Mine harmful gas early warning system and method based on WOA-WNN algorithm

Abstract

The invention provides a mine harmful gas early warning system and method based on a WOA-WNN algorithm, and relates to the technical field of coal mine safety. The system comprises an original data acquisition layer, an original data transmission layer and a data processing terminal, the data processing terminal comprises a classification early warning model, a PLC control module, a CPU industrial data processing chip, a second WiFi communication module, a power supply, an audible and visual alarm, a display screen, a power switch, a data acquisition layer switch, an alarm switch, an early warning switch and a WIFI transmission switch. According to the method, gas sensors are arranged at different places in the production process of a mine working face to collect and summarize original concentration data of harmful gas discharged from the different places, and a WOA-WNN algorithm is adopted to calculate harmful gas early warning values of the different places according to the summarized original concentration data. Therefore, classification early warning of the mine harmful gas is carried out, and the accuracy and reliability of classification early warning of the mine harmful gas are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of coal mine safety technology, and in particular to a mine harmful gas early warning system and method based on a WOA-WNN algorithm. Background Art

[0002] The threat posed by some harmful gases in the production environment of mine working faces to workers' health and production safety cannot be ignored. Therefore, it is of great significance to develop an efficient classification and early warning system for harmful gases in mine working faces. At present, traditional harmful gas monitoring systems are limited by problems such as detection sensitivity, response speed and accuracy. In the complex and changeable mine environment, traditional sensor technology is often difficult to provide accurate real-time monitoring data. Moreover, traditional data processing methods often rely on simple threshold settings or rule-based logical judgments, lack of in-depth mining and analysis of complex data, resulting in low warning accuracy and high false alarm rate. Existing mine working face harmful gas early warning systems usually lack intelligent features and cannot be flexibly adjusted and optimized according to actual environmental changes. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, the present invention proposes a mine harmful gas early warning system and method based on the WOA-WNN algorithm, aiming to overcome the limitations of traditional sensor technology, adapt to the complex environment of mine working faces, improve the accuracy and reliability of early warning, and realize intelligent management of mine safety.

[0004] The first aspect of the present invention proposes a mine harmful gas early warning system based on the WOA-WNN algorithm, the system comprising: an original data acquisition layer, an original data transmission layer and a data processing terminal;

[0005] The original data collection layer is used to collect and summarize the original concentration data of harmful gases emitted at different locations during the production process of the mine working face, and transmit the summarized original concentration data to the original data transmission layer;

[0006] The original data transmission layer is used to establish a channel connection between the original data acquisition layer and the data processing terminal, and transmit the received original concentration data to the data processing terminal;

[0007] The data processing terminal is used to adopt the wavelet neural network WOA-WNN algorithm improved based on the whale algorithm, and to classify and warn the harmful gases in the mine at different locations in the production process of the mine working face according to the received original concentration data;

[0008] The raw data acquisition layer includes: a plurality of sensor modules, a data monitoring station, a single-chip microcomputer module and a first WiFi communication module distributed at different locations in the mine working surface;

[0009] The sensor module is used to collect the original concentration data of harmful gases emitted at different locations during the production process of the mine working face, and transmit the collected original concentration data to the data monitoring station;

[0010] The data monitoring station is used to transmit all received original concentration data to the single-chip microcomputer module; and upload the original concentration data to the original data transmission layer;

[0011] The single chip microcomputer module is used to summarize all received original concentration data and transmit the summarized original concentration data to the first WiFi communication module;

[0012] The first WiFi communication module is used to transmit the aggregated original concentration data to the original data transmission layer via a WiFi signal;

[0013] The original data transmission layer includes: an underground WiFi base station and a data storage server;

[0014] The underground WiFi base station is used to receive the aggregated raw concentration data and transmit it to the data storage server;

[0015] The data storage server is used to store the original concentration data uploaded by the data monitoring station and the aggregated original concentration data from the underground WiFi base station, and transmit all the stored original concentration data to the data processing terminal;

[0016] The data processing terminal is a box, including: a classification warning model, a PLC control module, a CPU industrial data processing chip, a second WiFi communication module, a power supply, an audible and visual alarm, a display screen, a power switch, a data acquisition layer switch, an alarm switch, a warning switch and a WiFi transmission switch;

[0017] The classification warning model, PLC control module, CPU industrial data processing chip, second WiFi communication module and power supply are all embedded in the box; wherein the CPU industrial data processing chip is embedded with the classification warning model; the PLC control module controls the CPU industrial data processing chip, the second WiFi communication module, the power supply, the sound and light alarm, the display screen, the power switch, the data acquisition layer switch, the alarm switch, the warning switch and the WIFI transmission switch through wires for linkage control;

[0018] The second WIFI module is used to receive original concentration data from a data storage server;

[0019] The power supply is used to supply power to the data processing terminal;

[0020] The sound and light alarm is installed on the upper surface of the box body and is used to issue an alarm according to the classification warning results of harmful gases in the mine;

[0021] The display screen is installed on the front surface of the box body and is used to display the warning value of harmful gases in the mine;

[0022] The power switch, data acquisition layer switch, alarm switch, early warning switch and WIFI transmission switch are sequentially installed side by side on the front surface of the box and are located below the display screen; wherein the power switch is used to control the on / off state of the power supply; the data acquisition layer switch is used to control the on / off state of the sensor module in the original data acquisition layer through a wireless signal; the alarm switch is used to control the on / off state of the sound and light alarm; the early warning switch is used to control the on / off state of the CPU industrial data processing chip; the WIFI transmission switch is used to control the on / off state of the first WiFi communication module and the second WiFi communication module;

[0023] The CPU industrial data processing chip is used to build a classification warning model according to the WOA-WNN algorithm, and input the original concentration data received by the data processing terminal into the classification warning model to obtain the warning value of harmful gas in the mine, and send the obtained warning value to the display screen in the form of a data graph and display it; at the same time, according to the warning value, the position of the sensor module corresponding to the warning value is divided into a low-risk area, a medium-risk area, a high-risk area or a harmful gas abnormal outflow disaster area; when the position is a harmful gas abnormal outflow disaster area, the sound and light alarm is controlled to sound an alarm by transmitting a signal to the PLC control module;

