Neural Recipe Construction Using Dimensionality Reduction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The construction of dose-response relationships in processes like wire-and-cable recipe design and steel rolling often relies on trial and error, failing to effectively utilize accumulated experiential knowledge.

Innovation Solution

A recipe construction system incorporating a dimension reduction module, neural network module, search module, and determining module to process historical recipe information, train neural network parameters, and search for candidate recipes that meet specified physical property criteria, thereby automating the use of experiential knowledge.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If trial and error method is used to construct dose-response relationships, then flexibility in handling different processes is maintained, but time consumption and efficiency deteriorate

Engineering Contradiction:

Improveflexibility in handling different processesVSAvoidtime consumption in recipe construction

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The system performs preliminary action by pre-processing historical recipe information through dimensionality reduction and building trained neural network models in advance. When a new recipe construction task arises, the pre-built model can quickly process the query without requiring time-consuming trial and error, thus resolving the contradiction between maintaining flexibility and reducing time consumption.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system creates a virtual copy of the complex dose-response relationship through a neural network model. This copied model reproduces the behavior of the actual manufacturing process, allowing rapid experimentation and evaluation of different recipes in the virtual space without affecting real production, thereby reducing time consumption while maintaining adaptability.

Inventive Principle:

Principle #26Copying

2Device complexity

If trial and error method is used to construct dose-response relationships, then no complex computational infrastructure is needed, but knowledge accumulation and utilization efficiency deteriorate

Engineering Contradiction:

Improvecomputational infrastructure complexityVSAvoidutilization efficiency of accumulated experiential knowledge

Core Design Contradiction:

Device complexityVSLoss of information

Solution Approach 1:

The system implements feedback by training the neural network model on historical recipe information and using the learned patterns to guide new recipe construction. The model continuously improves by incorporating feedback from past successful recipes, enabling effective knowledge accumulation and utilization without requiring complex computational infrastructure.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system applies parameter changes by transforming the high-dimensional historical recipe data into a lower-dimensional feature space through dimensionality reduction techniques. This transformation preserves the essential information while reducing complexity, allowing the neural network to efficiently learn from accumulated knowledge without requiring excessive computational resources.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If dimensional reduction is applied to historical recipe information, then processing efficiency is improved, but information loss may occur

Engineering Contradiction:

Improveprocessing efficiency of recipe informationVSAvoidcomposition information accuracy

Core Design Contradiction:

ProductivityVSLoss of information

Solution Approach 1:

The system applies parameter changes by transforming the data from high-dimensional composition space to a lower-dimensional feature space using dimensionality reduction techniques. This transformation is designed to preserve the most important information related to physical properties while reducing processing complexity, thus improving productivity with minimal information loss.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The neural network model acts as an intermediary that learns the mapping between the reduced-dimensional features and the actual physical properties. This intermediary compensates for any information loss during dimensionality reduction by learning the underlying relationships, ensuring that the essential composition information is preserved and can be accurately predicted.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11829390B2Recipe construction system, recipe construction method, computer readable recording media with stored programs, and non-transitory computer program product

Publication Date: 2023.11.28 WALSIN LIHWA

AI summary

A recipe construction system, a recipe construction method, a computer readable recording media with stored programs, and a non-transitory computer program product are provided. A dimension reduction module obtains dimension-reduced composition information according to composition information of each piece of historical recipe information and a dimension reduction algorithm. A neural network module obtains a plurality of trained neural network parameters according to the composition information and the dimension reduction composition information. The neural network module obtains dimension-reduced initial composition information according to the trained neural network parameters and initial composition information. A search module searches the historical recipe information for a first quantity of pieces of candidate recipe information. A determining module determines whether physical property information of each piece of candidate recipe information meets a specification, and outputs solution recipe information that meets the specification.