Undermine sensor disaster risk assessment method based on four-dimensional information diffusion

Abstract

The invention relates to the technical field of under-mine disaster risk assessment, in particular to an under-mine sensor disaster risk assessment method based on four-dimensional information diffusion, which comprises the steps of obtaining step data according to historical reference data, arranging a sensor in a space structure according to the step data, and constructing a space monitoring field, acquiring sensor position data and disaster data; according to the method, a sensor disaster risk is defined as a fuzzy risk, then disaster information of all sensors affected by disasters is diffused to a whole space monitoring field constructed based on a disaster scene based on four-dimensional information diffusion, an obtained information matrix is processed into a fuzzy relation matrix, and then a sensor disaster risk curved surface is constructed. Therefore, the sensor disaster risks of all the sensors in the disaster scene are described. According to the invention, through identification and diffusion of the sensor disaster data in the space monitoring field, assessment and early warning are accurately carried out on the disaster risk under the mine.

Description

Technical Field

[0001] The present invention relates to the technical field of mine disaster risk assessment, and in particular to an underground mine sensor disaster risk assessment method based on four-dimensional information diffusion. Background Art

[0002] With socioeconomic development, the demand for various mineral resources is increasing. Mining operations are prone to disasters such as underground fires, floods, and gas explosions. To mitigate the losses caused by these disasters, researchers are using empirical methods, statistical analysis, and big data intelligent identification to assess disaster risk.

[0003] Empirical methods rely primarily on expert judgment and historical disaster data. Their assessments are typically qualitative and lack quantification, resulting in low risk assessment accuracy. Statistical analysis methods use mathematical models and statistical analysis to quantify disaster risk, such as by assessing risk based on parameters like disaster frequency and intensity. Big data intelligent identification primarily uses mathematical models and big data technologies, such as artificial intelligence and machine learning, to quantify disaster risk.

[0004] Statistical analysis and big data intelligent identification require the collection of large amounts of disaster data for analysis and training. However, in reality, available disaster data is relatively scarce, resulting in insufficient and incomplete disaster samples. In the face of insufficient disaster data, information diffusion technology can be used for research. Information diffusion technology is a fuzzy mathematical processing method that uses set values ​​on samples, optimizing the use of fuzzy information in samples to compensate for information deficiencies. Currently, information diffusion technology is widely used in one-dimensional and two-dimensional disaster scenarios, focusing on assessing the loss of life and property caused by disasters. It is mostly used in disaster scenarios such as floods, typhoons, and earthquakes, and has less application in disaster assessments of mines.

[0005] In the existing methods of assessing disaster risks using information diffusion technology, the focus is on identifying and processing the obtained data; however, the location setting and identification of sensors have not been deeply studied. As a result, the obtained data are too dispersed, resulting in reduced recognition accuracy, or too concentrated, resulting in low recognition efficiency.

[0006] Therefore, a mine sensor disaster risk assessment method based on four-dimensional information diffusion is proposed. Summary of the Invention

[0007] The purpose of the present invention is to provide a mine sensor disaster risk assessment method based on four-dimensional information diffusion, defining the sensor disaster risk as a fuzzy risk, and then based on four-dimensional information diffusion, diffusing the disaster information of all sensors that have been affected by the disaster to the entire spatial monitoring field constructed based on the disaster scenario, processing the obtained information matrix into a fuzzy relationship matrix, and then constructing a sensor disaster risk surface to describe the sensor disaster risk of all sensors in the disaster scenario.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A mine sensor disaster risk assessment method based on four-dimensional information diffusion includes:

[0010] S10. Identify the spatial structure of the target mine, match and collect historical warning data based on the spatial structure to obtain historical reference data; obtain step length data based on the historical reference data, set sensors in the spatial structure based on the step length data, and construct a spatial monitoring field;

[0011] S20. Collect data from sensors in the spatial monitoring field to obtain sensor location data and disaster data; classify sensors according to the disaster data to obtain monitoring points and sample points, and collect the monitoring point data and sample point data; identify the sample point data to obtain the diffusion coefficient;

[0012] S30. Construct a disaster diffusion identification model to identify the diffusion coefficient and sample point data, obtain the diffusion of the sample point data in the spatial monitoring field, and construct a risk information matrix;

[0013] S40. Normalizing the risk information matrix to obtain a fuzzy relationship matrix; the fuzzy relationship matrix includes fuzzy position data and fuzzy disaster data of the sensor;

[0014] S50. Processing the fuzzy location data using a membership function to obtain a location fuzzy set; performing a fuzzy synthesis operation on the location fuzzy set and a fuzzy relationship matrix using an approximate inference operator to obtain a fuzzy risk value list; defuzzifying the fuzzy risk value list using a centroid method to obtain a predicted risk value list;

[0015] S60. Construct a risk surface recognition model to recognize the predicted risk value list to obtain disaster risk surface data; and provide real-time disaster warning based on the risk disaster surface data.