[0024] The process of constructing the classification warning model according to the WOA-WNN algorithm is as follows: constructing a wavelet neural network consisting of an input layer, a hidden layer and an output layer; wherein the input layer and the output layer are responsible for the input and output of the original concentration data respectively; the hidden layer is used to calculate the warning value of the harmful gas in the mine according to the input original concentration data;

[0025] Assuming that the number of nodes in the input layer, hidden layer, and output layer are n, i, and m respectively, when the original concentration data is transferred from the input layer to the hidden layer, for any node j in the hidden layer, the original concentration data input in the node is calculated as:

[0026]

[0027] Where S j is the original concentration data input in node j; k represents the kth node in the input layer; ω kj is the input layer connection weight of node j; x k represents the original concentration data transmitted by the kth node in the input layer;

[0028] The Morlet wavelet is selected as the base wave, and the original concentration data input in the node j of the hidden layer is scaled by the wavelet basis at the node, and the output data of the node j is obtained as follows:

[0029]

[0030] h j =φ[(S j -b j ) / a j ]

[0031] Where φ(x) is the wavelet basis function; x represents the variable in the wavelet basis function; e is a natural constant; h j is the output data of the jth node in the hidden layer; b j is the smoothing factor of the wavelet basis function; a j is the scaling factor of the wavelet basis function;

[0032] The output data of all nodes in the hidden layer are transferred to the output layer, and the output data of each node in the output layer is calculated. For any node t in the output layer, the output data of node t is:

[0033]

[0034] where y t is the output data of node t; Represents the function of calculating the output data of the nodes in the output layer; ω jt is the hidden layer connection weight of node t;

[0035] According to the output data of all nodes in the output layer and the output layer connection weight ω of each node t Calculate the harmful gas warning value at the location of each sensor module;

[0036] The smoothing factor b of the wavelet basis function for any node j in the hidden layer of the wavelet neural network is j and the scaling factor a of the wavelet basis function j , and the hidden layer connection weight ω of any node t in the output layer jt And the output layer connection weight ω t As a set of parameters to be optimized, the whale algorithm is used to optimize all the parameters to be optimized, and the optimal solution of each set of parameters to be optimized is substituted into the wavelet neural network, and the wavelet neural network obtained after substituting the optimal solution is used as a classification warning model;

[0037] The process of using the whale algorithm to optimize all parameters to be optimized is as follows: for each group of parameters to be optimized, a number of whale individuals are randomly generated and the fitness of each whale individual is calculated, and the positions of all whale individuals are updated according to the whale individual with the highest fitness; a random number P is generated in the range of [0,1], and it is determined whether the random number P is less than a set threshold. If the random number P is less than the set threshold, it is determined whether the absolute value of the coefficient vector |A| is greater than or equal to 1; if the random number P is not less than the set threshold, the prey encirclement stage is entered, and the position of the optimal whale individual in the prey encirclement stage is calculated; if the random number P is less than the set threshold, then Determine whether the absolute value of the coefficient vector |A| is greater than or equal to 1. If |A|>=1, enter the random search for the optimal solution stage and calculate the position of the optimal whale individual in the random search for the optimal solution stage; if |A|<1, enter the surround optimal solution stage and calculate the position of the optimal whale individual in the surround optimal solution stage; determine whether the current iteration round has reached the preset number of iterations. If so, take the position of the optimal whale individual in the current round as the optimal solution for the group of parameters to be optimized; if not, update the positions of all whale individuals according to the position of the optimal whale individual in the current round and perform the next iteration until the preset number of iterations is reached;

[0038] The mine harmful gas early warning method based on the WOA-WNN algorithm proposed in the second aspect of the present invention is implemented by the mine harmful gas early warning system based on the WOA-WNN algorithm, and the method comprises the following steps:

[0039] Step 1: Obtain several locations to be detected on the working surface of the mine, and arrange a gas sensor for detecting harmful gases at each location to be detected;

[0040] Step 2: Turn on the power switch, data acquisition layer switch, alarm switch, early warning switch, and WIFI transmission switch on the box;

[0041] Step 3: By aligning the sampling time of the gas sensors, the original concentration data of each gas sensor at any time is obtained and transmitted to the data monitoring station;

[0042] Step 4: The data monitoring station transmits the original concentration data of each gas sensor at any time to the single-chip microcomputer module for aggregation, and transmits the aggregated original concentration data in turn through the WiFi communication module and the underground WiFi base station and uploads it to the data storage server for storage;

[0043] Step 5: The data processing terminal calls the original concentration data stored in the data storage server, and uses the original concentration data to calculate the harmful gas warning value of each location to be detected;

[0044] Step 6: According to the harmful gas warning value of each location to be detected, each location to be detected is divided into a low-risk area, a medium-risk area, a high-risk area or a harmful gas abnormal outflow disaster area. When the location to be detected is a harmful gas abnormal outflow disaster area, the CPU industrial processing chip controls the sound and light alarm to sound an alarm by transmitting a signal to the PLC control module;

[0045] The step 3 further comprises:

[0046] Step 301: setting a sampling interval, and collecting observation values ​​y of all gas sensors at different times according to the sampling interval;

[0047] Step 302: for each gas sensor q, obtain an observation curve y(q) of the gas sensor q by fitting according to the observation values ​​y of the gas sensor at different times;

[0048] Step 303: The value corresponding to any time in the observation curve y(q) of the gas sensor q is used as the original concentration data of the gas sensor q at any time and transmitted to the data monitoring station;

[0049] The step 5 further comprises:

[0050] The step 5 further comprises:

[0051] Step 501: The data processing terminal calls the original concentration data stored in the data storage server as a sample data set, and divides the sample data set into a sample training set and a sample verification set according to a preset ratio;

