Advanced adjustment method for required air volume of mine local ventilator

Abstract

The invention relates to a mine local fan required air volume advanced adjustment method, which comprises the following steps of: 1, establishing a physical model of a local fan air duct by using numerical simulation software, performing simulation operation, and generating local fan air duct opening air volume simulation data under various working conditions; 2, carrying out data screening processing on the generated air volume simulation data of the air duct opening of the local fan under each working condition; 3, training a neural network according to the screened data, and performing performance prediction on the air volume of the air duct opening by using verification set data in the trained neural network; and 4, target parameters are set according to the predicted air volume value, and the output air volume is adjusted in advance through a closed-loop PID feedback control mechanism. According to the advanced adjustment method for the needed air volume of the mine local ventilator, the prediction efficiency and precision of the air volume of an air duct opening of the ventilator are greatly improved.

Description

Technical Field

[0001] The invention relates to the technical field of coal mine ventilation, in particular to a method for pre-regulating required air volume of a local ventilator in a mine. Background Art

[0002] As the key driving force of the underground ventilation system, the core function of the local ventilator is to efficiently deliver fresh air to the excavation working face through the wind tube, thereby maintaining a good circulation of underground air and significantly reducing the risk of accidents caused by gas accumulation. Although the current underground ventilators in coal mines have adopted variable frequency technology, they still tend to operate at a relatively fixed frequency in actual operation. This operating mode often leads to unnecessary power consumption and fails to maximize the utilization of energy. In view of the complexity and variability of the underground environment, including the real-time fluctuations of key factors such as gas concentration, dust concentration, temperature and humidity, it is particularly important to accurately predict the air volume required by the local ventilator at the next moment. By introducing advanced prediction technology and combining these environmental factors for dynamic analysis, it is possible to achieve instant prediction and adjustment of the required air volume of the ventilator, and then drive the ventilator to perform variable frequency speed regulation to meet the actual demand with the most optimized air volume supply, thereby significantly saving energy consumption while ensuring safe production. In contrast, the traditional ventilator air volume adjustment method has certain limitations in response speed and prediction accuracy, and it is difficult to adapt to the rapid changes in the underground environment. Therefore, developing and implementing a system that can accurately predict the air volume required by local fans is not only an important innovation in coal mine ventilation technology, but also a key measure to improve the safety level of coal mine production and promote the achievement of energy conservation and emission reduction goals. It has far-reaching practical significance and application value.

[0003] In current technical practice, in order to monitor the safety conditions in mines, it is common to deploy a variety of sensors and personnel positioning systems at various ventilation demand points. These devices can instantly collect real-time data on gas concentration, coal dust generation and other key environmental parameters. Subsequently, the system will calculate the current air demand at each air-using location through the corresponding algorithm based on the collected parameter information. However, it is worth noting that this calculation process itself takes a certain amount of time, resulting in a certain time delay in the final air demand data relative to the real-time situation. This delay means that before the accurate air demand is calculated, there will be a time window, the so-called "gray period", during which the actual air volume in the mine may not match the actual demand, which may cause safety hazards and increase the risk of mine face operations. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art.

[0005] To this end, the present invention proposes a method for pre-adjusting the required air volume of a local ventilation fan in a mine, which greatly improves the prediction efficiency and accuracy of the air volume at the fan duct outlet.

[0006] The method for pre-adjusting the required air volume of a local ventilation fan in a mine according to an embodiment of the present invention comprises the following steps:

[0007] Step 1: Use numerical simulation to obtain simulation results: Use numerical simulation software to establish a physical model of the local fan duct, perform simulation operations, and generate simulation data of the air volume of the local fan duct outlet under various working conditions with different duct materials; wherein the parameters included in the physical model include air volume, gas concentration, oxygen concentration, temperature, and humidity;

[0008] The wind tube is circular, and the average velocity of the fluid in the circular wind tube is calculated as follows:

[0009]

[0010] The symbols in formula (1) mean:

[0011] v represents the average velocity of the circular wind tube fluid, in m / s;

[0012] α represents the kinematic viscosity coefficient of the fluid;

[0013] R e represents the Reynolds number;

[0014] r o Indicates the diameter of the circular air duct, in meters.

