CN-121859260-B - Method and system for predicting wind pressure drop of turning platform of turning ramp by improving BP algorithm

CN121859260BCN 121859260 BCN121859260 BCN 121859260BCN-121859260-B

Abstract

The invention relates to the technical field of mine ventilation, in particular to a method and a system for predicting the wind pressure drop of a turning platform of a turning ramp of an improved BP algorithm, which comprise the steps of obtaining an initial data set of ventilation of the turning platform of the turning ramp so as to obtain a standardized data set of the initial data set; the method comprises the steps of determining ventilation local resistance parameters according to a standardized data set, establishing a ventilation local resistance equation of a turn-back ramp turning platform, establishing a loss function by combining a neural network structure and the ventilation local resistance equation, training and evaluating the neural network structure by utilizing the loss function to obtain a candidate wind pressure drop prediction model, and determining a final turn-back ramp turning platform wind pressure drop prediction model according to an evaluation result of the candidate wind pressure drop prediction model to realize prediction analysis of the turn-back ramp turning platform wind pressure drop. According to the method, physical constraints are introduced into BP algorithm to construct a prediction model and combine with a model strategy, so that the method can adapt to the situation of different mine turning ramp to realize the prediction of the wind pressure drop of the complex turning ramp turning platform.

Inventors

WANG XIAODONG
JIA JUN
CHEN XIANGRU
WANG JUNJIE
LI ZHONGFU
YAN SHUO
LI LU

Assignees

昆明理工大学

Dates

Publication Date: 20260508
Application Date: 20260316

Claims (7)

