Coal mine underground CO sensor early warning diagnosis method and device

Abstract

The application discloses a coal mine underground CO sensor early warning diagnosis method and device, which comprises the following steps: a first step of data acquisition; a second step of acquiring a characteristic input vector; a third step of designing a topological structure of a back propagation neural network; a fourth step of creating a neural network model; and a fifth step of training the neural network model. By installing the device beside the CO sensor, each state of a fuel vehicle passing through the CO sensor can be recognized, early warning can be realized, and diagnosis on carbon monoxide alarm can be performed.

Description

Technical Field

[0001] The present invention relates to the technical field of monitoring devices, and in particular to an early warning diagnosis method and device for a CO sensor in an underground coal mine. Background Art

[0002] Currently, large equipment and consumables in mines are typically transported by fuel-powered trucks. However, these trucks generate large amounts of carbon monoxide in closed tunnels. When these trucks pass through tunnels equipped with CO sensors, the sensors trigger alarms. Currently, these trucks can only be manually marked on the host computer for advance warning, without any early warning. Furthermore, the host computer cannot intelligently diagnose the carbon monoxide alarm caused by the passage of fuel-powered trucks. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art.

[0004] To this end, the present invention proposes a coal mine underground CO sensor early warning diagnosis method and device. By installing the coal mine underground CO sensor early warning diagnosis device next to the CO sensor, the various states of the fuel vehicle passing by the CO sensor can be identified, thereby providing early warning and diagnosing the carbon monoxide alarm.

[0005] The method for early warning and diagnosis of CO2 in underground coal mines according to an embodiment of the present invention includes the following steps:

[0006] Step 1, data collection: Use the sound sensor, acceleration sensor, photoelectric sensor, NO sensor, and NO2 sensor to detect sound characteristics, acceleration characteristics, light intensity, nitric oxide, and nitrogen dioxide respectively. The data collection frequency of the sound sensor, acceleration sensor, photoelectric sensor, NO sensor, and NO2 sensor is f;

[0007] Step 2: Get the feature input vector: Select the time slice window t; First, perform FFT on the sound sensor and acceleration sensor to obtain the frequency domain characteristics of the sound and acceleration, and select the frequency value with the largest amplitude. As feature labels; Indicates the frequency value corresponding to the maximum amplitude after the sound signal is fast Fourier transformed; Indicates the frequency value corresponding to the maximum amplitude after the fast Fourier transform of the acceleration signal; secondly, the time slice window selects the maximum value of the sound sensor, acceleration sensor, photoelectric sensor, NO sensor, NO2 sensor, and CO sensor As feature labels; It represents the maximum value obtained by the sound sensor within time t; Indicates the maximum value obtained by the acceleration sensor within time t; It represents the maximum value obtained by the photoelectric sensor within time t; It represents the maximum value obtained by the nitric oxide sensor within time t; Indicates the maximum value obtained by the nitrogen dioxide sensor within time t; Represents the maximum value obtained by the carbon monoxide sensor within time t; Finally, set the input vector to ;

[0008] Step 3: Design the topology of the back propagation neural network: the topology includes the number of network layers and the number of nodes in each layer, wherein the network layers include an input layer, a hidden layer and an output layer; the number of nodes in the input layer is m=8, and the eigenvector is The number of nodes in the output layer is n=3, and the output layer vector is represented by [0, 1, 2], where 0 means no fuel vehicle is detected; 1 means a fuel vehicle is detected approaching, which is used as a warning mark; 2 means a fuel vehicle is detected passing by, which is used as a carbon monoxide alarm diagnosis;

[0009] Step 4: Create a neural network model;

[0010] Step 5: Train the neural network model.

[0011] According to an embodiment of the present invention, a coal mine underground CO sensor early warning and diagnostic device comprises a housing, an aerial plug module and a detection module, wherein the aerial plug module and the detection module are both integrated on the housing; the detection module comprises a sound sensor, an acceleration sensor, a photoelectric sensor, an NO sensor and a NO2 sensor; the aerial plug module is used to upload the data collected by the detection module to a host computer in real time and obtain CO sensor data; the sound sensor is used to detect sound characteristics, wherein the sound characteristics include the loudness and frequency of the roaring sound; the acceleration sensor is used to detect acceleration characteristics, wherein the acceleration characteristics include vibration amplitude and frequency; the photoelectric sensor is used to monitor light intensity; the NO sensor is used to detect the concentration of nitric oxide; and the NO2 sensor is used to detect the concentration of nitrogen dioxide.

