Microseismic event identification method, device and equipment and storage medium

Abstract

The invention discloses a micro-seismic event identification method, device and equipment and a storage medium, the disclosed micro-seismic event identification method is applied to a graph neural network, and the graph neural network is composed of a plurality of mutually connected nodes. According to the application, the response result of each sensor node is obtained by responding to the stress signal of the signal node through each sensor node, and the response result at least comprises waveform data and audio data; inputting each waveform data into a preset waveform prediction model to obtain an original waveform probability vector corresponding to each sensor node; inputting each piece of audio data into a preset audio prediction model to obtain an original audio probability vector corresponding to each sensor node; and according to each original waveform probability vector and each original audio probability vector, determining the probability that the stress signal is triggered by the preset micro-seismic event, thereby realizing automatic identification of the micro-seismic event without depending on artificial experience, effectively improving the identification speed of the micro-seismic event, and further improving the timeliness of micro-seismic early warning.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly to a microseismic event recognition method, device, equipment, and storage medium. Background Art

[0002] Mine microseisms are a type of ground pressure disaster caused by rock stress changes during mine exploitation. High-energy mine microseismic events can cause economic losses to mining enterprises and even endanger the lives of miners, seriously threatening the safe production of mines. Therefore, the monitoring and recognition of microseismic events are of great significance for ensuring property and life safety.

[0003] Currently, the main way to prevent high-energy microseismic events is to bury multiple sensors inside the mine. The sensors capture stress signals generated by stress changes and output corresponding waveform data and audio data. Technicians can identify microseismic events related to rock mass state changes by analyzing the waveform data and audio data, and then capture early warning signals of high-energy microseismic events based on the identified microseismic events, so as to take scientific and reasonable countermeasures to ensure the safe and stable progress of production activities.

[0004] However, since the stress signals monitored by the sensors are not only triggered by microseismic events but also by other types of events, such as blasting, noise, and mechanical vibrations, the above-mentioned manual method relies on manual experience and has a large workload, and cannot achieve rapid recognition of microseisms, resulting in low timeliness of microseismic early warning.

[0005] The above content is only used to assist in understanding the technical solution of this application, and does not represent an admission that the above content is prior art. Summary of the Invention

[0006] The main purpose of this application is to provide a microseismic event recognition method, device, equipment, and storage medium, aiming to solve the technical problem that the prior art relies on manual experience, has a large workload, and cannot achieve rapid recognition of microseisms, resulting in low timeliness of microseismic early warning.

[0007] To achieve the above object, this application provides a microseismic event recognition method. The method is applied to a graph neural network, which is composed of multiple interconnected nodes. The nodes include sensor nodes and signal nodes. The sensor nodes are composed of sensors, and the signal nodes are composed of the positions where stress signals are generated. The method includes: Each of the sensor nodes responds to the stress signal of the signal node to obtain a response result of each of the sensor nodes. The response result at least includes waveform data and audio data; Input each of the waveform data into a preset waveform prediction model to obtain an original waveform probability vector corresponding to each sensor node, where the original waveform probability vector represents the probability that the stress signal is triggered by each preset event; Input each of the audio data into a preset audio prediction model to obtain an original audio probability vector corresponding to each sensor node, where the original audio probability vector represents the probability that the stress signal is triggered by each of the preset events; Determine the probability that the signal node is triggered by a preset microseismic event according to each of the original waveform probability vectors and each of the original audio probability vectors.

[0008] In one embodiment, the step of determining the probability that the stress signal is triggered by a preset microseismic event according to each of the original waveform probability vectors and each of the original audio probability vectors includes: Obtain the environmental parameters of the signal node at the current moment, where the environmental parameters at least include temperature parameter, humidity parameter, air pressure parameter, and regional maximum principal stress value; Combine the original waveform probability vector and the original audio probability vector of the same sensor node according to the temperature parameter and the humidity parameter to obtain a combined probability vector corresponding to each sensor node; Aggregate each of the combined probability vectors according to the air pressure parameter and the regional maximum principal stress value to obtain a target probability vector; Determine the probability that the stress signal is triggered by a preset microseismic event based on the target probability vector.

[0009] In one embodiment, the step of combining the original waveform probability vector and the original audio probability vector of the same sensor node according to the temperature parameter and the humidity parameter to obtain a combined probability vector corresponding to each sensor node includes: Determine a temperature attenuation coefficient based on the temperature parameter through a preset temperature attenuation coefficient formula; Wherein, the preset temperature attenuation coefficient formula is: ; In the formula, is the signal node, t is the current moment, is the temperature attenuation coefficient of the signal node at the current moment, is the temperature parameter of the signal node at the current moment, is the first learnable attenuation coefficient, is the optimal working temperature of the sensor, is the temperature influence range parameter; Determine a humidity attenuation coefficient based on the humidity parameter through a preset humidity attenuation coefficient formula; Among them, the preset humidity attenuation coefficient formula is: ; In the formula, is the signal node, t is the current moment, is the humidity attenuation coefficient of the signal node at the current moment, is the humidity parameter of the signal node at the current moment, is the second learnable attenuation coefficient, is the learnable parameter, is the correction coefficient, is the humidity influence effective threshold; Determine the combined posterior probability vector corresponding to each sensor based on the temperature attenuation coefficient and the humidity attenuation coefficient through the preset probability formula; Among them, the preset probability formula is: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the combined posterior probability vector of the sensor node at the current moment, is the temperature attenuation coefficient of the signal node at the current moment, is the humidity attenuation coefficient of the signal node at the current moment, is the original waveform probability vector of the sensor node at the current moment, is the original audio probability vector of the sensor node at the current moment.

[0010] In one embodiment, the step of aggregating each combined posterior probability vector according to the air pressure parameter and the regional maximum principal stress value to obtain the target probability vector includes: Determine the air pressure attenuation coefficient based on the air pressure parameter through the preset air pressure attenuation coefficient formula; Among them, the preset air pressure attenuation coefficient formula is: ; In the formula, is the signal node, t is the current moment, is the air pressure attenuation coefficient of the signal node at the current moment, is the air pressure parameter of the signal node at the current moment, is the standard atmospheric pressure, is the sensitivity of the sensor node under the standard atmospheric pressure, is the pressure influence coefficient; Determine the maximum principal stress value of the sensor corresponding to each of the sensor nodes; Based on the regional maximum principal stress value and the maximum principal stress values of each of the sensors, determine the stress propagation attenuation coefficient corresponding to each of the sensor nodes through a preset stress propagation attenuation coefficient formula; Among them, the preset stress propagation attenuation coefficient formula is: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the stress propagation attenuation coefficient between the sensor node and the signal node at the current moment, is the regional maximum principal stress value, is the maximum principal stress value of the sensor, is the included angle degree between the regional maximum principal stress direction of the signal node and the line connecting the signal node and the sensor node, is the material attenuation coefficient of stress propagation, is the distance between the signal node and the sensor node; Determine the attention weight between each of the sensor nodes and the signal node according to the air pressure attenuation coefficient and each of the stress propagation attenuation coefficients; Aggregate each of the combined probability vectors according to each of the attention weights to obtain a target probability vector.

