Mine dust concentration real-time monitoring method

Abstract

The invention discloses a mine dust concentration real-time monitoring method, and belongs to the field of dust falling. The method comprises the following steps: acquiring dust data of each dust sensor and a preset mine tunnel model to obtain all adjacent dust sensors of any one dust sensor; acquiring an accumulated distance matrix between the variable quantity sequences of any one dust sensor and the adjacent dust sensor, and acquiring inflection points by utilizing the accumulated distance matrix; calculating the influence coefficient value of any dust sensor and the adjacent dust sensor by using the inflection point, performing weighted prediction by using the influence coefficient values of all the adjacent dust sensors of each dust sensor to obtain the dust concentration prediction value of each dust sensor, and calculating the dust concentration prediction value of each dust sensor according to the dust concentration prediction value of each dust sensor. And real-time monitoring of the mine dust concentration is completed. When the method is used for dust falling control, a stable and accurate dust concentration value prediction result can be obtained, so that a dust concentration monitoring result has higher robustness.

Description

Technical Field

[0001] The present invention relates to the technical field of dust suppression, and in particular to a real-time monitoring method for mine dust concentration. Background Art

[0002] During mine operations, mine dust poses a high risk to mine workers. It not only has a health impact on the workers, but also has a safety impact on the operation safety. Therefore, it is necessary to monitor the mine dust in real time and control the mine dust concentration within the operation requirements by using dust reduction equipment.

[0003] Traditional mine dust monitoring mainly relies on sensor data. Due to the complex tunnel structure and the time and space delays in dust diffusion, the monitoring results are seriously lagging, unable to reflect the laws of pollution propagation, and the monitoring results are large in error.

[0004] In order to solve the above problems, the existing technology relies on the historical data of each sensor data for data prediction. However, due to the many factors that cause dust generation in the mine operation content, such as the movement of mine cars in the mine, local ore body shedding and other factors, the dust concentration changes are highly random. If only relying on the historical data of each sensor data for data prediction, it is difficult to adapt to real-time working conditions, resulting in large errors in the prediction results. When using mine dust concentration monitoring for dust reduction, it is difficult to achieve stable mine dust monitoring and cause waste of dust reduction resources. Summary of the Invention

[0005] The purpose of the present invention is to provide a real-time monitoring method for dust concentration in a mine, so as to solve the problem that the existing technology cannot effectively realize real-time and stable monitoring of dust concentration.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for real-time monitoring of dust concentration in a mine, comprising the following steps:

[0008] S1: Obtain dust data from each dust sensor and a preset mine tunnel model. Based on the mine tunnel model and the coordinate values ​​of each dust sensor, obtain all neighboring dust sensors of any dust sensor. The dust data includes dust concentration values ​​and coordinate values.

[0009] S2: Calculate the change in dust concentration values ​​of each dust sensor at consecutive moments and construct a change sequence to obtain the maximum dust concentration change rate;

[0010] S3: Obtain a cumulative distance matrix between the variation sequence of any dust sensor and the variation sequence of a single adjacent dust sensor, and calculate the inflection point of the shortest distance path in the cumulative distance matrix;

[0011] S4: Obtain the collection time point and cumulative distance value of the inflection point, obtain the time difference and the cumulative distance value change rate, calculate the influence coefficient value of any dust sensor and a single adjacent dust sensor based on the time difference and cumulative distance value change rate and the maximum dust concentration change rate, and obtain the maximum value of all influence coefficient values ​​of any dust sensor;

[0012] S5: Perform weighted prediction using the influence coefficient values ​​of all neighboring dust sensors of each dust sensor to obtain a dust concentration prediction value of each dust sensor and perform dust reduction control.

[0013] Furthermore, the mine tunnel model in S1 is provided with GPS positioning information, which comes from the GPS locator installed on each dust sensor. The three-dimensional mine tunnel model is obtained through data mapping, and the longitude and latitude coordinate values ​​in the GPS data corresponding to different mapping positions are obtained through data calibration during mapping.

[0014] Furthermore, each dust sensor is arranged in the mine, and the collection frequency of each dust sensor is once per second. The dust concentration value is continuously collected by each dust sensor to obtain the dust concentration value collected by each dust sensor at a continuous moment, wherein the first The dust sensor is in the The dust concentration at the moment is .

[0015] Furthermore, the method for obtaining the maximum dust concentration change rate in S2 is:

[0016] S201: Obtain a dust concentration value sequence of a single dust concentration sensor within a time period, calculate a first-order difference sequence of the dust concentration value sequence, and similarly calculate a first-order difference sequence of a single dust sensor adjacent to the single dust concentration sensor;

[0017] S202: Obtaining the maximum value of the first-order difference sequence of adjacent single dust sensors;

[0018] S203: Obtain the mean of the first-order difference sequence of the dust concentration value sequence;

[0019] S204: Calculate the ratio of the maximum value in S202 to the average value in S203 as the maximum dust concentration change rate.

