CN-121009323-B - Industrial solid waste intelligent classification method based on big data

CN121009323BCN 121009323 BCN121009323 BCN 121009323BCN-121009323-B

Abstract

The application provides an industrial solid waste intelligent classification method based on big data, which comprises the steps of obtaining coal ash porosity and steel slag oxide layer data, analyzing surface microscopic characteristics, extracting light scattering intensity and infrared radiation characteristics, determining association expression of environmental parameters and coal ash and steel slag surface characteristics according to nonlinear interaction relation of humidity, temperature and surface characteristics, extracting classification characteristic semantics from the association expression of the environmental parameters and the coal ash and steel slag surface characteristics, constructing a classification characteristic set with reinforced coal ash and steel slag semantics, optimizing the recognition accuracy of the classification characteristics according to nonlinear response characteristics of the coal ash and steel slag extracted from the classification characteristic set, obtaining the optimized classification characteristic set, classifying solid waste types, generating a preliminary classification model, obtaining a classification error sample from the preliminary classification model, carrying out iterative optimization on the preliminary classification model, and outputting a solid waste classification result adapting to environmental change.

Inventors

CAI FENGHUA
Qiu Change
CHEN LIRONG

Assignees

梅州华立风实业有限公司

Dates

Publication Date: 20260508
Application Date: 20250811

Claims (7)