[0011] In an embodiment, the step of determining the attention weight between each of the sensor nodes and the signal node according to the air pressure attenuation coefficient and each of the stress propagation attenuation coefficients includes: Determine the attention coefficient between each of the sensor nodes and the signal node based on the air pressure attenuation coefficient and each of the stress propagation attenuation coefficients through a preset attention coefficient formula; Among them, the preset attention coefficient formula is: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the attention coefficient between the sensor node and the signal node at the current moment, is the air pressure attenuation coefficient of the signal node at the current moment, is the stress propagation attenuation coefficient between the sensor node and the signal node at the current moment, is the non-linear activation function, is the learnable parameter vector, Tis a matrix transpose operation, is a learnable weight matrix for shared linear transformation, is an environmental feature vector of the environment where the signal node is located, is an environmental feature vector of the environment where the sensor node is located; Normalize each of the attention coefficients through a preset attention weight formula to obtain the attention weights corresponding to each of the sensor nodes; Among them, the preset attention weight formula is: ; In the formula, is the signal node, j is the sensor node, is any sensor node that receives a stress signal, t is the current moment, is the attention weight between the sensor node and the signal node at the current moment, is the attention coefficient between the sensor node and the signal node at the current moment, is the environmental feature vector of the environment where the signal node is located, is the environmental feature vector of the environment where the sensor node is located, is the learnable weight matrix for shared linear transformation, is the set of all sensor nodes that receive stress signals, is the air pressure attenuation coefficient of the signal node at the current moment, is the stress propagation attenuation coefficient between the sensor node and the signal node at the current moment, is the attention coefficient between any sensor node that receives a stress signal and the signal node at the current moment, is the propagation attenuation coefficient between any sensor node that receives a stress signal and the signal node at the current moment.

[0012] In one embodiment, the step of aggregating the combined probability vectors according to the respective attention weights to obtain a target probability vector includes: Aggregate the combined probability vectors based on the respective attention weights through a first preset output formula to obtain a target probability vector; Among them, the first preset output formula is: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the target probability vector of the signal node at the current moment, is the activation function, is the attention weight between the sensor node and the signal node at the current moment, is the learnable weight matrix of the shared linear transformation, is the combined posterior probability vector of the sensor node at the current moment.

[0013] In one embodiment, the step of aggregating the combined posterior probability vectors according to the attention weights to obtain a target probability vector further includes: Aggregating the combined posterior probability vectors based on the attention weights through a second preset output formula to obtain a target probability vector; wherein, the second preset output formula is: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the target probability vector of the signal node at the current moment, is the activation function, is the attention weight between the sensor node and the signal node at the current moment, is the learnable weight matrix of the shared linear transformation, is the combined posterior probability vector of the sensor node at the current moment, Z is the number of attention heads.

[0014] In addition, to achieve the above object, the present application also proposes a microseismic event recognition device, and the device includes: A data acquisition module, configured to obtain the response results of each sensor node by responding to the stress signal of the signal node, and the response results at least include waveform data and audio data; A probability prediction module, configured to input each of the waveform data into a preset waveform prediction model to obtain an original waveform probability vector corresponding to each sensor node, and the original waveform probability vector represents the probability that the stress signal is triggered by each preset event; The probability prediction module is further configured to input each of the audio data into a preset audio prediction model to obtain an original audio probability vector corresponding to each sensor node, and the original audio probability vector represents the probability that the stress signal is triggered by each of the preset events; An event recognition module, configured to determine the probability that the stress signal is triggered by a preset microseismic event according to each of the original waveform probability vectors and each of the original audio probability vectors.

[0015] In addition, to achieve the above object, the present application also provides a microseismic event recognition device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, and the computer program is configured to implement the steps of the microseismic event recognition method as described above.

[0016] In addition, to achieve the above object, the present application also provides a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium, and when the computer program is executed by a processor, it implements the steps of the microseismic event recognition method as described above.

[0017] One or more technical solutions provided by the present application have at least the following technical effects: The present application is applied to a graph neural network, which is composed of a plurality of interconnected nodes. The nodes include sensor nodes and signal nodes. The sensor nodes are composed of sensors, and the signal nodes are composed of positions generating stress signals. The method includes: each sensor node responds to the stress signal of the signal node to obtain the response result of each sensor node, and the response result at least includes waveform data and audio data; inputting each waveform data into a preset waveform prediction model to obtain the original waveform probability vector corresponding to each sensor node, and the original waveform probability vector represents the probability that the stress signal is triggered by each preset event; inputting each audio data into a preset audio prediction model to obtain the original audio probability vector corresponding to each sensor node, and the original audio probability vector represents the probability that the stress signal is triggered by each preset event; determining the probability that the stress signal is triggered by a preset microseismic event according to each original waveform probability vector and each original audio probability vector. By constructing a graph neural network including sensor nodes and signal nodes, and adopting the environmental parameters of the signal nodes, the original waveform probability vector and the original audio probability vector representing the probability that the stress signal is triggered by each preset event are obtained through the preset waveform prediction model and the preset audio prediction model respectively, and then combining the original waveform probability vector and the original audio probability vector, the probability that the stress signal is triggered by a preset microseismic event is determined, so that the probability that the stress signal is triggered by each preset event can be predicted, and then the automatic recognition of microseismic events can be realized, without relying on manual experience, effectively improving the recognition speed of microseisms, and further improving the timeliness of microseismic early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings here are incorporated into the specification and form a part of this specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0019] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0020] Figure 1 It is a schematic flowchart of the first embodiment of the microseismic event recognition method of the present application; Figure 2 It is a schematic flowchart of the second embodiment of the microseismic event recognition method of the present application; Figure 3 It is a schematic flowchart of the third embodiment of the microseismic event recognition method of the present application; Figure 4 It is a block diagram of the module structure of the microseismic event recognition device of the present application; Figure 5 It is a schematic diagram of the hardware structure of the microseismic event recognition device of the present application.

[0021] The realization of the purpose of the present application, functional features and advantages will be further described with reference to the embodiments and the drawings. Specific embodiments

[0022] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0023] To better understand the technical solutions of the present application, the following will be described in detail in combination with the drawings of the specification and specific embodiments.

[0024] The main solution of the embodiment of the present application is applied to a graph neural network, which is composed of multiple interconnected nodes. The nodes include sensor nodes and signal nodes. The sensor nodes are composed of sensors, and the signal nodes are composed of the positions generating stress signals. The solution includes: each sensor node responds to the stress signal of the signal node to obtain the response results of each sensor node. The response results at least include waveform data and audio data; input each waveform data into a preset waveform prediction model to obtain the original waveform probability vector corresponding to each sensor node. The original waveform probability vector represents the probability that the stress signal is triggered by each preset event; input each audio data into a preset audio prediction model to obtain the original audio probability vector corresponding to each sensor node. The original audio probability vector represents the probability that the stress signal is triggered by each preset event; obtain the environmental parameters of the signal node at the current moment; determine the probability that the stress signal is triggered by a preset microseismic event according to the environmental parameters, each original waveform probability vector and each original audio probability vector.