[0012] The beneficial effects of the present invention are that the coal mine underground CO sensor early warning diagnosis method and device of the present invention can intelligently diagnose whether a fuel vehicle passes through the coal mine, thereby giving early warning to the CO sensor and realizing carbon monoxide alarm diagnosis; feature detection parameters are selected according to the movement characteristics of the fuel vehicle in the coal mine, and data fusion is performed, and the back propagation neural network algorithm is used for identification, which greatly improves the reliability of identification; it can be applied not only to the typical scenario of carbon monoxide alarm caused by the passage of fuel vehicles in the coal mine, but also to the scenario of carbon monoxide alarm caused by blasting in the coal mine.

[0013] According to one embodiment of the present invention, in step 3, the number of nodes in the hidden layer is , the empirical formula is:

[0014] (1)

[0015] The symbols in formula (1) mean:

[0016] Indicates the number of nodes in the input layer;

[0017] Indicates the number of nodes in the output layer;

[0018] represents the number of nodes in the hidden layer, .

[0019] According to one embodiment of the present invention, in step 3, the feature vector The corresponding weight vector is , the initial value of the weight is initialized using Xavier, the activation function of the hidden layer uses the sigmoid function, and the activation function of the output layer uses the softmax function.

[0020] According to one embodiment of the present invention, in step 4, the mean square error function is selected as the loss function, and the calculation formula is as follows:

[0021] (2)

[0022] The symbols in formula (2) mean:

[0023] represents the mean square error;

[0024] Indicates the number of nodes in the output layer;

[0025] Represents the total number of training sample sets;

[0026] Indicates the expected output value;

[0027] Indicates the actual output value;

[0028] Indicates the index number of the output unit.

[0029] According to one embodiment of the present invention, in step 5, the neural network model is trained. The specific process is:

[0030] Step 1: Given the input feature quantity and expected output, initialize the weights;

[0031] Step 2: Input the training sample set;

[0032] Step ③: Calculate the output of hidden layer and output layer units;

[0033] Step ④: Calculate the deviation between the expected value and the actual output;

[0034] Step 5: Determine whether the deviation meets the requirements:

[0035] When the deviation is judged to meet the requirements, the total error is judged to meet the requirements. If the total error meets the requirements, the neural network model training process is terminated; if the total error does not meet the requirements, the network parameters are optimized and the process returns to step ②.

[0036] When the judgment deviation does not meet the requirements, the network parameters are optimized and the process returns to step ②.

[0037] According to one embodiment of the present invention, after the data acquisition frequency f and the time slice window t are determined, a first timing time 1 / f of the timer and a second timing time t of the timer are set.

[0038] According to one embodiment of the present invention, when the first timing time 1 / f of the judgment timer arrives, each sensor collects data signals.

[0039] According to one embodiment of the present invention, when the second timing time t of the timer is reached, the data sequence of each sensor is obtained. , and then perform data filtering.

[0040] According to one embodiment of the present invention, the device further comprises a digital tube module, and the digital tube module is used to display the detection parameters.

[0041] Other features and advantages of the present invention will be described in the following description, and in part will 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.

[0042] 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

[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. 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 any creative work.

[0044] Figure 1 This is a schematic structural diagram of the CO sensor early warning and diagnostic device for underground coal mines according to the present invention;

[0045] Figure 2 It is a flowchart of feature input vector acquisition;

[0046] Figure 3 It is a flowchart of neural network model training.

[0047] The numbers in the figure are: 1. Shell; 2. Aviation plug-in module; 3. Digital tube module; 4. Sound sensor; 5. Acceleration sensor; 6. Photoelectric sensor; 7. NO sensor; 8. NO2 sensor. DETAILED DESCRIPTION

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only 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 making creative efforts shall fall within the scope of protection of the present invention.

[0049] In the description of the present invention, it should be understood that the terms "one side", "the other side", "both sides", "between", "middle", "upper end", "lower end", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0050] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "disposed" and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections, and may be directly connected or indirectly connected through an intermediate medium. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0051] The following describes in detail the coal mine underground CO sensor early warning diagnosis method and device according to the embodiment of the present invention with reference to the accompanying drawings, mainly addressing the problem that the passage of fuel vehicles in coal mines causes carbon monoxide alarms, but cannot provide early warning and intelligent diagnosis.

[0052] See Figure 1The CO2 sensor early warning and diagnostic device for an underground coal mine includes a housing 1, an air plug module 2, a digital tube module 3, and a detection module. The air plug module 2, the digital tube module 3, and the detection module are all integrated into the housing 1. The detection module includes an acoustic sensor 4, an acceleration sensor 5, a photoelectric sensor 6, an NOx sensor 7, and a NO2 sensor 8. It should be noted that the CO2 sensor is not integrated into the housing 1. The interface of the air plug module 2 of the underground coal mine CO2 sensor early warning and diagnostic device can be used to interact with the nearby CO2 sensor to obtain data.

