Batch Process Similarity Analysis for Golden Batch Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Batch processes in process engineering face challenges in identifying and reproducing optimal production conditions due to variations in process parameters, equipment conditions, and quality of raw materials, leading to inefficiencies and delays in problem detection and resolution.
Innovation Solution
A data-model-based method for determining the similarity of batch process steps using anomaly detection to optimize production processes by comparing current iterations with historic reference phases, allowing for automated analysis and cause identification without complex physical modeling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex physical modeling is used to analyze batch process variations, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent replaces complex physical modeling (mechanical/systematic approach) with data-driven machine learning models. Instead of using complex physical equations to model batch process variations, the system uses historical process data to train models that automatically identify patterns and deviations, thereby reducing modeling complexity while maintaining or improving measurement precision.
Solution Approach 2:
The patent creates a virtual copy of the batch process through data modeling. By copying historical process data and using it to train machine learning models, the system replicates the behavior of the physical process without needing to understand or model the underlying physics, thus simplifying the analytical approach while preserving accuracy.
2Manufacturing precision
If extensive data analysis of batch iterations is performed to identify golden batches, then manufacturing precision improves, but loss of time increases
Solution Approach 1:
The patent performs preliminary actions by continuously training machine learning models on historical batch data in advance. The models are pre-trained to recognize patterns of optimal performance (golden batches) and deviations, so when a new batch is executed, the system can quickly compare real-time data against the pre-trained models and immediately identify problems without requiring extensive post-batch analysis.
Solution Approach 2:
The system implements continuous feedback by comparing real-time batch data against the trained models and automatically identifying deviations. This feedback mechanism enables rapid detection of problems during batch execution, allowing operators to take corrective actions sooner rather than waiting for extensive post-batch analysis, thus reducing time loss while maintaining precision.
3Ease of operation
If manual analysis of batch processes is used, then ease of operation is maintained, but productivity decreases
Solution Approach 1:
The patent implements self-service by automating the batch analysis process through machine learning models. The system automatically compares batch data against trained models, identifies deviations, and suggests corrective actions without requiring manual intervention. This maintains ease of operation for users while dramatically improving productivity through automated, rapid analysis of batch processes.
Solution Approach 2:
The patent replaces manual analysis methods with automated machine learning-based analysis. Instead of operators manually examining batch data and identifying patterns, the system uses trained models to automatically detect deviations and optimize processes, thereby increasing productivity while keeping the user interface simple and easy to operate.
Data Source
AI summary
A method and system for improving a production process in a technical installation in which a process-engineering process having a process is implemented, where data records, which characterize an iteration of a process step and which have values of process variables, are recorded in a time-dependent manner and stored in a data memory, and, for each process step, the data records of an iteration are selected as a test phase, and the data records of at least one further iteration are selected as a reference phase, where similarity between the data records of the test phase and at least one reference phase is subsequently determined in pairs, where a calculated phase similarity measure is used to optimize the process such that multiple applications in process optimization, such as determining a “golden batch” and a root cause analysis of faulty batches by correlating with metadata, can be performed.