[0020] Furthermore, the calculation method of the influence coefficient value in S4 includes:

[0021] S401: Obtain a first-order difference sequence of a single dust concentration sensor, and similarly calculate the first-order difference sequence of a single dust sensor adjacent to the single dust concentration sensor;

[0022] S402: Calculate the cumulative distance matrix between the first-order difference sequences of the target dust sensor and its neighboring dust sensors, and calculate the shortest distance path of the cumulative distance matrix;

[0023] S403: Obtain a stable inflection point according to the shortest distance path;

[0024] S404: Obtain the time difference between the mean of the time point corresponding to the inflection point and the current moment, and obtain the ratio of the cumulative distance value in the cumulative distance matrix corresponding to the inflection point to the cumulative distance value calculated by the dynamic time warping algorithm;

[0025] S405: Calculate the influence coefficient value between a single dust concentration sensor and a single adjacent dust sensor using the negative correlation between the maximum dust concentration change rate and the time difference and the negative correlation between the ratio.

[0026] Furthermore, obtaining a stable inflection point in S403 includes:

[0027] a. Perform linear fitting on all two-dimensional data points along the shortest distance path using the least squares method to obtain a fitting function for the shortest distance path. Calculate the Euclidean distance from each two-dimensional data point along the shortest distance path to the fitting function to obtain a Euclidean distance sequence. Smooth the Euclidean distance sequence using the moving average method. Calculate the inflection point of the smoothed Euclidean distance sequence to obtain the two-dimensional data corresponding to the inflection point.

[0028] b. Similarly, obtain the inflection point calculation results for each of the previous X moments;

[0029] c. Obtain the maximum and minimum values ​​of the two-dimensional data of each inflection point in the first-order difference sequence of a single dust sensor and the first-order difference sequence of an adjacent single dust sensor, and calculate the difference between the maximum value and the minimum value;

[0030] d. If the response difference is less than the difference threshold, the current inflection point is determined to be stable.

[0031] Furthermore, the calculation method of the dust concentration prediction value in S5 includes:

[0032] S501: Calculate the maximum influence coefficient value of each dust sensor adjacent to each dust sensor using The normalization algorithm is used to normalize the data and use the weighted moving average method as the prediction weight value to predict the dust concentration of each dust sensor;

[0033] S502: Using the real-time dust concentration values ​​of the adjacent dust sensors to replace the historical data required by the weighted moving average method; using the normalized maximum influence coefficient value of each dust sensor adjacent to each dust sensor as a weight, and multiplying it with the dust concentration value of each dust sensor adjacent to each dust sensor at the time to be predicted, to obtain multiple multiplication results, and accumulating all the multiplication results to obtain the dust concentration prediction value of each dust sensor at the time to be predicted;

[0034] S503: Obtain the dust concentration prediction value and the actual dust concentration value of each dust sensor at the latest moment, and take the maximum value of the dust concentration prediction value and the actual dust concentration value as the final dust concentration value.

[0035] Furthermore, the calculation formula for the dust concentration prediction value in S5 is as follows:

[0036] For the first Moment The dust sensors are weighted to predict the Moment The dust concentration prediction value of each dust sensor is , , where is the number, indicating the The dust sensor next to the A dust sensor, For the The total number of all neighboring dust sensors of a dust sensor;

[0037] For the The dust sensor next to the A dust sensor for the The maximum influence coefficient value after normalization of the dust sensor. The larger the value, the higher the A dust sensor for the The greater the influence weight of the dust sensor in the dust concentration prediction; For the The dust sensor next to the The dust sensor is in the Dust concentration value at the moment;

[0038] Get the Moment Dust concentration prediction value of each dust sensor Hedi Moment The measured dust concentration value of each dust sensor ,Pick and The maximum value of is taken as the final dust concentration value. When the final dust concentration value is greater than the dust concentration threshold, the dust suppression device is turned on. When the final dust concentration value is less than or equal to the dust concentration threshold, the dust suppression device is not turned on.

[0039] Furthermore, the first-order difference is used to calculate the change in dust concentration value of each dust sensor, and the Dust concentration sensor The first-order difference sequence of , get the The dust sensor next to the Dust sensor, calculate the Dust sensor The first-order difference sequence of ;

[0040] Get The maximum value of The ratio of the means of The dust sensor is relative to the Maximum dust concentration change rate of a dust sensor ,Right now ,in, for The maximum value of for The mean of The larger the value, the The greater the change in the dust concentration value of the dust sensor, the The greater the impact of the dust sensor on dust concentration prediction.