[0052] Step 502: Set the first maximum number of iterations and the number of individuals in the whale population, input the sample training set into the classification warning model, and use the whale algorithm to iteratively optimize the classification warning model of the input sample training set to obtain the weight and optimal threshold of the classification warning model, and then substitute the obtained weight and optimal threshold into the classification warning model;

[0053] Step 503: setting the second maximum number of iterations, the number of neurons and the learning rate, using an empirical formula to determine the number of nodes in the hidden layer of the classification warning model, and obtaining the classification warning model under the current iteration round;

[0054] Step 504: using the classification warning model obtained in step 502 to calculate the harmful gas warning value corresponding to all sample data in the sample training set, and using the root mean square error RMSE to calculate the evaluation index of the current iteration round;

[0055] Step 505: Determine whether the evaluation index of the current iteration round is greater than the set threshold. If so, return to step 502 until the second maximum iteration round is reached, and use the classification warning model under the last iteration round as the trained classification warning model; if not, use the classification warning model under the current round as the trained classification warning model;

[0056] Step 506: input the sample verification set into the trained classification warning model for verification to obtain the final classification warning model;

[0057] Step 507: For any location to be detected, the data processing terminal calls the original concentration data of the location to be detected stored in the data storage server and inputs the final classification warning model to obtain the harmful gas warning value of the location to be detected.

[0058] The beneficial effects of adopting the above technical solution are:

[0059] The system of the present invention collects the original data of harmful gas concentration at the working face of the mine, and realizes wireless transmission of the data through a WiFi communication module, thereby ensuring the timeliness and reliability of data transmission.

[0060] The system of the present invention has powerful data processing and analysis capabilities. It applies a wavelet neural network algorithm (WNN) improved based on a whale optimization algorithm (WOA) to efficiently classify and process the collected data, thereby achieving in-depth mining and analysis of complex data, thereby providing accurate warning results and reducing the false alarm rate. Specifically, the WOA algorithm is used to optimize the parameters of a classification warning model constructed based on the WNN algorithm, thereby achieving the optimal parameter combination of the classification warning model, thereby improving the accuracy and reliability of the classification warning model.

[0061] The system of the present invention has intelligent features, that is, it can be dynamically adjusted and optimized according to the data collected in real time, ensuring that the early warning model can provide accurate early warnings under different conditions, ensuring the timeliness and effectiveness of the early warning system. The system of the present invention provides multi-level security protection by providing graded early warnings for the concentration of harmful gases, including low-risk areas, medium-risk areas, high-risk areas, and areas of disaster caused by abnormal outflow of harmful gases. When it is detected that the concentration of harmful gases exceeds the set threshold, the early warning system automatically triggers an audible and visual alarm to promptly remind on-site personnel to take corresponding measures to prevent the occurrence of safety accidents.

[0062] The system of the present invention adopts modular design, including three parts: data acquisition layer, original data transmission layer and original data processing layer, and each module is connected by a standardized interface, which is convenient for the installation, maintenance and expansion of the system. The data processing terminal integrates key components such as PLC control module, WiFi communication module, power supply, CPU industrial data processing chip, etc. to ensure the stable operation of the system.

[0063] In summary, the system of the present invention can operate stably in the complex environment of a mine working face, has strong adaptability, is suitable for intelligent management of mine safety, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 Schematic diagram of a mine harmful gas early warning system based on the WOA-WNN algorithm in this embodiment;

[0065] Figure 2 This is a schematic diagram of the layout of the embedded modules in the box of the data processing terminal in this embodiment;

[0066] Figure 3 This is a schematic diagram of the exterior of the data processing terminal in this embodiment;

[0067] Figure 4 Schematic diagram of the flow of the WOA-WNN algorithm in this implementation mode;

[0068] Figure 5 This is a flow chart of a mine harmful gas early warning method based on the WOA-WNN algorithm in this embodiment;

[0069] Figure 6 It is a hazard classification diagram of mine harmful gas warning in this implementation mode;

[0070] In the figure: 1-sensor module; 2-data monitoring station; 3-single-chip microcomputer module; 4-first WiFi communication module; 5-underground WiFi base station; 6-data storage server; 7-data processing terminal; 701-classification warning model; 702-PLC control module; 703-CPU industrial data processing chip; 704-second WiFi communication module; 705-power supply; 706-sound and light alarm; 707-display screen; 708-power switch; 709-data acquisition layer switch; 710-alarm switch; 711-warning switch; 712-WIFI transmission switch. DETAILED DESCRIPTION

[0071] In order to facilitate the understanding of the present application, the specific embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and embodiments. The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present application more thoroughly and comprehensively understood.

[0072] A mine harmful gas early warning system based on the WOA-WNN algorithm in this embodiment is as follows: Figure 1-3 As shown, the system includes: an original data acquisition layer, an original data transmission layer and a data processing terminal 7.

[0073] The original data collection layer is used to collect and summarize the original concentration data of harmful gases emitted at different locations during the production process of the mine working face, and transmit the summarized original concentration data to the original data transmission layer.

[0074] The raw data acquisition layer includes: a plurality of sensor modules 1 distributed at different locations in the mine working surface, a data monitoring station 2, a single chip computer module 3 and a first WiFi communication module 4.

[0075] The sensor module 1 is used to collect original concentration data of harmful gases emitted at different locations during the production process of the mine working face, and transmit the collected original concentration data to the data monitoring station 2.

[0076] The data monitoring station 2 is used to transmit all received original concentration data to the single chip computer module 3; and upload the original concentration data to the original data transmission layer.

[0077] The single chip computer module 3 is used to aggregate all received original concentration data and transmit the aggregated original concentration data to the first WiFi communication module 4 .

[0078] The first WiFi communication module 4 is used to transmit the aggregated original concentration data to the original data transmission layer via WiFi signals.

[0079] The original data transmission layer is used to establish a channel connection between the original data acquisition layer and the data processing terminal 7, and transmit the received original concentration data to the data processing terminal 7.

[0080] The original data transmission layer includes: an underground WiFi base station 5 and a data storage server 6.

