Mining dust concentration real-time monitoring and early warning system based on Internet of Things

Abstract

The invention relates to the technical field of data processing, in particular to a mining dust concentration real-time monitoring and early warning system based on the Internet of Things, the system comprises a processor and a memory, and the processor executes a computer program of the memory to realize the following steps: acquiring dust concentration data in a mine, obtaining a preset number of dust concentration data sequences; acquiring a template data sequence, and acquiring sequence representativeness evaluation values of all the dust concentration data sequences according to data fluctuation characteristics in the dust concentration data sequences and the correlation and data difference between the dust concentration data sequences and the template data sequence; screening at least one target data sequence from the preset number of dust concentration data sequences; and constructing a dust concentration prediction model according to all the target data sequences, and carrying out monitoring and early warning on dust concentration data in a future preset time length. According to the invention, the influence of extreme values in the dust concentration data is reduced, and the prediction capability of the dust concentration prediction model in the mine is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a real-time monitoring and early warning system for mining dust concentration based on the Internet of Things. Background Art

[0002] The IoT mining dust concentration real-time monitoring and early warning system is an intelligent system based on IoT technology for dust concentration monitoring, data collection, real-time analysis and early warning in the mine environment. The system aims to protect the health and safety of mine workers and promote the effective prevention and control of dust occupational hazards by efficiently and in real time monitoring the dust concentration in the air. Due to the harsh environment in the mine, the change of dust concentration is affected by various factors such as work activities and equipment operation, and the data pattern is complex. It is difficult to capture this complex pattern with real-time data alone. Therefore, how to accurately monitor the dust concentration in the mine environment in real time and issue early warnings to effectively protect the health and safety of mine workers is the main challenge currently faced.

[0003] In the traditional way, the real-time monitoring and early warning system of dust concentration in the Internet of Things mines builds a dust concentration prediction model by combining historical data to improve the system's ability to identify abnormal conditions, obtains the predicted value of dust concentration data according to the constructed dust concentration prediction model, and compares it with the dust concentration data threshold to complete the real-time monitoring and early warning of the current dust concentration. Because the dust concentration prediction model in the traditional way is built in combination with historical data, and because of the harsh environment in the mine (such as high humidity, high wind speed, strong corrosiveness, etc.), the data obtained by the dust concentration sensor will have more noise, and the impact of emergencies (such as equipment failure, blasting operations) will cause extreme values ​​in the dust concentration data. These dust concentration data are not universal, and the focus of dust concentration monitoring in mines is usually the dynamic changes of conventional concentrations. Therefore, these dust concentration data containing noise or emergencies will cause the dust concentration prediction model to overfit these special cases, thereby reducing the generalization ability under normal conditions and directly affecting the accuracy of the system.

[0004] Therefore, how to reduce the impact of extreme values ​​in dust concentration data and improve the predictive ability of dust concentration prediction models in mine environments has become an urgent problem to be solved. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a real-time monitoring and early warning system for mining dust concentration based on the Internet of Things to solve the problem of how to reduce the impact of extreme values ​​in dust concentration data and improve the predictive ability of the dust concentration prediction model in the mine environment.

[0006] In an embodiment of the present invention, a real-time monitoring and early warning system for mining dust concentration based on the Internet of Things is provided, comprising a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the following method is implemented: Obtain dust concentration data at each sampling moment in the mine, obtain an initial dust concentration data sequence within a preset sampling period, obtain a preset number of initial dust concentration data sequences, pre-process each of the initial dust concentration data sequences respectively, and obtain the preset number of dust concentration data sequences; According to the correlation between the preset number of dust concentration data sequences, a template data sequence is obtained, and for any dust concentration data sequence among the preset number of dust concentration data sequences, according to the data fluctuation characteristics in the any dust concentration data sequence, and the correlation and data difference between the any dust concentration data sequence and the template data sequence, a sequence representative evaluation value of the any dust concentration data sequence is obtained; Obtaining sequence representativeness evaluation values ​​of all dust concentration data sequences, and selecting at least one target data sequence from the preset number of dust concentration data sequences according to the sequence representativeness evaluation values ​​of all dust concentration data sequences; A dust concentration prediction model is constructed based on all target data sequences, and the dust concentration data within a preset time length in the future is monitored and warned using the dust concentration prediction model.

[0007] Preferably, obtaining the sequence representativeness evaluation value of any dust concentration data sequence according to the data fluctuation characteristics in any dust concentration data sequence, and the correlation and data difference between any dust concentration data sequence and the template data sequence includes: According to the data fluctuation characteristics in any dust concentration data sequence, obtaining the stability value of any dust concentration data sequence; Obtaining the Spearman correlation coefficient between any one of the dust concentration data sequences and the template data sequence, and the Spearman ranking of each dust concentration data in any one of the dust concentration data sequences and the template data sequence, and obtaining the time series feature correlation coefficient between any one of the dust concentration data sequences and the template data sequence according to the Spearman correlation coefficient between any one of the dust concentration data sequences and the template data sequence, and the time difference between any one of the dust concentration data sequences and the dust concentration data with the same Spearman ranking in the template data sequence; According to the data difference between any one of the dust concentration data sequences and the template data sequence, obtaining the feature similarity between any one of the dust concentration data sequences and the template data sequence; According to the stability value of any dust concentration data sequence, and the time series feature correlation coefficient and feature similarity between any dust concentration data sequence and the template data sequence, a sequence representativeness evaluation value of any dust concentration data sequence is obtained.