[0041] Further, after obtaining a stable inflection point, obtain continuous The stable inflection point in the secondary inflection point calculation corresponds to The ratio of the cumulative distance value of a two-dimensional data in the cumulative distance matrix to the cumulative distance value calculated by the dynamic time warping DTW algorithm , where For The traversal value of Calculation of the sub-inflection point, For the The dust sensor and Between the dust sensors The cumulative distance value of the two-dimensional data in the cumulative distance matrix in the secondary inflection point calculation, For the The dust sensor and Between the dust sensors The cumulative distance value calculated by the Dynamic Time Warping (DTW) algorithm in the secondary inflection point calculation, where the cumulative distance value calculated by the Dynamic Time Warping (DTW) algorithm is the maximum cumulative distance value corresponding to all data points in the shortest distance path;

[0042] Get The mean of the ratios in the secondary inflection point calculation , The smaller it is, the more stable the inflection point is. The dust concentration change at the first dust sensor is compared with the dust concentration change at the second dust sensor There is a large difference in dust concentration at each dust sensor. The dust sensor is the The proximity sensor of the dust sensor is The smaller the value, the greater the dust transmission rate will be when the dust is transmitted through the mine tunnel. The dust concentration at the first dust sensor is affected by the The greater the impact of dust concentration at each dust sensor;

[0043] Then get the first A dust sensor for the The influence coefficient value of each dust sensor :

[0044] , where exp is the exponential function, M is the hyperparameter, and the empirical value of M is 0.56;

[0045] For the The maximum dust concentration change rate of the dust sensor indicates the The greater the change in the dust concentration value of the dust sensor, the The greater the impact is when a dust sensor is used to predict dust concentration;

[0046] For the A dust sensor and The average of the ratios of the cumulative distance values ​​calculated from the stable inflection points between the dust sensors; The smaller it is, the more stable the inflection point is. The dust concentration change at the first dust sensor is compared with the dust concentration change at the second dust sensor There is a large difference in dust concentration at each dust sensor. The dust sensor is the The neighboring sensor of the dust sensor is used to predict the dust concentration of the dust sensor. The greater the impact of the dust sensor on dust concentration prediction, the greater the impact of the dust sensor on dust concentration prediction. Perform negative correlation mapping;

[0047] For the A dust sensor and The time difference between the mean of the stable inflection point calculation time points between the dust sensors and the current moment is the time difference of the current moment. The dust sensor to the The delay time of each dust sensor is longer. The smaller the value, the A dust sensor for the The greater the impact of each dust sensor, the more it is mapped using an exponential function;

[0048] In getting the The dust sensor next to the After the dust sensor, due to the Each dust sensor also has multiple dust sensors for its neighboring dust sensors, and obtains the first The multiple influence coefficient values ​​corresponding to each adjacent dust sensor of the dust sensor are The multiple influence coefficient values ​​corresponding to each adjacent dust sensor of the dust sensor take the maximum value.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] The real-time monitoring method for mine dust concentration of the present invention obtains the time delay difference of adjacent areas when dust propagates in the mine tunnel, calculates the influence coefficient value of dust propagation between different adjacent areas, measures the impact of dust propagation in the mine tunnel on the dust concentration of the adjacent areas when there is a time delay difference, and uses the influence coefficient value of the adjacent dust area and the dust concentration value of the adjacent area to predict the dust concentration value, so that when using dust monitoring data for dust reduction control, a stable and accurate dust concentration value prediction result can be obtained, and the dust concentration monitoring result has higher robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 Flowchart of the monitoring method of the present invention. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 are within the scope of protection of the present invention.

[0053] See also Figure 1In an embodiment of the present invention, a method for real-time monitoring of dust concentration in a mine is provided, which mainly includes the following steps:

[0054] S1: Obtain dust data of each dust sensor and a preset mine tunnel model, and obtain all neighboring dust sensors of any dust sensor based on the mine tunnel model and the coordinate values ​​of each dust sensor; the dust data includes dust concentration values ​​and coordinate values.

[0055] Specifically, multiple dust sensors are arranged in the mine, and the dust sensor collection frequency is once per second. The dust concentration value is continuously collected by each dust sensor to obtain the dust concentration value collected by each dust sensor at a continuous moment. The dust sensor is in the The dust concentration at the moment is .

[0056] A positioning device is installed on each dust sensor to obtain the location information of each dust sensor. In this embodiment, the positioning device is a GPS locator. The longitude and latitude of the GPS locator are collected as coordinate values ​​to obtain the coordinate value of each dust sensor.