[0025] In the process of manually identifying microseismic events, since the stress signals monitored by sensors are triggered not only by microseismic events, but also by other types of events such as blasting, noise, and mechanical vibration, the above-mentioned manual method relies on manual experience, involves a large amount of work, and cannot achieve rapid identification of microseismic events, resulting in low timeliness of microseismic early warning.

[0026] The present application provides a solution. By constructing a graph neural network including sensor nodes and signal nodes, using the environmental parameters of the signal nodes, and combining the environmental parameters with the original waveform probability vector and the original audio probability vector that can characterize the triggering probability of preset events, the probability that the stress signal is triggered by a preset microseismic event is determined. Thus, based on identifying the probabilities of stress signals being triggered by various preset events, automatic identification of microseismic events can be achieved, without relying on manual experience, effectively improving the identification speed of microseismic events, and further improving the timeliness of microseismic early warning.

[0027] Based on this, the present application provides a method for identifying microseismic events, referring to Figure 1 , Figure 1 which is a schematic flowchart of the first embodiment of the microseismic event identification method of the present application.

[0028] In this embodiment, it is applied to a graph neural network. The graph neural network is composed of multiple interconnected nodes. The nodes include sensor nodes and signal nodes. The sensor nodes are composed of sensors, and the signal nodes are composed of the positions where stress signals are generated. The microseismic event identification method includes steps S10 to S40: Step S10, each of the sensor nodes responds to the stress signal of the signal node to obtain the response results of each of the sensor nodes. The response results at least include waveform data and audio data.

[0029] It should be noted that the execution subject of the method in this embodiment can be a computing service device with functions of microseismic event identification, network communication, and program operation, such as a mobile phone, a tablet computer, a personal computer, etc., or other electronic devices that implement the same or similar functions, such as a device including an intelligent system for mine safety monitoring. Hereinafter, a microseismic event identification device (abbreviated as identification device) is used to illustrate this embodiment and the following embodiments.

[0030] It can be understood that the above sensors can be detection devices for capturing stress signals generated inside the mine. The stress signal can be a mechanical wave signal generated due to rock stress changes during mine mining. The stress changes can be triggered by corresponding events inside the mine, such as microseismic events, blasting events, noise events, mechanical vibration events, etc.

[0031] It should be noted that the three-dimensional spatial image of the mine address can be divided into multiple grid regions with different shapes according to the ore body address characteristics and the geological structure of the mining area, and the overall grid region composed of each grid region covers the entire mining area. Each sensor can be used as a sensor node in the grid region. Correspondingly, the position where the stress signal is generated can be used as a signal node in the grid region. Each sensor node and the signal node are connected by edges in the grid region to form a graph neural network.

[0032] For example, assuming the coordinates of the signal node are , and the coordinates of the sensor node are , then the distance between the signal node and any sensor node can be represented by a preset distance formula. The preset distance formula is: ; In the formula, is the signal node, j is the sensor node, is the distance between the signal node and the sensor node.

[0033] It can be understood that the waveform data can be a continuous change record of the stress signal in the time domain. After the sensor receives the stress signal, it can convert the stress signal into an electrical signal, and after analog-to-digital conversion, it obtains the digitized waveform data, which can characterize the changes of information such as the vibration amplitude, frequency, and phase of the stress over time.

[0034] It should be noted that the audio data can be the propagation situation of the sound in the air when the event corresponding to the stress signal occurs, including the intensity, frequency components, duration, etc. of the sound. After the sensor receives the stress signal, it can obtain the audio data by collecting the sound or acoustic wave signal when the event corresponding to the stress signal occurs.

[0035] In a specific implementation, when a stress signal appears in the internal grid region of the mine, that is, in the graph neural network, each sensor can receive the stress signal and identify it, and output the corresponding waveform data and audio data respectively. The waveform data and audio data output by each sensor are the response results. The above recognition device can communicate with each sensor, and after each sensor responds to the stress signal of the signal node, it obtains the response results of each sensor.

[0036] Step S20: Input each of the waveform data into a preset waveform prediction model to obtain the original waveform probability vector corresponding to each of the sensor nodes, where the original waveform probability vector characterizes the probability that the stress signal is triggered by each preset event.

[0037] Step S30: Input each of the audio data into a preset audio prediction model to obtain an original audio probability vector corresponding to each sensor node, where the original audio probability vector represents the probability that the stress signal is triggered by each of the preset events.

[0038] It should be noted that the preset waveform prediction model can be a machine learning model or a deep learning model for predicting the probability that a stress signal is triggered by different preset events based on waveform data.

[0039] It can be understood that the preset audio prediction model can be a machine learning model or a deep learning model for predicting the probability that a stress signal is triggered by different preset events based on audio data.

[0040] In a specific implementation, a large amount of waveform data and audio data of known event types can be collected. The collected waveform data and audio data are respectively preprocessed, such as normalization, denoising, resampling, etc. For waveform data, methods such as short-time Fourier transform and wavelet transform can be used to extract time-frequency domain features, and the time-frequency domain features such as amplitude, frequency, energy, duration, etc. Then, based on the extracted time-frequency domain features and the corresponding labels (constructed by event types), waveform training data is constructed; for audio data, methods such as Mel spectrogram and voice spectrogram can be used to convert the audio signal into image features, and the image features such as Mel frequency cepstral coefficients, spectrogram, zero-crossing rate, etc. Then, based on the extracted image features and the corresponding labels (constructed by event types), audio training data is constructed. The initial model is trained with the waveform training data to obtain the preset waveform prediction model, and the initial model is trained with the audio training data to obtain the preset audio prediction model. Among them, the initial model can be a convolutional neural network, a recurrent neural network, etc., and this embodiment does not limit this.

[0041] Further, after the above recognition device receives the waveform data and audio data fed back by each sensor at the current moment, it can call the preset waveform prediction model and the preset audio prediction model. Input each waveform data into the preset waveform prediction model to obtain an original waveform probability vector corresponding to each sensor, where the original waveform probability vector represents the probability that the stress signal is triggered by each preset event; input each audio data into the preset audio prediction model to obtain an original audio probability vector corresponding to each sensor, where the original audio probability vector represents the probability that the stress signal is triggered by each preset event.

[0042] It should be noted that the above-mentioned preset events correspond to the labels used in the training processes of the preset waveform prediction model and the preset audio prediction model. The number of specific event types can be determined according to actual needs. For example, when the labels used in the training processes of the above-mentioned preset waveform prediction model and the preset audio prediction model are microseismic events, blasting events, noise events, and mechanical vibration events, the corresponding preset events can be microseismic events, blasting events, noise events, and mechanical vibration events. Assuming the type of the preset event is N, for any sensor node, the obtained original waveform probability vector and original audio probability vector are expressed as follows: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the original waveform probability vector, is the original audio probability vector, is the original waveform probability that the stress signal is triggered by the first type of event, is the original waveform probability that the stress signal is triggered by the second type of event ， is the original waveform probability that the stress signal is triggered by the Nth type of event, is the original audio probability that the stress signal is triggered by the first type of event, is the original audio probability that the stress signal is triggered by the second type of event, is the original audio probability that the stress signal is triggered by the Nth type of event.

