CN-122022450-A - Intelligent mine pressure monitoring and early warning system based on AI algorithm and dynamic analysis method thereof

CN122022450ACN 122022450 ACN122022450 ACN 122022450ACN-122022450-A

Abstract

The invention relates to the technical field of ore pressure early warning, and discloses an ore pressure intelligent monitoring early warning system based on an AI algorithm and a dynamic analysis method thereof, wherein the ore pressure intelligent monitoring early warning system based on the AI algorithm comprises a distance sensitive spectrum information extraction module for obtaining distance sensitive spectrum information; the system comprises a radial sequence determining module, a radial graph generating module, a node level risk amount calculating module, an alarm triggering state determining module and a space positioning module, wherein the radial sequence determining module determines a radial sequence, the radial graph generating module generates a radial graph, the node level risk amount calculating module generates a node level risk amount, the alarm triggering state determining module determines an alarm triggering state, and the space positioning module identifies the maximum value position of a continuous risk field as a space positioning result. According to the invention, the distance sensitive spectrum information is extracted and is integrated with the medium dispersion characteristic, the association rule of the signal and the propagation distance is captured, the ordered radial graph is constructed by combining the energy centroid and the radial sequence, and the internal connection of the spatial distribution and the spectrum characteristics among the nodes is excavated through the neural network, so that the false alarm and missing report condition of the alarm is reduced, and the safety production guarantee capability is improved.

Inventors

ZHAO YUMING
Huang changshan
ZHAO GUOQI
DOU TAO
Zheng Lankuang
ZHANG PEI
CHEN LU
FENG QIANG
ZHOU JIAN
HAN FEI

Assignees

淮北矿业股份有限公司
四川航天电液控制有限公司

Dates

Publication Date: 20260512
Application Date: 20251228

Claims (10)