[0057] To construct a mine tunnel model, a three-dimensional physical model can be constructed as the mine tunnel model through tools such as BIM. The existing technology will not be repeated here. The coordinate value of each dust sensor is imported into the mine tunnel model. Through manual labeling, each dust sensor that can be directly connected to each dust sensor in multiple mine tunnels is regarded as a neighboring dust sensor of each dust sensor. In this embodiment, when there is no other dust sensor between two dust sensors in the shortest mine tunnel connected in the mine tunnel model, the two dust sensors are defined as adjacent dust sensors, and are each other's adjacent dust sensors, so each dust sensor will have one or more adjacent dust sensors.

[0058] Among them, the mine tunnel model with GPS positioning information is a three-dimensional mine tunnel model obtained through data surveying and mapping, and the longitude and latitude coordinate values ​​in the GPS data corresponding to different surveying positions are obtained through data calibration during surveying, so that the three-dimensional mine tunnel model has GPS positioning information, and the longitude and latitude coordinate values ​​in the GPS data corresponding to different model positions can be obtained in the three-dimensional mine tunnel model.

[0059] S2: Calculate the change in the dust concentration value of each dust sensor at consecutive moments and construct a change sequence to obtain the maximum dust concentration change rate.

[0060] Specifically, S1 can obtain the dust concentration value collected by each dust sensor at a continuous time, and set the time period The time period is 30 seconds, which can be used to obtain the dust sensor information in each time period. The dust concentration value sequence within.

[0061] No. Dust concentration sensors in the time period The dust concentration value sequence in , when the When the dust concentration of a dust concentration sensor changes, the dust concentration of the adjacent dust sensors will also change. However, dust propagation takes time and the mine tunnels are complex, resulting in a delay in dust propagation, and the dust concentration of the adjacent dust sensors cannot change in time.

[0062] In order to carry out dust reduction control during dust monitoring, it is necessary to predict the dust concentration in the mine tunnel in advance to prevent the problem of excessive dust concentration in a short period of time or excessive waste of dust reduction resources. Therefore, it is necessary to predict the dust concentration at the dust sensor. Since there are many factors that cause dust generation during operation, it is impossible to achieve accurate prediction based on the historical dust concentration of a single dust sensor alone. In addition, since the mine tunnels are connected, the dust concentrations at different locations in the mine tunnels affect each other.

[0063] In order to obtain the influence of the change of dust concentration value of each dust sensor on the dust concentration value of each dust sensor adjacent to it, the first-order difference is used to calculate the change of dust concentration value of each dust sensor, and the first Dust concentration sensor The first-order difference sequence of , get the The dust sensor next to the Dust sensor, calculate the Dust sensor The first-order difference sequence of .

[0064] Get The maximum value of The ratio of the means of The dust sensor is relative to the Maximum dust concentration change rate of a dust sensor ,Right now ,in, for The maximum value of for The mean of The larger the value, the The greater the change in the dust concentration value of the dust sensor, the The greater the impact of the dust sensor on dust concentration prediction.

[0065] S3: Obtain a cumulative distance matrix between the variation sequence of any dust sensor and the variation sequence of a single adjacent dust sensor, and calculate the inflection point of the shortest distance path in the cumulative distance matrix.

[0066] Specifically, the correlation performance between the difference sequences is used to represent the The dust concentration change of the dust sensor affects the The dust concentration value of each dust sensor is affected. After the dust concentration value of a dust sensor changes, due to the existence of delay, it will cause the first The dust concentration value of the first dust sensor cannot be compared with the The dust concentration values ​​of the dust sensors are highly correlated, and there are large differences in the dust concentration values, resulting in a decrease in correlation. In addition, different dusts are generated by different factors, resulting in different delay times.

[0067] In order to capture the delay of the dust concentration change every time the dust concentration changes, the influence of dust at any dust sensor on the dust concentration at the adjacent dust sensor can be represented by the delay feature.

[0068] Using the Dynamic Time Warping (DTW) algorithm, we calculate and The cumulative distance matrix between them is used to obtain the shortest distance path in the cumulative distance matrix. and There is a time delay between them, which results in that after a certain time point on the shortest distance path, and The correlation between them will change. The cumulative distance matrix and the method for obtaining the shortest distance path in the cumulative distance matrix are both well-known contents. The horizontal coordinate of the cumulative distance matrix is , the vertical axis is , the shortest distance path is composed of two-dimensional data points in the cumulative distance matrix, and the shortest distance path is and A two-dimensional data sequence consisting of the elements in .

[0069] The shorter the delay time, the The dust concentration at the first dust sensor has just changed. The dust concentration at the first dust sensor has not changed, so we can refer to the The dust concentration at the first dust sensor is used for the Dust concentration prediction at each dust sensor.

[0070] The longer the delay, the The dust is transmitted from the first dust sensor to the The dust concentration at the first dust sensor will be lower, because the dust in normal operation will naturally settle during the spreading process, so the first The change of dust concentration at the first dust sensor has an impact on the The smaller the impact of dust concentration changes at each dust sensor.