[0081] The underground WiFi base station 5 is used to receive the aggregated original concentration data and transmit it to the data storage server 6 .

[0082] The data storage server 6 is used to store the original concentration data uploaded by the data monitoring station 2 and the aggregated original concentration data from the underground WiFi base station 5 , and transmit all the stored original concentration data to the data processing terminal 7 .

[0083] In this embodiment, the original data transmission layer transmits the collected original concentration data to the data storage server 6 through the underground industrial ring network base station, that is, the underground WiFi base station 5 and saves it. At the same time, the data processing terminal 7 can call the data in the data storage server 6 at any time.

[0084] The data processing terminal 7 is used to adopt a wavelet neural network (WOA-WNN) algorithm improved based on the whale algorithm, and to perform classified warning of mine harmful gases at different locations in the production process of the mine working face according to the received original concentration data.

[0085] The data processing terminal 7 is a box, including: a classification warning model 701, a PLC control module 702, a CPU industrial data processing chip 703, a second WiFi communication module 704, a power supply 705, an audible and visual alarm 706, a display screen 707, a power switch 708, a data acquisition layer switch 709, an alarm switch 710, a warning switch 711 and a WiFi transmission switch 712;

[0086] The classification warning model 701, PLC control module 702, CPU industrial data processing chip 703, second WiFi communication module 704 and power supply 705 are all embedded in the box; wherein the classification warning model 701 is embedded in the CPU industrial data processing chip 703; the PLC control module 702 controls the CPU industrial data processing chip 703, the second WiFi communication module 704, the power supply 705, the sound and light alarm 706, the display screen 707, the power switch 708, the data acquisition layer switch 709, the alarm switch 710, the warning switch 711 and the WIFI transmission switch 712 through wires.

[0087] The second WIFI module 704 is used to receive original concentration data from the data storage server 6 .

[0088] The power supply 705 is used to supply power to the data processing terminal 7 .

[0089] In this embodiment, if Figure 2 As shown, the second WIFI module 704 can realize the wireless transmission of the original concentration data between the original data transmission layer and the data processing terminal 7.

[0090] The sound and light alarm 706 is installed on the upper surface of the box body and is used to issue an alarm according to the classification warning results of harmful gases in the mine.

[0091] The display screen 707 is installed on the front surface of the box body and is used to display the warning value of harmful gases in the mine.

[0092] The power switch 708, data acquisition layer switch 709, alarm switch 710, early warning switch 711 and WIFI transmission switch 712 are installed side by side on the front surface of the box in sequence and are located below the display screen 707; wherein the power switch 708 is used to control the switching state of the power supply 705; the data acquisition layer switch 709 is used to control the switching state of the sensor module 1 in the original data acquisition layer through a wireless signal; the alarm switch 710 is used to control the switching state of the sound and light alarm 706; the early warning switch 711 is used to control the switching state of the CPU industrial data processing chip 703; the WIFI transmission switch 712 is used to control the switching state of the first WiFi communication module 4 and the second WiFi communication module 704.

[0093] The CPU industrial data processing chip 703 is used to construct a classification warning model 701 according to the WOA-WNN algorithm, and input the original concentration data received by the data processing terminal 7 into the classification warning model 701 to obtain the warning value of harmful gas in the mine, and send the obtained warning value to the display screen 707 in the form of a data graph for display; at the same time, according to the warning value, the position of the sensor module corresponding to the warning value is divided into a low-risk area, a medium-risk area, a high-risk area or a harmful gas abnormal outflow disaster area; when the position is a harmful gas abnormal outflow disaster area, the signal is transmitted to the PLC control module 702 to control the sound and light alarm 706 to sound an alarm.

[0094] In this embodiment, the data processing terminal 7 is the core of the system, and the data is processed by the CPU industrial data processing chip 703. The CPU industrial processing chip 703 uses the WOA-WNN algorithm to calculate the warning value of harmful gas in the mine and displays it on the display screen 707 in the form of a data graph; at the same time, the warning value is compared with the set threshold value. If it exceeds the set threshold value, the location of the sensor module corresponding to the warning value is the disaster area of ​​abnormal outflow of harmful gas, and the signal is transmitted to the PLC control module 702 to control the sound and light alarm 706 to alarm.

[0095] In this embodiment, the data processing terminal is used to process the original concentration data of harmful gases in mines monitored by sensor modules at different locations, so as to prepare for data classification processing, and on this basis, use the wavelet neural network algorithm improved by the whale algorithm to establish a classification warning model; the original concentration data of harmful gases in mines are collected and transmitted to the classification warning model for analysis via WiFi communication, the data is pre-processed by relying on the established classification warning model and the optimal parameter combination of the classification warning model is obtained by using the whale algorithm, and the parameters are substituted into the classification warning model for training; the model is trained by training the training set, and the output result obtained by inputting the test set into the forecast model can finally give a classification warning for the abnormal outburst of harmful gases in the working face environment.

[0096] The process of constructing the classification warning model 701 according to the WOA-WNN algorithm is: constructing a wavelet neural network composed of an input layer, a hidden layer and an output layer; wherein the input layer and the output layer are responsible for the input and output of the original concentration data respectively; the hidden layer is used to calculate the warning value of harmful gases in the mine according to the input original concentration data.

[0097] Assuming that the number of nodes in the input layer, hidden layer, and output layer are n, i, and m respectively, when the original concentration data is transferred from the input layer to the hidden layer, for any node j in the hidden layer, the original concentration data input in the node is calculated as:

[0098]

[0099] Where S j is the original concentration data input in node j; k represents the kth node in the input layer; ω kj is the input layer connection weight of node j; x k represents the raw concentration data transmitted by the kth node in the input layer.

[0100] The Morlet wavelet is selected as the base wave, and the original concentration data input in the node j of the hidden layer is scaled by the wavelet basis at the node, and the output data of the node j is obtained as follows:

[0101]

[0102] h j =φ[(S j -b j ) / a j ] (3)

[0103] Where φ(x) is the wavelet basis function; x represents the variable in the wavelet basis function; e is a natural constant; h j is the output data of the jth node in the hidden layer; b jis the smoothing factor of the wavelet basis function; a j is the expansion factor of the wavelet basis function; where a j and b j The values ​​of are all consecutive numbers in the interval [1,2], for example, 1.21, 1.211, 1.2111.