[0016] The step length data acquisition process is as follows: collecting historical spatial structures, historical disaster types, historical step length data and historical warning accuracy of various types of mines in historical data to obtain historical warning data;

[0017] Based on the spatial structure of the target mine, the historical spatial structure of the mine in the historical warning data is similarly identified, and matching is performed based on the similarity to obtain historical reference data; the historical reference data includes historical disaster types, historical step data, and historical warning accuracy extracted from the historical warning data;

[0018] The historical reference data is divided according to the historical disaster types in the historical reference data to obtain multiple disaster type reference data; the similarity of the spatial structure of the mine is used as the data weight, and the historical step length data and historical warning accuracy in the disaster type reference data are identified to obtain disaster step length data;

[0019] According to the frequency of occurrence, scope of impact, duration, spread speed and losses caused by the disaster type, the disaster type coefficient is calculated, and the disaster step data is weighted according to the disaster type coefficient to obtain the step data of the target mine.

[0020] The step data is the position interval of the sensor in the spatial structure and the level interval of the disaster data, including the first step data, the second step data, the third step data and the fourth step data;

[0021] The first step data is the position interval of the X axis in the spatial structure; the second step data is the position interval of the Y axis in the spatial structure; the third step data is the position interval of the Z axis in the spatial structure; and the fourth step data is the level interval for classifying the disaster risk values ​​obtained by sensor monitoring.

[0022] The sensor position data is the position data of the sensor in the spatial monitoring field; the disaster data is the disaster level fed back by the sensor; the sensors are divided according to the disaster level to obtain monitoring points and sample points; the monitoring points are sensors that are not affected by the disaster; the sample points are sensors that have been affected by the disaster.

[0023] The diffusion coefficient is calculated based on the two-point proximity principle and the average distance assumption, and is obtained by identifying the sample point data set through the Newton iteration method;

[0024] The diffusion coefficient includes a first diffusion coefficient, a second diffusion coefficient, a third diffusion coefficient and a fourth diffusion coefficient; the first diffusion coefficient is the diffusion coefficient on the X-axis; the second diffusion coefficient is the diffusion coefficient on the Y-axis; the third diffusion coefficient is the diffusion coefficient on the Z-axis; and the fourth diffusion coefficient is the diffusion coefficient of the sensor disaster data.

[0025] The identification process of the disaster diffusion identification model is as follows: creating a thread pool pool, with a total of s threads in the thread pool; selecting s according to CPU parameters; creating an initial matrix based on the spatial monitoring field, and initializing the values ​​of all positions in the initial matrix to 0; selecting a monitoring point and making a judgment based on the distance between the monitoring point and the sample point; when the distance threshold is met, calculating the diffusion impact data of the sample point on the monitoring point; accumulating the diffusion impact data of all the sample points obtained to obtain the diffusion disaster data of the monitoring point; and updating the initial matrix based on the diffusion disaster data of the monitoring point.

[0026] After all monitoring points are identified, the risk information matrix is ​​updated.

[0027] The calculation formula for the diffusion impact data is:

[0028]

[0029] Among them, μ((x i ,y i ,z i ,r i ),(x′ q ,y r ′,z′ v ,r u ′)) represents the diffusion impact data of sample point i on the monitoring point; (x i ,y i ,z i ,r i ) represents the sample point data, where x i ,y i , z i Represents the sample point location data, r i Represents the disaster data of the sample point; (x′ q ,y r ′,z′ v ,r u ′) monitoring point data, where x′ q ,y r ′,z′ v , represents the monitoring point location data, r u ′ represents the disaster data of the monitoring point; h x , h y , h z , h r are the first diffusion coefficient, the second diffusion coefficient, the third diffusion coefficient and the fourth diffusion coefficient respectively; exp() represents an exponential function with the natural constant e as the base.

[0030] The calculation formula for the diffusion disaster data is:

[0031]

[0032] Among them, μ((x i ,y i ,z i ,r i ),(x′ q ,y r ′,z′ v ,r u ′)) represents the diffusion impact data of sample point i on the monitoring point; η qrvu represents the diffusion disaster data of the monitoring point; n represents the number of sample points that meet the distance threshold.

[0033] The risk surface recognition model identifies the predicted risk value list in the following process:

[0034] Calculate the predicted disaster risk value of the sample point according to the predicted risk value list;

[0035] If the difference between the predicted disaster risk value and the actual disaster risk value of the sample is not greater than the risk threshold, no adjustment is required; if the difference between the predicted disaster risk value and the actual disaster risk value of the sample is greater than the risk threshold, the predicted disaster risk value in the predicted risk value list is adjusted to obtain an updated disaster risk value; an adjusted risk list is obtained based on the updated disaster risk value;

[0036] Repeat the adjustment of the risk list until the gap is smaller than the risk threshold or the number of cycles is reached;

[0037] The disaster risk surface data is obtained based on the final adjusted risk list.

[0038] The adjustment process of updating the disaster risk value is as follows:

[0039]

[0040] Where η″ qrvu Represents the updated disaster risk value; η′ qrvu represents the predicted disaster risk value; ε represents the risk threshold; sr i Represents the actual disaster risk value of the sample at sample point i.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. The present invention collects historical mine warning data to obtain historical spatial structure, historical disaster types, historical step length data, and historical warning accuracy. Based on the similarity between the spatial structure of the target mine and the historical spatial structure, the method screens mine data with similar spatial structure to the target mine to serve as historical reference data. Based on the historical warning accuracy, historical disaster types, and historical step lengths in the historical reference data, as well as the similarity of the mine spatial structure, the method identifies the target mine and obtains optimal step length data. This optimal step length data provides a basis for spatial sensor placement and disaster level setting, effectively improving the accuracy of risk assessment.

