News

Switch to smart factories and preparedness for change

Industry 4.0 has been one of the most popular topics of the past few years, alongside with digitalization, and connecting manufacturing processes and accompanying data. Data are becoming the so-called “new crude oil”.

It is essential how they are obtained, structurally stored and then processed. This must be ensured through an appropriate system infrastructure (either locally in the corporate network or in the cloud) and an information system that is fully tailored to the processes and manufacturing.

It must not be overlooked that often certain segments of the manufacturing process have to be adapted in such a way as not to interrupt the data flow and their long-term storage. This is a fairly complex issue connecting several areas, departments and knowledge bases within a company, and that requires people to be ready for change and the introduction of the so-called digital culture, which means that everybody in the company – regardless of their role – has to be aware of the impact that process digitalization will introduce.

At Stroka Business Group we have been focusing on this issue for years, and most actively since 2018, when we were successful in obtaining funding for our IQ TPM 4.0 project. This is an advanced solution for electronically capturing, sending and storing readouts from different sensors that have been installed on manufacturing machines.

The core of the system is cloud storage for large data sets and the Data Explorer tool for advanced analytics. From the “programming” perspective, the key system elements include:

• Data storage in Azure cloud (Blob storage)
• A module for predictive maintenance through machine learning methods
• Advanced visualization, made using the Microsoft Power BI data visualization tool and an analytics database in the cloud, as part of the Azure Data Explorer service, which supports processing large amounts of data in real time

Figure 1 – Solution architecture from the point where data is captured and sent (left) to the point where it is stored, and in the end sent to the users (web or mobile app).

In order to build the IQ TPM 4.0 system, we needed knowledge from interdisciplinary areas (mechanical engineering, electrical engineering, computer engineering and economy). The solution consists of the hardware part, namely the module and the sensors. The IoT gateway routes the data to cloud storage, sending them at certain intervals to the Azure IoT HUB, where the data is then routed to the correct locations. The hardware part of the solution was implemented by our consortium partner Marovt.

Challenges

During project development we faced many issues that we prefer to call challenges:

1. Learning about work processes in manufacturing and company organization

One of the key factors for successfully establishing the system was learning about the organization and work at Marovt. The first part of the project included workshops at which we learned about their manufacturing processes, toured the factory and examined the machines and work processes up close. We also discussed the limitations related to sensor installation and the start of data capture. Next we reviewed their server architecture and the solutions they use (internal ERP, CRM, a dedicated solution for monitoring the manufacturing process, etc.). This provided us with an insight into the data, the options for setting up connections, and the state of the IT infrastructure at the company.

2. Collaboration during the implementation

As mentioned above, this issue comprises several areas, which made it crucial to establish collaboration between different people – both within our company (planning, development, machine learning and advanced analytics, graphic design and data visualization), as well as between the consortium partners. Regular Friday coordinations were conducted using Microsoft Teams, followed by regular live meetings, where we analyzed the past period and set clear goals for the future.

3. Preparedness for the approach to agile development and prototyping

With some more complex parts of the system we were not able to fully forecast their operations or possible limitations, so we used a fast and agile approach to software development and prototyping. For machine learning, which is used to predict any delays and interruptions in machine operation, this proved to be a highly suitable approach. The construction of the model for monitored machine learning (data collection and labelling) was completed through several repetitions. This allowed us to further improve the success rate of delay classification and detection. The result is a procedure for data processing and machine learning for each individual machine. We concluded that there are too many variables that effect the operation of each individual machine and its behavior (the worker operating the machine, the tools used and their quality, the complexity of the manufactured good, the surrounding temperature, etc.).

4. Accepting that digitalization and evolution to Industry 4.0 does not yield immediate results

In general, the first phase of the digitalization of manufacturing requires network infrastructure in the manufacturing halls. Considering the specifics of manufacturing this alone may be a demanding project. Then work has to begin on several fronts:

• Joint planning with the client and the definition of the Computerized Maintenance Management System (CMMS) so that it is tailored to the existing manufacturing processes in the company, where possible.

• Networking the machines, installing sensors and modules, and establishing data capture and routing for later processing.
• Informing the employees of new developments in the company:

• maintenance service for verifying the operation of the sensors
• the internal IT department for verifying the accuracy of the captured data
• employees who actually use the system

• Joint workshops for the development team and experts in maintenance (machine and electrical engineers) for a detailed analysis of the machines and obtaining the know-how for constructing the model for machine learning, and establishing the advanced analytics (e.g. events on the manufacturing line, realized production, values captured through sensors, etc.).

We only covered a few of the issues that are part of the pilot implementation in such an approach in a manufacturing enterprise system. It required quite a lot of input from both sides. Cooperation is always the key and correct approach to resolving any issues that – let’s be realistic – always crop up when a new information system is implemented in a company.

Results

The outcome of the development and the partnership is the IQ TPM 4.0 solution that provides manufacturing machine networking through its hardware part (both new and older models), using a dedicated module to which sensors connect (the selection of sensors depends on the type of manufacturing, and in our case was the production of usage sensors for electricity, energy, vibration, loudness, moisture and temperature). The data is sent to the cloud through the IoT Gateway, where it is stored in long-term dedicated storage (Blob storage is used for this purpose in Azure).

The Azure Data Explorer analytics service ensures fast queries, as it is tightly connected with the Power BI visualization tool. Using this database, we also prepare the data for the construction of the machine learning model. In our case this approach is used for predicting delays on the machines and their energy use.

In the next article we will focus on the issues related to machine learning, the procedures for predicting maintenance in manufacturing using the case of predicting delays, and with other challenges that we faced during the IQ TPM 4.0 project. We will also detail the effects that such a system has on the company’s operations and the efficiency of asset utilization.