[0043] Step S40, determine the probability that the stress signal is triggered by the preset microseismic event according to each of the original waveform probability vectors and each of the original audio probability vectors.

[0044] It should be noted that the above-mentioned preset microseismic event can be a preset event used to characterize microseismic. This preset microseismic event can be included in the preset event.

[0045] In a specific implementation, when the above-mentioned recognition device obtains the original waveform probability vectors and original audio probability vectors of each sensor, for the original waveform probability vector and the original audio probability vector of the same sensor, the elements of the same event type in the two are then arithmetically averaged, so as to fuse the original waveform probability vector and the original audio probability vector of the sensor to obtain a fused probability vector. The above process can be repeated to obtain the fused probability vectors of each sensor, and then the fused probability vectors of all sensors are averaged to obtain a target probability vector, where each element in the target probability vector represents the probability that the stress signal is triggered by each preset event. In the target probability vector, the event triggering the stress signal can be the element with the highest probability. If the probability of triggering by the preset microseismic event is the highest, it is determined that the above stress signal is triggered by the preset microseismic event.

[0046] This embodiment is applied to a graph neural network, which is composed of multiple interconnected nodes. The nodes include sensor nodes and signal nodes. The sensor nodes are composed of sensors, and the signal nodes are composed of the positions where stress signals are generated. The method includes: each sensor node responds to the stress signal of the signal node to obtain the response result of each sensor node, and the response result at least includes waveform data and audio data; input each waveform data into a preset waveform prediction model to obtain the original waveform probability vector corresponding to each sensor node, and the original waveform probability vector represents the probability that the stress signal is triggered by each preset event; input each audio data into a preset audio prediction model to obtain the original audio probability vector corresponding to each sensor, and the original audio probability vector represents the probability that the stress signal is triggered by each preset event; determine the probability that the stress signal is triggered by a preset microseismic event according to each original waveform probability vector and each original audio probability vector. In this embodiment, by constructing a graph neural network including sensor nodes and signal nodes, using the environmental parameters of the signal nodes, the original waveform probability vector and the original audio probability vector corresponding to the probability that the stress signal is triggered by each preset event are obtained respectively through the preset waveform prediction model and the preset audio prediction model, and then combining the original waveform probability vector and the original audio probability vector, the probability that the stress signal is triggered by the preset microseismic event is determined, so that the probability that the stress signal is triggered by each preset event can be predicted, and then the automatic recognition of microseismic events can be realized, without relying on manual experience, effectively improving the recognition speed of microseisms, and further improving the timeliness of microseismic early warning.

[0047] Based on the first embodiment of the present application, the second embodiment of the present application is proposed. In the second embodiment of the present application, the same or similar content as that in the above first embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 2 , Figure 2 which is a schematic flowchart of the second embodiment of the microseismic event recognition method of the present application.

[0048] In this embodiment, step S40 includes steps S41 to S44: Step S41, obtaining the environmental parameters of the signal node at the current moment, where the environmental parameters at least include temperature parameters, humidity parameters, air pressure parameters, and the regional maximum principal stress value.

[0049] It should be noted that the above regional maximum principal stress value can be the maximum stress among the three principal axis stresses of the internal stress state of the signal node when it is stressed.

[0050] In a specific implementation, corresponding environmental sensors for collecting environmental parameters can be additionally set at the positions where each node is located, such as capacitive humidity sensors, thermistor temperature sensors, semiconductor air pressure sensors, etc. The above identification device can communicate with each environmental sensor, and while obtaining the response results of each sensor, obtain the initial environmental parameters fed back by each environmental sensor, and then perform normalization, time alignment, etc. on each initial environmental parameter to obtain the environmental parameters for subsequent use.

[0051] Step S42, combining the original waveform probability vector and the original audio probability vector of the same sensor node according to the temperature parameter and the humidity parameter to obtain a combined probability vector corresponding to each sensor node.

[0052] In a specific implementation, for any sensor, the above identification device can correct its original waveform probability vector and the original audio probability vector according to the temperature parameter and humidity parameter corresponding to the sensor. Specifically, the weights corresponding to the temperature parameter and the humidity parameter can be preset, and the original waveform probability vector and the original audio probability vector are weighted and averaged, so that the obtained combined probability vector can reflect the environment where the sensor node is located and improve the accuracy of the combined probability vector.

[0053] In a feasible implementation manner, step S42 includes: Step S421, determining a temperature attenuation coefficient based on the temperature parameter through a preset temperature attenuation coefficient formula.

[0054] Among them, the preset temperature attenuation coefficient formula is: ; In the formula, is the signal node, t is the current moment, is the temperature attenuation coefficient of the signal node at the current moment, is the temperature parameter of the signal node at the current moment, is the first learnable attenuation coefficient, is the optimal operating temperature of the sensor, is the temperature influence range parameter.

[0055] In a specific implementation, in a high-temperature environment, the sensor is vulnerable to noise interference, resulting in errors in the waveform data output by the sensor, and further leading to the accuracy of the original waveform probability vector. Therefore, a preset temperature attenuation coefficient formula can be constructed in advance. After the above recognition device obtains the waveform data of each sensor and the temperature parameter of the signal node at the current moment, it can read the first learnable attenuation coefficient, the optimal working temperature of the sensor, and the temperature influence range parameter saved in advance, and substitute the temperature parameter, the first learnable attenuation coefficient, the optimal working temperature of the sensor, and the temperature influence range parameter into the above preset temperature attenuation coefficient formula to obtain the temperature attenuation coefficient of the signal node at the current moment. Through this temperature attenuation coefficient, the influence of temperature on the waveform data can be reduced, and the accuracy of the original waveform probability vector can be improved.

[0056] It should be understood that the above first learnable attenuation coefficient can be a parameter applicable to the preset temperature attenuation coefficient formula, and technicians can adjust it according to relevant test data to improve the accuracy of the preset temperature attenuation coefficient formula.

[0057] Step S422, determining a humidity attenuation coefficient based on the humidity parameter through a preset humidity attenuation coefficient formula.

[0058] Among them, the preset humidity attenuation coefficient formula is: ; In the formula, is the signal node, t is the current moment, is the humidity attenuation coefficient of the signal node at the current moment, is the humidity parameter of the signal node at the current moment, is the second learnable attenuation coefficient, is the learnable parameter, is the correction coefficient, is the humidity influence activation threshold.

[0059] In a specific implementation, in a mine environment, high humidity can lead to increased energy loss during the propagation of audio signals, resulting in errors in the audio data output by the sensor, and further leading to the accuracy of the original audio probability vector. Therefore, a preset humidity attenuation coefficient formula can be constructed in advance. After the above recognition device obtains the audio data of each sensor and the humidity parameter of the signal node at the current moment, it can read the second learnable attenuation coefficient, learnable parameter, correction coefficient, and humidity influence activation threshold saved in advance, and substitute the humidity parameter, the first learnable attenuation coefficient, the second learnable attenuation coefficient, the learnable parameter, the correction coefficient, and the humidity influence activation threshold into the above preset humidity attenuation coefficient formula to obtain the humidity attenuation coefficient of the signal node at the current moment. Through this humidity attenuation coefficient, the influence of humidity on the audio data can be reduced, and the accuracy of the original audio probability vector can be improved.