[0071] To find and For data points where the correlation between the two-dimensional data points changes, the least squares method is used to perform linear fitting on all the two-dimensional data points of the shortest distance path to obtain the fitting function of the shortest distance path. The Euclidean distance value of each two-dimensional data point in the shortest distance path to the fitting function is calculated to obtain a Euclidean distance value sequence. The moving average method is used to smooth the data of the Euclidean distance value sequence. The inflection point of the smoothed Euclidean distance value sequence is calculated. The two-dimensional data corresponding to the inflection point is and The moving average method takes a time window length of 5, which can be adjusted by the implementer according to the specific implementation scenario.

[0072] If the There is no change in dust concentration at the dust sensor, resulting in The dust concentration at the first dust sensor is The dust concentrations of the dust sensors have always been highly correlated. and The correlation between them has not changed, which will lead to the inflection point of the calculation being unstable. It is necessary to judge the stability of the obtained inflection point to prevent the dust concentration prediction from being too biased during dust settling.

[0073] In this embodiment, the stability judgment process of the inflection point is as follows: obtaining the two-dimensional data corresponding to the inflection point, obtaining the two-dimensional data in the and The time values ​​corresponding to the data points are obtained continuously In the calculation of the secondary inflection point, the two-dimensional data obtained is and The maximum and minimum values ​​of the corresponding serial numbers are respectively, if the two-dimensional data is and The difference between the maximum value of the sequence number and the minimum value of the sequence number is less than , it means that the current inflection point is stable, otherwise, it is considered that the inflection point is unstable and there is no need to predict the dust concentration of the adjacent dust sensors. Second inflection point calculation Take the empirical value of 5, which can be adjusted by the implementer according to the specific implementation scenario. is the sequence number difference threshold, The empirical value is 3, which can be adjusted by the implementer according to the specific implementation scenario.

[0074] By calculating the stable inflection point, the adaptive judgment of the delay time is realized. The dust concentration at the dust sensor is measured When the dust concentration value at the dust sensor is used to predict dust precipitation, it can have higher robustness, making the first The dust concentration value prediction results at each dust sensor are more stable and effective, reducing the waste of dust reduction resources during dust monitoring.

[0075] S4: Obtain the collection time point and cumulative distance value of the inflection point, obtain the time difference and the cumulative distance value change rate, calculate the influence coefficient value of any dust sensor and a single adjacent dust sensor based on the time difference and the cumulative distance value change rate and the maximum dust concentration change rate, and obtain the maximum value of all influence coefficient values ​​of any dust sensor.

[0076] Specifically, after obtaining a stable inflection point, obtain continuous The stable inflection point in the secondary inflection point calculation corresponds to Two-dimensional data in The time point sequence in the dimension, and the number of data in the time point sequence is , get the mean of the time points in the time point sequence, and calculate the time difference between the mean of the time points and the current moment , the time difference is the current From the dust sensor to the The delay time at each dust sensor is calculated, wherein the time difference calculated by two time points is a well-known content and will not be described in detail here.

[0077] After obtaining a stable inflection point, obtain continuous The stable inflection point in the secondary inflection point calculation corresponds to The ratio of the cumulative distance value of a two-dimensional data in the cumulative distance matrix to the cumulative distance value calculated by the dynamic time warping (DTW) algorithm , where For The traversal value of Calculation of the sub-inflection point, For the The dust sensor and Between the dust sensors The cumulative distance value of the two-dimensional data in the cumulative distance matrix in the secondary inflection point calculation, For the The dust sensor and Between the dust sensors The cumulative distance value calculated by the dynamic time warping (DTW) algorithm in the secondary inflection point calculation, wherein the cumulative distance value calculated by the dynamic time warping (DTW) algorithm is the maximum value of the cumulative distance values ​​corresponding to all data points in the shortest distance path.

[0078] Get The mean of the ratios in the secondary inflection point calculation , The smaller it is, the more stable the inflection point is. The dust concentration change at the first dust sensor is compared with the dust concentration change at the second dust sensor There is a large difference in dust concentration at each dust sensor, but due to the The dust sensor is the The proximity sensor of the dust sensor, so The smaller the value, the greater the dust transmission rate will be when the dust is transmitted through the mine tunnel. The dust concentration at the first dust sensor is affected by the The greater the dust concentration at each dust sensor, the greater the impact.

[0079] Then get the first A dust sensor for the The influence coefficient value of each dust sensor :

[0080] , where exp is the exponential function, M is the hyperparameter, and M takes an empirical value of 0.56, which can be adjusted by the implementer according to the specific implementation scenario.

[0081] For the The maximum dust concentration change rate of the dust sensor indicates the The greater the change in the dust concentration value of the dust sensor, the The greater the impact of the dust sensor on dust concentration prediction.

