Product Performance Prediction Using Device Outlier Correction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional product performance prediction methods in industrial production lines face inaccuracies due to unknown differences between real and simulated environments, leading to discrepancies between predicted and actual product performance, especially during device exceptions.

Innovation Solution

A product performance prediction modeling method that utilizes device outlier data and machine learning to establish a mapping relationship between device outlier data and product performance, enabling accurate predictions through a system comprising sensors, a cloud platform, and a computer device for real-time prediction and model updates.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If simulation software is used to perform simple performance verification, then the prediction process is simple and fast, but the prediction accuracy deteriorates due to unknown differences between real and simulated environments

Engineering Contradiction:
Improveprediction speedVSAvoidprediction accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent introduces an intermediary correction mechanism that bridges the gap between simulation and reality. A correction model is trained using actual production data to learn the differences between simulated and real device behaviors. This correction model then adjusts simulation results to match real-world outcomes, maintaining fast prediction speeds while significantly improving accuracy.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent dynamically adjusts simulation parameters based on learned deviations from real data. By continuously updating correction parameters using actual production data, the system adapts the simulation model to reflect real-world conditions more accurately, resolving the contradiction between using simple simulation (fast) and achieving high accuracy.

Inventive Principle:
Principle #35Parameter changes

2Device complexity

If traditional simulation methods are used, then device complexity is low and processing is simple, but reliability of prediction deteriorates during device exceptions

Engineering Contradiction:
Improvesystem simplicityVSAvoidprediction reliability during exceptions
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent performs preliminary actions by pre-training correction models using historical production data before actual prediction needs arise. The system learns from past device exceptions and normal operations, building a knowledge base that can be quickly applied during new prediction scenarios. This preliminary learning ensures reliability during device exceptions without requiring complex real-time processing.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements feedback mechanisms where actual production data continuously feeds back into the correction model. The system monitors prediction accuracy and uses real-world outcomes to refine and update the correction parameters, creating a self-improving system that becomes more reliable over time while maintaining relatively simple architecture.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3811241B1Product performance prediction modeling method and apparatus, computer device, computer-readable storage medium, and product performance prediction method and prediction system
Publication Date: 2023.08.30 SIEMENS AG
  • EP3811241B1 patent drawingFigure 1
  • EP3811241B1 patent drawingFigure 2
  • EP3811241B1 patent drawingFigure 3~4

AI summary

Provided are a product performance prediction modeling method and apparatus, a product performance prediction method, a product performance prediction system, a computer device, and a storage medium. The product performance prediction modeling method includes: acquiring first sample data, wherein the first sample data includes device outlier data generated in a process of manufacturing a product by a device; acquiring a production line configuration simulation parameter of a production line where the device is located, and product information of the product manufactured by the production line; selecting a simulation model to perform simulation test on the performance of the product, so as to obtain product performance simulation data; and inputting the device outlier data, the production line configuration simulation parameter, the product information and the product performance simulation data into a machine learning model to perform machine learning training, so as to obtain a product performance prediction model. The foregoing product performance prediction modeling method and apparatus, product performance prediction method, product performance prediction system, computer device and storage medium can accurately predict product performances during device exception.