[0043] 2. This invention expands the application of information diffusion technology from two-dimensional to three-dimensional, enabling four-dimensional disaster risk assessment of underground mine sensor risks. The coordinates of three-dimensional spatial points plus the disaster risk value can be used to construct a four-dimensional sample. This four-dimensional sample is then evaluated, resulting in a four-dimensional disaster risk assessment, enabling accurate assessment of underground mine disaster risk.

[0044] 3. The present invention constructs a disaster diffusion model to identify the diffusion coefficient and sample point data, traverses and identifies the monitoring points through multi-threading technology, and makes judgments based on the distance between the monitoring point and the sample point; when the distance meets the threshold, the diffusion impact data of the sample point on the monitoring point is calculated; the diffusion impact data of all the sample points obtained by calculation are accumulated to obtain the diffusion disaster data of the monitoring point; after all the monitoring points are identified, the risk information matrix is ​​updated; the risk information matrix can be used to accurately and quickly identify the impact of the disaster information of the sample point on the monitoring point, effectively improving the accuracy and efficiency of identifying underground risks.

[0045] 4. The present invention performs a normalization operation on the risk information matrix to obtain a fuzzy relationship matrix; performs a fuzzy identification process on the fuzzy relationship matrix to obtain a fuzzy risk value list; defuzzifies the fuzzy risk value list through the center of gravity method to obtain a predicted risk value list; and then uses an algorithm to correct the risk surface to make the calculated disaster risk value and the true risk value of the sample more accurate; thereby effectively improving the accuracy of underground mine risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 Schematic diagram of a flow chart of a mine sensor disaster risk assessment method based on four-dimensional information diffusion according to the present invention;

[0047] Figure 2 A schematic diagram of a fuzzy synthesis scene according to the present invention;

[0048] Figure 3 Schematic diagram of disaster risk of the present invention. DETAILED DESCRIPTION

[0049] 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.

[0050] Example 1

[0051] This paper proposes a mine sensor disaster risk assessment method based on four-dimensional information diffusion, such as Figure 1 Shown, including:

[0052] S10. Identify the spatial structure of the target mine, match and collect historical warning data based on the spatial structure to obtain historical reference data; obtain step length data based on the historical reference data, set sensors in the spatial structure based on the step length data, and construct a spatial monitoring field. The step length data acquisition process is as follows:

[0053] Collect historical spatial structure, historical disaster types, historical step length data and historical warning accuracy of various types of mines in historical data to obtain historical warning data;

[0054] Based on the spatial structure of the target mine, the historical spatial structure of the mine in the historical warning data is similarly identified, and matching is performed based on the similarity to obtain historical reference data; the historical reference data includes historical disaster types, historical step data, and historical warning accuracy extracted from the historical warning data;

[0055] The historical reference data is divided according to the historical disaster types in the historical reference data to obtain multiple disaster type reference data; the similarity of the spatial structure of the mine is used as the data weight, and the historical step length data and historical warning accuracy in the disaster type reference data are identified to obtain disaster step length data;

[0056] According to the frequency of occurrence, scope of impact, duration, spread speed and losses caused by the disaster type, the disaster type coefficient is calculated, and the disaster step data is weighted according to the disaster type coefficient to obtain the step data of the target mine.

[0057] The step data is the position interval of the sensor in the spatial structure and the level interval of the disaster data; including the first step data, the second step data, the third step data and the fourth step data;

[0058] The first step data is the position interval of the X axis in the spatial structure; the second step data is the position interval of the Y axis in the spatial structure; the third step data is the position interval of the Z axis in the spatial structure; and the fourth step data is the level interval for classifying the disaster risk values ​​obtained by sensor monitoring.

[0059] The present invention collects historical mine warning data to obtain historical spatial structure, historical disaster types, historical step length data, and historical warning accuracy. Based on the similarity between the spatial structure of the target mine and the historical spatial structure, the method filters the data of mines with similar spatial structures to the target mine to serve as historical reference data. Based on the historical warning accuracy, historical disaster types, and historical step lengths in the historical reference data, as well as the similarity of the mine spatial structure, the method identifies the target mine and obtains the optimal step length data. This optimal step length data provides a basis for spatial sensor placement and disaster level setting, effectively improving the accuracy of risk assessment.

[0060] S20. Collect data from sensors in the spatial monitoring field to obtain sensor location data and disaster data; classify sensors according to the disaster data to obtain monitoring points and sample points, and collect monitoring point data and sample point data; identify the sample point data to obtain the diffusion coefficient.

[0061] The sensor location data is the sensor's position within the spatial monitoring field; the disaster data is the disaster level reported by the sensor. Sensors are classified according to the disaster level to obtain monitoring points and sample points. Monitoring points are sensors unaffected by the disaster; sample points are sensors affected by the disaster. Disaster risk values ​​are analyzed based on the parameters identified by the sensors. The disaster risk values ​​are then classified based on the fourth step data to obtain disaster data.

