Calcination Throughput Prediction for Plasterboard Process Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating calcination throughput in plasterboard production are unreliable due to difficulties in calculating the time derivative of stucco mass in silos, leading to inaccurate predictions and inefficient production control.
Innovation Solution
A computer-implemented method using a trained gradient boosting tree regressor model to predict calcination throughput in real-time based on calcination parameters, such as mill inlet pressure and calcination temperature, and a configuration optimization process to adjust variables for optimal throughput.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional methods are used to estimate calcination throughput by calculating the time derivative of stucco mass in silos, then the estimation process can be implemented, but the reliability of the estimation is poor due to calculation difficulties and inaccuracies
Solution Approach 1:
The patent replaces the traditional mechanical/mathematical calculation method (time derivative of stucco mass) with a machine learning-based prediction system. The trained model uses multiple calcination parameters as inputs to predict throughput, substituting the unreliable derivative calculation with a data-driven approach that achieves both reliability and precision.
Solution Approach 2:
The patent introduces multiple intermediate parameters (mill inlet pressure, calcination temperature, gas consumption, output aperture position, mill outlet pressure, combustion air temperature, rock feeding rate, recirculation air flow, main fan power, stucco cooler outlet temperature, and mill pressure) as mediators between the calcination process and throughput estimation. These intermediaries provide comprehensive process information to the prediction model, enabling accurate throughput forecasting without direct mass derivative calculation.
2Productivity
If real-time calcination throughput estimation is implemented using a trained model, then production control efficiency is improved, but the system complexity increases due to model training and parameter collection requirements
Solution Approach 1:
The patent performs preliminary action by training the machine learning model offline using historical calcination data before deployment. This pre-training phase captures the complex relationships between process parameters and throughput, allowing the model to provide accurate real-time predictions during operation without requiring complex real-time computations, thus balancing productivity improvement with manageable system complexity.
Solution Approach 2:
The trained model serves multiple functions: it estimates calcination throughput, identifies optimal parameter configurations, and supports production planning and control decisions. This multi-functionality justifies the initial system complexity investment by providing comprehensive production optimization capabilities from a single predictive system.
Data Source
Figure 1
Figure 2
Figure 2a
AI summary
A computer-implemented method is provided which enables a user to utilise historical data to optimise the configuration to a drying environment.