1. Intelligent mine pressure monitoring and early warning system based on AI algorithm, which is characterized by comprising: The distance sensitive spectrum information extraction module is used for acquiring original vibration signals of the sensing nodes, converting the original vibration signals into power spectrum density, converting the power spectrum density into normalized power spectrum, and calculating the variance of the normalized power spectrum under the power of frequency by combining the medium dispersion index to acquire the distance sensitive spectrum information of each sensing node; the radial sequence determining module is used for determining an energy centroid according to node energy and space coordinates of each sensing node, calculating radial distance of each sensing node relative to the energy centroid, obtaining passive radial coordinates, arranging the sensing nodes according to the passive radial coordinates in an ascending order, and determining a radial sequence; The radial graph generation module establishes a connecting edge between adjacent sensing nodes according to a radial sequence to construct a path graph structure, calculates an edge weight according to the radial interval between the adjacent nodes, takes distance sensitive spectrum information and passive radial coordinates as node characteristics, and generates a radial graph; The node level risk amount calculation module inputs the radial graph into a neural network, extracts the difference value of the sensitive spectrum information between adjacent nodes as a message, performs node state evolution by combining the edge weight and the radial interval, and outputs the node level risk amount of each sensing node; The alarm triggering state determining module is used for carrying out path integration on the node level risk quantity along the passive radial coordinate to obtain a radial integration index, and comparing the index with a preset threshold value to determine an alarm triggering state; And the space positioning module is used for projecting the node level risk quantity to the target space domain and reconstructing a continuous risk field, and identifying the maximum position of the continuous risk field as a space positioning result and outputting the maximum position of the continuous risk field.
2. The intelligent mine pressure monitoring and early warning system based on the AI algorithm according to claim 1, wherein the original vibration signals of all the sensing nodes are obtained, and the analysis duration and the frequency analysis bandwidth consisting of a lower frequency limit and an upper frequency limit are determined; Performing integral transformation on the original vibration signal to obtain the energy distribution condition of the original vibration signal in the frequency analysis bandwidth, and obtaining the power spectral density; calculating a full frequency domain integral value of the power spectrum density in the frequency analysis bandwidth, and dividing the power spectrum density and the full frequency domain integral value to obtain a normalized power spectrum; And performing power processing on the frequency variable by using the medium dispersion index to obtain a frequency power function, calculating a second-order accumulated moment and a first-order accumulated moment of the frequency power function under the weighting of the normalized power spectrum, performing subtraction operation on the squares of the second-order accumulated moment and the first-order accumulated moment to obtain the variance of the normalized power spectrum under the frequency power, and determining the variance as distance sensitive spectrum information.
3. The intelligent mine pressure monitoring and early warning system based on the AI algorithm as set forth in claim 1, wherein the intelligent mine pressure monitoring and early warning system based on the AI algorithm is characterized by obtaining node energy and space coordinates of each sensing node, multiplying the space coordinates of each sensing node by the node energy of the sensing node to obtain node weighted space positions of each sensing node, calculating vector sums of all node weighted space positions to obtain space weighted sums, calculating numerical sums of all node energies to obtain total energy value, dividing the space weighted sums by the total energy value to obtain energy centroid; and comparing the values of the passive radial coordinates of the sensing nodes, and arranging the sensing nodes according to the sequence from small to large to obtain a radial sequence.
4. The intelligent mine pressure monitoring and early warning system based on the AI algorithm as claimed in claim 1, wherein two adjacent sensing nodes are connected in pairs and the direction is determined according to the arrangement sequence of the sensing nodes in the radial sequence, a directed connected edge set is established, and a path diagram structure is constructed; For each directional continuous edge in the path diagram structure, calculating the difference value between the passive radial coordinates of the sensing nodes at the end positions of the directional continuous edge and the passive radial coordinates of the sensing nodes at the start positions of the directional continuous edge to obtain radial intervals, calculating the numerical sum of all the radial intervals in the path diagram structure to obtain a radial span total value, and dividing the radial interval corresponding to each directional continuous edge by the radial span total value to obtain edge weights; And combining the distance sensitive spectrum information of each sensing node with the passive radial coordinates of each sensing node to construct node characteristics, and collecting each sensing node, the directed edge collection, the edge weight and the node characteristics to generate a radial graph.
5. The intelligent mine pressure monitoring and early warning system based on the AI algorithm as claimed in claim 1, wherein the distance sensitive spectrum information and the passive radial coordinates of each sensing node are set as initial node states of each sensing node; And calculating a passive radial coordinate difference value between the end point sensing node and the starting point sensing node of each directional connecting edge in the path diagram structure, obtaining a radial interval, calculating a distance sensitive spectrum information difference value between the starting point sensing node and the end point sensing node of the directional connecting edge, obtaining a spectrum information difference value, and multiplying the edge weight and the spectrum information difference value corresponding to the directional connecting edge and a vector containing a numerical value I and the radial interval to obtain a message vector.
6. The intelligent mine pressure monitoring and early warning system based on the AI algorithm as set forth in claim 5, wherein for each sensing node, all message vectors pointing to the sensing node are accumulated to obtain an aggregate message vector; Setting a fixed weight matrix, carrying out multiplication operation and summation on the current node state and the aggregate message vector by using the fixed weight matrix to obtain an evolution input item, and carrying out soft addition function operation on the evolution input item to finish node state updating; Setting a fixed weight vector, calculating an inner product of the fixed weight vector and the node state after the set times are updated, obtaining a risk mapping value, performing soft-add function operation on the risk mapping value, and outputting node level risk quantity of each sensing node.
7. The intelligent mine pressure monitoring and early warning system based on the AI algorithm as claimed in claim 1, wherein according to the arrangement sequence of each sensing node in the radial sequence, two adjacent sensing nodes are extracted, the numerical sum of node level risk amounts of the two sensing nodes is calculated, and the calculation of dividing by two is carried out, so that a segmented risk average value is obtained; Subtracting the passive radial coordinates of the sensor nodes arranged in front from the passive radial coordinates of the sensor nodes arranged in back in the two adjacent sensor nodes to obtain the sectional radial distance; multiplying the sectional risk mean value by the sectional radial distance to obtain sectional risk integral contribution, and accumulating all sectional risk integral contribution generated in the whole path range to obtain a radial integral index; Acquiring a preset threshold value, judging whether the radial integral index reaches the preset threshold value, determining the alarm triggering state as an opening alarm under the condition that the radial integral index is larger than or equal to the preset threshold value, and determining the alarm triggering state as a closing alarm under the condition that the radial integral index is smaller than the preset threshold value.
8. The intelligent mine pressure monitoring and early warning system based on the AI algorithm according to claim 1, wherein the space coordinates of all the sensing nodes and the node level risk quantity of each sensing node are obtained, a target space domain is set, the space geometric distance between any two sensing nodes is calculated to form a distance value set, the median of the distance value set is extracted, and the median is determined as the bandwidth; Selecting a space position point in a target space domain, calculating the geometric spacing between the space position point and the space coordinates of the sensing node aiming at each sensing node, dividing the geometric spacing by the bandwidth to obtain a ratio, performing square operation on the ratio, dividing the square operation result by two, taking a negative value, and performing exponential operation with a natural constant as a base to obtain a weight coefficient corresponding to the space position point of the sensing node.
9. The intelligent mine pressure monitoring and early warning system based on the AI algorithm of claim 8, wherein the node level risk amount of each sensing node is multiplied by a weight coefficient corresponding to a spatial position point of the sensing node to obtain a weighted risk component, the sum operation is performed on all weighted risk components to obtain a risk weighted sum value, the sum operation is performed on all weight coefficients to obtain a weight sum value, and the risk weighted sum value is divided by the weight sum value to obtain a risk evaluation value of the spatial position point; And respectively calculating corresponding risk evaluation values for all spatial position points in the target spatial domain, completing reconstruction of a continuous risk field, searching spatial coordinates with maximum risk evaluation values in the continuous risk field, determining the spatial coordinates as a spatial positioning result, and outputting the spatial coordinates.
10. The intelligent mine pressure monitoring and dynamic analyzing method based on the AI algorithm is characterized in that the intelligent mine pressure monitoring and early warning system based on the AI algorithm as claimed in any one of claims 1 to 9 is executed, and comprises the following steps: Step S201, obtaining an original vibration signal of a sensing node, converting the original vibration signal into power spectrum density, converting the power spectrum density into normalized power spectrum, and calculating the variance of the normalized power spectrum under the power of frequency by combining a medium dispersion index to obtain distance sensitive spectrum information of each sensing node; step S202, determining an energy mass center according to node energy and space coordinates of each sensing node, calculating radial distance of each sensing node relative to the energy mass center, obtaining passive radial coordinates, arranging the sensing nodes according to the passive radial coordinates in ascending order, and determining a radial sequence; step S203, establishing a connecting edge between adjacent sensing nodes according to a radial sequence to construct a path diagram structure, calculating an edge weight according to the radial interval between the adjacent nodes, and generating a radial diagram by taking distance sensitive spectrum information and passive radial coordinates as node characteristics; step S204, inputting the radial graph into a neural network, extracting the difference value of the sensitive spectrum information between adjacent nodes as a message, carrying out node state evolution by combining the edge weight and the radial interval, and outputting the node level risk quantity of each sensing node; Step S205, carrying out path integration on node level risk quantity along passive radial coordinates to obtain a radial integration index, comparing the index with a preset threshold value, and determining an alarm triggering state; and S206, projecting the node level risk amount to a target space domain, reconstructing a continuous risk field, and identifying the maximum position of the continuous risk field as a space positioning result and outputting.

