Industrial Process Quality Prediction Across Multiple Processing Stations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing industrial processes, such as mining, face challenges in optimizing product quality prediction due to siloed operational strategies and the computational demands of creating a material flow digital twin, making it difficult to reliably estimate final product quality and other parameters like energy consumption and CO2 emissions in real-time.
Innovation Solution
A computer-implemented method that trains or retracts a prediction model using geological and processing data from multiple stations within an industrial process, enabling real-time or near real-time prediction of product quality through machine learning techniques like regression, support vector machines, or deep learning, specifically using neural networks, to improve operational planning and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a material flow digital twin is created to enable connectivity and homogenization of data across processing stations, then the ability to predict final product quality improves, but the computational demand becomes highly demanding and reliability decreases for large-scale industrial processes
Solution Approach 1:
The patent segments the complex material flow digital twin into multiple independent prediction models, each responsible for a specific processing station or operational aspect. This segmentation reduces the computational complexity of each individual model while maintaining the overall predictive capability across the entire industrial process by combining results from multiple specialized models.
Solution Approach 2:
The patent applies partial action by implementing prediction models only for the most critical processing stations or parameters that have the greatest impact on final product quality. This selective approach reduces computational demand while still achieving reliable predictions by focusing resources on the most influential factors in the material flow.
2Ease of operation
If siloed operational strategies are used for each separate processing station, then operational simplicity is maintained, but conflicting strategies result in suboptimal overall process efficiency
Solution Approach 1:
The patent implements universal prediction models that can be applied across multiple processing stations and different operational contexts. These models serve multiple functions by predicting various product quality parameters and can be used for both monitoring and optimization purposes, replacing the need for separate siloed strategies while maintaining ease of operation through a unified approach.
Solution Approach 2:
The patent introduces feedback mechanisms where prediction results from individual processing station models are fed back into the overall process control system. This feedback enables coordination between previously siloed operations, allowing the system to optimize overall productivity by adjusting operational parameters based on predicted outcomes from multiple stations while keeping individual station operations simple.
Data Source
AI summary
A computer-implemented method is provided. The method includes receiving geological data of a material and processing data referring to a plurality of processing stations of an industrial process for manufacturing a product from the material; receiving, for the geological data and the processing data, corresponding product quality data of the manufactured product; and training or retraining a prediction model for the industrial process to determine predicted product quality data for the geological data and the processing data