Background

Along with the profound digitalization trend in manufacturing and warehousing industries, the concept of digital twin has gained increasing attention. Digital twin is a virtual representation of a real-world system such as a physical site, asset, system, or process. It combines data sets across multiple sources, from design data to Internet of Things data. Feeding these data into digital twin software built on top of a real-time 3D platform, it simulates the behavior of a system, exactly as it would perform in the real world and enables decision-making. Digital twin has become the primary method for planning, analyzing, and optimizing the business layout and processes. Simulation software packages are widely accepted tools to implement digital twin platforms. Although these packages can capture the dynamics of business processes in detail, they fail to utilize AI to its’ full potential and process the data to reveal actionable insights. In addition, applying AI to the simulation packages requires different skillsets with expert knowledge which is challenging for many enterprises. Therefore, new solutions need to be developed to address the current limitations of such a digital system.

Technology Overview

Researchers at Northeastern University have developed a digital twin model, called TwinAI. TwinAI links simulation platforms with an additional analytical programing language, which extends the simulation capability and greatly enhances its application. In this invention, an Application Programming Interface (API) is developed that connects a simulation software with an open-source programming language such as Python to perform a wide range of advanced data analytics such as experimentation, optimization, data mining, artificial intelligence machine learning, reinforcement learning, statistical analysis, and data visualization. TwinAI provides real-time information since it uses a live stream of data from embedded platforms and devices which enables the Digital Twin to be adaptable and reflect changes in the system’s state. The data-driven approach of TwinAI allows the simulation model to be automatically created or updated without user intervention. This eliminates the need for reconfiguration and makes the experimentation faster. TwinAI uses prime AI technology such as Deep Reinforcement Learning for future self-learning, self-optimizing, and other self-driving abilities to autonomously optimize operations and solve errors in the processes. The TwinAI model can solve problems in various fields, including healthcare, manufacturing, warehousing, and pharma.

Benefits

Real-time optimization and management
Data-driven
Autonomous
Capability to handle a large amount of data with different types (date, numerical, string, etc.)
Independent performance from the Algorithm

Applications

Process automatization, routing robots, and layout design in warehousing
Schedule jobs, improve material handling, and resource utilization in manufacturing
Control the size distribution of the particles during crystallization and production in the pharmaceutical process

Opportunity

Testing and validating MVP with pilot partner(s)
Connecting with prospective customers

Digital Twin Model Supporting Decision in Operation and Process

IP Status

Seeking