1. The method for predicting the wind pressure drop of the turning platform of the turning ramp by improving the BP algorithm is characterized by comprising the following steps of: Acquiring an initial data set of ventilation of a turning-back type ramp turning platform, and setting a data standardization processing model to process the initial data set to obtain a standardized data set of ventilation of the turning-back type ramp turning platform; determining ventilation local resistance parameters according to the standardized data set, and establishing a ventilation local resistance equation of the turn-back ramp turning platform based on the ventilation local resistance parameters; Building a neural network structure, building a loss function by combining the neural network structure and the ventilation local resistance equation, and training and evaluating the neural network structure by utilizing the loss function based on the standardized data set to obtain a candidate wind pressure drop prediction model; Determining a final reentrant ramp turning platform wind pressure drop prediction model according to an evaluation result of the candidate wind pressure drop prediction model, and realizing predictive analysis of reentrant ramp turning platform wind pressure drop based on the final reentrant ramp turning platform wind pressure drop prediction model Said determining ventilation localized resistance parameters from said normalized data set comprises; Constructing a ventilation local resistance parameter expression; determining a ventilation local resistance parameter from the ventilation local resistance parameter expression and the normalized data set; the ventilation local resistance parameter expression satisfies the following relationship: , Wherein, the In order to support the degree of protrusion, For the distance from the outermost edge of the supporting structure to the center of the tunnel section, The distance from the wall surface to the circle center is originally excavated for the roadway; , Wherein, the As a coefficient of the relative height difference, For the longitudinal height difference of the turning platform, The roadway width; The establishing a ventilation local resistance equation of the turn-back ramp turning platform based on the ventilation local resistance parameter comprises the following steps of; Obtaining a support protrusion degree and a relative height difference coefficient by using the ventilation local resistance parameter expression; Fitting and determining a basic local resistance coefficient and each influence coefficient by adopting a multiple linear regression method based on original historical data in an initial data set; Establishing a ventilation local resistance equation of the turn-back type ramp turning platform based on the support protrusion degree, the relative height difference coefficient and the influence coefficient; the ventilation local resistance equation satisfies the following relationship; , Wherein, the For a local wind pressure drop reference calculated based on a physical equation, As a basis for the local drag coefficient, Is that Is used for the influence coefficient of (a), In order to support the degree of protrusion, Is that Is used for the influence coefficient of (a), As a coefficient of the relative height difference, Is that And Is used to determine the co-influence coefficient of (c), In order to achieve a gas flow density, Is the airflow velocity; the building of the neural network structure and the building of the loss function by combining the neural network structure and the ventilation local resistance equation comprise the following steps: determining the layer number and the node number of the neural network, introducing a BP algorithm nonlinear activation function, constructing a neural network topology structure, and initializing the weight and the bias parameters of the neural network; introducing a physical constraint equation, and analyzing an airflow continuity equation residual error and an air pressure drop residual error according to the physical constraint equation; establishing a loss function by combining the neural network structure, the airflow continuity equation residual error, the wind pressure drop residual error and the ventilation local resistance equation; the loss function satisfies the following relationship: , Wherein, the As a function of the loss, Predicted values for neural networks And true observations The mean square error between the two, In order to balance the weight of the weight, For the purpose of reducing the residual error weight of the wind pressure, As a residual of the physical consistency of the wind pressure drop, For the continuous residual weighting of the air flow, Is a continuous residual of the gas flow.
2. The method for predicting wind pressure drop of a turn-back ramp turning platform for improving BP algorithm according to claim 1, wherein the obtaining the initial data set of the turn-back ramp turning platform ventilation comprises: Acquiring a roadway three-dimensional model, and acquiring three-dimensional space data of a turn-back type ramp turning platform based on the roadway three-dimensional model; collecting ventilation information of the folding ramp turning platform, and obtaining ventilation parameter data of the folding ramp turning platform in a period of time according to the ventilation information; and integrating the three-dimensional space data and the ventilation parameter data to obtain an initial data set of ventilation of the folding ramp turning platform.
3. The method for predicting the wind pressure drop of a turn-back ramp turning platform by improving a BP algorithm according to claim 2, wherein the setting a data standardization processing model to process the initial data set to obtain a standardized data set of the turn-back ramp turning platform ventilation comprises: Analyzing the mean value and standard deviation of the initial data set based on the initial data set; Establishing a data standardization processing model according to the mean value and the standard deviation; Processing the initial data set by using the data standardization processing model to obtain a standardized data set for ventilation of the folding ramp turning platform; the data standardization processing model meets the following relations: , Wherein, the In order to normalize the data it is, As the original data in the initial data set, Is the mean value of the two values, Is the standard deviation.
4. The method for predicting wind pressure drop of a turn-back ramp turning platform with improved BP algorithm according to claim 1, wherein the training and evaluating the neural network structure based on the standardized data set and using the loss function to obtain a candidate wind pressure drop prediction model comprises: Building a training model aimed at minimizing the loss function; Adopting an iterative optimization algorithm, carrying out iterative solution on the training model based on the standardized data set, and updating the weight and the bias parameters of the neural network until convergence conditions are met; Setting a verification evaluation function to evaluate the performance of the model in the iterative process; analyzing the predicted performance effect of the wind pressure drop prediction model according to the verification evaluation function; And determining a candidate wind pressure drop prediction model based on the minimized result and the predicted performance effect.
5. The method for predicting wind pressure drop of a turn-back ramp turning platform based on the BP algorithm of claim 4, wherein determining the final turn-back ramp turning platform wind pressure drop prediction model based on the evaluation result of the candidate wind pressure drop prediction model comprises: Extracting different input feature combinations from the standardized data set; Selecting a wind pressure drop prediction model based on the input feature combination; the wind pressure drop prediction model satisfies the following relationship: , Wherein, the In order to be a wind pressure drop prediction model, In order to input the data it is possible, For geometric feature vectors extracted based on the roadway three-dimensional model, For the width of the roadway, For the longitudinal height difference of the turning platform, In order to support the degree of protrusion, For the flow rate of the air stream, Is the air flow density, in particular The vector data obtained after the feature extraction of the three-dimensional model at least comprises one or more of turning radius, turning angle, roadway center line length and area reduction rate, and the geometric features can quantitatively represent the complex influence of the space morphology of the turning platform on the airflow field.
6. The method for predicting wind pressure drop of a turn-back ramp turning platform based on the improved BP algorithm of claim 5, wherein said selecting a wind pressure drop prediction model based on said input feature combination comprises: Based on different input characteristic combinations, respectively training to obtain a plurality of candidate wind pressure drop prediction models, and constructing a candidate model set; Constructing a hierarchical decision tree model, and taking the prediction performance index and the physical constraint residual value output by the verification evaluation function as discrimination nodes of the hierarchical decision tree model; And screening and analyzing the candidate model set by utilizing the layered decision tree model, and determining an output performance optimal model as the final reentrant ramp turning platform wind pressure drop prediction model.
7. A system for predicting the wind pressure drop of a turn-around ramp turning platform for improving a BP algorithm, wherein the system comprises a processor, an input device, an output device and a memory, the processor, the input device, the output device and the memory being interconnected, wherein the memory is configured to store a computer program comprising program instructions, the processor being configured to invoke the program instructions to perform the method for predicting the wind pressure drop of a turn-around ramp turning platform for improving the BP algorithm according to any one of claims 1 to 6.