[0015] Step 2, data processing: data screening and processing are performed on the generated local ventilator air volume simulation data of the air duct outlet under each working condition, and the default data is inserted;

[0016] Step 3: Establish a neural network model, use the neural network model to predict the air volume at the air duct outlet, and the program automatically selects the best: train the neural network based on the screened data, and use the validation set data in the trained neural network to predict the performance of the air volume at the air duct outlet, and then get the predicted air volume value;

[0017] The neural network model includes an input layer, a hidden layer, and an output layer. The gas concentration, oxygen concentration, temperature, and humidity are used as the input layer, and the position of the average wind speed circle from the top of the tunnel is used as the output layer. m ] is the input sample, [Y1, Y2, Y3, …Y n ] is the output sample,

[0018] The output of each unit in the hidden layer is calculated as follows:

[0019] Oj =f(ΣW ij X i -θ j ) (2)

[0020] The symbols in formula (2) mean:

[0021] O j Represents the output results of each unit in the hidden layer;

[0022] f represents a nonlinear function;

[0023] W ij Represents the weight matrix of the hidden layer;

[0024] θ j represents the neuron threshold;

[0025] X i Represents an input sample;

[0026] The output result of the output layer is calculated as follows:

[0027] Y j =f(∑T JK O j -θ k ) (3)

[0028] The symbols in formula (3) mean:

[0029] Y j Represents the output result of the output layer;

[0030] f represents a nonlinear function;

[0031] T JK Represents the weight matrix of the output layer;

[0032] O j Represents the output results of each unit in the hidden layer;

[0033] θ k represents a neuron threshold;

[0034] Step 4: Use the optimal program to pre-adjust the air volume at the air duct outlet: set the target parameters according to the predicted air volume value, and use the closed-loop PID feedback control mechanism to pre-adjust the output air volume;

[0035] The predicted air volume value is calculated as follows:

[0036] U(t)=Kp*e(t)+Ki*∫e(t)dt+Kd*de(t) / dt (4)

[0037] The symbols in formula (4) mean:

[0038] U(t) represents the output of the controller at time t;

[0039] e(t) represents the difference between the system output and the expected value, that is, the deviation;

[0040] Kp, Ki, and Kd are the three key parameters of the PID controller, where Kp is the proportional gain, Ki is the integral gain, and Kd is the differential gain;

[0041] ∫e(t)dt represents the deviation integral from time 0 to time t;

[0042] de(t) / dt represents the derivative of the deviation;

[0043] The calculation formula of e(t) is as follows:

[0044] e(t)=y(t)-r(t) (5)

[0045] The symbols in formula (5) mean:

[0046] y(t) is the system output;

[0047] r(t) is the expected value.

[0048] The beneficial effect of the present invention is that by adopting typical operating conditions based on artificial settings as input to train the neural network model, the model can efficiently predict and cover the wind volume data under a wide range of operating conditions, shorten the prediction cycle, enhance the simulation efficiency, and achieve a leap in measurement accuracy.

[0049] According to one embodiment of the present invention, in step 1, the physical model uses gas concentration, temperature, humidity and dust concentration as input variables, and the air volume required by the local ventilation fan at a future time is used as the output variable of the physical model.

[0050] According to one embodiment of the present invention, in step 3, the processed data is screened to train a neural network, and the optimal weights and optimal thresholds in the neural network are determined through the training process.

[0051] According to one embodiment of the present invention, in step 4, the closed-loop PID feedback control mechanism automatically adjusts the wind speed of the ventilator based on the difference between the actual value and the predicted value of the current air volume, thereby achieving advance regulation of the air volume.

[0052] According to one embodiment of the present invention, the calculation formula of the air volume is as follows:

[0053] Q=S×V (6)

[0054] The symbols in formula (6) mean:

[0055] Q represents air volume;

[0056] S represents the cross-sectional area of ​​the fan;

[0057] V represents the fan speed.

[0058] According to one embodiment of the present invention, in step 3, the neural network integrates a BP neural network, a DBN neural network and a DNN neural network.

[0059] According to one embodiment of the present invention, the filtered data is used to train three models, namely, BP neural network, DBN neural network and DNN neural network. Each neural network can learn the relationship between the required air volume of the fan and the influencing factors from the filtered data.

[0060] According to one embodiment of the present invention, in step 1, the numerical simulation software is CFD numerical simulation software.

[0061] According to one embodiment of the present invention, an evaluation index is used to compare the effects of three algorithms: BP neural network, DBN neural network and DNN neural network.

[0062] According to an embodiment of the present invention, the evaluation index is the degree of fitting or the root mean square error.