[0062] The diffusion coefficient is calculated based on the two-point proximity principle and the average distance assumption, and is obtained by identifying the sample point data set through the Newton iteration method;

[0063] The diffusion coefficient includes a first diffusion coefficient, a second diffusion coefficient, a third diffusion coefficient and a fourth diffusion coefficient; the first diffusion coefficient is the diffusion coefficient on the X-axis; the second diffusion coefficient is the diffusion coefficient on the Y-axis; the third diffusion coefficient is the diffusion coefficient on the Z-axis; and the fourth diffusion coefficient is the diffusion coefficient of the sensor disaster data.

[0064] The present invention collects sensor location data and disaster data of sensors, divides the sensor types according to the disaster data, and obtains monitoring points and sample points; the monitoring points are sensors that have not been affected by disasters; the sample points are sensors that have been affected by disasters; and then identification is performed based on the sample point data to obtain the diffusion coefficient, which provides a basis for risk diffusion identification.

[0065] S30. Construct a disaster diffusion identification model to identify the diffusion coefficient and sample point data, obtain the diffusion of the sample point data in the spatial monitoring field, and construct a risk information matrix.

[0066] The identification process of the disaster diffusion identification model is as follows: create a thread pool pool, with a total of s threads in the thread pool; the selection of s is determined according to the CPU parameters; create an initial matrix according to the spatial monitoring field, and the values ​​of all positions in the initial matrix are initialized to 0; select monitoring points, and make judgments based on the distance between the monitoring points and the sample points; when the distance threshold is met, calculate the diffusion impact data of the sample points on the monitoring points; accumulate the diffusion impact data of all the sample points obtained by measurement to obtain the diffusion disaster data of the monitoring points; update the initial matrix according to the diffusion disaster data of the monitoring points; after all the monitoring points are identified, update the risk information matrix.

[0067] The calculation formula for the diffusion impact data is:

[0068]

[0069] Among them, μ((x i ,y i ,z i ,r i ),(x′ q ,y r ′,z′ v ,r u ′)) represents the diffusion impact data of sample point i on the monitoring point; (x i ,y i ,z i ,r i ) represents the sample point data, where x i ,y i , z i Represents the sample point location data, r i Represents the disaster data of the sample point; (x′ q ,y r ′,z′ v ,r u ′) monitoring point data, where x′ q ,y r ′,z′ v , represents the monitoring point location data, r u′ represents the disaster data of the monitoring point; h x , h y , h z , h r are the first diffusion coefficient, the second diffusion coefficient, the third diffusion coefficient and the fourth diffusion coefficient respectively; exp() represents an exponential function with the natural constant e as the base.

[0070] The calculation formula for the diffusion disaster data is:

[0071]

[0072] Among them, μ((x i ,y i ,z i ,r i ),(x′ q ,y r ′,z′ v ,r u ′)) represents the diffusion impact data of sample point i on the monitoring point; η qrvu represents the diffusion disaster data of the monitoring point; n represents the number of sample points that meet the distance threshold.

[0073] The present invention constructs a disaster diffusion model to identify the diffusion coefficient and sample point data, traverses and identifies the monitoring points, and makes judgments based on the distance between the monitoring point and the sample point; when the distance meets the threshold, the diffusion impact data of the sample point on the monitoring point is calculated; the diffusion impact data of all the sample points obtained by calculation are accumulated to obtain the diffusion disaster data of the monitoring point; after all the monitoring points are identified, the risk information matrix is ​​updated; the risk information matrix can accurately identify the impact of the disaster information of the sample point on the monitoring point.

[0074] S40. Perform a normalization operation on the risk information matrix to obtain a fuzzy relationship matrix; the fuzzy relationship matrix includes the fuzzy position data and fuzzy disaster data of the sensor.

[0075] S50. Process the fuzzy position data through a membership function to obtain a position fuzzy set; perform a fuzzy synthesis operation on the position fuzzy set and the fuzzy relationship matrix through an approximate reasoning operator to obtain a fuzzy risk value list; defuzzify the fuzzy risk value list through a centroid method to obtain a predicted risk value list.

[0076] S60. Construct a risk surface recognition model to recognize the predicted risk value list to obtain disaster risk surface data; and provide real-time disaster warning based on the risk disaster surface data.

[0077] The risk surface identification model identifies the predicted risk value list in the following process: calculating the predicted disaster risk value of the sample point according to the predicted risk value list; if the difference between the predicted disaster risk value and the actual disaster risk value of the sample is not greater than the risk threshold, no adjustment is required; if the difference between the predicted disaster risk value and the actual disaster risk value of the sample is greater than the risk threshold, adjusting the predicted disaster risk value in the predicted risk value list to obtain an updated disaster risk value; obtaining an adjusted risk list according to the updated disaster risk value; repeating the operation on the adjusted risk list until the difference is less than the risk threshold or the number of cycles is reached; and obtaining disaster risk surface data according to the final adjusted risk list.

