work, that is, how do we create the model and the asset?

(V) Introduction to the basic concept of IoT digital twinning

Before we move on to model creation and asset presentation, let’s introduce the basic concept of IoT Digital Twin.

We believe that there is a real-time and accurate mapping of objects in the physical world in the digital world, which can organize the actual device data and some other data to form the Jason model and become a carrier.

Above is a conceptual picture of our digital twin.

First, the data twin can be divided into model and asset. The model is equivalent to a Java class in the development process, representing a template of a class. Generating an asset after instantiation, which is equivalent to a New Class, generates an object. An object corresponds to an asset.

At the same time, models are divided into two types. The first type is attributes, which can be further divided into three types:

The first is static configuration properties, which do not need to be reported by the device and are not likely to change, such as product model, device type, etc.

The second is the measurement data attributes, the measurement data attributes need to be reported by the equipment. Generally speaking, that is, data analysis is not available, the need for others to give the system data. This includes properties reported by the device, and may include properties read from a third party’s business system, which are considered to be measured properties;

The third is to analyze task attributes, which need to be further calculated after the data is reported.

For the last task analysis property, there are corresponding tasks to configure and to calculate. In this process, it is equivalent to the loading and configuration of the algorithm: the data is analyzed first, and then the back-end computing engine loads the configured business logic. There are currently three types of analysis task attributes:

The first is a transformation calculation: for a simple example, suppose the creation contains two properties, a and b, and we require that during the process a+b=c, then this is a transformation calculation. The transformation property is required to be real-time, and the data timestamp of the two values AB is the same;

The second is aggregation operation: aggregation is the calculation of a time dimension. Suppose an average temperature of the past five minutes is required. If the equipment reports data every five seconds, then an average of all reported data within five minutes is needed, which is equivalent to an aggregation operation in the time dimension.

The third type is flow computation: flow computation is used in complex situations where logic cannot be expressed by a simple if /else. For example, when an asset reports a number of parameters, the system needs to calculate a result from those parameters and then return the asset, then the flow calculation functions as a calculator. The function of stream computation 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 above is an overall modeling concept, based on these concepts, we can better understand the content of the following.

II. Asset modeling operation

(I) Equipment modeling: SMT production line printing machine equipment

When building a digital asset model for things in the physical world, you must define the asset model before you create the asset. Generally speaking, there are seven kinds of equipment in a production line. Let’s take one of the printing presses as an example to explain how the equipment is modeled.

First, the configuration of the properties. For the printing press, our three attributes are:

Static configuration properties: the ideal printing time of the product and the model of the equipment

Measurement data attributes: printing speed, demoulding speed, printing height

Analyze task attributes: time utilization, performance utilization, pass rate, OEE

The analysis task properties also have the following calculation configuration:

Transition calculation: Calculate time utilization, calculate performance utilization, calculate OEE, and determine temperature state

Aggregate calculation: calculate the actual working hours, calculate the actual working hours, and calculate the pass rate

Stream calculation: Not in use for SMT scenarios

The following figure shows the page of property editing, including static configuration, measurement data, and analysis tasks for your reference.

The following figure shows a complete sample with all the parameters equipped. You can see that there are about 70 attributes in it, and these attributes all simulate some properties of the real industry. All the data in the figure below, including the sample and format, are from the actual production data of the southern factory, so they are relatively real.

With the screenshots below, we hope to illustrate how the analysis task of the printing press is configured. In the case of “conversion calculation”, just read the reported temperature value and make an expression to determine, for example, if the temperature is greater than 25 or less than 35, then it is considered normal. Copy the result of the judgment to the application, and the application can use the result directly.

The following figure shows the configured analysis tasks. As you can see, we currently have 11 analysis tasks configured, including the calculation of capital rate, performance utilization, pass rate, OEE, and various status judgments of the kinds mentioned earlier.

(II) Production line modeling: SMT production line

Having said “equipment modeling”, let’s talk about “production line modeling”.

Production line modeling is actually the same concept as equipment modeling, and the models are similar. However, the static attributes and measurement data attributes are not configured for the time being, because the production line is relatively simple and mainly focuses on the value of OEE, that is, the analysis of task attributes, including the four indicators related to OEE, as well as transformation calculation, aggregate calculation and stream calculation.

Analyze that the configuration of the task attributes is consistent with the device production line, so I will not repeat this explanation.

An example of a device asset allocation diagram for a printing press is shown below:

Next, let’s look at how the production line assets are constructed. As shown in the figure below, production line assets are divided into three layers:

The first layer is the factory (parent assets);

The second layer is the production line (sub-assets);

The third layer is devices (sub-assets).

Production lines and equipment also have models, and the three-tier model forms a “parent-child” number of assets. Assets come from the model and are instantiated from the model. At the same time, when the model is instantiated as an asset, the hierarchy can be specified according to the business scenario and the assets are independent of each other.

The following diagram shows the constructed asset tree. Compared to the logical diagram in the previous diagram, this is an example diagram. The figure shows that an electronics plant has three SMT production lines, and each line has seven SMT devices

(III) OEE related indicator configuration (equipment & production line)

Let’s take a look at how each metric of the device is calculated, as shown in the figure below. Let’s take the “product pass rate” (gray below) as an example.

As mentioned above, pass rate =[quantity of qualified output]/[quantity of output]*100%. The “TS_Sum” in the table represents a sequential sum, which means you can sum the output over a time range, such as the output over a period of five minutes. Other indicators are calculated in a similar way to the pass rate, so they are not described here.

The index calculation process of production line and equipment is similar, but the difference lies in the source of data. Production line data is derived from subassets, not generated by the production line itself. Because the data between the “father and son” of the asset can be quoted from each other, while the production line itself does not report any data.

(4) Asset operation monitoring

After all product creation and property configuration are completed, you can click Publish to publish and run the model. The model itself is a static process at the time of definition and is activated once it is published. According to the task analysis logic defined in the preceding sequence, the system will automatically calculate and obtain real-time results for reporting. All of the data can be seen in the figure below.

In addition to the above data presentation mode, you can also display the data into line charts, heat maps, curves and so on according to the needs of the business. It is easier to analyze the graphic presentation mode and get the results you want. An example figure is shown below.

To experience a production line modeling process, you can go to Huawei Cloud IoT Data Analysis Service (https://www.huaweicloud.com/p…

Third, summary

From the above explanation and introduction, we can draw the following conclusions:

• Real-time and accurate mapping of objects in the physical world into the digital world, the carrier of organizing data & models, is the digital twin of IoT field;
• Organize the data & model around a concrete physical object and define a digital model, that is, the digital modeling process in the IoT field;
• Device digital twin model consists of attribute and task analysis.
• When building an asset model for things in the physical world, you define the model before you create the asset
• The benefits of object-oriented modeling: encapsulation, inheritance, composition, reuse efficiency and extensibility.

Huawei cloud IoT data analysis service is based on the Internet of Things asset model, integrating IoT data integration, cleaning, storage, analysis and visualization, providing one-stop service for IoT data developers, lowering the development threshold, shortening the development cycle, quickly realizing the realization of IoT data value, and making the digital transformation and upgrading of factories “within reach”.

Let’s experience it together

Click on the attention, the first time to understand Huawei cloud fresh technology ~