[0104] The output data of all nodes in the hidden layer are transferred to the output layer, and the output data of each node in the output layer is calculated. For any node t in the output layer, the output data of node t is:

[0105]

[0106] where y t is the output data of node t; Represents the function of calculating the output data of the nodes in the output layer; ω jt is the hidden layer connection weight of node t.

[0107] According to the output data of all nodes in the output layer and the output layer connection weight ω of each node t Calculate the harmful gas warning value at the location of each sensor module.

[0108] In this embodiment, in the process of using the WOA algorithm to obtain the key parameters in the WNN algorithm, the wavelet neural network mainly has four parameters involved in the calculation, which are the smoothing factor b of the wavelet basis function, j With the expansion factor a j And the hidden layer connection weights ω of each node in the output layer jt The connection weight ω with the output layer t The values ​​of these four parameters will directly affect the performance of the wavelet neural network. The existing algorithm can be used to calculate the output data of all nodes in the output layer and the output layer connection weights of each node ω. t Calculate the harmful gas warning value at the location of each sensor module.

[0109] The smoothing factor b of the wavelet basis function for any node j in the hidden layer of the wavelet neural network is j and the scaling factor a of the wavelet basis function j , and the hidden layer connection weight ω of any node t in the output layer jt And the output layer connection weight ω t As a group of parameters to be optimized, the whale algorithm is used to optimize all the parameters to be optimized, the optimal solution of each group of parameters to be optimized is substituted into the wavelet neural network, and the wavelet neural network obtained after substituting the optimal solution is used as a classification warning model.

[0110] In this embodiment, the process of optimizing all parameters to be optimized using the whale algorithm requires two steps of algorithms, namely: searching for the optimal solution of encirclement and searching for the optimal solution by updating the spiral position.

[0111] The process of optimizing all parameters to be optimized by using the whale algorithm is as follows: for each group of parameters to be optimized, a number of whale individuals are randomly generated and the fitness of each whale individual is calculated, and the positions of all whale individuals are updated according to the whale individual with the highest fitness; a random number P is generated in the range of [0,1], and it is determined whether the random number P is less than a set threshold;

[0112] If the random number P is not less than the set threshold, the process enters the stage of surrounding the prey. The stage of surrounding the prey is:

[0113]

[0114] in is the distance vector between the current whale individual and the optimal whale individual; is the coefficient vector; is the position vector of the best whale individual in the current iteration round; is the position of the current whale individual in the current iteration round; is the position of the current whale individual in the next iteration round; is the coefficient vector.

[0115] The coefficient vector for:

[0116]

[0117] in is the coefficient vector in the iterative process; is a random vector, and is in the range of [0,1].

[0118]

[0119] Where a is the coefficient in the iteration process, and a decreases linearly from 2 to 0 during the iteration process; t is the current iteration round; t max is the maximum number of iterations.

[0120] In this embodiment, the threshold is set to 0.5.

[0121] If the random number P is less than the set threshold, then determine whether the absolute value of the coefficient vector |A| is greater than or equal to 1. If |A|>=1, then enter the random search for the optimal solution stage; the random search for the optimal solution stage is:

[0122]

[0123] in is a random whale position vector in the current iteration.

[0124] If |A|<1, the optimal solution encirclement phase is entered; the optimal solution encirclement phase is:

[0125]

[0126]

[0127] Where p is a random number between [0, 1]; k is a correction coefficient, and in this implementation, k is set to be a random number between [1, 2]; is the distance vector between the current whale individual and the optimal whale individual in the stage of surrounding the optimal solution; b is a constant representing the shape of the logarithmic spiral; l is a random number between [0,1]; is the position vector of other whale individuals except the current whale individual.

[0128] In this embodiment, if Figure 4 As shown in the figure, the random search for the optimal solution stage corresponds to searching for prey, and the encirclement of the optimal solution stage corresponds to capturing prey. In the encirclement of the optimal solution stage, the optimal solution is searched by updating the spiral position, and a spiral equation is established between the solution and the position of the optimal solution to find the optimal threshold, that is, the optimal solution obtained is the threshold required by the WNN algorithm.

[0129] Determine whether the current iteration round has reached the preset number of iterations. If so, the position of the best whale individual in the current round is used as the optimal solution for the group of parameters to be optimized; if not, the positions of all whale individuals are updated according to the position of the best whale individual in the current round and the next iteration is performed until the preset number of iterations is reached.

[0130] A mine harmful gas early warning method based on the WOA-WNN algorithm of this embodiment is implemented by the mine harmful gas early warning system based on the WOA-WNN algorithm, such as Figure 5 As shown, the method comprises the following steps:

[0131] Step 1: Obtain several locations to be detected on the working surface of the mine, and arrange a gas sensor for detecting harmful gases at each location to be detected.

[0132] In this embodiment, a hydrogen sulfide gas sensor for detecting hydrogen sulfide gas, one of the harmful gases in mines, is selected as the sensor module 1, and several sensor modules 1 are arranged at the upper corner of the return air, the working face support frame arm, and the gas (oil-type gas) extraction drilling site.

[0133] Step 2: Turn on the power switch, data acquisition layer switch, alarm switch, early warning switch, and WIFI transmission switch on the box.

[0134] Step 3: By aligning the sampling time of the gas sensors, the original concentration data of each gas sensor at any time is obtained and transmitted to the data monitoring station 2.

[0135] Step 301: Set a sampling interval, and collect observation values ​​y of all gas sensors at different times according to the sampling interval.

[0136] Step 302: For each gas sensor q, an observation curve y(q) of the gas sensor q is obtained by fitting according to the observation values ​​y of the gas sensor at different times.

[0137] In this embodiment, the original concentration data of the gas sensor q at any time can be obtained by fitting the observation curve y(q).

