Multi-Level Feature Selection for Dioxin Detection in MSWI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting dioxin (DXN) emission concentration in municipal solid waste incineration processes are inefficient, particularly due to high-dimensional data challenges, collinearity among variables, and the complexity of the dioxin generation and emission mechanisms, leading to suboptimal control and high toxicity risks.
Innovation Solution
A multi-level feature selection method is employed, dividing the incineration process into sub-processes, using correlation coefficients and mutual information to select first-level features, followed by genetic algorithm-based partial least squares (GAPLS) for redundancy processing and statistical screening to obtain third-level features, which are then used to establish a DXN detection model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If online direct detection method is used for DXN emission concentration, then detection speed and responsiveness are improved (lag time reduced to minute/second), but measurement precision and reliability deteriorate due to high-dimensional data challenges and collinearity among variables
Solution Approach 1:
The patent segments the high-dimensional feature space by dividing the incineration process into multiple sub-processes (drying, combustion, cooling, etc.) and selecting key features from each sub-process. This segmentation reduces the dimensionality of input data while preserving the essential information needed for accurate DXN concentration prediction, thereby maintaining measurement precision while enabling faster online detection.
Solution Approach 2:
The patent extracts the most critical features from hundreds of process variables using correlation analysis and mutual information calculations. By identifying and extracting only the features with the strongest relationships to DXN emission concentration, the model achieves accurate predictions with reduced computational complexity, resolving the contradiction between detection speed and precision.
2Measurement precision
If hundreds of process variables are used for DXN detection modeling, then measurement completeness is improved, but device complexity and computational burden increase
Solution Approach 1:
The patent applies partial action by selecting only a subset of the most relevant features from the hundreds of available process variables. Through correlation analysis and mutual information calculations, the method identifies and uses only the critical features needed for accurate DXN prediction, avoiding the computational burden of processing all variables while maintaining detection accuracy.
Solution Approach 2:
The patent transforms the high-dimensional parameter space by changing the representation of process variables into a reduced set of composite features or selected key parameters. This parameter transformation maintains the essential information for DXN detection while significantly reducing model complexity and computational requirements.
3Ease of manufacture
If small sample data is used for modeling, then data acquisition difficulty is reduced, but manufacturing precision and model reliability deteriorate due to insufficient training data
Solution Approach 1:
The patent performs preliminary feature selection and data preprocessing before model training to maximize the information content of small samples. By pre-calculating correlations and mutual information, and selecting the most informative features in advance, the method ensures that each training sample contributes maximally to model learning, thereby achieving good prediction accuracy even with limited data.
Solution Approach 2:
The patent introduces feature selection metrics (correlation coefficients, mutual information values) as intermediaries between the raw process data and the prediction model. These intermediaries transform the small sample data into a more informative representation, enabling the model to learn effective patterns even with limited training samples.
Data Source
AI summary
A method for detecting a dioxin emission concentration of a municipal solid waste incineration process based on multi-level feature selection. A grate furnace-based MSWI process is divided into a plurality of sub-processes. A correlation coefficient value, a mutual information value and a comprehensive evaluation value between each of original input features of the sub-processes and the DXN emission concentration are obtained, thereby obtaining first-level features. The first-level features are selected and statistically processed by adopting a GAPLS-based feature selection algorithm and according to redundancy between different features, thereby obtaining second-level features. Third-level features are obtained according to the first-level features and statistical results of the second-level features. A PLS algorithm-based DXN detection model is established based on model prediction performance and the third-level features. The obtained PLS algorithm-based DXN detection model is applied to detect the DXN emission concentration of the MSWI process.