[0060] It should be understood that the above second learnable attenuation coefficient can be a parameter applicable to the preset humidity attenuation coefficient formula, and technicians can adjust it according to relevant test data to improve the accuracy of the preset humidity attenuation coefficient formula.

[0061] Step S423, determine the combined posterior probability vector corresponding to each sensor based on the temperature attenuation coefficient and the humidity attenuation coefficient through a preset probability formula.

[0062] Among them, the preset probability formula is: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the combined posterior probability vector of the sensor node at the current moment, is the temperature attenuation coefficient of the signal node at the current moment, is the humidity attenuation coefficient of the signal node at the current moment, is the original waveform probability vector of the sensor node at the current moment, is the original audio probability vector of the sensor node at the current moment.

[0063] In a specific implementation, for the same sensor node, after the above recognition device obtains the temperature attenuation coefficient and the humidity attenuation coefficient, it can substitute the temperature attenuation coefficient, the humidity attenuation coefficient, the original waveform probability vector, and the original audio probability vector into the preset probability formula constructed in advance to obtain the combined posterior probability vector corresponding to the sensor node, thereby realizing the correction of the original waveform probability vector and the original audio probability vector through the temperature attenuation coefficient and the humidity attenuation coefficient, reducing the influence of temperature and humidity on the original waveform probability vector and the original audio probability vector, and improving the accuracy of the combined posterior probability vector.

[0064] It should be understood that for each sensor node, the above-mentioned preset probability formula can be called to obtain the combined posterior probability vector corresponding to each.

[0065] Step S43: Aggregate each of the combined posterior probability vectors according to the air pressure parameter and the regional maximum principal stress value to obtain a target probability vector.

[0066] In a specific implementation, after the above-mentioned identification device obtains the combined posterior probability vectors of each sensor node, it can update the edge weights of the connections between each sensor node and the signal node through the air pressure parameter and the maximum principal stress value of the air pressure. Each edge weight can be used as a reliability weight for measuring the combined posterior probability vector corresponding to each sensor node. Then, a weighted average is performed based on each edge weight and the combined posterior probability vectors of each sensor node to obtain a target probability vector. Compared with obtaining the target probability vector by taking the average of the fused posterior probability vectors of all sensors, updating the edge weights of the connections between each sensor node and the signal node through the air pressure parameter and the maximum principal stress value of the air pressure effectively improves the accuracy of the target probability vector, and thus improves the identification accuracy of microseismic events.

[0067] Step S44: Determine the probability that the stress signal is triggered by a preset microseismic event based on the target probability vector.

[0068] In a specific implementation, after the above-mentioned identification device obtains the target probability vector, the element corresponding to the preset microseismic event in the target probability vector is used as the probability that the stress signal is triggered by the preset microseismic event. That is, in the target probability vector, if the element of the probability triggered by the preset microseismic event is the highest, it is determined that the above-mentioned stress signal is triggered by the preset microseismic event.

[0069] In this embodiment, by obtaining the environmental parameters of the signal node at the current moment, the environmental parameters at least include temperature parameter, humidity parameter, air pressure parameter, and regional maximum principal stress value; combining the original waveform probability vector and the original audio probability vector of the same sensor node according to the temperature parameter and the humidity parameter to obtain the combined posterior probability vector corresponding to each sensor node; aggregating each combined posterior probability vector according to the air pressure parameter and the regional maximum principal stress value to obtain a target probability vector; determining the probability that the stress signal is triggered by a preset microseismic event based on the target probability vector, thereby reducing the influence of temperature and humidity on the sensor response result and improving the accuracy of microseismic identification.

[0070] Based on the first embodiment and the second embodiment of the present application, the third embodiment of the present application is proposed. In the third embodiment of the present application, the content that is the same as or similar to the above-mentioned first embodiment and second embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 3 , Figure 3Schematic flowchart of the third embodiment of the microseismic event recognition method of the present application.

[0071] In this embodiment, step S43 includes steps S431 to S434: Step S431, determining the air pressure attenuation coefficient based on the air pressure parameter through a preset air pressure attenuation coefficient formula.

[0072] Wherein, the preset air pressure attenuation coefficient formula is: ; In the formula, is the signal node, t is the current moment, is the air pressure attenuation coefficient of the signal node at the current moment, is the air pressure parameter of the signal node at the current moment, is the standard atmospheric pressure, is the sensitivity of the sensor node under the standard atmospheric pressure, is the pressure influence coefficient.

[0073] In a specific implementation, in a mine environment, the sensitivity of the sensor will be affected by the air pressure. Since the air pressure inside the mine is relatively high, the sensitivity of the buried sensor may be reduced. Therefore, a preset air pressure attenuation coefficient formula can be constructed in advance. After the above recognition device obtains the air pressure parameter, it can read the pre-saved standard atmospheric pressure and the sensitivity of the sensor node under the standard atmospheric pressure, and then substitute the air pressure parameter, the standard atmospheric pressure, and the sensitivity of the sensor node under the standard atmospheric pressure into the above preset air pressure attenuation coefficient formula to obtain the air pressure attenuation coefficient. By using this air pressure attenuation coefficient, the influence of the air pressure on the sensor sensitivity can be reduced, thereby improving the accuracy of microseismic recognition.

[0074] Step S432, determining the maximum principal stress value of the sensor corresponding to each sensor node.

[0075] It should be noted that the above maximum principal stress value of the sensor can be the maximum stress among the three principal axis stresses of the internal stress state of the sensor node when it is stressed.

[0076] Step S433, determining the stress propagation attenuation coefficient corresponding to each sensor node based on the regional maximum principal stress value and the maximum principal stress value of each sensor through a preset stress propagation attenuation coefficient formula.

[0077] Wherein, the preset stress propagation attenuation coefficient formula is: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the stress propagation attenuation coefficient between the sensor node and the signal node at the current moment, is the maximum principal stress value of the area, is the maximum principal stress value of the sensor, is the included angle degree between the maximum principal stress direction of the area of the signal node and the line connecting the signal node and the sensor node, is the material attenuation coefficient of stress propagation, is the distance between the signal node and the sensor node.

[0078] In a specific implementation, a preset propagation attenuation coefficient formula can be constructed in advance. For any sensor node, after the above recognition device obtains the maximum principal stress value of the area and the maximum principal stress value of the sensor, it can read the included angle degree between the maximum principal stress direction of the area of the signal node and the line connecting the signal node and the sensor node, and the material attenuation coefficient of stress propagation that are pre-saved, and substitute the maximum principal stress value of the area, the maximum principal stress value of the sensor, the included angle degree, and the material attenuation coefficient of stress propagation into the above preset stress propagation attenuation coefficient formula to obtain the stress propagation attenuation coefficient between the sensor node and the signal node at the current moment. For other sensor nodes, repeating the above process can obtain the stress propagation attenuation coefficients between each sensor and the signal node.

