共用方式為

Manufacturing data solutions and Factory Operations Agent in Azure AI reference architecture

  • 發行項

Global manufacturing companies today face unprecedented complexity in managing their operations across vast and fragmented data estates. They need to collect and analyze data from different hardware and software, in the cloud and at the edge, supported by many ISVs and vendors. In this environment, simply getting a holistic picture of operations is challenging; acting swiftly and effectively on that information is impossible.

Manufacturers know that the modernization of their data estates, combined with the massive potential of AI to produce insights and solve problems, is a critical step to secure the future of their business. However, they require assistance in consolidating and interpreting the data produced by their operations. They also need to effectively, safely, and accurately utilize AI to address their specific needs promptly.

Manufacturing data solutions in Microsoft Fabric gather data from various sources such as MES, machines, sensors, PLCs, and applications. This data is then structured into a manufacturing-specific data model and graphs, which facilitate data-driven applications like AI assistants for operational visibility.

Current industry challenges

Manufacturers wanting to digitally transform their factories often encounter challenges due to the dispersion of data across numerous disparate systems. They also lack specialized technical personnel, like data scientists or AI experts, who can handle the intricacies and large amounts of data integration required to establish a manufacturing data foundation.

In some cases, manufacturers can prove the value of AI in a proof-of-concept but don’t have the technology infrastructure, data foundation, and resources they need to scale AI solutions across the entire factory network.

Benefits of Manufacturing data solutions and Factory Operations Agent in Azure AI

Manufacturing data solutions in Microsoft Fabric and the Factory Operations Agent in Azure AI can help you:

  • Accelerate time to value with prebuilt data ingestion plugins to leading MES providers, factory applications, and IoT management platforms.

  • Build a semantic graph across various manufacturing processes.

  • Use generative AI to bring the power of natural language interfaces  to address specific manufacturing use cases such as quality issue resolution, root cause analysis and more.

  • Accelerate the deployment of generative AI solutions across operations with repeatability using your unified manufacturing data foundation on the Microsoft Cloud for Manufacturing.

Manufacturing data solutions and Factory Operations Agent in Azure AI - Reference Architecture

Microsoft Cloud for Manufacturing provides an overview of common and recommended implementation technology choices, terminology, technology principles, common configuration environments, and composition of applicable services.

Screenshot shows reference architecture for Manufacturing data solutions.

Download a printable PDF of this reference architecture diagram.

The key stages/data flows are as follows:

On-premises data sources

The architecture includes systems that generate Operational Technology (OT) data, such as Programmable Logic Controllers (PLCs), Distributed Control Systems (DCS), or full Manufacturing Execution Systems (MES). It can also include related IT data sources, such as Enterprise Resource Planning (ERP) systems and associated databases.

Data ingestion, normalization, and contextualization

OT and IT data need to be ingested, normalized, and contextualized before it can be used in the cloud. For use in Manufacturing data solutions, one of the formats the data is sent to the cloud is the OPC UA (IEC 62541) open-standard format. OPC UA is the leading industrial interoperability standard for industrial data modeling and secure communication from sensor to cloud. The source device or system supplies this data along with the corresponding metadata. This data is then transmitted to Microsoft Fabric.

At preview, Manufacturing data solutions supports the use of multiple systems for connecting, normalizing, and contextualizing data offered by both Microsoft and ISV partners.

Azure IoT Operations, currently in public preview, is Microsoft’s comprehensive solution to capture asset data, process it at the edge, and send it to the cloud with Azure Arc–enabled services. IoT Operations is used for connectivity with the industrial edge at low latency and high fidelity, built to support powerful cloud analytics, apps, and digital twins. For more information, see Azure IoT Operations and how to deploy its capabilities and services.

Azure Data Gateway is an on-premises component required for accessing on-premises databases behind corporate firewalls in a secure way. It can be deployed in Microsoft Edge gateway systems and configured to send on-premises data to Azure Data Factory, Azure Logic Apps or directly to Microsoft Fabric. It comes with over 50 connectors to various on-premises systems.

Microsoft ISV Partner Litmus, which offers a prebuilt portfolio of 275+ industrial data connectors, teamed with Microsoft to enable seamless delivery of standardized OPC UA data from the edge into Fabric for use in Manufacturing data solutions as well. Litmus also offers a tool, Litmus Edge Manager, to provision, configure and manage connectors, available on the Azure Marketplace.

After data is ingested at the edge, it can be brought to the cloud via several paths:

Data transformation in Manufacturing data solutions

Data that lands into Manufacturing data solutions gets transformed into a refined or silver layer based on the manufacturing industry data model using the ISA95 data protocol.

ISA95 data protocol, an international standard that facilitates the integration of enterprise and control systems in manufacturing. It includes a layered model to represent the hierarchy and relationships within a manufacturing organization. It enables the Manufacturing data solutions knowledge graph to represent factory data in a highly accurate, nuanced, and timely manner. This data in turn provides the foundation for the Factory Operations Agent to deliver valuable insight into factory operations.

Factory Operations Agent in Azure AI

The Factory Operations Agent in Azure AI accelerates partners’ successful adoption of the Microsoft Copilot stack by offering out-of-box templates, connectors, and skills, packaged in an OpenAI standard plugin.

The agent provides:

  • Domain-specific prompt templates, data grounding skills, and connectors.

  • Thorough validation of the agent's outputs, via post-processing, to avoid incorrect information.

  • Responsible AI guardrails to ensure only relevant, safe, responsible responses are given within the manufacturing setting.

Partners can also create and register their own prompt templates, skills, and connectors as desired, to extend the capabilities of the Factory Operations Agent.

Extensibility

Manufacturing data solutions and Factory Operations Agent in Azure AI are designed with extensibility in mind, empowering partners and customers to build further capabilities using the Microsoft ecosystem. For this purpose, Microsoft created Accelerators in Copilot Studio to enable quicker development using agents.

Data applications: The Manufacturing data solutions can power various robust data tools. These tools include Fabric Notebooks, Power BI views, and Power Apps from Microsoft, as well as any ISV data platforms and applications such as Sight Machine’s Data Foundation for Manufacturing, Litmus Edge Platform, etc.

Copilot and Agent experiences: The Factory Operations Agent in Azure AI is a foundation for building powerful generative AI applications. These applications span various user experiences, from building other new solutions in Microsoft Copilot Studio to collaborating in Microsoft Teams and Microsoft 365 Copilots.

Next steps

Get started with Manufacturing data solutions (preview) in Microsoft Fabric.