AgensGraph Use Case #11: Insulation Material Prescription System
Establishment of Graph Model-based Machine Learning Data Analysis
Insulated cables transfer electrical signals or power from one device to the other. Cables need to be tailored for a specific purpose. For example, these insulated cables can be used for building wires for lighting, power and control circuits and the high voltage power cables transmit alternate and direct current power. These high-voltage or high-temperature applications have strict requirements in regards to mechanical, chemical and electrical stability.
The insulated cables used for specific needs consist of different chemicals. Making the precise insulated cables in a suitable environment has always been a challenge for many cable manufacturing companies. Company L was one of the biggest industries in the field, but even the most experienced researchers encountered trials and errors when trying to satisfy the needs of the end-users.
Challenges of an Insulated Cable Manufacturing Company
The researchers of Company L faced the following challenges when making the insulation material requested by their customers.
Having no proper system to store and manage data, the work progress depended heavily on the researchers’ experience. They had to rely on their instincts built upon the experience of countless tests. If one of those researchers is to retire, that would mean the disappearance of knowledge the said researcher acquired over the years.
Company L wasn’t the only cable manufacturing company in the market so they had to lower the price of every individual material if they wanted to stay in the competition. In some cases when a needed material was not available or produced anymore, the researchers had to find new, cost-efficient material as a replacement.
Company L experienced difficulty in sharing R&D data results internally because the departments within the company all had different standards in managing data. This means that the data produced through various experiments were also not standardized. Performing hundreds of operations were followed by an almost equal amount of errors.
Managing a long list of compound data into tables is an inefficient method when it comes to analyzing the relationships between the materials and resulting outcomes.
As a result, Company L sought a graph DBMS optimized for relationship-centric data storage and implemented AgensGraph.
The Data-driven Insulation Material Prescription System
The model of Insulation Material Prescription System runs with graph DBMS and machine learning. Company L chose AgensGraph to reduce data processing time and further advance the automation of the analysis model.
AgensGraph is a multi-model graph DBMS integrated from a relational database system. All the data collected and tested by Company L are migrated to AgensGraph without complication. AgensGraph’s flexible schema makes it easy to add and store relational data and serves a critical role in visualizing them.
Once the data are transferred, the next step is to extract the codes of the stored data using AgensGraph’s algorithms. These codes are then calculated with the analysis tools like Python to draw the data of ingredients and characteristic features. The machine learning model collects all data and predicts a design of experiments (DoE) with a high success rate of producing insulation material with specific characteristics.
AgensGraph’s Precision Reduces Production Time & Money
The difficulty of finding a DoE with a high success rate was as difficult as finding a needle in the haystack. But applying AgensGraph as a medium to store and show data in a hierarchical graph form has made DoE deducting process significantly faster than before.
With the newly implemented data analysis model within the Insulation Material Prescription System, Company L is now able to extract the exact ingredients with required attributes.
For example, if Company L struggled to produce a frost-resistant rubber material in the past, the analysis model will now be able to predict which composition of materials will be durable for a cold environment. Since there are greatly reduced tests than before, there would be less waste of materials, which leads to saving more money and time. With this new technology, Company L’s competitiveness among other insulated cable companies will significantly grow.
If the project with Company L expands further, Bitnine will further establish a big data platform for more projects, which will lead to the management of more data from other researchers and departments.
▶ Download the full use case of AgensGraph: Insulation Material Prescription System.
※ AgensGraph Community Edition is licensed under Apache 2.0, so please feel free to use and send us an inquiry/idea about our products to email@example.com or our Github