[0079] Step S434, determine the attention weights between each of the sensor nodes and the signal node according to the air pressure attenuation coefficient and each of the stress propagation attenuation coefficients.

[0080] In a specific implementation, the above recognition device can determine the edge weights of the lines connecting each sensor node and the signal node according to the air pressure attenuation coefficient and each stress propagation attenuation coefficient, and this weight is the attention weight.

[0081] In a feasible implementation manner, step S434 includes: steps S4341~S4342: S4341, determine the attention coefficients between each of the sensor nodes and the signal node based on the air pressure attenuation coefficient and each of the stress propagation attenuation coefficients through a preset attention coefficient formula.

[0082] Among them, the preset attention coefficient formula is: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the attention coefficient between the sensor node and the signal node at the current moment, is the air pressure attenuation coefficient of the signal node at the current moment, is the stress propagation attenuation coefficient between the sensor node and the signal node at the current moment, is a non-linear activation function, is a learnable parameter vector, T is the matrix transpose operation, is the learnable weight matrix of the shared linear transformation, is the environmental feature vector of the environment where the signal node is located, is the environmental feature vector of the environment where the sensor node is located.

[0083] It should be noted that the environmental feature vector of the environment where the above signal node is located may include various environmental parameters at the signal node, and the environmental feature vector of the environment where the above sensor node is located may include various environmental parameters at the sensor node.

[0084] In a specific implementation, a preset attention coefficient formula can be pre-constructed. For any sensor node, after the above recognition device determines the air pressure attenuation coefficient and the stress propagation attenuation coefficient, it can obtain the non-linear activation function, the learnable parameter vector, the learnable weight matrix of the shared linear transformation, the environmental feature vector of the environment where the signal node is located, and the environmental feature vector of the environment where the sensor node is located, and substitute the air pressure attenuation coefficient, the stress propagation attenuation coefficient, the non-linear activation function, the learnable parameter vector, the learnable weight matrix of the shared linear transformation, the environmental feature vector of the environment where the signal node is located, and the environmental feature vector of the environment where the sensor node is located into the above preset attention coefficient formula to obtain the attention coefficient between the sensor node and the signal node at the current moment. For other sensor nodes, by repeating the above process, the attention coefficients between each sensor and the signal node can be obtained.

[0085] S4342, standardize each of the attention coefficients through a preset attention weight formula to obtain the attention weights corresponding to each of the sensor nodes; Among them, the preset attention weight formula is: ; In the formula, is the signal node, j is the sensor node, is any sensor node that receives the stress signal, t is the current moment, is the attention weight between the sensor node and the signal node at the current moment, is the attention coefficient between the sensor node and the signal node at the current moment, is the environmental feature vector of the environment where the signal node is located, is the environmental feature vector of the environment where the sensor node is located, is the learnable weight matrix of the shared linear transformation, A set of sensor nodes for stress signals triggered by all received signal nodes The air pressure attenuation coefficient of the signal node at the current moment The stress propagation attenuation coefficient between the sensor node and the signal node at the current moment The attention coefficient between any sensor node receiving the stress signal and the signal node at the current moment The propagation attenuation coefficient between any sensor node receiving the stress signal and the signal node at the current moment

[0086] In a specific implementation, a preset attention weight formula can be pre-constructed. The above recognition device can obtain the attention coefficients of each sensor node (i.e., all sensor nodes receiving stress signals). For the attention coefficient of any sensor node, it can be normalized by combining the attention coefficients of all sensor nodes to obtain the attention weight of any sensor node. Specifically, for any sensor node, the above recognition device can substitute the environmental feature vector of the environment where the signal node is located, the environmental feature vector of the environment where the sensor node is located, the learnable weight matrix of the shared linear transformation, and the attention coefficients of each sensor node into the above preset attention weight formula, so as to normalize the attention coefficient of the any sensor node and obtain the attention weight of the any sensor node. For other sensor nodes, by repeating the above process, the attention coefficients of each sensor can be normalized to obtain the attention weights of each sensor node.

[0087] Step S435, aggregate each combined probability vector according to each of the attention weights to obtain a target probability vector.

[0088] In a specific implementation, the above recognition device can aggregate each combined probability vector through the attention weights corresponding to each sensor node, so that the obtained target probability vector combines multi-modal data such as temperature parameters, humidity, air pressure, stress, waveform data, and audio data, effectively improving the reliability of the discrimination results of a single sensor, and further improving the accuracy of microseismic signal recognition by aggregating each combined probability vector of each sensor through the attention weights.

[0089] In a feasible implementation manner, step S435 includes step S4351: Step S4351, aggregate each combined probability vector based on each of the attention weights through a first preset output formula to obtain a target probability vector.

[0090] Among them, the first preset output formula is: ; In the formula, is the signal nodej is a sensor node, t is the current moment, is the target probability vector of the signal node at the current moment, is the activation function, is the attention weight between the sensor node and the signal node at the current moment, is the learnable weight matrix of the shared linear transformation, is the combined posterior probability vector of the sensor node at the current moment.

[0091] In a specific implementation, a first preset output formula can be pre-constructed. The above recognition device can substitute the activation function, attention weight, learnable weight matrix of the shared linear transformation, and the combined posterior probability vector of each sensor node into the above first preset output formula to aggregate the combined posterior probability vectors of each sensor node to obtain the target probability vector.

[0092] In another feasible implementation manner, step S435 includes step S4352: Step S4352: Aggregate each of the combined posterior probability vectors based on each of the attention weights through a second preset output formula to obtain the target probability vector.

[0093] Among them, the second preset output formula is: ; In the formula, is the signal node, j is the sensor node, t is the current moment, is the target probability vector of the signal node at the current moment, is the activation function, is the attention weight between the sensor node and the signal node at the current moment, is the learnable weight matrix of the shared linear transformation, is the combined posterior probability vector of the sensor node at the current moment, Z is the number of attention heads.

[0094] In a specific implementation, in order to improve the accuracy of the target probability vector, a multi-head attention mechanism can be pre-used to introduce multiple groups of learnable weight matrices of the shared linear transformation to construct the second preset output formula. The above recognition device can substitute the activation function, attention weight, learnable weight matrix of the shared linear transformation, number of attention heads, and the combined posterior probability vector of each sensor node into the above second preset output formula to obtain the accuracy of the target probability vector, thereby improving the accuracy of the target probability vector.

[0095] In this embodiment, the air pressure attenuation coefficient is determined based on the air pressure parameter through a preset air pressure attenuation coefficient formula; the maximum principal stress value of each sensor node is determined; the stress propagation attenuation coefficient corresponding to each sensor node is determined based on the regional maximum principal stress value and the maximum principal stress values of each of the sensors through a preset propagation attenuation coefficient formula; the attention weight between each sensor node and the signal node is determined according to the air pressure attenuation coefficient and each stress propagation attenuation coefficient; and the target probability vector is obtained by aggregating each combined posterior probability vector according to each attention weight, thereby realizing the integration of multi-modal data into the graph neural network and effectively improving the recognition accuracy of microseismic events.