[0008] Preferably, obtaining the stability value of any dust concentration data sequence according to the data fluctuation characteristics in any dust concentration data sequence includes: Obtaining the maximum value and the minimum value in any dust concentration data sequence, and calculating the first difference between the maximum value and the minimum value in any dust concentration data sequence; Obtaining the average value and median value of any dust concentration data sequence, and calculating the absolute value of the first difference between the average value and the median value; An inverse proportional normalization process is performed on the sum of the first difference and the absolute value of the first difference to obtain a stability value of any dust concentration data sequence.

[0009] Preferably, the step of obtaining the time series characteristic correlation coefficient between any dust concentration data sequence and the template data sequence according to the Spearman correlation coefficient between any dust concentration data sequence and the template data sequence, and the time difference between any dust concentration data sequence and dust concentration data with the same Spearman ranking in the template data sequence, comprises: Respectively obtain the time intervals between the sampling moments of two dust concentration data with the same Spearman ranking in any dust concentration data sequence and the template data sequence, obtain the corresponding accumulated value of the time interval, and perform inverse proportional normalization processing on the accumulated value of the time interval to obtain a first inverse proportional normalized value; According to the Spearman correlation coefficient between any dust concentration data sequence and the template data sequence, the first inverse proportional normalization value and the mean of the constant 1, the time series characteristic correlation coefficient between any dust concentration data sequence and the template data sequence is obtained.

[0010] Preferably, obtaining the degree of feature similarity between any dust concentration data sequence and the template data sequence according to the data difference between any dust concentration data sequence and the template data sequence includes: respectively calculating the absolute values ​​of the differences between two dust concentration data at the same position in any dust concentration data sequence and the template data sequence, and correspondingly obtaining the accumulated value of the absolute values ​​of the differences, obtaining the maximum value and the minimum value in the template data sequence, calculating the second difference between the maximum value and the minimum value in the template data sequence, calculating the second absolute value of the difference between the first difference and the second difference, and performing inverse proportional normalization processing on the accumulated value of the absolute values ​​of the differences and the addition result of the second absolute value of the differences, to obtain a second inverse proportional normalized value; Obtaining a first variance of any dust concentration data sequence, obtaining a second variance of the template data sequence, and performing inverse proportional normalization processing on an absolute value of a difference between the first variance and the second variance to obtain a third inverse proportional normalized value; The degree of feature similarity between any dust concentration data sequence and the template data sequence is obtained according to the average of the second inverse proportional normalized value and the third inverse proportional normalized value.

[0011] Preferably, the obtaining of the sequence representativeness evaluation value of any dust concentration data sequence according to the stability value of any dust concentration data sequence, and the time series feature correlation coefficient and feature similarity between any dust concentration data sequence and the template data sequence comprises: The sequence representativeness evaluation value of any dust concentration data sequence is obtained according to the stability value of any dust concentration data sequence, the time series feature correlation coefficient between any dust concentration data sequence and the template data sequence, and the average value of the feature similarity.

[0012] Preferably, the step of selecting at least one target data sequence from the preset number of dust concentration data sequences according to the sequence representativeness evaluation values ​​of all dust concentration data sequences includes: Set a sequence representative evaluation value interval, and select dust concentration data sequences whose sequence representative evaluation values ​​are within the sequence representative evaluation value interval, and record them as target data sequences.

[0013] Preferably, obtaining the template data sequence according to the correlation between the preset number of dust concentration data sequences includes: The preset number of dust concentration data sequences are fitted to obtain a template data sequence.

[0014] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The present invention obtains dust concentration data at each sampling moment in a mine, obtains an initial dust concentration data sequence within a preset sampling period, obtains a preset number of initial dust concentration data sequences, pre-processes each of the initial dust concentration data sequences respectively, and obtains the preset number of dust concentration data sequences; obtains a template data sequence according to the correlation between the preset number of dust concentration data sequences, and for any dust concentration data sequence in the preset number of dust concentration data sequences, obtains a sequence representative judgment value of any dust concentration data sequence according to data fluctuation characteristics in the any dust concentration data sequence, and a correlation and data difference between the any dust concentration data sequence and the template data sequence; obtains sequence representative judgment values ​​of all dust concentration data sequences, and selects at least one target data sequence from the preset number of dust concentration data sequences according to the sequence representative judgment values ​​of all dust concentration data sequences; constructs a dust concentration prediction model according to all target data sequences, and uses the dust concentration prediction model to monitor and warn dust concentration data within a preset time length in the future. Among them, the sequence representativeness judgment value of any dust concentration data sequence obtained through the data fluctuation characteristics in any dust concentration data sequence, as well as the correlation and data difference between any dust concentration data sequence and the template data sequence, can reflect the extent to which the dust concentration data sequence is affected by noise and emergencies, and can also reflect the overall trend and distribution characteristics of the dust concentration data sequence. By screening and obtaining a target data sequence that is more in line with the dust concentration data change trend in the actual scenario and is less affected by noise and emergencies, a dust concentration prediction model is constructed, the influence of extreme values ​​in the dust concentration data is reduced, and the prediction ability of the dust concentration prediction model in the mine environment is improved, thereby accurately monitoring and warning the dust concentration in the mine environment in real time, thereby effectively protecting the health and safety of mine workers. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0016] Figure 1 This is a flow chart of a method for real-time monitoring and early warning of mining dust concentration based on the Internet of Things provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0017] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, but should not be understood as limiting the present disclosure.