[0063] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description and the drawings.

[0064] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0066] Figure 1 is a neural network diagram,topological structure of the present invention;

[0067] Figure 2 It is the overall flow chart of the present invention;

[0068] Figure 3 It is the algorithm PID closed-loop control diagram of the present invention. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] The following is a detailed description of the method for pre-adjusting the required air volume of a local ventilation fan in a mine according to an embodiment of the present invention with reference to the accompanying drawings.

[0071] See Figure 2 The method for pre-adjusting the required air volume of the local ventilation fan in a mine includes the following steps:

[0072] Step 1: Use numerical simulation to obtain simulation results: Use numerical simulation software to establish a physical model of the local fan duct, perform simulation operations, and generate simulation data of the air volume of the local fan duct outlet under various working conditions with different duct materials; wherein the parameters included in the physical model include air volume, gas concentration, oxygen concentration, temperature, and humidity;

[0073] The wind tube is circular, and the average velocity of the fluid in the circular wind tube is calculated as follows:

[0074]

[0075] The symbols in formula (1) mean:

[0076] v represents the average velocity of the circular wind tube fluid, in m / s (meters per second);

[0077] α represents the kinematic viscosity coefficient of the fluid;

[0078] R e represents the Reynolds number;

[0079] r o Indicates the diameter of the circular air duct, in m (meter);

[0080] Step 2, data processing: data screening and processing are performed on the generated local ventilator air volume simulation data of the air duct outlet under each working condition, and the default data is inserted;

[0081] Step 3: Establish a neural network model and use it to predict the air volume at the air duct outlet. The program automatically selects the best: train the neural network based on the filtered data, and use the validation set data in the trained neural network to predict the air volume at the air duct outlet, and then get the predicted air volume value; Figure 1 shown.

[0082] The present invention uses the four environmental parameters of gas concentration, temperature, humidity and dust concentration as input variables of the neural network, and uses the air volume required by the local ventilator at the next moment as the output. The established air volume prediction model is as follows: Figure 1 shown.

[0083] The neural network model includes an input layer, a hidden layer, and an output layer. The gas concentration, oxygen concentration, temperature, and humidity are used as the input layer, and the position of the average wind speed circle from the top of the tunnel is used as the output layer. m ] is the input sample, [Y1, Y2, Y3, …Y n ] is the output sample,

[0084] The output of each unit in the hidden layer is calculated as follows:

[0085] O j =f(∑W ij X i -θ j ) (2)

[0086] The symbols in formula (2) mean:

[0087] O j Represents the output results of each unit in the hidden layer;

[0088] f represents a nonlinear function;

[0089] W ij Represents the weight matrix of the hidden layer;

[0090] θ j represents the neuron threshold;

[0091] X i Represents an input sample;

[0092] The output result of the output layer is calculated as follows:

[0093] Y j =f(∑T JK O j -θ k ) (3)

[0094] The symbols in formula (3) mean:

[0095] Y j Represents the output result of the output layer;

[0096] f represents a nonlinear function;

[0097] T JK Represents the weight matrix of the output layer;

[0098] O j Represents the output results of each unit in the hidden layer;

[0099] θ k represents a neuron threshold;

[0100] Step 4: Use the optimal program to pre-adjust the air volume at the air duct outlet: set the target parameters according to the predicted air volume value, and use the closed-loop PID feedback control mechanism to pre-adjust the output air volume; Figure 3 shown.

[0101] The present invention uses the required air volume at the next moment as the set air volume of the local ventilator, introduces a closed-loop feedback air volume, and regulates the required air volume in advance. The established required air volume advance closed-loop regulation is as follows: Figure 3 shown.

[0102] The predicted air volume value is calculated as follows:

[0103] U(t)=Kp*e(t)+Ki*∫e(t)dt+Kd*de(t) / dt (4)

[0104] The symbols in formula (4) mean:

[0105] U(t) represents the output of the controller at time t;

[0106] e(t) represents the difference between the system output and the expected value, that is, the deviation;

[0107] Kp, Ki, and Kd are the three key parameters of the PID controller, where Kp is the proportional gain, Ki is the integral gain, and Kd is the differential gain;

[0108] ∫e(t)dt represents the deviation integral from time 0 to time t;

[0109] de(t) / dt represents the derivative of the deviation;

[0110] The calculation formula of e(t) is as follows:

[0111] e(t)=y(t)-r(t) (5)

[0112] The symbols in formula (5) mean:

[0113] y(t) is the system output;

[0114] r(t) is the expected value.

