Multivariate solid waste proportion optimization method based on coupling of matrix operation and machine learning

Patent metadata

Field	Value
Publication number	US-2025190885-A1
Application number	US-202418665044-A
Country	US
Kind code	A1
Filing date	May 15, 2024
Priority date	Dec 12, 2023
Publication date	Jun 12, 2025
Grant date	—

Abstract

The disclosure provides a multivariate solid waste proportion optimization method based on a coupling of matrix operation and machine learning, including following steps: obtaining proportioning feedstock, where the proportioning feedstock include solid waste and boiler/industrial kiln working feedstock; using a machine learning model embedded optimization algorithm to obtain a set of characteristic parameter requirements of the proportioning feedstock; according to the set of characteristic parameter requirements of the proportioning feedstock, optimizing the blending ratio of the solid wastes and the boiler/industrial kiln working feedstock to obtain an optimized blending ratio of the solid wastes and the boiler/industrial kiln working feedstock; according to the optimized blending ratio of the solid wastes and the boiler/industrial kiln working feedstock, obtaining an optimized set of solid waste characteristic parameter requirements; completing solid waste proportion based on the optimized set of solid waste characteristic parameter requirements and the set of solid waste characteristic parameters.

First claim

What is claimed is:

1 . A multivariate solid waste proportion optimization method based on a coupling of matrix operation and machine learning, comprising:
obtaining initial characteristic parameters and proportioning feedstock, wherein the proportioning feedstock comprise solid waste and boiler/industrial kiln working feedstock; using a machine learning model embedded optimization algorithm to obtain a set of characteristic parameter requirements of the proportioning feedstock; obtaining an optimal blending ratio of the solid wastes and the boiler/industrial kiln working feedstock according to a mathematical relationship between the set of the characteristic parameter requirements of the proportioning feedstock, characteristic parameter requirements of the solid wastes and characteristic parameter requirements of the boiler/industrial kiln working feedstock; obtaining an optimal set of solid waste characteristic parameter requirements according to the optimal blending ratio of the solid wastes and the boiler/industrial kiln working feedstock; and completing solid waste proportion based on the optimal set of the solid waste characteristic parameter requirements and a set of solid waste characteristic parameters.

2 . The multivariate solid waste proportion optimization method based on the coupling of matrix operation and machine learning according to claim 1 , wherein the initial characteristic parameters comprise:
the set of solid waste characteristic parameters, a set of the characteristic parameter requirements of the solid wastes and a set of characteristic parameter requirements of the boiler/industrial kiln working feedstock.

3 . The multivariate solid waste proportion optimization method based on the coupling of matrix operation and machine learning according to claim 1 , wherein obtaining the set of the characteristic parameter requirements of the proportioning feedstock comprises:
constructing an initial machine learning model; optimizing the initial machine learning model to obtain a final machine learning model; based on the final machine learning model, combining with the optimization algorithm, and obtaining the set of the characteristic parameter requirements of the proportioning feedstock.

4 . The multivariate solid waste proportion optimization method based on the coupling of matrix operation and machine learning according to claim 3 , wherein steps of obtaining the final machine learning model comprise:
carrying out data preprocessing on the proportioning feedstock, training the initial machine learning model by using preprocessed data, and then carrying out a cross-validation on the models to obtain optimal hyperparameters of an algorithm; and obtaining the final machine learning model based on obtained optimal hyperparameters and by combining input parameters and output performance indexes of the initial machine learning model.

5 . The multivariate solid waste proportion optimization method based on the coupling of matrix operation and machine learning according to claim 3 , wherein based on the final machine learning model and combining with the optimization algorithm, obtaining the set of the characteristic parameter requirements of the proportioning feedstock comprises:
obtaining a preset search space according to boundary conditions of the input parameters; and according to the final machine learning model, using the optimization algorithm and a heuristic search strategy, automatically searching in the preset search space, and obtaining the set of the characteristic parameter requirements of the proportioning feedstock through iteration.

6 . The multivariate solid waste proportion optimization method based on the coupling of matrix operation and machine learning according to claim 2 , wherein obtaining an optimized blending ratio comprises:
constructing a blending ratio formula according to the set of characteristic parameter requirements of the proportioning feedstock, the set of the characteristic parameter requirements of the solid wastes and the set of characteristic parameter requirements of the boiler/industrial kiln working feedstock; substituting a maximum value and a minimum value of each subset in the set of the characteristic parameter requirements of the solid wastes and a corresponding subset in the set of characteristic parameter requirements of the boiler/industrial kiln working feedstock into the blending ratio formula, and optimizing the set of the characteristic parameter requirements of the solid wastes and the set of characteristic parameter requirements of the boiler/industrial kiln working feedstock to obtain the optimized blending ratio.

7 . The multivariate solid waste proportion optimization method based on the coupling of matrix operation and machine learning according to claim 6 , wherein the blending ratio formula is:

XW
+

(

1
-
X

)

⁢
C

=
Z

,

X represents a proportion of total solid waste in proportioning feedstock, and an unit is %; W represents the set of the characteristic parameter requirements of the solid wastes; C represents the set of characteristic parameter requirements of the boiler/industrial kiln working feedstock; and Z represents the set of characteristic parameter requirements of the proportioning feedstock.

8 . The multivariate solid waste proportion optimization method based on the coupling of matrix operation and machine learning according to claim 6 , wherein based on the optimal set of the optimal solid waste characteristic parameter requirements and the set of the solid waste characteristic parameters, completing the solid waste proportion comprises:
obtaining a set of solid waste blending ratio according to the optimal set of the solid waste characteristic parameter requirements and the set of the solid waste characteristic parameters; obtaining an allowable solid waste blending ratio range according to the maximum and minimum values of the set of solid waste blending ratio and the optimal set of the solid waste characteristic parameter requirements, and completing the solid waste proportion.

Assignees

• Univ Tianjin Commerce
• Univ Tianjin

Inventors

• Chen Guanyi
• Sun Yunan
• Tao Junyu
• Wu Wenzhu
• Wu Shuang
• Chen Guandong
• Chen Chao

Classifications

• G06Q10/04 Primary

  Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" (market predictions or forecasting for commercial activities G06Q30/0202) · CPC title

• G06Q10/30 Primary

  Administration of product recycling or disposal · CPC title