[0018] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0019] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0020] The embodiment of the present invention provides a real-time monitoring and early warning system for mining dust concentration based on the Internet of Things, including a processor and a memory, wherein the processor executes a computer program in the memory to implement a real-time monitoring and early warning method for mining dust concentration based on the Internet of Things, such as Figure 1 As shown, the method comprises the following steps: Step S101, obtain the dust concentration data at each sampling moment in the mine, obtain the initial dust concentration data sequence within the preset sampling period, obtain a preset number of initial dust concentration data sequences, pre-process each of the initial dust concentration data sequences respectively, and obtain the preset number of dust concentration data sequences.

[0021] The dust concentration data at each sampling moment in the mine is collected by a dust concentration sensor to obtain the initial dust concentration data sequence within the preset sampling period, and a preset number of initial dust concentration data sequences are obtained. Normally, the sampling frequency of the dust concentration sensor is 1-10 Hz. In this embodiment, the sampling frequency of the dust concentration sensor is set to 1 Hz, the sampling period is 24 hours, the preset number is 4, and 4 initial dust concentration data sequences are obtained. There is no limitation here and it can be set according to the specific implementation scenario.

[0022] After obtaining the four initial dust concentration data sequences, since there may be data missing or duplication due to equipment or transmission problems during the sampling process, each initial dust concentration data sequence needs to be preprocessed. The specific steps of the preprocessing are: (1) Use the linear interpolation method to process the missing data in the initial dust concentration sequence; (2) Use the threshold filtering method to remove the duplicate data in the initial dust concentration data sequence; (3) For each initial dust concentration data sequence, align the sampling time within the sampling period and ensure that the dust concentration data in each initial dust concentration data sequence are arranged in ascending order according to the sampling time; Among them, the linear interpolation method and the threshold filtering method belong to the existing technology and will not be repeated here. After preprocessing each initial dust concentration data sequence, four dust concentration data sequences are obtained.

[0023] Step S102: obtaining a template data sequence based on the correlation between the preset number of dust concentration data sequences, and obtaining a sequence representativeness judgment value of any dust concentration data sequence among the preset number of dust concentration data sequences based on the data fluctuation characteristics in the any dust concentration data sequence, and the correlation and data difference between the any dust concentration data sequence and the template data sequence.

[0024] In order to accurately monitor and warn of dust concentration in the mine environment in real time and effectively protect the health and safety of mine workers, the IoT mine dust concentration real-time monitoring and early warning system traditionally builds a dust concentration prediction model by combining historical data to improve the system's ability to identify abnormal conditions. The dust concentration data prediction value within a preset time length in the future is obtained according to the constructed dust concentration prediction model, and the dust concentration data prediction value is compared with the preset dust concentration data threshold to complete real-time monitoring and early warning of dust concentration. However, the traditional dust concentration prediction model is constructed based on historical data. The harsh environment in mines (such as high humidity, high wind speed, strong corrosion, etc.) will cause more noise in the data obtained by dust concentration sensors. At the same time, the impact of emergencies (such as equipment failure, blasting operations) will cause extreme values ​​in the dust concentration data. These dust concentration data are not universal, and the focus of dust concentration monitoring in mines is usually the dynamic changes of conventional concentrations. Therefore, these dust concentration data containing noise or emergencies will cause the dust concentration prediction model to overfit to these special cases, thereby reducing the generalization ability under normal conditions and directly affecting the accuracy of the early warning.

[0025] Therefore, it is necessary to reduce the impact of extreme values ​​in dust concentration data caused by harsh environments and emergencies, and to improve the predictive ability of dust concentration prediction models in mine environments.

[0026] In order to solve the above problems, it is necessary to screen out the target data sequences whose dust concentration data are less affected by noise or sudden events and have the overall change trend and distribution characteristics of dust concentration data from a preset number of dust concentration data sequences, and build a dust concentration prediction model through the screened target data sequences to improve the prediction ability of the dust concentration prediction model in the mine environment.

[0027] Since the dust concentration data caused by harsh environment and emergencies include dust concentration data sequences with extreme values, there will be obvious long-tail distribution characteristics in the data distribution (that is, although the number of data with large values ​​is small, they account for a large proportion of the overall data), which makes the dust concentration data sequence less stable. Therefore, dust concentration data sequences with higher stability (that is, less affected by harsh environment and emergencies) can be screened out based on the data fluctuation characteristics in the dust concentration data sequence.