[0115] In the first step, the physical model uses gas concentration, temperature, humidity and dust concentration as input variables, and the air volume required by the local ventilation fan at a future time is used as the output variable of the physical model.

[0116] In the third step, the processed data is filtered to train the neural network, and the optimal weight and threshold in the neural network are determined through the training process. When a certain weight makes the error between the actual required air volume and the predicted air volume less than 5%, the weight is determined to be the optimal weight; when a certain threshold makes the predicted air volume error less than 5%, the threshold is determined to be the optimal threshold.

[0117] Before training begins, weights and thresholds are usually initialized randomly. The initialization range can be set based on experience. For example, in a BP neural network, the threshold value range is usually [-1, 1], and the weight value range can be a random number between 0 and 1. Through multiple iterations of the training process, the weights and thresholds are continuously adjusted to make the network output closer to the actual value.

[0118] In step 4, the closed-loop PID feedback control mechanism automatically adjusts the wind speed of the fan based on the difference between the actual value and the predicted value of the current air volume to achieve advance regulation of the air volume.

[0119] The air volume calculation formula is as follows:

[0120] Q=S×V (6)

[0121] The symbols in formula (6) mean:

[0122] Q represents air volume;

[0123] S represents the cross-sectional area of ​​the fan;

[0124] V represents the fan speed.

[0125] In step 3, the neural network integration includes BP neural network, DBN neural network and DNN neural network.

[0126] The filtered data are used to train three models: BP neural network, DBN neural network and DNN neural network. Each neural network can learn the relationship between the required air volume of the fan and the influencing factors from the filtered data. The three neural network models, BP neural network, DBN neural network and DNN neural network, are used for screening at the same time. The network model that accurately predicts the regulatory relationship between the required air volume of the fan and the external factors of gas concentration, oxygen concentration and temperature and humidity is selected.

[0127] In the first step, the numerical simulation software is CFD numerical simulation software.

[0128] The air volume demand of local ventilation fans in mines is directly affected by the complex environment underground, especially the four key factors of gas concentration, temperature, humidity and dust concentration, which have a significant impact on ventilation demand. In order to accurately predict and adjust the air volume required by local ventilation fans at the next moment, a prediction model is designed. The prediction model uses gas concentration, temperature, humidity and dust concentration as input variables, and the air volume required by local ventilation fans at future moments as the output variable of the prediction model.

[0129] In order to achieve this prediction function, neural network technology is used. The optimal weights and thresholds in the neural network are determined through the training process. This process enables the neural network to learn the complex relationship between the input environmental parameters and the required air volume, thereby achieving accurate prediction of the air volume demand. In order to further improve the response speed and stability of the system, a closed-loop PID feedback control mechanism is also introduced. This closed-loop PID feedback control mechanism can automatically adjust the wind speed of the fan based on the difference between the actual value and the predicted value of the current air volume, and realize the advance adjustment of the air volume. Through the closed-loop action of the PID controller, the system can quickly respond to changes in the underground environment and ensure that the ventilation volume is always maintained at the most suitable level to ensure the safety and efficiency of mine operations.

[0130] Based on the actual collected offline data, the physical model of the local fan air supply system was accurately constructed using advanced numerical simulation software. This physical model is designed to simulate the air volume distribution and flow state in the wind tube under different working conditions, and generate a series of detailed data files related to air volume, wind tube performance, etc. Subsequently, these simulation results were deeply processed and analyzed, and the correlation information between key air volume parameters and system operation characteristics was extracted through scientific data processing methods.

[0131] On this basis, the neural network model is used as a prediction tool, and the previously processed data is used as training samples to train and optimize the neural network. After sufficient learning, the neural network model can achieve high-precision prediction of the air volume of local fans under different working conditions. The realization of this prediction capability not only provides a reliable theoretical basis for the design and optimization of local fans, but also provides important practical guidance for the variable frequency operation of ventilation systems, aiming to improve the energy efficiency, response speed and overall operation stability of the system.