[0082] For the A dust sensor and The average of the ratios of the accumulated distance values ​​calculated from the stable inflection points between the dust sensors. The smaller it is, the more stable the inflection point is. The dust concentration change at the first dust sensor is compared with the dust concentration change at the second dust sensor There is a large difference in dust concentration at each dust sensor, but due to the The dust sensor is the The neighboring sensor of the dust sensor, so when predicting the dust concentration of dust fall, The greater the impact of the dust concentration prediction, the more dust sensors are used. Therefore, the exponential function is used to Perform negative correlation mapping.

[0083] For the A dust sensor and The time difference between the mean of the stable inflection point calculation time points between the dust sensors and the current moment is the time difference of the current moment. The dust sensor to the The delay time of each dust sensor is longer. The smaller the value, the A dust sensor for the The greater the impact of each dust sensor, the exponential function is used for mapping in order to prevent the time difference from being 0.

[0084] In getting the The dust sensor next to the After the dust sensor, due to the Each dust sensor also has multiple dust sensors for its neighboring dust sensors, and obtains the first The multiple influence coefficient values ​​corresponding to each adjacent dust sensor of the dust sensor are The multiple influence coefficient values ​​corresponding to each adjacent dust sensor of the dust sensor take the maximum value.

[0085] S5: Perform weighted prediction using the influence coefficient values ​​of all neighboring dust sensors of each dust sensor to obtain a dust concentration prediction value of each dust sensor and perform dust reduction control.

[0086] Specifically, for The maximum influence coefficient value of each adjacent dust sensor is calculated using Normalization algorithm normalizes the data as the When predicting the dust concentration of each dust sensor, the prediction weight value of each dust sensor is used to calculate the dust concentration of the first dust sensor using the weighted moving average method. The dust concentration prediction of dust sensors is performed using The sum of all prediction weight values ​​obtained by the normalization algorithm is 1.

[0087] Among them, when using the weighted moving average method, the real-time dust concentration value of the adjacent dust sensor is used to replace the historical data required in the weighted moving average method, so that when facing sudden dust generation, dust prediction can be made in advance, dust reduction preparation can be achieved in advance, dust reduction effect can be improved, and real-time monitoring of dust in the mine tunnel can be carried out.

[0088] For the first Moment The dust sensors are weighted to predict the Moment The dust concentration prediction value of each dust sensor is , , where is the number, indicating the The dust sensor next to the A dust sensor, For the The total number of all neighboring dust sensors of a dust sensor.

[0089] For the The dust sensor next to the A dust sensor for the The maximum influence coefficient value after normalization of the dust sensor. The larger the value, the higher the A dust sensor for the The greater the influence weight of the dust sensor in dust concentration prediction.

[0090] For the The dust sensor next to the The dust sensor is in the The dust concentration value at the moment.

[0091] Get the Moment Dust concentration prediction value of each dust sensor Hedi Moment The measured dust concentration value of each dust sensor ,Pick and The maximum value of is taken as the final dust concentration value, because the dust concentration of the adjacent dust sensor may decrease after dust precipitation, making the dust concentration prediction value lower than the measured dust concentration value.

[0092] When the final dust concentration value is greater than the dust concentration threshold, the dust reduction equipment is turned on. When the final dust concentration value is less than or equal to the dust concentration threshold, the dust reduction equipment is not turned on, thereby realizing real-time monitoring of dust concentration. Among them, the dust reduction equipment in this embodiment is a water spray dust reduction equipment, which can be replaced by the implementer according to the specific implementation scenario.

[0093] To sum up: the present invention provides a real-time monitoring method for mine dust concentration, which obtains the time delay difference of adjacent areas when dust propagates in the mine tunnel, calculates the influence coefficient value of dust propagation between different adjacent areas, measures the impact of dust propagation in the mine tunnel on the dust concentration in the adjacent area when there is a time delay difference, and uses the influence coefficient value of the adjacent dust area and the dust concentration value of the adjacent area to predict the dust concentration value, so that when using dust monitoring data for dust reduction control, it can have a stable and accurate dust concentration value prediction result, thereby improving the stability of real-time monitoring of dust concentration.