[0096] This application also provides a microseismic event recognition device. Please refer to Figure 4 , Figure 4 , which is the module structure block diagram of the microseismic event recognition device of this application. The microseismic event recognition device includes: A data acquisition module 10, configured to obtain the response results of each of the sensor nodes by each sensor node responding to the stress signal of the signal node. The response results at least include waveform data and audio data.

[0097] A probability prediction module 20, configured to input each of the waveform data into a preset waveform prediction model to obtain the original waveform probability vector corresponding to each sensor node, where the original waveform probability vector represents the probability that the stress signal is triggered by each preset event.

[0098] The probability prediction module 20 is further configured to input each of the audio data into a preset audio prediction model to obtain the original audio probability vector corresponding to each sensor node, where the original audio probability vector represents the probability that the stress signal is triggered by each of the preset events.

[0099] An event recognition module 30, configured to determine the probability that the stress signal is triggered by a preset microseismic event according to each of the original waveform probability vectors and each of the original audio probability vectors.

[0100] The microseismic event recognition device provided by this application adopts the microseismic event recognition method in the above embodiment, which can solve the technical problems in the prior art that rely on manual experience, have a large workload, cannot quickly recognize microseisms, and result in low timeliness of microseismic early warning. Compared with the prior art, the beneficial effects of the microseismic event recognition device provided by this application are the same as those of the microseismic event recognition method provided in the above embodiment, and other technical features in the microseismic event recognition device are the same as the features disclosed in the above microseismic event recognition method, and will not be elaborated here.

[0101] The present application provides an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned security verification method applied to a head-mounted display device or a wearable device.

[0102] Reference is made below Figure 5 , Figure 5 which is a schematic hardware structure diagram of the microseismic event recognition device of the present application. The microseismic event recognition device involved in the embodiments of the present application may include, but is not limited to, mobile terminals such as laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The microseismic event recognition device shown is only an example and should not impose any limitations on the functions and usage scope of the embodiments of the present application.

[0103] As Figure 5As shown, the microseismic event recognition device may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of the microseismic event recognition device are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems may be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 may allow the microseismic event recognition device to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows a microseismic event recognition device having various systems, it should be understood that it is not required to implement or have all the shown systems. More or fewer systems may be alternatively implemented or had.

[0104] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program may be downloaded and installed from a network through the communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the methods of the embodiments disclosed in the present application are executed.

[0105] The microseismic event recognition device provided in the present application adopts the microseismic event recognition method in the above embodiment, and can solve the technical problems in the prior art that rely on manual experience, have a large workload, cannot achieve fast recognition of microseisms, and result in low timeliness of microseismic early warning. Compared with the prior art, the beneficial effects of the microseismic event recognition device provided in the present application are the same as those of the microseismic event recognition method provided in the above embodiment, and other technical features in the microseismic event recognition device are the same as those disclosed in the method of the previous embodiment, which will not be elaborated here.

[0106] It should be understood that each part disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0107] As mentioned above, the above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

[0108] This application provides a storage medium having computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the microseismic event recognition method in the above embodiments.

[0109] The computer-readable storage medium provided by this application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM: Random Access Memory), read-only memory (ROM: Read Only Memory), erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program, and the program can be used by or combined with an instruction execution system, system or device. The program code contained on the computer-readable storage medium can be transmitted by any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0110] The above computer-readable storage medium can be included in the microseismic event recognition device; or it can exist alone without being assembled into the microseismic event recognition device.

[0111] The above computer-readable storage medium carries one or more programs. When the one or more programs are executed by the microseismic event recognition device, the microseismic event recognition device is caused to: obtain the response results of each sensor node by means of the stress signals of the response signal nodes through each sensor node, and the response results at least include waveform data and audio data; input each waveform data into a preset waveform prediction model to obtain the original waveform probability vector corresponding to each sensor node, and the original waveform probability vector represents the probability that the stress signal is triggered by each preset event; input each audio data into a preset audio prediction model to obtain the original audio probability vector corresponding to each sensor node, and the original audio probability vector represents the probability that the stress signal is triggered by each preset event; determine the probability that the signal node is triggered by a preset microseismic event according to each original waveform probability vector and each original audio probability vector.

[0112] Computer program code for performing the operations of the present application may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may execute entirely on the user's computer, partially on the user's computer, execute as a stand-alone software package, execute partially on the user's computer and partially on a remote computer, or execute entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (for example, by connecting through an Internet service provider using the Internet).

[0113] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system for performing the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0114] The modules involved in the embodiments of this application can be implemented in software or in hardware. Among them, the name of the module does not constitute a limitation on the unit itself in some cases.

[0115] The readable storage medium provided in this application is a computer-readable storage medium. The computer-readable storage medium stores computer-readable program instructions (i.e., computer programs) for executing the above microseismic event recognition method, which can solve the technical problems in the prior art that rely on manual experience, have a large workload, cannot achieve rapid recognition of microseisms, and result in low timeliness of microseismic early warning. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the microseismic event recognition method provided in the above embodiments, and will not be elaborated here.

[0116] The above are only some embodiments of this application, and do not limit the patent scope of this application accordingly. Any equivalent structural transformation made by using the content of the specification and drawings of this application under the technical concept of this application, or direct / indirect application in other related technical fields, is included in the patent protection scope of this application.

Claims

1. A microseismic event recognition method, characterized in that The method is applied to a graph neural network, which is composed of multiple interconnected nodes. The nodes include sensor nodes and signal nodes. The sensor nodes are composed of sensors, and the signal nodes are composed of positions generating stress signals. The method includes: Each of the sensor nodes responds to the stress signal of the signal node to obtain a response result of each of the sensor nodes. The response result at least includes waveform data and audio data; Input each of the waveform data into a preset waveform prediction model to obtain an original waveform probability vector corresponding to each of the sensor nodes. The original waveform probability vector represents the probability that the stress signal is triggered by each preset event; Input each of the audio data into a preset audio prediction model to obtain an original audio probability vector corresponding to each of the sensor nodes. The original audio probability vector represents the probability that the stress signal is triggered by each of the preset events; Determine the probability that the signal node is triggered by a preset microseismic event according to each of the original waveform probability vectors and each of the original audio probability vectors.

2. The microseismic event recognition method according to claim 1, wherein The step of determining the probability that the stress signal is triggered by a preset microseismic event according to each of the original waveform probability vectors and each of the original audio probability vectors includes: Obtain the environmental parameters of the signal node at the current moment. The environmental parameters at least include temperature parameters, humidity parameters, air pressure parameters, and the regional maximum principal stress value; Combine the original waveform probability vector and the original audio probability vector of the same sensor node according to the temperature parameter and the humidity parameter to obtain a combined probability vector corresponding to each of the sensor nodes; Aggregate each of the combined probability vectors according to the air pressure parameter and the regional maximum principal stress value to obtain a target probability vector; Determine the probability that the stress signal is triggered by a preset microseismic event based on the target probability vector.