[0132] In order to build a comprehensive neural network learning mechanism, three different neural network algorithms - BP neural network (back propagation neural network), DBN neural network (deep belief network) and DNN neural network (deep neural network) - are integrated into a unified framework. This framework aims to optimize each network model through training data to achieve accurate prediction of the required air volume of the ventilator. The specific component process is as follows:

[0133] 1. First, the collected training data will be input into this integrated neural network learning mechanism. These training data will be used to train the BP neural network, DBN neural network and DNN neural network models respectively, to ensure that each neural network can learn the complex relationship between the fan air volume demand and influencing factors from the training data. After the training is completed, an independent validation set data will be used to evaluate the prediction performance of the three network models. Specifically, the input data of the validation set will be sent to the three trained neural networks respectively, and each neural network will generate corresponding air volume demand prediction results.

[0134] Second, we will use evaluation indicators to compare the effects of the three algorithms: BP neural network, DBN neural network, and DNN neural network. The evaluation indicators are the degree of fit or the root mean square error. Specifically, in order to objectively compare the effects of the three algorithms, we will use a series of evaluation indicators, such as the degree of fit (such as R 2 The prediction results are comprehensively analyzed by using the following indicators: R, R-value, root mean square error (RMSE), etc. These indicators will help quantify the performance of each algorithm in terms of prediction accuracy, stability, etc. The degree of fit is an important indicator for evaluating how well a statistical model fits the data. In regression analysis, R 2 The coefficient of determination is one of the most commonly used goodness of fit indicators. 2 The value represents the percentage of the total variability in the data that is explained by the model and is a measure of how well the model's predictions match the actual observed values. 2 The value ranges from 0 to 1. Root Mean Square Error (RMSE) is a commonly used indicator for evaluating the prediction performance of regression models. It measures the square root of the average of the sum of squares of the differences between the model predictions and the actual observations, so it reflects the overall deviation between the model predictions and the actual observations.

[0135] Table 1 Comparison of prediction results

[0136]

[0137] 3. Finally, based on the above evaluation results, the algorithm with the best performance will be selected as the final prediction tool.

[0138] This selection process will ensure that the forecasting system can predict the ventilator air demand in the most accurate and reliable way, providing strong support for the optimization and management of the ventilation system.

[0139] Aiming at the problems of hysteresis and low efficiency in regulating the required air volume of traditional mine local ventilators, the present invention proposes a method for predicting the required air volume based on numerical simulation and neural network.

[0140] The method for predicting required air volume based on numerical simulation and neural network of the present invention has the following advantages:

[0141] First, in order to improve the prediction efficiency and accuracy of the air volume at the fan duct outlet, the present invention adopts a method of training a neural network model based on artificially set typical working conditions as input. This neural network model can efficiently predict and cover the air volume data under a wide range of working conditions, effectively replacing the traditional cumbersome processes such as manual adjustment and frequency conversion control. This scheme significantly shortens the prediction cycle, enhances the simulation efficiency, and achieves a leap in measurement accuracy. However, although CFD (computational fluid dynamics) numerical simulation is known for its powerful mathematical calculation model and can accurately simulate complex processes such as fluid diffusion and heat transfer, it faces many challenges when used alone, including complex operation, large workload, slow simulation speed, long numerical analysis time, and limited coverage of typical working conditions. In contrast, the neural network model simulates the computing mechanism of the human brain and shows its significant advantages in processing large amounts of information, fast calculation and solving complex problems. It is not only objective and true, but also can achieve high-precision prediction of future air volume changes in a simpler and faster way, thereby showing great application potential and value in the field of fan air volume prediction.

[0142] 2. Using neural networks based on numerical simulation to predict the air volume of local ventilators can save a lot of numerical simulation steps, speed up the simulation, comprehensively cover various working conditions, and reduce human influence.

[0143] 3. The neural network algorithm is organized into a neural network algorithm learning machine to shorten the time of algorithm regulation.