[0028] However, the dust concentration data sequence for constructing the dust concentration prediction model must also reflect the overall change trend and distribution characteristics of the dust concentration data. Since the preset number of dust concentration data sequences obtained cover a longer time period, the data is richer and can capture a variety of dynamic change trends. Therefore, based on the correlation between the preset number of dust concentration data sequences, a template data sequence that can reflect the overall change trend and distribution characteristics of the dust concentration data can be obtained. Then, based on the correlation and data differences between the dust concentration data sequence and the template data sequence, a dust concentration data sequence that better matches the template data sequence (that is, better represents the overall change trend and distribution characteristics of the dust concentration data) is screened out.

[0029] To sum up, for any dust concentration data sequence among the preset number of dust concentration data sequences, the sequence representativeness judgment value of any dust concentration data sequence can be obtained according to the data fluctuation characteristics in any dust concentration data sequence, as well as the correlation and data difference between any dust concentration data sequence and the template data sequence. Then, through the sequence representativeness judgment value of each dust concentration data sequence, the dust concentration data sequence with a higher degree of stability (i.e., less affected by harsh environment and emergencies) and more matched with the template data sequence (i.e., better representing the overall change trend and distribution characteristics of the dust concentration data) can be screened out.

[0030] Among them, according to the correlation between a preset number of dust concentration data sequences, the specific method for obtaining a template data sequence that can reflect the overall change trend and distribution characteristics of the dust concentration data is: polynomial fitting is performed on the preset number of dust concentration data sequences to obtain a template data sequence. Polynomial fitting belongs to the existing technology and will not be repeated here.

[0031] Furthermore, according to the data fluctuation characteristics in any dust concentration data sequence, as well as the correlation and data difference between any dust concentration data sequence and the template data sequence, the method for obtaining the sequence representativeness evaluation value of any dust concentration data sequence is as follows: (1) Obtaining a stability value of any dust concentration data sequence according to data fluctuation characteristics in any dust concentration data sequence.

[0032] Specifically, the maximum value and the minimum value in any dust concentration data sequence are obtained, and a first difference between the maximum value and the minimum value in any dust concentration data sequence is calculated; Obtaining the average value and median value of any dust concentration data sequence, and calculating the absolute value of the first difference between the average value and the median value; An inverse proportional normalization process is performed on the sum of the first difference and the absolute value of the first difference to obtain a stability value of any dust concentration data sequence.

[0033] In one embodiment, taking the jth dust concentration data sequence as an example, the maximum value and the minimum value in the jth dust concentration data sequence, as well as the average value and the median value of the jth dust concentration data sequence are obtained, and the stability value of the jth dust concentration data sequence is calculated:

[0034] in, is the stability value of the j-th dust concentration data sequence; is the maximum value in the jth dust concentration data sequence; is the minimum value in the jth dust concentration data sequence; is the average value of the jth dust concentration data series; is the median value of the jth dust concentration data sequence; j is the sequence number of the dust concentration data sequence; | | is the absolute value sign; exp(-) is the inverse proportional normalization function.

[0035] It should be noted that is the difference between the maximum and minimum values ​​in the jth dust concentration data sequence, The larger the value is, the larger the fluctuation range of the dust concentration data in the j-th dust concentration data sequence is, the more likely it is that extreme values ​​affected by harsh environments and emergencies will appear, and the smaller the stability value of the j-th dust concentration data sequence is. represents the difference between the mean and median values ​​of the jth dust concentration data series, The larger the value is, the more likely it is that the average value of the j-th dust concentration data sequence is affected by harsh environments and emergencies, and the smaller the stability value of the j-th dust concentration data sequence is; the larger the stability value of the j-th dust concentration data sequence is, the greater the stability of the j-th dust concentration data sequence is, and the more it conforms to the results of subsequent screening, and the greater the sequence representativeness judgment value of the j-th dust concentration data sequence is.

[0036] (2) Obtaining the Spearman correlation coefficient between any dust concentration data sequence and the template data sequence, as well as the Spearman ranking of each dust concentration data in any dust concentration data sequence and the template data sequence; and obtaining the time series feature correlation coefficient between any dust concentration data sequence and the template data sequence based on the Spearman correlation coefficient between any dust concentration data sequence and the template data sequence, as well as the time difference between dust concentration data with the same Spearman ranking in any dust concentration data sequence and the template data sequence.

[0037] Since the dust concentration in mines is affected by many factors, its data usually changes monotonically and nonlinearly. The Spearman correlation coefficient can effectively capture the monotonic relationship between the data without considering whether its change is linear, and it is more robust to outliers and noise. Therefore, it is very suitable for the correlation calculation between the dust concentration data sequence and the template data sequence. However, because the dust concentration data is time series data, the Spearman correlation coefficient only considers the consistency of the data in the monotonic relationship when calculating, and ignores the information of the data time scale. Therefore, in order to better capture the correlation between the template data sequence and each dust concentration data sequence, the time difference between the dust concentration data is introduced when calculating the Spearman correlation coefficient, so as to better capture the dynamic characteristics of the dust concentration data changing over time.