[0078] The adjustment process of updating the disaster risk value is as follows:

[0079]

[0080] Where η″ qrvu Represents the updated disaster risk value; η′ qrvu represents the predicted disaster risk value; ε represents the risk threshold; sr i Represents the actual disaster risk value of the sample at sample point i.

[0081] The present invention obtains step data based on historical reference data, sets sensors in the spatial structure according to the step data, constructs a spatial monitoring field, and obtains sensor position data and disaster data; defines sensor disaster risk as a fuzzy risk, and then based on four-dimensional information diffusion, diffuses the disaster information at all sensors that have been affected by the disaster to the entire spatial monitoring field constructed based on the disaster scenario, processes the obtained information matrix into a fuzzy relationship matrix, and then constructs a sensor disaster risk surface to obtain the sensor disaster risk of all sensors, so as to accurately and quickly evaluate and warn of underground mine disaster risks.

[0082] Example 2

[0083] In order to verify the effectiveness of the mine sensor disaster risk assessment method based on four-dimensional information diffusion described in the present invention, it is tested according to actual scenarios. The testing process includes:

[0084] S10. Identify the spatial structure of the target mine, match and collect historical warning data based on the spatial structure to obtain historical reference data; obtain step data based on the historical reference data, set sensors in the spatial structure based on the step data, and construct a spatial monitoring field.

[0085] The step length data acquisition process is as follows: collecting historical spatial structures, historical disaster types, historical step length data and historical warning accuracy of various types of mines in historical data to obtain historical warning data;

[0086] Based on the spatial structure of the target mine, the historical spatial structure of the mine in the historical warning data is similarly identified, and matching is performed based on the similarity to obtain historical reference data; the historical reference data includes historical disaster types, historical step data, and historical warning accuracy extracted from the historical warning data;

[0087] The historical reference data is divided according to the historical disaster types in the historical reference data to obtain multiple disaster type reference data; the similarity of the spatial structure of the mine is used as the data weight, and the historical step length data and historical warning accuracy in the disaster type reference data are identified to obtain disaster step length data;

[0088] According to the frequency of occurrence, scope of impact, duration, spread speed and losses caused by the disaster type, the disaster type coefficient is calculated, and the disaster step data is weighted according to the disaster type coefficient to obtain the step data of the target mine.

[0089] The step data is the position interval of the sensor in the spatial structure and the level interval of the disaster data; including the first step data, the second step data, the third step data and the fourth step data;

[0090] The first step data is the position interval of the X axis in the spatial structure; the second step data is the position interval of the Y axis in the spatial structure; the third step data is the position interval of the Z axis in the spatial structure; and the fourth step data is the level interval for classifying the disaster risk values ​​obtained by sensor monitoring.

[0091] S20. Collect data from sensors in the spatial monitoring field to obtain sensor location data and disaster data; classify sensors according to the disaster data to obtain monitoring points and sample points, and collect monitoring point data and sample point data; identify the sample point data to obtain the diffusion coefficient.

[0092] The sensor position data is the position data of the sensor in the space monitoring field; the disaster data is the disaster level fed back by the sensor;

[0093] Sensors are divided according to the disaster level to obtain monitoring points and sample points; the monitoring points are sensors that are not affected by the disaster; and the sample points are sensors that have been affected by the disaster.

[0094] The diffusion coefficient is calculated based on the two-point proximity principle and the average distance assumption, and is obtained by identifying the sample point data set through the Newton iteration method;

[0095] The diffusion coefficient includes a first diffusion coefficient, a second diffusion coefficient, a third diffusion coefficient and a fourth diffusion coefficient; the first diffusion coefficient is the diffusion coefficient on the X-axis; the second diffusion coefficient is the diffusion coefficient on the Y-axis; the third diffusion coefficient is the diffusion coefficient on the Z-axis; and the fourth diffusion coefficient is the diffusion coefficient of the sensor disaster data.

[0096] S30. Construct a disaster diffusion identification model to identify the diffusion coefficient and sample point data, obtain the diffusion of the sample point data in the spatial monitoring field, and construct a risk information matrix.

[0097] The identification process of the disaster diffusion identification model is as follows:

[0098] Create a thread pool pool, which has s threads in total; the selection of s is determined by the CPU parameters;

[0099] Creating an initial matrix according to the spatial monitoring field, wherein the values ​​of all positions of the initial matrix are initialized to 0;

[0100] Select a monitoring point and make a judgment based on the distance between the monitoring point and the sample point; when the distance threshold is met, calculate the diffusion impact data of the sample point on the monitoring point; accumulate the diffusion impact data of all the sample points to obtain the diffusion disaster data of the monitoring point; update the initial matrix based on the diffusion disaster data of the monitoring point;

[0101] After all monitoring points are identified, the risk information matrix is ​​updated.