[0138] Step 303: The value corresponding to any time in the observation curve y(q) of the gas sensor q is used as the original concentration data of the gas sensor q at any time and transmitted to the data monitoring station 2.

[0139] In this embodiment, the sampling time of the sensor module 1 is aligned, and the sampling interval is once every 1 second after the alignment time. The data monitoring station 2 obtains the observation values ​​of each sensor module 1 at the same time, as shown in Table 1.

[0140] Table 1 Data observation values ​​of each sensor module

[0141]

[0142] Step 4: The data monitoring station 2 transmits the original concentration data of each gas sensor at any time to the single-chip computer module 3 for aggregation, and transmits the aggregated original concentration data in turn through the WiFi communication module 4 and the underground WiFi base station 5 and uploads it to the data storage server 6 for storage.

[0143] In this embodiment, the raw data collected by the sensor module 1 is integrated by the single chip computer module 3 and transmitted to the WiFi communication module 4, and wirelessly transmitted to the WiFi base station 5, completing the uploading of the monitoring data obtained by the data monitoring station 2 and uploading it to the data storage server 6 for storage.

[0144] Step 5: The data processing terminal 7 calls the original concentration data stored in the data storage server 6, and uses the original concentration data to calculate the harmful gas warning value of each location to be detected.

[0145] Step 501: the data processing terminal 7 calls the original concentration data stored in the data storage server 6 as a sample data set, and divides the sample data set into a sample training set and a sample verification set according to a preset ratio.

[0146] In this embodiment, real-time sample training and sample verification are performed by calling the original concentration data in the data storage server 6, wherein the sample training set is 80% of the sample data set, the sample verification set is 20% of the sample data set, and the number of warning data finally obtained is the same as the number of data in the sample verification set. After obtaining the warning value, the location of the gas sensor is classified and warned according to the warning value, the warning location of the warning value, that is, the location of the gas sensor, and the danger level.

[0147] Step 502: Set the first maximum number of iterations and the number of individuals in the whale population, input the sample training set into the classification warning model, and use the whale algorithm to iteratively optimize the classification warning model of the input sample training set to obtain the weights and optimal thresholds of the classification warning model, and then substitute the obtained weights and optimal thresholds into the classification warning model.

[0148] Step 503: Set the second maximum number of iterations, the number of neurons and the learning rate, use the empirical formula to determine the number of nodes in the hidden layer of the classification warning model, and obtain the classification warning model under the current iteration round.

[0149] The number of nodes in the hidden layer of the classification warning model is expressed as:

[0150]

[0151] Where z represents the number of nodes in the hidden layer; m is the number of nodes in the input layer; k is the number of nodes in the output layer; and a is a constant between [1, 10].

[0152] Step 504: Utilize the classification warning model obtained in step 502 to calculate the hazardous gas warning values ​​corresponding to all sample data in the sample training set, and use the root mean square error (RMSE) to calculate the evaluation index of the current iteration round.

[0153] The calculation method of the evaluation index is:

[0154]

[0155] Where n is the number of sample data in the sample training set; i is the i-th sample data in the sample training set; y i is the original concentration data corresponding to the i-th sample data; is the harmful gas warning value obtained by the classification warning model for the i-th sample data.

[0156] In this embodiment, the sample training needs to substitute the sample training set into the classification warning model 701, and use the whale algorithm for iterative training, set the first maximum number of iterations to 1000 times, the number of individuals in the whale population to 30, and use the whale algorithm to optimize the original concentration data of different locations to be detected, and obtain the weight and optimal threshold of the classification warning model 701 suitable for the hydrogen sulfide monitoring environment, and substitute the weight and optimal threshold into the classification warning model 701. Set the second maximum number of iterations to 1000 times, the number of neurons to 6, the learning rate to 0.01, and use formula (7) to calculate the number of nodes in the hidden layer of the classification warning model. In general, to ensure the accuracy of training, a is set to 2, m is set to 2, and k is set to 2, and the number of nodes z in the hidden layer is 4. The RMSE corresponding to each location to be detected in this embodiment is shown in Table 2. The smaller the RMSE, the closer the harmful gas warning value is to the true value, that is, the better the effect of the classification warning model.

[0157] Table 2 RMSE values ​​of different locations to be detected

[0158] Sensor location RMSE value Mine working face return air upper corner 0.0023698 Working face support arm 0.002347 Gas (oil-type gas) extraction drilling site 0.0023591

[0159] Step 505: Determine whether the evaluation index of the current iteration round is greater than the set threshold. If so, return to step 502 until the second maximum iteration round is reached, and use the classification warning model under the last iteration round as the trained classification warning model; otherwise, use the classification warning model under the current round as the trained classification warning model.

[0160] In this embodiment, if the calculated RMSE value is greater than 0.5, the process returns to step 502; if the calculated RMSE value is less than 0.5, the classification warning model in the current round is used as the trained classification warning model.

[0161] Step 506: Input the sample verification set into the trained classification warning model for verification to obtain the final classification warning model.

[0162] Step 507: For any location to be detected, the data processing terminal 7 calls the original concentration data of the location to be detected stored in the data storage server 6 and inputs the final classification warning model to obtain the harmful gas warning value of the location to be detected.

[0163] Step 6: According to the harmful gas warning value of each location to be detected, each location to be detected is divided into a low-risk area, a medium-risk area, a high-risk area or a harmful gas abnormal outflow disaster area. When the location to be detected is a harmful gas abnormal outflow disaster area, the CPU industrial processing chip controls the sound and light alarm to sound an alarm by transmitting a signal to the PLC control module.

[0164] The method for dividing each location to be detected into a low-risk area, a medium-risk area, a high-risk area or a harmful gas abnormal outburst disaster area according to the harmful gas warning value of each location to be detected is as follows: if the harmful gas warning value of a location to be detected is in the interval [0,6.6], the location to be detected is determined as a low-risk area; the unit of the value in the interval is parts per million ppm; if the harmful gas warning value of a location to be detected is in the interval [6.7,49.9], the location to be detected is determined as a medium-risk area; if the harmful gas warning value of a location to be detected is in the interval [50,99.9], the location to be detected is determined as a high-risk area; if the harmful gas warning value of a location to be detected is in the interval [100,+∞], the location to be detected is determined as a harmful gas abnormal outburst disaster area.