[0038] Specifically, the time intervals between the sampling moments of two dust concentration data with the same Spearman ranking in any dust concentration data sequence and the template data sequence are respectively obtained, and the accumulated values ​​of the time intervals are correspondingly obtained, and the accumulated values ​​of the time intervals are inversely normalized to obtain a first inversely normalized value; According to the Spearman correlation coefficient between any dust concentration data sequence and the template data sequence, the first inverse proportional normalization value and the mean of the constant 1, the time series characteristic correlation coefficient between any dust concentration data sequence and the template data sequence is obtained.

[0039] In one embodiment, taking the j-th dust concentration data sequence as an example, firstly, the Spearman correlation coefficient between the j-th dust concentration data sequence and the template data sequence, and the Spearman ranking of each dust concentration data in the j-th dust concentration data sequence and the template data sequence are obtained. The Spearman correlation coefficient and the Spearman ranking belong to the prior art and are not described here in detail. Then, the time interval between the sampling moments of two dust concentration data with the same Spearman ranking in the j-th dust concentration data sequence and the template data sequence is obtained respectively. According to the Spearman correlation coefficient between the j-th dust concentration data sequence and the template data sequence and the time interval between the sampling moments of two dust concentration data with the same Spearman ranking in the j-th dust concentration data sequence and the template data sequence, the time series feature correlation coefficient between the j-th dust concentration data sequence and the template data sequence is calculated:

[0040] in, is the time series characteristic correlation coefficient between the jth dust concentration data sequence and the template data sequence; is the Spearman correlation coefficient between the jth dust concentration data sequence and the template data sequence; is the time interval between the sampling moments of the i-th dust concentration data in the j-th dust concentration data sequence and the dust concentration data with the same Spearman rank in the template data sequence; j is the sequence number of the dust concentration data sequence; i is the sequence number of the dust concentration data in the j-th dust concentration data sequence; n is the number of dust concentration data in the j-th dust concentration data sequence; | | is the absolute value sign; exp(-) is the inverse proportional normalization function; 1 is a constant.

[0041] It should be noted that the larger the Spearman correlation coefficient between the jth dust concentration data sequence and the template data sequence, the greater the consistency of the jth dust concentration data sequence and the template data sequence in the monotonic relationship, and the larger the correlation coefficient of the time series characteristics between the jth dust concentration data sequence and the template data sequence; It represents the cumulative value of the time interval between the sampling moments of the two dust concentration data with the same Spearman ranking in the j-th dust concentration data sequence and the template data sequence, The smaller it is, the higher the temporal dynamics of the j-th dust concentration data sequence and the template data sequence are in sync, and the greater the correlation coefficient between the temporal characteristics of the j-th dust concentration data sequence and the template data sequence is; the larger the correlation coefficient between the temporal characteristics of the j-th dust concentration data sequence and the template data sequence is, the more the j-th dust concentration data sequence conforms to the results of subsequent screening, and the greater the sequence representativeness judgment value of the j-th dust concentration data sequence is.

[0042] (3) Obtaining a degree of feature similarity between any one of the dust concentration data sequences and the template data sequence based on a data difference between the any one of the dust concentration data sequences and the template data sequence.

[0043] Specifically, the absolute values ​​of the differences between any one of the dust concentration data sequences and two dust concentration data at the same position in the template data sequence are calculated respectively, and the corresponding accumulated values ​​of the absolute values ​​of the differences are obtained, the maximum value and the minimum value in the template data sequence are obtained, the second difference between the maximum value and the minimum value in the template data sequence is calculated, the second absolute value of the difference between the first difference and the second difference is calculated, and the accumulated value of the absolute value of the difference and the addition result of the second absolute value of the difference are inversely normalized to obtain a second inversely normalized value; Obtaining a first variance of any dust concentration data sequence, obtaining a second variance of the template data sequence, and performing inverse proportional normalization processing on an absolute value of a difference between the first variance and the second variance to obtain a third inverse proportional normalized value; The degree of feature similarity between any dust concentration data sequence and the template data sequence is obtained according to the average of the second inverse proportional normalized value and the third inverse proportional normalized value.

[0044] In one embodiment, taking the jth dust concentration data sequence as an example, first obtain the maximum value and the minimum value in the jth dust concentration data sequence, calculate the difference between the maximum value and the minimum value in the jth dust concentration data sequence, record it as the first difference, obtain the maximum value and the minimum value in the template data sequence, calculate the difference between the maximum value and the minimum value in the template data sequence, record it as the second difference; then obtain the variance of the jth dust concentration data sequence, record it as the first variance, obtain the variance of the template data sequence, record it as the second variance; according to the difference in dust concentration data between the jth dust concentration data sequence and the template data sequence, the first difference, the second difference, the first variance and the second variance, calculate the feature similarity between the jth dust concentration data sequence and the template data sequence:

[0045] in, is the characteristic similarity between the jth dust concentration data sequence and the template data sequence; is the i-th dust concentration data in the j-th dust concentration data sequence; is the i-th dust concentration data in the template data sequence; is the first difference; is the second difference; is the first variance; is the second variance; i is the sequence number of the dust concentration data in the j-th dust concentration data sequence and the template data sequence; j is the sequence number of the dust concentration data sequence; n is the number of dust concentration data in the j-th dust concentration data sequence; | | is the absolute value sign; exp(-) is the inverse proportional normalization function.