Description

Method and system for predicting wind pressure drop of turning platform of turning ramp by improving BP algorithm Technical Field The invention relates to the technical field of mine ventilation, in particular to a method and a system for predicting the wind pressure drop of a turning platform of a turning ramp by improving a BP algorithm. Background The foldback ramp is a core channel of a mine in the mine ventilation system, and the ventilation local resistance characteristic of the foldback ramp has great influence on the overall energy consumption of the mine ventilation system, so that the foldback ramp is an important link of optimization, energy saving and consumption reduction of the ventilation system. At present, when the wind pressure drop of a folding ramp turning platform is calculated, an empirical formula such as a local resistance coefficient method is mainly relied on, a related manual is consulted to obtain a local resistance coefficient, and parameters such as wind speed, air density and the like are combined to calculate the local resistance coefficient, however, the traditional method has a plurality of limitations in practical application, namely complex nonlinear association exists among geometric characteristics such as turning angle, curvature radius, roadway width ratio, section size and the like, the actual wind pressure drop condition is difficult to accurately reflect only according to the empirical formula calculation, besides the geometric parameters, the supporting prominence, relative height difference coefficient, air density, air flow disturbance and other environmental and structural factors can influence the wind pressure drop, the calculation difficulty and complexity are further increased, the empirical formula is summarized under specific working conditions, the adaptability to the complex and changeable actual working conditions is poor, the completely matched empirical formula is difficult to find in practical application to accurately calculate, the situation is limited by on-site conditions, the timeliness of a measuring result is low, the requirements of dynamic adjustment and intelligent ventilation system control cannot be met in time, and the intelligent ventilation requirement cannot be well adapted to intelligent ventilation development. Therefore, a multidimensional prediction method based on BP algorithm is required to promote intelligent development of mine ventilation, and accurate prediction of the local resistance of mine ventilation is further realized. Disclosure of Invention Aiming at the defects of the existing method and the defects of practical application, in order to realize efficient and accurate prediction of the wind pressure of the turning platform of the folding ramp, the multi-dimensional characteristics are effectively fused, and the continuous optimization and practical application of the intelligent model are realized. The invention provides a method for predicting wind pressure drop of a turning platform of a turning ramp with an improved BP algorithm, which comprises the following steps of obtaining an initial data set of the ventilation of the turning platform of the turning ramp, setting a data standard processing model to process the initial data set to obtain a standardized data set of the ventilation of the turning platform of the turning ramp, determining a ventilation local resistance parameter according to the standardized data set, establishing a ventilation local resistance equation of the turning platform of the turning ramp based on the ventilation local resistance parameter, establishing a neural network structure, combining the neural network structure with the ventilation local resistance equation, establishing a loss function, training and evaluating the neural network structure based on the standardized data set and utilizing the loss function to obtain a candidate wind pressure drop prediction model, determining a final turning platform wind pressure drop prediction model according to an evaluation result of the candidate wind pressure drop prediction model, and realizing prediction analysis of the wind pressure drop of the turning platform of the turning ramp based on the final turning platform wind pressure drop prediction model. The method comprises the steps of data preprocessing, model construction, training, and the like, can fully consider the complex characteristics of the ventilation system of the folding ramp turning platform, and finally obtains the optimal wind pressure drop prediction model through the combination of physical constraint and data driving and the multi-model screening strategy, and can more accurately capture the influence of various factors in the ventilation system on wind pressure drop, thereby remarkably improving the accuracy of the prediction result. Optionally, the acquiring of the initial data set of the ventilation of the folding ramp turning platform comprises acquiring a roadwa