[0165] In this embodiment, the harmful gas warning values ​​corresponding to all sensor modules are shown in Table 3, and the hazard classification of the mine harmful gas warning at the location to be detected where each sensor module is located is as follows: Figure 6 shown.

[0166] Table 3 Harmful gas warning values ​​corresponding to all sensor modules

[0167]

[0168] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

1. A mine harmful gas early warning system based on WOA-WNN algorithm, characterized in that: The system includes: an original data collection layer, an original data transmission layer and a data processing terminal; The original data collection layer is used to collect and summarize the original concentration data of harmful gases emitted at different locations during the production process of the mine working face, and transmit the summarized original concentration data to the original data transmission layer; The original data transmission layer is used to establish a channel connection between the original data acquisition layer and the data processing terminal, and transmit the received original concentration data to the data processing terminal; The data processing terminal is used to adopt the wavelet neural network WOA-WNN algorithm improved based on the whale algorithm, and to classify and warn of harmful gases in mines at different locations in the production process of the mine working face according to the received original concentration data.

2. The mine harmful gas early warning system based on the WOA-WNN algorithm according to claim 1 is characterized in that: The raw data acquisition layer includes: a plurality of sensor modules, a data monitoring station, a single-chip microcomputer module and a first WiFi communication module distributed at different locations in the mine working surface; The sensor module is used to collect the original concentration data of harmful gases emitted at different locations during the production process of the mine working face, and transmit the collected original concentration data to the data monitoring station; The data monitoring station is used to transmit all received original concentration data to the single-chip microcomputer module; and upload the original concentration data to the original data transmission layer; The single chip microcomputer module is used to summarize all received original concentration data and transmit the summarized original concentration data to the first WiFi communication module; The first WiFi communication module is used to transmit the aggregated original concentration data to the original data transmission layer via WiFi signals.

3. The mine harmful gas early warning system based on the WOA-WNN algorithm according to claim 2 is characterized in that: The original data transmission layer includes: an underground WiFi base station and a data storage server; The underground WiFi base station is used to receive the aggregated raw concentration data and transmit it to the data storage server; The data storage server is used to store the original concentration data uploaded by the data monitoring station and the aggregated original concentration data from the underground WiFi base station, and transmit all the stored original concentration data to the data processing terminal.

4. The mine harmful gas early warning system based on the WOA-WNN algorithm according to claim 3 is characterized in that: The data processing terminal is a box, including: a classification warning model, a PLC control module, a CPU industrial data processing chip, a second WiFi communication module, a power supply, an audible and visual alarm, a display screen, a power switch, a data acquisition layer switch, an alarm switch, a warning switch and a WiFi transmission switch; The classification warning model, PLC control module, CPU industrial data processing chip, second WiFi communication module and power supply are all embedded in the box; wherein the CPU industrial data processing chip is embedded with the classification warning model; the PLC control module controls the CPU industrial data processing chip, the second WiFi communication module, the power supply, the sound and light alarm, the display screen, the power switch, the data acquisition layer switch, the alarm switch, the warning switch and the WIFI transmission switch through wires for linkage control; The second WIFI module is used to receive original concentration data from a data storage server; The power supply is used to supply power to the data processing terminal; The sound and light alarm is installed on the upper surface of the box body and is used to issue an alarm according to the classification warning results of harmful gases in the mine; The display screen is installed on the front surface of the box body and is used to display the warning value of harmful gases in the mine; The power switch, data acquisition layer switch, alarm switch, early warning switch and WIFI transmission switch are installed side by side in sequence on the front surface of the box and are located below the display screen; wherein the power switch is used to control the on-off state of the power supply; the data acquisition layer switch is used to control the on-off state of the sensor module in the original data acquisition layer through a wireless signal; the alarm switch is used to control the on-off state of the sound and light alarm; the early warning switch is used to control the on-off state of the CPU industrial data processing chip; the WIFI transmission switch is used to control the on-off state of the first WiFi communication module and the second WiFi communication module.

5. The mine harmful gas early warning system based on the WOA-WNN algorithm according to claim 4 is characterized in that: The CPU industrial data processing chip is used to build a classification warning model according to the WOA-WNN algorithm, and input the original concentration data received by the data processing terminal into the classification warning model to obtain the warning value of harmful gas in the mine, and send the obtained warning value to the display screen in the form of a data graph for display; at the same time, according to the warning value, the position of the sensor module corresponding to the warning value is divided into a low-risk area, a medium-risk area, a high-risk area or a harmful gas abnormal outflow disaster area; when the position is a harmful gas abnormal outflow disaster area, the sound and light alarm is controlled to sound an alarm by transmitting a signal to the PLC control module.