[0102] The calculation formula for the diffusion impact data is:

[0103]

[0104] Among them, μ((x i ,y i ,z i ,r i ),(x′ q ,y r ′,z′ v ,r u ′)) represents the diffusion impact data of sample point i on the monitoring point; (x i ,y i ,z i ,r i ) represents the sample point data, where x i ,y i , z i Represents the sample point location data, r i Represents the disaster data of the sample point; (x′ q ,y r ′,z′v ,r u ′) monitoring point data, where x′ q ,y r ′,z′ v , represents the monitoring point location data, r u ′ represents the disaster data of the monitoring point; h x , h y , h z , h r are the first diffusion coefficient, the second diffusion coefficient, the third diffusion coefficient and the fourth diffusion coefficient respectively; exp() represents an exponential function with the natural constant e as the base.

[0105] The calculation formula for the diffusion disaster data is:

[0106]

[0107] Among them, μ((x i ,y i ,z i ,r i ),(x′ q ,y r ′,z′ v ,r u ′)) represents the diffusion impact data of sample point i on the monitoring point; η qrvu represents the diffusion disaster data of the monitoring point; n represents the number of sample points that meet the distance threshold.

[0108] The present invention collects sensor location data and disaster data of sensors, divides the sensor types according to the disaster data, and obtains monitoring points and sample points; the monitoring points are sensors that have not been affected by disasters; the sample points are sensors that have been affected by disasters; and then identification is performed based on the sample point data to obtain the diffusion coefficient, which provides a basis for risk diffusion identification.

[0109] The present invention designs an algorithm pseudo code for the identification process of a disaster diffusion identification model to realize multi-process risk matrix calculation based on four-dimensional information diffusion.

[0110] The pseudo code of the disaster diffusion identification model algorithm is shown in Algorithm 1:

[0111]

[0112] Wherein, Δx represents the first step length data, Δy represents the second step length data, Δz represents the third step length data; α represents the distance threshold.

[0113] S40. Perform a normalization operation on the risk information matrix to obtain a fuzzy relationship matrix; the fuzzy relationship matrix includes the fuzzy position data and fuzzy disaster data of the sensor.

[0114] S50. Process the fuzzy position data through a membership function to obtain a position fuzzy set; perform a fuzzy synthesis operation on the position fuzzy set and the fuzzy relationship matrix through an approximate reasoning operator to obtain a fuzzy risk value list; defuzzify the fuzzy risk value list through a centroid method to obtain a predicted risk value list.

[0115] Get sensor s j Position data (x j ,y j ,z j ), use the membership function to fuzzify the location data and obtain the location fuzzy set The calculation formula of the position fuzzy set is:

[0116]

[0117] in, is the processed position fuzzy set, (x j ,y j ,z j ) is the position data of the sensor; x′ Q , y′ G , z′ V is the position data of other sensors; f(x j ,x′ Q ), f(y j ,y′ G ), f(z j ,z′ V ), are the membership functions of position data in different dimensions respectively; Q, G, V are the spatial range data of the spatial monitoring field.

[0118] The approximate inference operator is then used to calculate the sensor's degree of membership in relation to R, where R represents the spatial field of sensor disaster risk, i.e., the distribution range of disaster data. The sensor's sensor disaster risk value is represented using the Zadeh notation method and defuzzified using the center of gravity method to obtain a list of predicted risk values.

[0119] This paper proposes a fuzzy synthesis method based on four-dimensional information diffusion. The algorithm pseudo code is shown in Algorithm 2:

[0120]

[0121] The specific process is to perform a loop operation on each sample point of the sample point data B:

[0122] First, the coordinates of the monitoring points around the sample point are calculated according to the step data. Then, the corresponding elements in the fuzzy relationship matrix are taken out according to the obtained coordinates. The taken out elements are then fuzzy synthesized with the approximate reasoning operator to obtain the fuzzy risk value and store it in the fuzzy risk value list.

[0123] In data synthesis, only the monitoring points around the sample point are used for fuzzy synthesis. This is because the monitoring point closest to the sample point has the greatest impact on it. The other monitoring points are farther away from the sample point, and the risk information value spread to the sample point is small and almost negligible. Therefore, in the fuzzy synthesis process, the sample point is only synthesized with the surrounding monitoring points. Figure 2 The figure shows a scene in which a sample point and eight surrounding monitoring points are fuzzy synthesized.

[0124] S60. Construct a risk surface recognition model to recognize the predicted risk value list to obtain disaster risk surface data; and provide real-time disaster warning based on the risk disaster surface data.

[0125] Using Algorithm 2, we obtain the disaster risk values ​​for all sensors to be evaluated. The location data of all sensors and the corresponding sensor disaster data can be combined to form an initial disaster risk surface. However, this initial surface may differ from the actual sample output surface. That is, under the same sample input location data conditions, the disaster data described by this initial surface may be higher or lower than the sample output.

[0126] To this end, we propose a disaster risk surface adjustment method based on four-dimensional information diffusion. During each iteration, this method adjusts the list of predicted risk values ​​based on the difference between the described risk and the actual risk value, thereby making the sample output surface more closely resemble the description of the disaster risk. This disaster risk surface adjustment method based on four-dimensional information diffusion is implemented through a risk surface identification model.