[0046] It should be noted that It represents the accumulated value of the absolute value of the difference between the j-th dust concentration data sequence and the two dust concentration data with the same position in the template data sequence. The smaller it is, the closer the j-th dust concentration data sequence is to the template data sequence in terms of data changes, and the greater the similarity between the j-th dust concentration data sequence and the template data sequence; It represents the difference between the fluctuation range of the j-th dust concentration data sequence and the fluctuation range of the template data sequence. The smaller it is, the more similar the fluctuation range of the dust concentration data in the j-th dust concentration data sequence is to that in the template data sequence, and the greater the feature similarity between the j-th dust concentration data sequence and the template data sequence; the greater the feature similarity between the j-th dust concentration data sequence and the template data sequence, the more the j-th dust concentration data sequence conforms to the results of subsequent screening, and the greater the sequence representativeness judgment value of the j-th dust concentration data sequence.

[0047] (4) Obtaining a sequence representativeness evaluation value of any dust concentration data sequence based on a stability value of any dust concentration data sequence and a time series feature correlation coefficient and feature similarity between any dust concentration data sequence and the template data sequence.

[0048] Specifically, the sequence representativeness evaluation value of any dust concentration data sequence is obtained according to the stability value of any dust concentration data sequence, the time series feature correlation coefficient between any dust concentration data sequence and the template data sequence, and the average value of the feature similarity.

[0049] In one implementation, taking the j-th dust concentration data sequence as an example, the sequence representativeness evaluation value of the j-th dust concentration data sequence is calculated:

[0050] in, is the representative judgment value of the jth dust concentration data sequence; is the stability value of the j-th dust concentration data sequence; is the time series characteristic correlation coefficient between the jth dust concentration data sequence and the template data sequence; is the feature similarity between the jth dust concentration data sequence and the template data sequence; j is the sequence number of the dust concentration data sequence.

[0051] It should be noted that the greater the stability value of the j-th dust concentration data sequence, the greater the stability of the j-th dust concentration data sequence, the more it conforms to the results of subsequent screening, and the greater the sequence representativeness judgment value of the j-th dust concentration data sequence; the greater the temporal feature correlation coefficient between the j-th dust concentration data sequence and the template data sequence, the more the j-th dust concentration data sequence conforms to the results of subsequent screening, and the greater the sequence representativeness judgment value of the j-th dust concentration data sequence; the greater the feature similarity between the j-th dust concentration data sequence and the template data sequence, the more the j-th dust concentration data sequence conforms to the results of subsequent screening, and the greater the sequence representativeness judgment value of the j-th dust concentration data sequence. The greater the sequence representativeness judgment value of the j-th dust concentration data sequence, the less the j-th dust concentration data sequence is affected by noise and emergencies, and the j-th dust concentration data sequence can better represent the overall change trend and distribution characteristics of dust concentration data.

[0052] At this point, the sequence representativeness evaluation value of the j-th dust concentration data sequence is obtained.

[0053] Step S103, obtaining the sequence representativeness evaluation values ​​of all dust concentration data sequences, and selecting at least one target data sequence from the preset number of dust concentration data sequences according to the sequence representativeness evaluation values ​​of all dust concentration data sequences.

[0054] After obtaining the sequence representativeness judgment value of the j-th dust concentration data sequence, the sequence representativeness judgment values ​​of all dust concentration data sequences are obtained according to the method for obtaining the sequence representativeness judgment value of the j-th dust concentration data sequence. Then, based on the sequence representativeness judgment values ​​of all dust concentration data sequences, at least one target data sequence is screened that is less affected by noise and sudden events and can better represent the overall change trend and distribution characteristics of the dust concentration data.

[0055] Specifically, a sequence representative evaluation value interval is set, and dust concentration data sequences whose sequence representative evaluation values ​​are within the sequence representative evaluation value interval are screened and recorded as target data sequences.

[0056] In one embodiment, the sequence representative evaluation value interval is set to (0.7, 1). There is no limitation here and it can be set according to the specific implementation scenario. Taking the j-th dust concentration data sequence as an example, if the sequence representative evaluation value of the j-th dust concentration data sequence is within (0.7, 1), the j-th dust concentration data sequence is recorded as the target data sequence. Similarly, the dust concentration data sequences with sequence representative evaluation values ​​within (0.7, 1) are screened to obtain at least one target data sequence.

[0057] At this point, at least one target data sequence is screened out from the preset number of dust concentration data sequences.

[0058] Step S104: construct a dust concentration prediction model based on all target data sequences, and use the dust concentration prediction model to monitor and warn dust concentration data within a preset time length in the future.