6. The mine harmful gas early warning system based on the WOA-WNN algorithm according to claim 5 is characterized in that: The process of constructing the classification warning model according to the WOA-WNN algorithm is as follows: constructing a wavelet neural network consisting of an input layer, a hidden layer and an output layer; wherein the input layer and the output layer are responsible for the input and output of the original concentration data respectively; the hidden layer is used to calculate the warning value of the harmful gas in the mine according to the input original concentration data; Assuming that the number of nodes in the input layer, hidden layer, and output layer are n, i, and m respectively, when the original concentration data is transferred from the input layer to the hidden layer, for any node j in the hidden layer, the original concentration data input in the node is calculated as: Where S j is the original concentration data input in node j; k represents the kth node in the input layer; ω kj is the input layer connection weight of node j; x k represents the original concentration data transmitted by the kth node in the input layer; The Morlet wavelet is selected as the base wave, and the original concentration data input in the node j of the hidden layer is scaled by the wavelet basis at the node, and the output data of the node j is obtained as follows: h j =φ[(S j -b j ) / a j ] Where φ(x) is the wavelet basis function; x represents the variable in the wavelet basis function; e is a natural constant; h j is the output data of the jth node in the hidden layer; b j is the smoothing factor of the wavelet basis function; a j is the scaling factor of the wavelet basis function; The output data of all nodes in the hidden layer are transferred to the output layer, and the output data of each node in the output layer is calculated. For any node t in the output layer, the output data of node t is: where y t is the output data of node t; Represents the function of calculating the output data of the nodes in the output layer; ω jt is the hidden layer connection weight of node t; According to the output data of all nodes in the output layer and the output layer connection weight ω of each node t Calculate the harmful gas warning value at the location of each sensor module; The smoothing factor b of the wavelet basis function for any node j in the hidden layer of the wavelet neural network is j and the scaling factor a of the wavelet basis function j , and the hidden layer connection weight ω of any node t in the output layer jt And the output layer connection weight ω t As a group of parameters to be optimized, the whale algorithm is used to optimize all the parameters to be optimized, the optimal solution of each group of parameters to be optimized is substituted into the wavelet neural network, and the wavelet neural network obtained after substituting the optimal solution is used as a classification warning model.

7. The mine harmful gas early warning system based on the WOA-WNN algorithm according to claim 6 is characterized in that: The process of optimizing all parameters to be optimized by using the whale algorithm is as follows: for each group of parameters to be optimized, a number of whale individuals are randomly generated and the fitness of each whale individual is calculated, and the positions of all whale individuals are updated according to the whale individual with the highest fitness; a random number P is generated in the range of [0,1], and it is determined whether the random number P is less than a set threshold value. If the random number P is less than the set threshold value, it is determined whether the absolute value of the coefficient vector |A| is greater than or equal to 1; If the random number P is not less than the set threshold, the prey encirclement stage is entered, and the position of the best whale individual in the prey encirclement stage is calculated; if the random number P is less than the set threshold, it is determined whether the absolute value of the coefficient vector |A| is greater than or equal to 1. If |A|>=1, it enters the random search for the optimal solution stage, and calculates the position of the best whale individual in the random search for the optimal solution stage; if |A|<1, it enters the encirclement of the optimal solution stage, and calculates the position of the best whale individual in the encirclement of the optimal solution stage; it is determined whether the current iteration round reaches the preset number of iterations. If so, the position of the best whale individual in the current round is used as the optimal solution of the group of parameters to be optimized; if not, the positions of all whale individuals are updated according to the position of the best whale individual in the current round and the next iteration is performed until the preset number of iterations is reached.

8. A mine harmful gas early warning method based on the WOA-WNN algorithm, which is implemented by the mine harmful gas early warning system based on the WOA-WNN algorithm, characterized in that: The method comprises the following steps: Step 1: Obtain several locations to be detected on the working surface of the mine, and arrange a gas sensor for detecting harmful gases at each location to be detected; Step 2: Turn on the power switch, data acquisition layer switch, alarm switch, early warning switch, and WIFI transmission switch on the box; Step 3: By aligning the sampling time of the gas sensors, the original concentration data of each gas sensor at any time is obtained and transmitted to the data monitoring station; Step 4: The data monitoring station transmits the original concentration data of each gas sensor at any time to the single-chip microcomputer module for aggregation, and transmits the aggregated original concentration data in turn through the WiFi communication module and the underground WiFi base station and uploads it to the data storage server for storage; Step 5: The data processing terminal calls the original concentration data stored in the data storage server, and uses the original concentration data to calculate the harmful gas warning value of each location to be detected; Step 6: According to the harmful gas warning value of each location to be detected, each location to be detected is divided into a low-risk area, a medium-risk area, a high-risk area or a harmful gas abnormal outflow disaster area. When the location to be detected is a harmful gas abnormal outflow disaster area, the CPU industrial processing chip controls the sound and light alarm to sound an alarm by transmitting a signal to the PLC control module.

9. The mine harmful gas early warning method based on WOA-WNN algorithm according to claim 8, characterized in that: The step 3 further comprises: Step 301: setting a sampling interval, and collecting observation values ​​y of all gas sensors at different times according to the sampling interval; Step 302: for each gas sensor q, obtain an observation curve y(q) of the gas sensor q by fitting according to the observation values ​​y of the gas sensor at different times; Step 303: The value corresponding to any time in the observation curve y(q) of the gas sensor q is used as the original concentration data of the gas sensor q at any time and transmitted to the data monitoring station.

10. A mine harmful gas early warning method based on WOA-WNN algorithm according to claim 9, characterized in that: The step 5 further comprises: Step 501: The data processing terminal calls the original concentration data stored in the data storage server as a sample data set, and divides the sample data set into a sample training set and a sample verification set according to a preset ratio; Step 502: Set the first maximum number of iterations and the number of individuals in the whale population, input the sample training set into the classification warning model, and use the whale algorithm to iteratively optimize the classification warning model of the input sample training set to obtain the weight and optimal threshold of the classification warning model, and then substitute the obtained weight and optimal threshold into the classification warning model; Step 503: setting the second maximum number of iterations, the number of neurons and the learning rate, using an empirical formula to determine the number of nodes in the hidden layer of the classification warning model, and obtaining the classification warning model under the current iteration round; Step 504: using the classification warning model obtained in step 502 to calculate the harmful gas warning value corresponding to all sample data in the sample training set, and using the root mean square error RMSE to calculate the evaluation index of the current iteration round; Step 505: Determine whether the evaluation index of the current iteration round is greater than the set threshold. If so, return to step 502 until the second maximum iteration round is reached, and use the classification warning model under the last iteration round as the trained classification warning model; if not, use the classification warning model under the current round as the trained classification warning model; Step 506: input the sample verification set into the trained classification warning model for verification to obtain the final classification warning model; Step 507: For any location to be detected, the data processing terminal calls the original concentration data of the location to be detected stored in the data storage server and inputs the final classification warning model to obtain the harmful gas warning value of the location to be detected.

Images