[0144] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and inventive concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for pre-adjusting the required air volume of a local ventilation fan in a mine, characterized in that: The following steps are involved: Step 1: Use numerical simulation to obtain simulation results: Use numerical simulation software to establish a physical model of the local fan duct, perform simulation operations, and generate simulation data of the air volume of the local fan duct outlet under various working conditions with different duct materials; wherein the parameters included in the physical model include air volume, gas concentration, oxygen concentration, temperature, and humidity; The wind tube is circular, and the average velocity of the fluid in the circular wind tube is calculated as follows: The symbols in formula (1) mean: v represents the average velocity of the circular wind tube fluid, in m / s; α represents the kinematic viscosity coefficient of the fluid; R e represents the Reynolds number; r o Indicates the diameter of the circular air duct, in meters. Step 2, data processing: data screening and processing are performed on the generated local ventilator air volume simulation data of the air duct outlet under each working condition, and the default data is inserted; Step 3: Establish a neural network model, use the neural network model to predict the air volume at the air duct outlet, and the program automatically selects the best: train the neural network based on the screened data, and use the validation set data in the trained neural network to predict the performance of the air volume at the air duct outlet, and then get the predicted air volume value; The neural network model includes an input layer, a hidden layer, and an output layer. The gas concentration, oxygen concentration, temperature, and humidity are used as the input layer, and the position of the average wind speed circle from the top of the tunnel is used as the output layer. m ] is the input sample, [Y1, Y2, Y3, ...Y n ] is the output sample, The output of each unit in the hidden layer is calculated as follows: Oh j =f(ΣW ij X i -θ j ) (2) The symbols in formula (2) mean: O j Represents the output results of each unit in the hidden layer; f represents a nonlinear function; W ij Represents the weight matrix of the hidden layer; θ j represents the neuron threshold; X i Represents an input sample; The output result of the output layer is calculated as follows: Y j =f(ΣT JK Oh j -θ k ) (3) The symbols in formula (3) mean: Y j Represents the output result of the output layer; f represents a nonlinear function; T JK Represents the weight matrix of the output layer; O j Represents the output results of each unit in the hidden layer; θ k represents a neuron threshold; Step 4: Use the optimal program to pre-adjust the air volume at the air duct outlet: set the target parameters according to the predicted air volume value, and use the closed-loop PID feedback control mechanism to pre-adjust the output air volume; The predicted air volume value is calculated as follows: U(t)=Kp*e(t)+Ki*∫e(t)dt+Kd*de(t) / dt (4) The symbols in formula (4) mean: U(t) represents the output of the controller at time t; e(t) represents the difference between the system output and the expected value, that is, the deviation; Kp, Ki, and Kd are the three key parameters of the PID controller, where Kp is the proportional gain, Ki is the integral gain, and Kd is the differential gain; ∫e(t)dt represents the deviation integral from time 0 to time t; de(t) / dt represents the derivative of the deviation; The calculation formula of e(t) is as follows: e(t)=y(t)-r(t) (5) The symbols in formula (5) mean: y(t) is the system output; r(t) is the expected value.

2. The method for pre-adjusting the required air volume of a local ventilation fan in a mine according to claim 1, characterized in that: In the first step, the physical model uses gas concentration, temperature, humidity and dust concentration as input variables, and the air volume required by the local ventilation fan at a future time is used as the output variable of the physical model.

3. The method for pre-adjusting the required air volume of a mine local ventilator according to claim 1, characterized in that: In the third step, the processed data is used to train the neural network, and the optimal weights and thresholds in the neural network are determined through the training process.

4. The method for pre-adjusting the required air volume of a local ventilation fan in a mine according to claim 1, characterized in that: In step 4, the closed-loop PID feedback control mechanism automatically adjusts the wind speed of the ventilator based on the difference between the actual value and the predicted value of the current air volume, thereby achieving advanced regulation of the air volume.

5. The method for pre-adjusting the required air volume of a local ventilation fan in a mine according to claim 1, characterized in that: The calculation formula of the air volume is as follows: Q=S×V (6) The symbols in formula (6) mean: Q represents air volume; S represents the cross-sectional area of ​​the fan; V represents the fan speed.

6. The method for pre-adjusting the required air volume of a local ventilation fan in a mine according to claim 1, characterized in that: In step 3, the neural network integrates a BP neural network, a DBN neural network and a DNN neural network.

7. The method for pre-adjusting the required air volume of a local ventilation fan in a mine according to claim 6, characterized in that: The screened data are used to train three models, namely, BP neural network, DBN neural network and DNN neural network. Each neural network can learn the relationship between the required air volume of the ventilator and the influencing factors from the screened data.

8. The method for pre-adjusting the required air volume of a local ventilation fan in a mine according to claim 1, characterized in that: In the first step, the numerical simulation software is CFD numerical simulation software.

9. The method for pre-adjusting the required air volume of a local ventilation fan in a mine according to claim 1, characterized in that: Evaluation indicators are used to compare the effects of three algorithms: BP neural network, DBN neural network and DNN neural network.

10. The method for pre-adjusting the required air volume of a local ventilation fan in a mine according to claim 9, characterized in that: The evaluation index is the degree of fitting or the root mean square error.

Images