For Immediate Release posted on November 9, 2021
The latest version of Simio Software is packed with diverse features to assist Simio users with getting the best from their simulation and digital twin applications. Simio’s ability to train neural networks and support the import and use of ONNX files is the defining feature of the latest version.
The majority of the new features are designed to meet our customers’ requests. Simio is excited to announce these new customer-driven features that include:
- The ability to export error logs. Simio users can now export large data sets of errors to third party software applications. Exporting error logs to other applications empower users to simplify searches on extensive error logs.
- The option of importing data for a data table directly from an Excel Table object and the ability to import data for a data table directly from an Excel PivotTable object, not just an Excel Worksheet.
- A new Portal Users Guide that documents a user’s experience in Simio’s Portal Product. Simio users can now access model results and provide access to results to other users. Users can also copy and download plans to Desktop.
Simio is also delighted to announce a significant milestone with this latest update. Simio Software is the first discrete event simulation software to have embedded support for neural networks. The embedded support allows users to use neural networks for inference in a model’s logic, capture training data, and train neural network models to achieve accurate results.
Simio supports the Open Neural Network Exchange (ONNX) drive to promote the use of open artificial intelligence standards in simulation and digital twin modeling. ONNX provides definitions of built-in operators and standard types to enable AI developers to use models within diverse frameworks and compilers including the Simio Software.
Simio empowers users to create feedforward neural networks directly in Simio without any coding required. Simio enables the training of the developed neural network and the choice of exporting the neural network as an ONNX file. Simio also creates an enabling virtual environment for importing ONNX files of trained neural networks. Developers can evaluate the performance of imported ONNX files, as well as, use them to improve the inference capabilities of simulation and digital twin models built with Simio.
Simio’s support for ONNX files expands the development of open AI standards and their utilization and adoption rate across diverse industrial sectors. Customers who are compliant with ONNX standards will gain access to a powerful, intelligent object-based simulation and digital twin modeling software that facilitates interoperability. Free interoperability ensures Simio users can build and train neural networks with Simio, export them as ONNX files, and deploy them on any stack or application.
According to Eric Howard, Vice President of Marketing, “The neural network features in Simio are designed to support the use of neural networks for decision-making within the model, as well as, to record synthetic training data.” He continued by stating that “Simio users can now develop, train neural networks and evaluate ONNX-based applications or machine learning algorithms to understand their impact before implementation.”
For more information about Simio Software 14.230, you can examine the release notes here.