[0127] The risk surface recognition model identifies the predicted risk value list in the following process:

[0128] Calculate the predicted disaster risk value of the sample point according to the predicted risk value list;

[0129] If the difference between the predicted disaster risk value and the actual disaster risk value of the sample is not greater than the risk threshold, no adjustment is required; if the difference between the predicted disaster risk value and the actual disaster risk value of the sample is greater than the risk threshold, the predicted disaster risk value in the predicted risk value list is adjusted to obtain an updated disaster risk value; an adjusted risk list is obtained based on the updated disaster risk value;

[0130] Repeat the adjustment of the risk list until the gap is smaller than the risk threshold or the number of cycles is reached;

[0131] The disaster risk surface data is obtained based on the final adjusted risk list.

[0132] In this experiment, sensors that have been affected by disasters are randomly selected from all sensors in the scene as sample points. The corresponding risk levels are generated using a normal distribution. The normal distribution used is Where μ = 0.5, σ = 0.1. The final calculated risk map is as follows Figure 3 shown.

[0133] After obtaining the final disaster risk value, in order to allow emergency personnel to respond to the disaster quickly, the disaster risk value needs to be written in the form of risk level. This method provides emergency personnel with a more intuitive warning result. As shown in Table 1 and Table 2. Table 1 first row ω i Indicates the calculated risk value, the second line r i Represents the true risk value of the sample.

[0134] The following data is the first 11 results extracted from the results.

[0135] Table 1 Disaster risk values

[0136] <![CDATA[ω i ]]> 0.665 0.51 0.599 0.741 0.7 0.4 0.552 0.502 0.5 0.534 0.5 <![CDATA[r i ]]> 0.676 0.54 0.598 0.724 0.687 0.402 0.595 0.485 0.49 0.541 0.514

[0137] Table 2 Disaster risk levels

[0138] <![CDATA[ω i ]]> Ⅶ Ⅵ Ⅶ Ⅷ Ⅷ Ⅴ Ⅶ Ⅵ Ⅵ Ⅵ Ⅵ <![CDATA[r i ]]> Ⅶ Ⅵ Ⅶ Ⅷ Ⅷ Ⅴ Ⅶ Ⅵ Ⅵ Ⅵ Ⅵ

[0139] Table 2 shows the results after interval fuzzification. In real-world scenarios, disaster levels are typically presented as a range, not a specific value. Therefore, after deriving a specific disaster risk value, the results should be presented as intervals. This paper selects 10 points equally spaced from 0 to 1 and assigns risk levels as follows: For a point x, the interval [x - 0.5, x + 0.5) is considered a risk level range.

[0140] As shown in Tables 1 and 2 above, Table 1 compares the disaster risk values ​​calculated using the proposed method with the actual disaster risk values ​​for the sample, revealing a very small difference. Furthermore, after interval fuzzification, the two disaster risk values ​​fall within the same interval, indicating that their disaster risk levels are the same. Furthermore, the program execution time for this method is approximately 1.5 seconds, making it highly efficient. Experimental results demonstrate the effectiveness and efficiency of this method, enabling it to quickly assess the risk of a disaster, assign a risk level, and provide early warnings to emergency response personnel.

[0141] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A mine sensor disaster risk assessment method based on four-dimensional information diffusion, characterized in that: include: S10. Identify the spatial structure of the target mine, match and collect historical warning data based on the spatial structure, and obtain historical reference data; Step length data is obtained based on historical reference data, and sensors are set in the spatial structure based on the step length data to construct a spatial monitoring field; S20. Collect data from sensors in the spatial monitoring field to obtain sensor location data and disaster data; classify sensors according to the disaster data to obtain monitoring points and sample points, and collect the monitoring point data and sample point data; identify the sample point data to obtain the diffusion coefficient; S30. Construct a disaster diffusion identification model to identify the diffusion coefficient and sample point data, obtain the diffusion of the sample point data in the spatial monitoring field, and construct a risk information matrix; S40. Performing a normalization operation on the risk information matrix to obtain a fuzzy relationship matrix; The fuzzy relationship matrix includes fuzzy position data and fuzzy disaster data of the sensor; S50. Processing the fuzzy position data by a membership function to obtain a position fuzzy set; performing a fuzzy synthesis operation on the position fuzzy set and the fuzzy relationship matrix by an approximate reasoning operator to obtain a fuzzy risk value list; Defuzzifying the fuzzy risk value list by using a center of gravity method to obtain a predicted risk value list; S60. Construct a risk surface recognition model to recognize the predicted risk value list to obtain disaster risk surface data; and provide real-time disaster warning based on the risk disaster surface data.

2. The mine sensor disaster risk assessment method based on four-dimensional information diffusion according to claim 1 is characterized by: The step length data acquisition process is as follows: Collect historical spatial structure, historical disaster types, historical step length data and historical warning accuracy of various types of mines in historical data to obtain historical warning data; Based on the spatial structure of the target mine, the historical spatial structure of the mine in the historical warning data is similarly identified, and matching is performed based on the similarity to obtain historical reference data; the historical reference data includes historical disaster types, historical step data, and historical warning accuracy extracted from the historical warning data; Dividing the historical reference data according to the historical disaster types in the historical reference data to obtain multiple disaster type reference data; The similarity of the spatial structure of the mine is used as the data weight to identify the historical step length data and historical warning accuracy in the disaster type reference data to obtain the disaster step length data; According to the frequency of occurrence, scope of impact, duration, spread speed and losses caused by the disaster type, the disaster type coefficient is calculated, and the disaster step data is weighted according to the disaster type coefficient to obtain the step data of the target mine.