[0059] After obtaining the target data sequence, all target data sequences are used to construct and analyze feature engineering. According to the characteristics of dust concentration data, a suitable prediction model, such as LSTM, is selected. The prediction model is trained using the target data sequence to obtain a dust concentration prediction model. The dust concentration prediction model is run on the cloud server to predict the dust concentration data at each sampling moment in the next 24 hours. The predicted value of dust concentration data at each sampling moment in the next 24 hours is obtained. Then, according to the obtained target data sequence, regional characteristic standards, and national and industry safety standards, the dust concentration data threshold is set. The dust concentration threshold is divided into dangerous There are three levels of dust concentration data threshold, warning level dust concentration data threshold and normal level dust concentration data threshold, among which the danger level dust concentration data threshold is greater than the warning level dust concentration data threshold, and the warning level dust concentration data threshold is greater than the normal level dust concentration data threshold. The dust concentration data prediction value at each sampling moment in the next 24 hours is compared with the dust concentration data threshold. When the dust concentration data prediction value at any sampling moment in the dust concentration data prediction value at each sampling moment in the next 24 hours is greater than the dust concentration data threshold, the corresponding level of alarm processing is performed and the relevant operators are notified. Among them, the next 24 hours is the future preset time length set in this embodiment, which is not limited here and can be set according to the specific implementation scenario.

[0060] In one embodiment, the threshold value of the hazardous dust concentration data is set to , the warning level dust concentration data threshold is , the normal dust concentration data threshold is There is no restriction here and it can be set according to the specific implementation scenario. Taking the dust concentration data prediction value at the xth sampling time in the next 24 hours as an example, when the dust concentration data prediction value at the xth sampling time is less than or equal to When the dust concentration data at the xth sampling moment is determined to be within the normal range; when the predicted value of the dust concentration data at the xth sampling moment is greater than and less than or equal to When the dust concentration data prediction value at the xth sampling moment is greater than and less than or equal to When the dust concentration data prediction value at the xth sampling moment is greater than When this happens, a serious warning is required, and the operator is notified to stop the operation and fully initiate dust removal and ventilation measures.

[0061] It should be noted that the focus of the present invention is to reduce the impact of extreme values ​​in dust concentration data caused by noise and emergencies, to screen target data sequences that are less affected by noise and emergencies and that can better represent the overall change trend and distribution characteristics of dust concentration data, and to improve the prediction ability of the dust concentration prediction model in the mine environment. It is an existing technology to construct a dust concentration prediction model based on all target data sequences and use the dust concentration prediction model to monitor and warn dust concentration data within a preset time length in the future, which will not be repeated here.

[0062] In summary, the embodiment of the present invention obtains dust concentration data at each sampling moment in a mine, obtains an initial dust concentration data sequence within a preset sampling period, obtains a preset number of initial dust concentration data sequences, pre-processes each of the initial dust concentration data sequences respectively, and obtains the preset number of dust concentration data sequences; obtains a template data sequence according to the correlation between the preset number of dust concentration data sequences, and for any dust concentration data sequence in the preset number of dust concentration data sequences, obtains a sequence representative judgment value of any dust concentration data sequence according to the data fluctuation characteristics in the any dust concentration data sequence, and the correlation and data difference between the any dust concentration data sequence and the template data sequence; obtains sequence representative judgment values ​​of all dust concentration data sequences, and selects at least one target data sequence from the preset number of dust concentration data sequences according to the sequence representative judgment values ​​of all dust concentration data sequences; constructs a dust concentration prediction model according to all target data sequences, and uses the dust concentration prediction model to monitor and warn dust concentration data within a preset time length in the future. Among them, the sequence representativeness judgment value of any dust concentration data sequence obtained through the data fluctuation characteristics in any dust concentration data sequence, as well as the correlation and data difference between any dust concentration data sequence and the template data sequence, can reflect the extent to which the dust concentration data sequence is affected by noise and emergencies, and can also reflect the overall trend and distribution characteristics of the dust concentration data sequence. By screening and obtaining a target data sequence that is more in line with the dust concentration data change trend in the actual scenario and is less affected by noise and emergencies, a dust concentration prediction model is constructed, the influence of extreme values ​​in the dust concentration data is reduced, and the prediction ability of the dust concentration prediction model in the mine environment is improved, thereby accurately monitoring and warning the dust concentration in the mine environment in real time, thereby effectively protecting the health and safety of mine workers.

[0063] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A real-time monitoring and early warning system for mining dust concentration based on the Internet of Things, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the following method is implemented: Obtain dust concentration data at each sampling moment in the mine, obtain an initial dust concentration data sequence within a preset sampling period, obtain a preset number of initial dust concentration data sequences, pre-process each of the initial dust concentration data sequences respectively, and obtain the preset number of dust concentration data sequences; According to the correlation between the preset number of dust concentration data sequences, a template data sequence is obtained, and for any dust concentration data sequence among the preset number of dust concentration data sequences, according to the data fluctuation characteristics in the any dust concentration data sequence, and the correlation and data difference between the any dust concentration data sequence and the template data sequence, a sequence representative evaluation value of the any dust concentration data sequence is obtained; Obtaining sequence representativeness evaluation values ​​of all dust concentration data sequences, and selecting at least one target data sequence from the preset number of dust concentration data sequences according to the sequence representativeness evaluation values ​​of all dust concentration data sequences; A dust concentration prediction model is constructed based on all target data sequences, and the dust concentration data within a preset time length in the future is monitored and warned using the dust concentration prediction model.