[0094] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A real-time monitoring method for mine dust concentration, characterized by: The following steps are involved: S1: Obtain dust data from each dust sensor and a preset mine tunnel model. Based on the mine tunnel model and the coordinate values ​​of each dust sensor, obtain all neighboring dust sensors of any dust sensor. The dust data includes dust concentration values ​​and coordinate values. S2: Calculate the change in dust concentration values ​​of each dust sensor at consecutive moments and construct a change sequence to obtain the maximum dust concentration change rate; S3: Obtain a cumulative distance matrix between the variation sequence of any dust sensor and the variation sequence of a single adjacent dust sensor, and calculate the inflection point of the shortest distance path in the cumulative distance matrix; S4: Obtain the collection time point and cumulative distance value of the inflection point, obtain the time difference and the cumulative distance value change rate, calculate the influence coefficient value of any dust sensor and a single adjacent dust sensor based on the time difference and cumulative distance value change rate and the maximum dust concentration change rate, and obtain the maximum value of all influence coefficient values ​​of any dust sensor; S5: Perform weighted prediction using the influence coefficient values ​​of all neighboring dust sensors of each dust sensor to obtain a dust concentration prediction value of each dust sensor and perform dust reduction control.

2. A method for real-time monitoring of mine dust concentration according to claim 1, characterized in that: The mine tunnel model in S1 is equipped with GPS positioning information. The GPS positioning information comes from the GPS locator installed on each dust sensor. The three-dimensional mine tunnel model is obtained through data mapping, and the longitude and latitude coordinate values ​​in the GPS data corresponding to different mapping positions are obtained through data calibration during mapping.

3. A method for real-time monitoring of mine dust concentration according to claim 2, characterized in that: Each dust sensor is arranged in the mine, and the collection frequency of each dust sensor is once per second. The dust concentration value is continuously collected by each dust sensor to obtain the dust concentration value collected by each dust sensor at a continuous moment. The dust sensor is in the The dust concentration at the moment is .

4. A method for real-time monitoring of mine dust concentration according to claim 3, characterized in that: The method for obtaining the maximum dust concentration change rate in S2 is: S201: Obtain a dust concentration value sequence of a single dust concentration sensor within a time period, calculate a first-order difference sequence of the dust concentration value sequence, and similarly calculate a first-order difference sequence of a single dust sensor adjacent to the single dust concentration sensor; S202: Obtaining the maximum value of the first-order difference sequence of adjacent single dust sensors; S203: Obtain the mean of the first-order difference sequence of the dust concentration value sequence; S204: Calculate the ratio of the maximum value in S202 to the average value in S203 as the maximum dust concentration change rate.

5. A method for real-time monitoring of mine dust concentration according to claim 4, characterized in that: The calculation method of the influence coefficient value in S4 includes: S401: Obtain a first-order difference sequence of a single dust concentration sensor, and similarly calculate the first-order difference sequence of a single dust sensor adjacent to the single dust concentration sensor; S402: Calculate the cumulative distance matrix between the first-order difference sequences of the target dust sensor and its neighboring dust sensors, and calculate the shortest distance path of the cumulative distance matrix; S403: Obtain a stable inflection point according to the shortest distance path; S404: Obtain the time difference between the mean of the time point corresponding to the inflection point and the current moment, and obtain the ratio of the cumulative distance value in the cumulative distance matrix corresponding to the inflection point to the cumulative distance value calculated by the dynamic time warping algorithm; S405: Calculate the influence coefficient value between a single dust concentration sensor and a single adjacent dust sensor using the negative correlation between the maximum dust concentration change rate and the time difference and the negative correlation between the ratio.

6. A method for real-time monitoring of mine dust concentration according to claim 5, characterized in that: Obtaining a stable inflection point in S403 includes: a. Perform linear fitting on all two-dimensional data points along the shortest distance path using the least squares method to obtain a fitting function for the shortest distance path. Calculate the Euclidean distance from each two-dimensional data point along the shortest distance path to the fitting function to obtain a Euclidean distance sequence. Smooth the Euclidean distance sequence using the moving average method. Calculate the inflection point of the smoothed Euclidean distance sequence to obtain the two-dimensional data corresponding to the inflection point. b. Similarly, obtain the inflection point calculation results for each of the previous X moments; c. Obtain the maximum and minimum values ​​of the two-dimensional data of each inflection point in the first-order difference sequence of a single dust sensor and the first-order difference sequence of an adjacent single dust sensor, and calculate the difference between the maximum value and the minimum value; d. If the response difference is less than the difference threshold, the current inflection point is determined to be stable.

7. A method for real-time monitoring of mine dust concentration according to claim 6, characterized in that: The calculation method of the dust concentration prediction value in S5 includes: S501: Calculate the maximum influence coefficient value of each dust sensor adjacent to each dust sensor using The normalization algorithm is used to normalize the data and use the weighted moving average method as the prediction weight value to predict the dust concentration of each dust sensor; S502: Using the real-time dust concentration values ​​of the adjacent dust sensors to replace the historical data required by the weighted moving average method; using the normalized maximum influence coefficient value of each dust sensor adjacent to each dust sensor as a weight, and multiplying it with the dust concentration value of each dust sensor adjacent to each dust sensor at the time to be predicted, to obtain multiple multiplication results, and accumulating all the multiplication results to obtain the dust concentration prediction value of each dust sensor at the time to be predicted; S503: Obtain the dust concentration prediction value and the actual dust concentration value of each dust sensor at the latest moment, and take the maximum value of the dust concentration prediction value and the actual dust concentration value as the final dust concentration value.