1. An intelligent classification method for industrial solid wastes based on big data is characterized by comprising the following steps: Acquiring the data of the porosity of the fly ash and the oxide layer of the steel slag, analyzing the microscopic characteristics of the surface, and extracting the light scattering intensity and the infrared radiation characteristics; Carrying out infrared spectrum analysis on the light scattering intensity and the infrared radiation characteristics to obtain spectrum data, extracting characteristic peaks of a target wave band from the preprocessed spectrum data, and calculating peak value offset and intensity change caused by environmental parameter change to obtain characteristic offset under the environmental parameter change; Aiming at characteristic offset, a dynamic association knowledge graph of fly ash and steel slag surface characteristics is constructed, wherein nodes of the knowledge graph comprise characteristic offset nodes, humidity nodes, temperature nodes, fly ash surface characteristic nodes and steel slag surface characteristic nodes extracted from historical data, the weight value of a connecting edge between the nodes is determined through a Pearson correlation coefficient, if the absolute value of the correlation coefficient is larger than a preset threshold value, a connecting edge is established between the corresponding nodes, the dynamic association knowledge graph is used as a data and structure basic training multi-node association model, node characteristics of the knowledge graph are subjected to aggregation operation through a graph packing network and node representation are updated to obtain node embedding, a sliding window method is adopted for dividing time sequence data, dynamic change trend of the characteristic offset is captured, the node embedding and the dynamic change trend are input into a long-short-term memory network, and a characteristic offset prediction sequence is output; Determining the association expression of the environmental parameters and the surface characteristics of the fly ash and the steel slag according to the nonlinear interaction relation of humidity, temperature and the surface characteristics; Extracting classification characteristic semantics from the association expression of the environmental parameters and the surface characteristics of the fly ash and the steel slag, and constructing a classification characteristic set with reinforced fly ash and steel slag semantics; optimizing the recognition accuracy of the classification features according to the nonlinear response characteristics of the fly ash and the steel slag extracted from the classification feature set to obtain an optimized classification feature set, classifying the solid waste types, and generating a preliminary classification model; And obtaining a classification error sample from the preliminary classification model, performing iterative optimization on the preliminary classification model to obtain a final classification model, and outputting a solid waste classification result adapting to environmental changes.
2. The intelligent classification method for industrial solid waste based on big data according to claim 1, wherein the steps of obtaining the data of the porosity of the fly ash and the oxide layer of the steel slag, analyzing the microscopic characteristics of the surface, extracting the light scattering intensity and the infrared radiation characteristics, and comprises the following steps: The method comprises the steps of obtaining a two-dimensional cross-section image of fly ash particles, identifying a pore area and a solid area, calculating the proportion of the pore area to the total area to obtain porosity, carrying out energy spectrum analysis on the surface of a steel slag sample, determining the thickness of an oxide layer according to the content distribution of oxygen elements, irradiating the surfaces of the fly ash and the steel slag sample with laser, collecting scattered light signals through a photoelectric detector array, recording the scattered light intensity of each angle, calculating an infrared radiation intensity value according to the corresponding relation between the intensity of an absorption peak and the thickness of the oxide layer, extracting the maximum intensity value of the scattering angle in a preset range as a light scattering intensity characteristic, and extracting the average radiation intensity value of the preset wavelength range in the infrared absorption spectrum as an infrared radiation characteristic.
3. The intelligent classification method for industrial solid waste based on big data according to claim 1, wherein the infrared spectrum analysis is performed on the light scattering intensity and the infrared radiation characteristic to obtain spectrum data, the characteristic peak value of the target wave band is extracted from the preprocessed spectrum data, the peak value offset and the intensity change caused by the environmental parameter change are calculated, and the characteristic offset under the environmental parameter change is obtained, and the method comprises the following steps: The method comprises the steps of obtaining light scattering intensity and infrared radiation characteristics, carrying out frequency domain conversion on the light scattering intensity and the infrared radiation characteristics to obtain spectrum data, carrying out denoising treatment on the spectrum data by adopting a moving average filter, carrying out baseline correction by adopting a polynomial fitting method to obtain a spectrum curve after baseline correction, identifying characteristic peaks in a target wave band from the spectrum curve after baseline correction by adopting a second derivative method, and recording wavelength positions and peak intensities of the characteristic peaks.
4. The intelligent classification method for industrial solid waste based on big data according to claim 1, wherein the determining the association representation of the environmental parameter and the surface characteristics of the fly ash and the steel slag according to the nonlinear interaction relation of humidity, temperature and the surface characteristics comprises the following steps: the method comprises the steps of extracting a humidity main effect value, a temperature main effect value and a humidity temperature interaction effect value in the nonlinear interaction relation, calculating the contribution ratio of each effect to the surface characteristic change through normalization processing to obtain an environment parameter coupling strength index, constructing a three-dimensional tensor, decomposing the tensor to obtain a characteristic vector, determining the influence mode of an environment condition combination on the surface characteristic, calculating a probability density function of the influence mode through a kernel density estimation method, dividing an environment parameter space into different response areas to generate an environment parameter partition mapping relation, calculating the representative value and frequency weight of each partition in the partition mapping relation to obtain a comprehensive response index, and constructing the association expression.
5. The intelligent classification method for industrial solid wastes based on big data according to claim 1, wherein the steps of extracting classification feature semantics from the associated representation of environmental parameters and the surface characteristics of the fly ash and the steel slag, and constructing a classification feature set with enhanced semantics of the fly ash and the steel slag comprise: Analyzing element values of the association expression, converting the element values into descriptive feature words to generate initial classification feature semantics, mapping parameter combinations of the association expression into semantic vectors, calculating cosine similarity among nodes to obtain a semantic similarity matrix, dividing nodes in the semantic similarity matrix through a semantic segmentation method based on a threshold value to generate semantic categories and labels, identifying node categories containing coal ash and steel slag features to form feature subsets, adding high-frequency feature words and basic feature words, and outputting the classification feature sets.
6. The intelligent classification method for industrial solid waste based on big data according to claim 1, wherein the optimizing the identification accuracy of the classification features according to the nonlinear response characteristics of the fly ash and the steel slag extracted from the classification feature set, obtaining the optimized classification feature set, classifying the solid waste types, and generating the preliminary classification model comprises the following steps: Extracting nonlinear response characteristics in the classification feature set, calculating feature differentiation, removing redundant features to obtain a simplified feature set, adjusting feature weights of the simplified feature set through an information gain method to generate weighted feature vectors, constructing an optimized classification feature set, constructing a decision tree through a random forest algorithm, calculating classification consistency ratio, judging stability, and outputting a preliminary classification model containing classification rules.
7. The intelligent classification method for industrial solid waste based on big data according to claim 1, wherein the steps of obtaining a classification error sample from the preliminary classification model, performing iterative optimization on the preliminary classification model to obtain a final classification model, and outputting a solid waste classification result adapting to environmental changes comprise: Extracting error samples of the preliminary classification model and environmental parameter data thereof to generate an error sample set, calculating residual errors through a gradient lifting decision tree algorithm, iteratively optimizing the preliminary classification model to obtain a final classification model, inputting real-time environmental parameters, calculating classification reliability indexes, marking samples to be updated, adjusting node weights of a knowledge graph through an online gradient descent method, updating association strength values and edge connection relations, and outputting solid waste classification results.