[0053] Among them, the CO sensor is a carbon monoxide sensor; the NO sensor 7 is a nitric oxide sensor; and the NO2 sensor is a nitrogen dioxide sensor.

[0054] The aerial plug module 2 is the power supply interface and communication interface of the device. It can upload the collected data to the host computer in real time and obtain the CO sensor data nearby through the communication interface. That is, the aerial plug module 2 is used to upload the data collected by the detection module to the host computer in real time and obtain the CO sensor data. The sound sensor module 4 is used to detect sound characteristics, including the loudness and frequency of the roar. The acceleration sensor module 5 is used to detect acceleration characteristics, including vibration amplitude and frequency. The photoelectric sensor module 6 is used to monitor light intensity. The NO sensor 7 is used to detect the concentration of nitric oxide; the NO2 sensor 8 is used to detect the concentration of nitrogen dioxide. The digital tube module 3 is used to realize the display of detection parameters, and can display the real-time values ​​of the detection module in turn to facilitate human-computer interaction.

[0055] When a fuel truck travels through a tunnel, it often exhibits the following characteristics: 1. The internal combustion engine of a fuel truck emits a rumbling sound during operation. Due to the Doppler effect, when the truck approaches a CO sensor, the low-frequency signal primarily changes to a high-frequency signal; when the truck moves away from the CO sensor, the high-frequency signal primarily changes to a low-frequency signal. The present invention uses an acoustic sensor 4 to detect the characteristics of the rumbling sound, including the loudness and frequency of the rumbling sound. 2. When a fuel truck passes through a chamber equipped with a CO sensor, the suspended CO sensor may vibrate. The present invention uses an acceleration sensor 5 to detect the vibration amplitude and frequency. 3. Mines are often dark, and fuel trucks typically operate with their lights on. The present invention uses a photoelectric sensor 6 to monitor light intensity. 4. While traveling, fuel trucks produce not only incompletely combusted carbon monoxide but also significant nitrogen oxides, such as nitric oxide and nitrogen dioxide. The present invention uses an NO sensor 7 and a NO2 sensor 8 to detect these characteristic gases (nitric oxide and nitrogen dioxide).

[0056] See Figure 2, a coal mine underground CO sensor early warning diagnosis method, comprising the following steps:

[0057] Step 1, data acquisition: Use various sensors to collect data in real time, with an acquisition frequency f, preferably 1000 Hz; specifically, use a sound sensor, an acceleration sensor, a photoelectric sensor, a NO sensor, and a NO2 sensor to detect sound characteristics, acceleration characteristics, light intensity, nitric oxide, and nitrogen dioxide, respectively, and the data acquisition frequency of the sound sensor, the acceleration sensor, the photoelectric sensor, the NO sensor, and the NO2 sensor is f;

[0058] Step 2: Get the feature input vector:

[0059] Select a time slice window t = 1s (t represents time, s represents seconds). After determining the data acquisition frequency f and the time slice window t, set the first timer duration 1 / f and the second timer duration t. Set and enable the timer interrupt. The timer is installed in the MCU module within the CO2 sensor early warning and diagnostic device in the coal mine.

[0060] When it is determined that the first timing time 1 / f of the timer has arrived (ie, the first timing time 1 / f of the timer is interrupted), each sensor collects data signals.

[0061] When the second timing time t of the timer is reached (i.e., the timer interrupts the processing at the second timing time t), the data sequence of each sensor is obtained. Then the processor inside the underground CO2 sensor early warning and diagnosis device of the coal mine performs data filtering, sets the data collection completion flag flag = 1, and then exits the interrupt.

[0062] The main function main() first determines whether the data collection completion flag is 1: if the flag is not 1, the process ends; if the flag is 1, the data collection completion flag is cleared first, and then the following operations are performed;

[0063] Perform FFT on the sound sensor data sequence and the acceleration sensor data sequence to obtain the sound frequency and acceleration frequency corresponding to the maximum amplitude: First, perform FFT on the sound sensor and the acceleration sensor to obtain the frequency domain characteristics of the sound and acceleration, and select the frequency value with the largest amplitude. As feature labels; Indicates the frequency value corresponding to the maximum amplitude after the sound signal is fast Fourier transformed; Indicates the frequency value corresponding to the maximum amplitude of the acceleration signal after fast Fourier transform;

