CN-121999903-A - Method, device and system for predicting gas content based on multi-layer perceptron model

CN121999903ACN 121999903 ACN121999903 ACN 121999903ACN-121999903-A

Abstract

The invention relates to the field of coal mines, in particular to a method, a device and a system for predicting gas content based on a multi-layer perceptron model. The method comprises the steps of inputting gas related data, preprocessing the input data, training and verifying the model by utilizing a multi-layer perceptron model through a training data set and a verification data set until a preset termination condition is met, inputting the characteristic data to be predicted, and outputting a gas content prediction result based on trained model parameters. According to the invention, high-precision prediction of the gas content is realized through the multi-layer perceptron model, the prediction accuracy is remarkably improved, meanwhile, the calculation efficiency is greatly optimized, and reliable and real-time decision support is provided for coal mine safety monitoring.

Inventors

CUI FAN
SUN ZHENYAN
ZHANG LEI
YUAN ZHONGFENG
ZHANG QINGHUA
ZHANG SHUAI
ZHANG GUIXIN
WANG RAN
CHEN GONGHUA
MO LIANHONG
Ren yanchuan
XU SHURONG

Assignees

中国矿业大学(北京)
贵州安晟能源有限公司
中煤科工集团重庆研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260127

Claims (8)

1. A method for predicting gas content based on a multi-layer perceptron model is characterized by comprising the following steps: s1, inputting gas related data, wherein the data comprise characteristic data and tag data; s2, preprocessing input data; S3, model training and verification are carried out through a training data set and a verification data set by utilizing the multi-layer perceptron model until a preset termination condition is met; S4, inputting feature data to be predicted; S5, outputting a gas content prediction result based on the trained model parameters.
2. The method for predicting gas content based on the multi-layer perceptron model of claim 1, wherein the characteristic data comprises at least one of coal seam thickness, coal seam burial depth and gas emission amount, and the tag data is gas content data.
3. The method for predicting gas content based on a multi-layer perceptron model of claim 1, wherein said preprocessing comprises: carrying out standardized processing on the data to eliminate dimension differences; the normalized data is proportionally partitioned into a training data set and a validation data set.
4. The method for predicting gas content based on a multi-layer perceptron model of claim 1, wherein said model training and validation comprises: Updating model parameters through cyclic training until a loss value in gas content verification reaches a set threshold value; and storing the final multi-layer perceptron model training parameters.
5. The method for predicting gas content based on a multi-layer perceptron model of claim 1, wherein the type of characteristic data to be predicted is consistent with the type of characteristic data entered in step S1.
6. The method for predicting gas content based on the multi-layer perceptron model of claim 1, wherein said step S5 comprises: carrying out standardized processing on input characteristic data; leading in trained model parameters for prediction; And predicting and outputting a predicted gas content value.
7. The device of the method for predicting the gas content based on the multi-layer perceptron model is characterized in that: the apparatus for performing the method of predicting gas content as claimed in any one of claims 1-6, comprising: the data acquisition module ‌ is used for acquiring data of the thickness of the coal bed, the burial depth of the coal bed and the gas emission and corresponding gas content data; The data storage module ‌ is used for storing the acquired original data and the standardized processed data; the data processing module ‌ is used for carrying out standardization processing on the original data and dividing the processed data into a training data set and a verification data set; The model training module ‌ is configured to perform cyclic training and verification by using the training data set and the verification data set based on the multi-layer perceptron model, and update model parameters until a preset loss threshold condition is satisfied; the parameter storage module ‌ is used for storing the model parameters of the multi-layer perceptron after training; the prediction calculation module ‌ is used for carrying out gas content prediction calculation according to the input characteristic data to be predicted and the stored model parameters; And the output module ‌ is used for outputting a predicted gas content result.
8. A system for predicting gas content based on a multi-layer perceptron model, comprising: At least one processor ‌ for performing data processing and model calculation tasks; a memory ‌ communicatively coupled to the at least one processor, storing instructions executable by the processor, the instructions being executable by the at least one processor and causing the at least one processor to perform the method of predicting gas content as recited in any one of claims 1-6; The input device ‌ is used for gas related data; And the output device ‌ is used for outputting the predicted gas content result.

Description

Method, device and system for predicting gas content based on multi-layer perceptron model Technical Field The invention relates to the field of coal mines, in particular to a method, a device and a system for predicting gas content based on a multi-layer perceptron model. Background Gas disasters have historically been a core challenge in restricting coal mine safety production, and its prediction method has been a hotspot in research in the field. Traditional gas prediction relies mainly on the field experience of engineering technicians, combined with static mathematical models to make inferences. Although this approach plays an important role at the time, there are significant limitations on the accuracy and precision of the prediction. In recent years, with the breakthrough of the Internet of things and sensor technology, advanced means such as distributed optical fiber monitoring and laser spectrum detection are applied, so that dynamic real-time acquisition of key parameters such as gas concentration, pressure and flow is realized, and the accuracy of monitoring data is remarkably improved. However, gas prediction still faces two major core problems, namely that firstly, the unique adsorption gas storage mechanism of gas interacts with diversified coal mining processes to increase the complexity of the construction of a prediction model, and secondly, the prior art system is difficult to synchronously realize early warning of disasters and effective prediction of the whole working face range. The method is based on the fact that the gas content is closely related to various geological and engineering factors such as coal seam thickness, burial depth, gas emission quantity and the like, and the factors are directly and indirectly related to each other in a complicated and changeable manner, so that the method is a deep cause of difficulty and complexity in gas prediction work. It is worth noting that this situation is coming as machine learning technology is becoming more mature. Machine learning is particularly good at dealing with complex regression prediction problems under multi-feature, multi-constraint conditions. In view of the fact that gas prediction itself is a typical scenario highly correlated with multi-source information and multiple conditions, it is possible to implement fast and accurate gas prediction by using machine learning technology, and represents an important development direction in the future of the field. Disclosure of Invention The invention aims to overcome the defects of the prior art and provides a method, a device and a system for predicting gas content based on a multi-layer perceptron model. In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: a method for predicting gas content based on a multi-layer perceptron model comprises the following steps: s1, inputting gas related data, wherein the data comprise characteristic data and tag data; s2, preprocessing input data; S3, model training and verification are carried out through a training data set and a verification data set by utilizing the multi-layer perceptron model until a preset termination condition is met; S4, inputting feature data to be predicted; S5, outputting a gas content prediction result based on the trained model parameters. Further, the characteristic data comprise at least one of data related to gas content such as coal seam thickness, coal seam burial depth, gas emission amount and the like, and the tag data are gas content data. Further, the preprocessing includes: carrying out standardization treatment on the data by using a standard deviation standardization method to eliminate dimension differences; the normalized data is proportionally partitioned into a training data set and a validation data set. Further, the model training and validation includes: Updating model parameters through cyclic training until a loss value in gas content verification reaches a set threshold value; and storing the final multi-layer perceptron model training parameters. Further, the type of the feature data to be predicted is consistent with the type of the feature data input in step S1. Further, the step S5 includes: carrying out standardization processing on the input characteristic data by using a standard deviation standardization method; leading in trained model parameters for prediction; And predicting and outputting a predicted gas content value. The invention also provides a device for predicting the gas content based on the multilayer perceptron model, which is used for the method for predicting the gas content and comprises the following steps: the data acquisition module ‌ is used for acquiring data of the thickness of the coal bed, the burial depth of the coal bed and the gas emission and corresponding gas content data; The data storage module ‌ is used for storing the acquired original data and the standardized processed data; the data processing module ‌ is