Description

Intelligent mine pressure monitoring and early warning system based on AI algorithm and dynamic analysis method thereof Technical Field The invention relates to the technical field of ore pressure early warning, in particular to an intelligent ore pressure monitoring early warning system based on an AI algorithm and a dynamic analysis method thereof. Background With the increasing depth of mining, geological conditions are more complex, mine pressure appears more frequently, and mine pressure disasters become one of main risks threatening the safe production of mines. In order to prevent the mine pressure disasters, the mine needs to master the mine pressure dynamics in real time by means of the mine pressure monitoring and early warning technology. The traditional mine pressure monitoring mostly adopts a single signal threshold judgment mode, only evaluates risks through the amplitude of a vibration signal, does not consider the correlation characteristic of signal frequency spectrum characteristics and propagation distance, ignores the influence of medium dispersion on the signals, and causes the risk evaluation to lack comprehensiveness and accuracy. Meanwhile, the traditional technology is not combined with the energy of the sensing node and the space coordinate to construct orderly association, only single node signals are analyzed independently, and the propagation rule of the mine pressure risk along the space radial direction cannot be captured, so that the alarm triggering condition is set on one side, false alarm is easily caused by local signal fluctuation, or false alarm is easily caused by non-integration of global information. In addition, traditional monitoring can only acquire the risk data of discrete nodes, the discrete data cannot be fused into a continuous risk field, the core position of an ore pressure risk source is difficult to accurately position, the pertinence of on-site prevention and control measures is lacked, the potential risks are difficult to rapidly and effectively treat, and great hidden danger is brought to mine safety production. Disclosure of Invention The invention provides an intelligent mine pressure monitoring and early warning system based on an AI algorithm and a dynamic analysis method thereof, which solve the technical problems in the background technology. The invention provides an intelligent mine pressure monitoring and early warning system based on an AI algorithm, which comprises the following steps: The distance sensitive spectrum information extraction module is used for acquiring original vibration signals of the sensing nodes, converting the original vibration signals into power spectrum density, converting the power spectrum density into normalized power spectrum, and calculating the variance of the normalized power spectrum under the power of frequency by combining the medium dispersion index to acquire the distance sensitive spectrum information of each sensing node; the radial sequence determining module is used for determining an energy centroid according to node energy and space coordinates of each sensing node, calculating radial distance of each sensing node relative to the energy centroid, obtaining passive radial coordinates, arranging the sensing nodes according to the passive radial coordinates in an ascending order, and determining a radial sequence; The radial graph generation module establishes a connecting edge between adjacent sensing nodes according to a radial sequence to construct a path graph structure, calculates an edge weight according to the radial interval between the adjacent nodes, takes distance sensitive spectrum information and passive radial coordinates as node characteristics, and generates a radial graph; The node level risk amount calculation module inputs the radial graph into a neural network, extracts the difference value of the sensitive spectrum information between adjacent nodes as a message, performs node state evolution by combining the edge weight and the radial interval, and outputs the node level risk amount of each sensing node; The alarm triggering state determining module is used for carrying out path integration on the node level risk quantity along the passive radial coordinate to obtain a radial integration index, and comparing the index with a preset threshold value to determine an alarm triggering state; And the space positioning module is used for projecting the node level risk quantity to the target space domain and reconstructing a continuous risk field, and identifying the maximum position of the continuous risk field as a space positioning result and outputting the maximum position of the continuous risk field. The invention provides an intelligent mine pressure monitoring dynamic analysis method based on an AI algorithm, which comprises the following steps: Step S201, obtaining an original vibration signal of a sensing node, converting the original vibration signal into power spectrum density,