Interface Firmware Pipeline for Local ML Signal Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional industrial automation systems face increased communication latency and load due to the transmission of process data to controllers and edge devices, leading to higher costs and complexity when implementing machine learning applications, especially when field devices are not from the same provider.

Innovation Solution

Implement machine learning applications on interface devices by dividing them into logical components connected in a pipeline, generating code blocks using a translator, and deploying these blocks on the interface device, creating virtual ports for data transmission, thereby reducing the need for extensive data transfer and simplifying engineering.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If machine learning applications are run on controllers or edge devices, then AI processing capability is improved, but communication latency and load on interface devices increase

Engineering Contradiction:
ImproveAI processing capabilityVSAvoidcommunication latency
Core Design Contradiction:
Extent of automationVSLoss of time

Solution Approach 1:

The patent segments the machine learning application into multiple logical components (data collection, preprocessing, model inference, postprocessing) that can be distributed across different devices. The interface device executes the data collection and preprocessing components locally, while only transmitting processed results to controllers, thereby reducing communication load and latency while maintaining AI processing capability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a new dimension of execution by enabling machine learning applications to run directly on interface devices, which traditionally only performed data transmission. This dimensional shift from centralized controller-based execution to distributed interface device execution reduces communication overhead and latency.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Loss of energy

If machine learning applications are executed on field devices, then communication load is reduced, but engineering complexity increases due to multiple AI frameworks

Engineering Contradiction:
Improvecommunication loadVSAvoidengineering complexity
Core Design Contradiction:
Loss of energyVSDevice complexity

Solution Approach 1:

The patent creates a universal execution environment on interface devices that can handle multiple machine learning frameworks and models through a common architecture. The logical component pipeline provides a framework-agnostic interface, allowing the same application structure to work with different AI models without requiring separate engineering efforts for each field device type.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent introduces an intermediary layer of logical components that acts as a mediator between the diverse field devices and the machine learning models. This standardized intermediate representation simplifies engineering by providing a uniform interface, eliminating the need to directly integrate with multiple different AI frameworks on each field device.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Extent of automation

If process data is transmitted to controllers for AI processing, then centralized control is maintained, but communication path load increases leading to higher costs

Engineering Contradiction:
Improvecentralized controlVSAvoiddata transmission volume
Core Design Contradiction:
Extent of automationVSQuantity of substance

Solution Approach 1:

The patent extracts the data preprocessing and feature extraction functions from the centralized controller and places them on the interface device. Only the essential processed data and model inference results are transmitted to the controller, significantly reducing the volume of data transmitted while maintaining centralized control over the overall system behavior.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent performs preliminary data processing, filtering, and feature extraction at the interface device before transmission to the controller. This preliminary action reduces the amount of raw data that needs to be transmitted, decreasing communication path load and associated costs while preserving the ability to perform centralized analysis when needed.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3809214B1Method for configuring an interface device and interface device thereby
Publication Date: 2025.07.30 SIEMENS AG
  • EP3809214B1 patent drawingFigure 1
  • EP3809214B1 patent drawingFigure 2
  • EP3809214B1 patent drawingFigure 3

AI summary

The present disclosure describes a method for configuring an interface device connected to a control device and a field device. The method comprises receiving a first machine learning application comprising a plurality of logical components connected in a pipeline, the first machine learning application for analyzing a signal from the field device using a first machine learning model, generating a plurality of code blocks using a translator based on the plurality of logical components of the first machine learning application, connecting the plurality of code blocks in accordance with the pipeline of the first machine learning application, to generate a first output from the signal from the field device, and deploying the connected code blocks on a firmware of the interface device including creating a virtual port connectable to the control device, the virtual port for transmitting the first output to the control device.