3. The microseismic event recognition method according to claim 2, wherein The step of combining the original waveform probability vector and the original audio probability vector of the same sensor node according to the temperature parameter and the humidity parameter to obtain a combined probability vector corresponding to each of the sensor nodes includes: Determine a temperature attenuation coefficient based on the temperature parameter through a preset temperature attenuation coefficient formula; Among them, the preset temperature attenuation coefficient formula is: ； Wherein, is a signal node, t is the current moment, is the temperature attenuation coefficient of the signal node at the current moment, is the temperature parameter of the signal node at the current moment, is the first learnable attenuation coefficient, is the optimal operating temperature of the sensor, is the temperature influence range parameter; Determine a humidity attenuation coefficient based on the humidity parameter through a preset humidity attenuation coefficient formula; Among them, the preset humidity attenuation coefficient formula is: ； Wherein, is a signal node, t is the current moment, is the humidity attenuation coefficient of the signal node at the current moment, is the humidity parameter of the signal node at the current moment, is the second learnable attenuation coefficient, is a learnable parameter, is a correction coefficient, is the humidity influence effective threshold; Determine a combined probability vector corresponding to each of the sensors according to the temperature attenuation coefficient and the humidity attenuation coefficient through a preset probability formula; Among them, the preset probability formula is: ； Wherein, is a signal node, j is a sensor node, t is the current moment, is the combined posterior probability vector of the sensor node at the current moment, is the temperature attenuation coefficient of the signal node at the current moment, is the humidity attenuation coefficient of the signal node at the current moment, is the original waveform probability vector of the sensor node at the current moment, is the original audio probability vector of the sensor node at the current moment.

4. The microseismic event recognition method according to claim 2, characterized in that The step of aggregating each of the combined probability vectors according to the air pressure parameter and the regional maximum principal stress value to obtain a target probability vector includes: Determine an air pressure attenuation coefficient based on the air pressure parameter through a preset air pressure attenuation coefficient formula; Among them, the preset air pressure attenuation coefficient formula is: ； Wherein, is a signal node, t is the current moment, is the air pressure attenuation coefficient of the signal node at the current moment, is the air pressure parameter of the signal node at the current moment, is the standard atmospheric pressure, is the sensitivity of the sensor node under the standard atmospheric pressure, is the pressure influence coefficient; Determine the maximum principal stress value of each of the sensor nodes; Determine the stress propagation attenuation coefficient corresponding to each sensor node based on the maximum principal stress value of the region and the maximum principal stress values of each sensor through a preset stress propagation attenuation coefficient formula; Among them, the preset stress propagation attenuation coefficient formula is: ； In the formula, is the signal node, j is the sensor node, t is the current moment, is the stress propagation attenuation coefficient between the sensor node and the signal node at the current moment, is the maximum principal stress value of the region, is the maximum principal stress value of the sensor, is the included angle degree between the maximum principal stress direction of the region of the signal node and the connection line between the signal node and the sensor node, is the material attenuation coefficient of stress propagation, is the distance between the signal node and the sensor node; Determine the attention weight between each sensor node and the signal node according to the air pressure attenuation coefficient and each stress propagation attenuation coefficient; Aggregate the combined posterior probability vectors according to each attention weight to obtain a target probability vector.

5. The microseismic event recognition method according to claim 4, wherein The step of determining the attention weight between each sensor node and the signal node according to the air pressure attenuation coefficient and each stress propagation attenuation coefficient includes: Determine the attention coefficient between each sensor node and the signal node based on the air pressure attenuation coefficient and each stress propagation attenuation coefficient through a preset attention coefficient formula; Among them, the preset attention coefficient formula is: ； Wherein, is a signal node, j is a sensor node, t is the current moment, is the attention coefficient between the sensor node and the signal node at the current moment, is the air pressure attenuation coefficient of the signal node at the current moment, is the stress propagation attenuation coefficient between the sensor node and the signal node at the current moment, is a non-linear activation function, is a learnable parameter vector, T is a matrix transpose operation, is a learnable weight matrix of a shared linear transformation, is an environmental feature vector of the environment where the signal node is located, is an environmental feature vector of the environment where the sensor node is located; Normalize each attention coefficient through a preset attention weight formula to obtain the attention weight corresponding to each sensor node; Among them, the preset attention weight formula is: ； In the formula, is a signal node, j is a sensor node, is any sensor node that receives a stress signal, t is the current moment, is the attention weight between the sensor node and the signal node at the current moment, is the attention coefficient between the sensor node and the signal node at the current moment, is the environmental feature vector of the environment where the signal node is located, is the environmental feature vector of the environment where the sensor node is located, is the learnable weight matrix of the shared linear transformation, is the set of all sensor nodes that receive stress signals, is the air pressure attenuation coefficient of the signal node at the current moment, is the stress propagation attenuation coefficient between the sensor node and the signal node at the current moment, is the attention coefficient between any sensor node that receives a stress signal and the signal node at the current moment, is the propagation attenuation coefficient between any sensor node that receives a stress signal and the signal node at the current moment.

6. The microseismic event recognition method according to claim 5, characterized in that The step of aggregating the combined posterior probability vectors according to each attention weight to obtain a target probability vector includes: Aggregate the combined posterior probability vectors according to each attention weight through a first preset output formula to obtain a target probability vector; Among them, the first preset output formula is: ； Wherein, is a signal node, j is a sensor node, t is the current moment, is the target probability vector of the signal node at the current moment, is an activation function, is the attention weight between the sensor node and the signal node at the current moment, is a learnable weight matrix of the shared linear transformation, is the combined posterior probability vector of the sensor node at the current moment.

7. The microseismic event recognition method according to claim 5, wherein, The step of aggregating the combined posterior probability vectors according to each attention weight to obtain a target probability vector further includes: Aggregate the combined posterior probability vectors according to each attention weight through a second preset output formula to obtain a target probability vector; Among them, the second preset output formula is: ； Wherein, is a signal node, j is a sensor node, t is the current moment, is the target probability vector of the signal node at the current moment, is an activation function, is the attention weight between the sensor node and the signal node at the current moment, is a learnable weight matrix of the shared linear transformation, is the combined posterior probability vector of the sensor node at the current moment, Z is the number of attention heads.

8. A microseismic event recognition device, characterized in that, The device includes: A data acquisition module, configured to obtain the response results of each sensor node by responding to the stress signal of the signal node, and the response results at least include waveform data and audio data; A probability prediction module, configured to input each waveform data into a preset waveform prediction model to obtain an original waveform probability vector corresponding to each sensor node, and the original waveform probability vector represents the probability that the stress signal is triggered by each preset event; The probability prediction module is further configured to input each audio data into a preset audio prediction model to obtain an original audio probability vector corresponding to each sensor node, and the original audio probability vector represents the probability that the stress signal is triggered by each preset event; An event recognition module, configured to determine the probability that the stress signal is triggered by a preset microseismic event according to each original waveform probability vector and each original audio probability vector.

9. A microseismic event recognition device, characterized in that, The microseismic event recognition device includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, and the computer program is configured to implement the steps of the microseismic event recognition method according to any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the microseismic event recognition method according to any one of claims 1 to 7 are implemented.

Images