[0064] Find the maximum value of each sensor data sequence: Secondly, select the maximum value of the sound sensor, acceleration sensor, photoelectric sensor, NO sensor, NO2 sensor, and CO sensor in the time slice window As feature labels; It represents the maximum value obtained by the sound sensor within time t; Indicates the maximum value obtained by the acceleration sensor within time t; It represents the maximum value obtained by the photoelectric sensor within time t; It represents the maximum value obtained by the nitric oxide sensor within time t; Indicates the maximum value obtained by the nitrogen dioxide sensor within time t; Indicates the maximum value obtained by the carbon monoxide sensor within time t;

[0065] Composition of feature input vector: Finally, set the input vector to ;

[0066] Step 3: Design the topology of the back-propagation neural network: The topology includes the number of network layers and the number of nodes in each layer. The network layers include an input layer, a hidden layer, and an output layer.

[0067] The number of nodes in the input layer is m=8, and the eigenvector is The number of nodes in the output layer is n=3, and the output layer vector is represented by [0,1,2], where 0 means no fuel vehicle is detected; 1 means a fuel vehicle is detected approaching, which is used as a warning mark; 2 means a fuel vehicle is detected passing by, which is used as a carbon monoxide alarm diagnosis;

[0068] In step 3, the number of nodes in the hidden layer is , the empirical formula is:

[0069] (1)

[0070] The symbols in formula (1) mean:

[0071] Indicates the number of nodes in the input layer;

[0072] Indicates the number of nodes in the output layer;

[0073] represents the number of nodes in the hidden layer, .

[0074] Eigenvector The corresponding weight vector is , the initial value of the weight is initialized using Xavier, the activation function of the hidden layer uses the sigmoid function, and the activation function of the output layer uses the softmax function.

[0075] Step 4: Create a neural network model;

[0076] In step 4, the mean square error function is selected as the loss function, and the calculation formula is as follows:

[0077] (2)

[0078] The symbols in formula (2) mean:

[0079] represents the mean square error;

[0080] Indicates the number of nodes in the output layer;

[0081] Represents the total number of training sample sets;

[0082] Indicates the expected output value;

[0083] Indicates the actual output value;

[0084] Indicates the index number of the output unit.

[0085] The sample set refers to randomly selecting 30% of the samples in the complete data set as the test sample set, and the remaining 70% of the samples as the training sample set.

[0086] Step 5: Train the neural network model:

[0087] In step 5, the ADAM algorithm is selected to update the network parameters to improve the accuracy and generalization ability of the model.

[0088] See Figure 3 , training the neural network model, the specific process is:

[0089] Step 1: Given the input feature quantity and expected output, initialize the weights;

[0090] Step 2: Input the training sample set;

[0091] Step ③: Calculate the output of hidden layer and output layer units;

[0092] Step ④: Calculate the deviation between the expected value and the actual output;

[0093] Step 5: Determine whether the deviation meets the requirements:

[0094] When the deviation is judged to meet the requirements, the total error is judged to meet the requirements. If the total error meets the requirements, the neural network model training process is terminated; if the total error does not meet the requirements, the network parameters are optimized and the process returns to step ②.

[0095] When the judgment deviation does not meet the requirements, the network parameters are optimized and the process returns to step ②.

[0096] The above are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solutions and inventive concepts of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for early warning diagnosis of CO2 sensor in coal mines, characterized in that: A CO sensor early warning and diagnostic device is installed next to the CO sensor in a coal mine to identify and provide early warning when a fuel vehicle passes by. The device includes the following steps: Step 1, data collection: Use the sound sensor, acceleration sensor, photoelectric sensor, NO sensor, and NO2 sensor to detect sound characteristics, acceleration characteristics, light intensity, nitric oxide, and nitrogen dioxide respectively. The data collection frequency of the sound sensor, acceleration sensor, photoelectric sensor, NO sensor, and NO2 sensor is f; Step 2: Get the feature input vector: Select the time slice window t; First, perform FFT on the sound sensor and acceleration sensor to obtain the frequency domain characteristics of the sound and acceleration, and select the frequency value with the largest amplitude. As feature labels; Indicates the frequency value corresponding to the maximum amplitude after the sound signal is fast Fourier transformed; Indicates the frequency value corresponding to the maximum amplitude after the fast Fourier transform of the acceleration signal; secondly, the time slice window selects the maximum value of the sound sensor, acceleration sensor, photoelectric sensor, NO sensor, NO2 sensor, and CO sensor As feature labels; It represents the maximum value obtained by the sound sensor within time t; Indicates the maximum value obtained by the acceleration sensor within time t; It represents the maximum value obtained by the photoelectric sensor within time t; It represents the maximum value obtained by the nitric oxide sensor within time t; Indicates the maximum value obtained by the nitrogen dioxide sensor within time t; Represents the maximum value obtained by the carbon monoxide sensor within time t; Finally, set the input vector to ; Step 3: Design the topology of the back propagation neural network: the topology includes the number of network layers and the number of nodes in each layer, wherein the network layers include an input layer, a hidden layer and an output layer; the number of nodes in the input layer is m=8, and the eigenvector is The number of nodes in the output layer is n=3, and the output layer vector is represented by [0, 1, 2], where 0 means no fuel vehicle is detected; 1 means a fuel vehicle is detected approaching, which is used as a warning mark; 2 means a fuel vehicle is detected passing by, which is used as a carbon monoxide alarm diagnosis; Step 4: Create a neural network model; Step 5: Train the neural network model.