3. The mine sensor disaster risk assessment method based on four-dimensional information diffusion according to claim 1 is characterized by: The step data is the position interval of the sensor in the spatial structure and the level interval of the disaster data, including the first step data, the second step data, the third step data and the fourth step data; The first step data is the position interval of the X axis in the spatial structure; the second step data is the position interval of the Y axis in the spatial structure; the third step data is the position interval of the Z axis in the spatial structure; and the fourth step data is the level interval for classifying the disaster risk values ​​obtained by sensor monitoring.

4. The mine sensor disaster risk assessment method based on four-dimensional information diffusion according to claim 1 is characterized by: The sensor position data is the position data of the sensor in the spatial monitoring field; the disaster data is the disaster level fed back by the sensor; the sensors are divided according to the disaster level to obtain monitoring points and sample points; the monitoring points are sensors that are not affected by the disaster; the sample points are sensors that have been affected by the disaster.

5. The method for assessing underground mine sensor disaster risk based on four-dimensional information diffusion according to claim 1, characterized in that: The diffusion coefficient is calculated based on the two-point proximity principle and the average distance assumption, and is obtained by identifying the sample point data set through the Newton iteration method; The diffusion coefficients include a first diffusion coefficient, a second diffusion coefficient, a third diffusion coefficient, and a fourth diffusion coefficient; The first diffusion coefficient is the diffusion coefficient on the X-axis; The second diffusion coefficient is the diffusion coefficient on the Y axis; the third diffusion coefficient is the diffusion coefficient on the Z axis; The fourth diffusion coefficient is the diffusion coefficient of sensor disaster data.

6. The mine sensor disaster risk assessment method based on four-dimensional information diffusion according to claim 1 is characterized by: The identification process of the disaster diffusion identification model is as follows: Create a thread pool pool, which has s threads in total; the selection of s is determined by the CPU parameters; Creating an initial matrix according to the spatial monitoring field, wherein the values ​​of all positions of the initial matrix are initialized to 0; Select monitoring points and make judgments based on the distance between the monitoring points and the sample points; when the distance threshold is met, calculate the diffusion impact data of the sample points on the monitoring points; The diffusion impact data of all sample points obtained by measurement are accumulated to obtain the diffusion disaster data of the monitoring point; Update the initial matrix based on the diffusion disaster data of the monitoring points; After all monitoring points are identified, the risk information matrix is ​​updated.

7. The method for assessing underground mine sensor disaster risk based on four-dimensional information diffusion according to claim 6, characterized in that: The calculation formula for the diffusion impact data is: Among them, μ((x i ,y i ,z i ,r i ),(x′ q ,y′ r ,z′ v ,r′ u )) represents the diffusion impact data of sample point i on the monitoring point; (x i ,y i ,z i ,r i ) represents the sample point data, where x i ,y i , z i Represents the sample point location data, r i Represents the disaster data of the sample point; (x′ q ,y r ′,z′ v ,r u ′) monitoring point data, where x′ q ,y r ′,z′ v , represents the monitoring point location data, r u ′ represents the disaster data of the monitoring point; h x , h y , h z , h r are the first diffusion coefficient, the second diffusion coefficient, the third diffusion coefficient and the fourth diffusion coefficient respectively; exp() represents an exponential function with the natural constant e as the base.

8. The mine sensor disaster risk assessment method based on four-dimensional information diffusion according to claim 6 is characterized by: The calculation formula for the diffusion disaster data is: Among them, μ((x i ,y i ,z i ,r i ),(x′ q ,y r ′,z′ v ,r u ′)) represents the diffusion impact data of sample point i on the monitoring point; η qrvu represents the diffusion disaster data of the monitoring point; n represents the number of sample points that meet the distance threshold.

9. The method for assessing underground mine sensor disaster risk based on four-dimensional information diffusion according to claim 1, characterized in that: The risk surface recognition model identifies the predicted risk value list in the following process: Calculate the predicted disaster risk value of the sample point according to the predicted risk value list; If the difference between the predicted disaster risk value and the actual disaster risk value of the sample is not greater than the risk threshold, no adjustment is required; if the difference between the predicted disaster risk value and the actual disaster risk value of the sample is greater than the risk threshold, the predicted disaster risk value in the predicted risk value list is adjusted to obtain an updated disaster risk value; Adjust the risk list based on the updated disaster risk values; Repeat the adjustment of the risk list until the gap is smaller than the risk threshold or the number of cycles is reached; The disaster risk surface data is obtained based on the final adjusted risk list.

10. The mine sensor disaster risk assessment method based on four-dimensional information diffusion according to claim 9, characterized in that: The adjustment process of updating the disaster risk value is as follows: Where η″ qrvu Represents the updated disaster risk value; η′ qrvu represents the predicted disaster risk value; ε represents the risk threshold; sr i Represents the actual disaster risk value of the sample at sample point i.

Images