Description

Industrial solid waste intelligent classification method based on big data Technical Field The invention relates to the technical field of information, in particular to an industrial solid waste intelligent classification method based on big data. Background Industrial solid waste treatment is a key field for promoting green manufacturing and recycling economy, and is important in reducing environmental pollution, improving resource utilization efficiency and achieving sustainable development targets. With the deep industrialization, solid waste is various and complex in components, and improper treatment can cause soil and water pollution and even ecological crisis, so that a high-efficiency intelligent classification technology is needed to support accurate resource utilization. Currently, solid waste classification methods rely on single physical or chemical detection means, such as classification by particle size distribution or chemical component analysis, but these methods often show instability when facing complex environmental changes, especially in dynamic environment interaction, it is difficult to capture nonlinear changes of solid waste surface characteristics, resulting in degradation of classification accuracy. For example, it is difficult in the prior art to accurately distinguish between solid waste types having abrupt changes in surface characteristics due to changes in ambient humidity, which limits the adaptability and reliability of the classification system. In this field, a core challenge is how to effectively capture the dynamic correlation between the microscopic properties of solid waste surfaces and environmental factors. Microscopic characteristics of solid waste surfaces, such as porosity or oxide layer state, can vary nonlinearly due to environmental humidity, temperature, etc. For example, the surface porosity of fly ash can cause significant changes in light scattering intensity under humidity fluctuations, which directly affect the accuracy of optical detection. The abrupt change in light scattering intensity further leads to instability in classification feature extraction, as it is difficult for existing methods to dynamically correlate these nonlinear responses with environmental parameters. For example, in a real industrial scenario, when the humidity of the process plant changes rapidly from low to high, the optical signal of the fly ash may change suddenly, and the existing classification system cannot adjust the feature extraction strategy in real time, so that the classification error rate increases. Therefore, how to construct a knowledge system capable of dynamically associating environmental parameters with solid waste surface characteristics, and improving the recognition capability and stability of a classification system to nonlinear response by optimizing a feature extraction process becomes a key problem. Disclosure of Invention The invention provides an industrial solid waste intelligent classification method based on big data, which mainly comprises the following steps: Acquiring the data of the porosity of the fly ash and the oxide layer of the steel slag, analyzing the microscopic characteristics of the surface, and extracting the light scattering intensity and the infrared radiation characteristics; Carrying out infrared spectrum analysis on the light scattering intensity and the infrared radiation characteristics to obtain spectrum data, extracting characteristic peaks of a target wave band from the preprocessed spectrum data, and calculating peak value offset and intensity change caused by environmental parameter change to obtain characteristic offset under the environmental parameter change; Aiming at the characteristic offset, a dynamic association knowledge graph of the surface characteristics of the fly ash and the steel slag is constructed, a multi-node association model is trained, a characteristic offset dynamic change trend captured through a time sequence analysis method is input into the multi-node association model, and a nonlinear interaction relation of humidity, temperature and the surface characteristics is output; Determining the association expression of the environmental parameters and the surface characteristics of the fly ash and the steel slag according to the nonlinear interaction relation of humidity, temperature and the surface characteristics; Extracting classification characteristic semantics from the association expression of the environmental parameters and the surface characteristics of the fly ash and the steel slag, and constructing a classification characteristic set with reinforced fly ash and steel slag semantics; optimizing the recognition accuracy of the classification features according to the nonlinear response characteristics of the fly ash and the steel slag extracted from the classification feature set to obtain an optimized classification feature set, classifying the solid waste types, and generating a prelimina