8. A method for real-time monitoring of mine dust concentration according to claim 7, characterized in that: The calculation formula for the predicted value of dust concentration in S5 is as follows: For the first Moment The dust sensors are weighted to predict the Moment The dust concentration prediction value of each dust sensor is , , where is the number, indicating the The dust sensor next to the A dust sensor, For the The total number of all neighboring dust sensors of a dust sensor; For the The dust sensor next to the A dust sensor for the The maximum influence coefficient value after normalization of the dust sensor. The larger the value, the higher the A dust sensor for the The greater the influence weight of the dust sensor in the dust concentration prediction; For the The dust sensor next to the The dust sensor is in the Dust concentration value at the moment; Get the Moment Dust concentration prediction value of each dust sensor Hedi Moment The measured dust concentration value of each dust sensor ,Pick and The maximum value of is taken as the final dust concentration value. When the final dust concentration value is greater than the dust concentration threshold, the dust suppression device is turned on. When the final dust concentration value is less than or equal to the dust concentration threshold, the dust suppression device is not turned on.

9. The method for real-time monitoring of mine dust concentration according to claim 5, characterized in that: Use the first-order difference to calculate the change in dust concentration value of each dust sensor, and calculate the Dust concentration sensor The first-order difference sequence of , get the The dust sensor next to the Dust sensor, calculate the Dust sensor The first-order difference sequence of ; Get The maximum value of The ratio of the means of The dust sensor is relative to the Maximum dust concentration change rate of a dust sensor ,Right now ,in, for The maximum value of for The mean of The larger the value, the The greater the change in the dust concentration value of the dust sensor, the The greater the impact of the dust sensor on dust concentration prediction.

10. A method for real-time monitoring of mine dust concentration according to claim 9, characterized in that: After obtaining a stable inflection point, obtain continuous The stable inflection point in the secondary inflection point calculation corresponds to The ratio of the cumulative distance value of a two-dimensional data in the cumulative distance matrix to the cumulative distance value calculated by the dynamic time warping DTW algorithm , where For The traversal value of Calculation of the sub-inflection point, For the The dust sensor and Between the dust sensors The cumulative distance value of the two-dimensional data in the cumulative distance matrix in the secondary inflection point calculation, For the The dust sensor and Between the dust sensors The cumulative distance value calculated by the Dynamic Time Warping (DTW) algorithm in the secondary inflection point calculation, where the cumulative distance value calculated by the Dynamic Time Warping (DTW) algorithm is the maximum cumulative distance value corresponding to all data points in the shortest distance path; Get The mean of the ratios in the secondary inflection point calculation , The smaller it is, the more stable the inflection point is. The dust concentration change at the first dust sensor is compared with the dust concentration change at the second dust sensor There is a large difference in dust concentration at the dust sensor, and the 𝑗th dust sensor is The proximity sensor of the dust sensor is The smaller the value, the greater the dust transmission rate will be when the dust is transmitted through the mine tunnel. The dust concentration at the first dust sensor is affected by the The greater the impact of dust concentration at each dust sensor; Then get the first A dust sensor for the The influence coefficient value of each dust sensor : , where exp is the exponential function, M is the hyperparameter, and the empirical value of M is 0.56; For the The maximum dust concentration change rate of the dust sensor indicates the The greater the change in the dust concentration value of the dust sensor, the The greater the impact is when a dust sensor is used to predict dust concentration; For the A dust sensor and The average of the ratios of the cumulative distance values ​​calculated from the stable inflection points between the dust sensors; The smaller it is, the more stable the inflection point is. The dust concentration change at the first dust sensor is compared with the dust concentration change at the second dust sensor There is a large difference in dust concentration at each dust sensor due to the The dust sensor is the The neighboring sensor of the dust sensor is used to predict the dust concentration of the dust sensor. The greater the impact of the dust sensor on dust concentration prediction, the greater the impact of the dust sensor on dust concentration prediction. Perform negative correlation mapping; For the A dust sensor and The time difference between the mean of the stable inflection point calculation time points between the dust sensors and the current moment is the time difference of the current moment. The dust sensor to the The delay time of each dust sensor is longer. The smaller the value, the A dust sensor for the The greater the impact of each dust sensor, the more it is mapped using an exponential function; In getting the The dust sensor next to the After the dust sensor, get the The multiple influence coefficient values ​​corresponding to each adjacent dust sensor of the dust sensor are The multiple influence coefficient values ​​corresponding to each adjacent dust sensor of the dust sensor take the maximum value.

Images