2. The real-time monitoring and early warning system for mining dust concentration based on the Internet of Things according to claim 1 is characterized in that: The obtaining of a sequence representative evaluation value of any dust concentration data sequence according to the data fluctuation characteristics in any dust concentration data sequence, and the correlation and data difference between any dust concentration data sequence and the template data sequence includes: According to the data fluctuation characteristics in any dust concentration data sequence, obtaining the stability value of any dust concentration data sequence; Obtaining the Spearman correlation coefficient between any one of the dust concentration data sequences and the template data sequence, and the Spearman ranking of each dust concentration data in any one of the dust concentration data sequences and the template data sequence, and obtaining the time series feature correlation coefficient between any one of the dust concentration data sequences and the template data sequence according to the Spearman correlation coefficient between any one of the dust concentration data sequences and the template data sequence, and the time difference between any one of the dust concentration data sequences and the dust concentration data with the same Spearman ranking in the template data sequence; According to the data difference between any one of the dust concentration data sequences and the template data sequence, obtaining the feature similarity between any one of the dust concentration data sequences and the template data sequence; According to the stability value of any dust concentration data sequence, and the time series feature correlation coefficient and feature similarity between any dust concentration data sequence and the template data sequence, a sequence representativeness evaluation value of any dust concentration data sequence is obtained.

3. The real-time monitoring and early warning system for mining dust concentration based on the Internet of Things according to claim 2 is characterized in that: The step of obtaining the stability value of any dust concentration data sequence according to the data fluctuation characteristics in any dust concentration data sequence includes: Obtaining the maximum value and the minimum value in any dust concentration data sequence, and calculating the first difference between the maximum value and the minimum value in any dust concentration data sequence; Obtaining the average value and median value of any dust concentration data sequence, and calculating the absolute value of the first difference between the average value and the median value; An inverse proportional normalization process is performed on the sum of the first difference and the absolute value of the first difference to obtain a stability value of any dust concentration data sequence.

4. The real-time monitoring and early warning system for mining dust concentration based on the Internet of Things according to claim 2 is characterized in that: The step of obtaining the time series characteristic correlation coefficient between any dust concentration data sequence and the template data sequence according to the Spearman correlation coefficient between any dust concentration data sequence and the template data sequence, and the time difference between any dust concentration data sequence and dust concentration data with the same Spearman ranking in the template data sequence, comprises: Respectively obtain the time intervals between the sampling moments of two dust concentration data with the same Spearman ranking in any dust concentration data sequence and the template data sequence, obtain the corresponding accumulated value of the time interval, and perform inverse proportional normalization processing on the accumulated value of the time interval to obtain a first inverse proportional normalized value; According to the Spearman correlation coefficient between any dust concentration data sequence and the template data sequence, the first inverse proportional normalization value and the mean of the constant 1, the time series characteristic correlation coefficient between any dust concentration data sequence and the template data sequence is obtained.

5. The real-time monitoring and early warning system for mining dust concentration based on the Internet of Things according to claim 3 is characterized in that: The obtaining, according to the data difference between any dust concentration data sequence and the template data sequence, the degree of feature similarity between any dust concentration data sequence and the template data sequence comprises: respectively calculating the absolute values ​​of the differences between two dust concentration data at the same position in any dust concentration data sequence and the template data sequence, and correspondingly obtaining the accumulated value of the absolute values ​​of the differences, obtaining the maximum value and the minimum value in the template data sequence, calculating the second difference between the maximum value and the minimum value in the template data sequence, calculating the second absolute value of the difference between the first difference and the second difference, and performing inverse proportional normalization processing on the accumulated value of the absolute values ​​of the differences and the addition result of the second absolute value of the differences, to obtain a second inverse proportional normalized value; Obtaining a first variance of any dust concentration data sequence, obtaining a second variance of the template data sequence, and performing inverse proportional normalization processing on an absolute value of a difference between the first variance and the second variance to obtain a third inverse proportional normalized value; The degree of feature similarity between any dust concentration data sequence and the template data sequence is obtained according to the average of the second inverse proportional normalized value and the third inverse proportional normalized value.

6. The real-time monitoring and early warning system for mining dust concentration based on the Internet of Things according to claim 2 is characterized in that: The obtaining of a sequence representativeness evaluation value of any dust concentration data sequence according to the stability value of any dust concentration data sequence and the time series feature correlation coefficient and feature similarity between any dust concentration data sequence and the template data sequence includes: The sequence representativeness evaluation value of any dust concentration data sequence is obtained according to the stability value of any dust concentration data sequence, the time series feature correlation coefficient between any dust concentration data sequence and the template data sequence, and the average value of the feature similarity.

7. The real-time monitoring and early warning system for mining dust concentration based on the Internet of Things according to claim 1 is characterized in that: The step of selecting at least one target data sequence from the preset number of dust concentration data sequences according to the sequence representativeness evaluation values ​​of all dust concentration data sequences includes: Set a sequence representative evaluation value interval, and select dust concentration data sequences whose sequence representative evaluation values ​​are within the sequence representative evaluation value interval, and record them as target data sequences.

8. The real-time monitoring and early warning system for mining dust concentration based on the Internet of Things according to claim 1 is characterized in that: The step of obtaining a template data sequence according to the correlation between the preset number of dust concentration data sequences includes: The preset number of dust concentration data sequences are fitted to obtain a template data sequence.

Images