2. The method for early warning diagnosis of CO2 sensor in underground coal mines according to claim 1, characterized in that: In step 3, the number of nodes in the hidden layer is , the empirical formula is: （1） The symbols in formula (1) mean: Indicates the number of nodes in the input layer; Indicates the number of nodes in the output layer; represents the number of nodes in the hidden layer, .

3. The method for early warning diagnosis of CO2 sensor in underground coal mines according to claim 2, characterized in that: In step 3, the eigenvector The corresponding weight vector is , the initial value of the weight is initialized using Xavier, the activation function of the hidden layer uses the sigmoid function, and the activation function of the output layer uses the softmax function.

4. The method for early warning diagnosis of CO2 in underground coal mines according to claim 1, characterized in that: In step 4, the mean square error function is selected as the loss function, and the calculation formula is as follows: （2） The symbols in formula (2) mean: represents the mean square error; Indicates the number of nodes in the output layer; Represents the total number of training sample sets; Indicates the expected output value; Indicates the actual output value; Indicates the index number of the output unit.

5. The method for early warning diagnosis of CO2 sensor in underground coal mines according to claim 1, characterized in that: In step 5, the neural network model is trained. The specific process is: Step 1: Given the input feature quantity and expected output, initialize the weights; Step 2: Input the training sample set; Step ③: Calculate the output of hidden layer and output layer units; Step ④: Calculate the deviation between the expected value and the actual output; Step 5: Determine whether the deviation meets the requirements: When the deviation is judged to meet the requirements, the total error is judged to meet the requirements. If the total error meets the requirements, the neural network model training process is terminated. If the total error does not meet the requirements, the network parameters are optimized and the process returns to step ②. When the judgment deviation does not meet the requirements, the network parameters are optimized and the process returns to step ②.

6. The method for early warning diagnosis of CO2 sensor in underground coal mines according to claim 1, characterized in that: After determining the data acquisition frequency f and the time slice window t, set the timer first timing time 1 / f and the timer second timing time t.

7. The method for early warning diagnosis of CO2 sensor in underground coal mines according to claim 1, characterized in that: When the first timing time 1 / f of the judgment timer arrives, each sensor collects data signals.

8. The method for early warning diagnosis of CO2 sensor in underground coal mines according to claim 1, characterized in that: When the second timing time t of the judgment timer arrives, the data sequence of each sensor is obtained , and then perform data filtering.

9. A device for early warning diagnosis of CO2 sensor in a coal mine, which implements the method for early warning diagnosis of CO2 sensor in a coal mine according to claim 1, characterized in that: The device comprises a housing (1), an air plug module (2), and a detection module, wherein the air plug module (2) and the detection module are both integrated on the housing (1); the detection module comprises a sound sensor (4), an acceleration sensor (5), a photoelectric sensor (6), an NO sensor (7), and a NO2 sensor (8); The aerial plug module (2) is used to upload the data collected by the detection module to the host computer in real time and obtain the CO sensor data; The sound sensor (4) is used to detect sound characteristics, wherein the sound characteristics include the loudness and frequency of the roaring sound; The acceleration sensor (5) is used to detect acceleration characteristics, wherein the acceleration characteristics include vibration amplitude and frequency; The photoelectric sensor (6) is used to monitor light intensity; The NO sensor (7) is used to detect the concentration of nitric oxide; The NO2 sensor (8) is used to detect the concentration of nitrogen dioxide.

10. The CO sensor early warning and diagnostic device for underground coal mines according to claim 9, characterized in that: The device further comprises a digital tube module (3), and the digital tube module (3) is used to realize the display of detection parameters.

Images