Three-dimensional calculation analysis method for excavation with or without support

Abstract

The present invention provides a three-dimensional calculation and analysis method for excavation with or without support, which relates to the field of underground engineering and includes the following steps: S1, data acquisition and preprocessing: obtaining geological information of on-site monitoring equipment and construction areas; collecting multi-source data in the current construction process based on the on-site monitoring equipment of the underground powerhouse cavern group and the geological information of the construction area, including construction stage data, surrounding rock monitoring data and reinforcement measure parameters; S2, data cleaning and feature extraction: performing preprocessing operations on the collected multi-source data, and extracting feature vectors based on the preprocessed multi-source data. First, it can accurately analyze the deformation law of the surrounding rock of the high-stress underground powerhouse cavern group under the excavation with or without support; second, it significantly improves the accuracy and reliability of stress analysis; third, it accurately determines the distribution law of the damage zone; fourth, it realizes the effective monitoring and analysis of the evolution law of microseismic source parameters.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of underground engineering, in particular to a three-dimensional calculation and analysis method for excavation with or without support. BACKGROUND

[0002] With the increasing demand for energy, the demand for underground powerhouse construction in the fields of hydropower, nuclear power and the like is increasing. However, under high ground stress conditions, the surrounding rock of the underground powerhouse cavern group presents a series of unstable mechanical behaviors after excavation.

[0003] At present, although the deformation, stress and failure zone distribution of the high ground stress underground powerhouse cavern group have been understood to some extent, i.e., they conform to the specific high ground stress underground powerhouse excavation deformation and failure zone distribution law, in actual engineering, when the deformation and stress of the three major caverns are in a seemingly stable and controllable state, it brings great difficulties to the subsequent monitoring and analysis work. This seemingly stable state often hides the potential instability factors and their development trend in the surrounding rock, making it extremely complex to calculate and analyze the deformation, stress and failure zone distribution law of the surrounding rock under the reinforcement state and the microseismic source parameter evolution law. The traditional monitoring and analysis method is difficult to accurately capture these key information, so as to provide accurate and reliable basis for formulating scientific and reasonable reinforcement measures, and it is difficult to meet the long-term stable and safe operation demand of the high ground stress underground powerhouse cavern group. SUMMARY

[0004] In view of the above-mentioned deficiencies of the prior art, the technical problem to be solved by the present application is to provide a three-dimensional calculation and analysis method for excavation with or without support, which can accurately capture the deformation, stress and failure zone distribution of the three major caverns, and analyze the high ground stress underground powerhouse excavation deformation and failure zone distribution law, so as to provide the basis for corresponding reinforcement measures to ensure long-term safety.

[0005] To solve the above technical problems, the technical scheme adopted by the present application is as follows: the present application provides a three-dimensional calculation and analysis method for excavation with or without support, comprising the following steps:

[0006] S1, data acquisition and preprocessing: obtaining the geological information of the field monitoring equipment and the construction area;

[0007] According to the field monitoring equipment and the geological information of the construction area of the underground powerhouse cavern group, the multi-source data in the current construction process are collected, including construction stage data , surrounding rock monitoring data and reinforcement measure parameters ;

[0008] S2, data cleaning and feature extraction: performing preprocessing operation on the collected multi-source data, and extracting feature vectors according to the preprocessed multi-source data;

[0009] S3, model prediction and analysis: the preprocessed multi-source data is integrated to form a complete input sample , and input into the model , analysis is performed to obtain the prediction result.

[0010] In the preferred scheme, the model is a pre-trained integrated model, and the integrated model integrates long short-term memory network (LSTM), convolutional neural network (CNN), support vector machine (SVM), and Transformer, and further includes the following steps:

[0011] S4, construction monitoring and early warning: the prediction result is integrated and feature encoded, and the prediction result is input into the Transformer. The input layer of the Transformer encodes the data in the prediction result to capture the sequence order, and then the data enters multiple encoder layers to capture the association between different prediction results. The output is a reweighted integrated feature representation, which is then input into a feedforward neural network composed of two linear transformation layers and a ReLU activation function for nonlinear transformation and feature extraction. After processing by multiple encoder layers, the data is mapped to a comprehensive early warning indicator by a linear projection layer in the output layer , the weights of which are optimized by the backpropagation algorithm during training to determine the best mapping relationship between the input prediction results and the comprehensive early warning indicator , and based on the comprehensive early warning indicator , the warning level is determined by comparing it with a threshold value.

[0012] Each encoder layer includes a multi-head self-attention mechanism and a feedforward neural network.

[0013] In the preferred scheme, in step S2, the preprocessing operation includes cleaning, removing outliers and error data, and is used to integrate and standardize the multi-source data to have a unified format and dimension.

[0014] The feature vectors extracted from the surrounding rock monitoring data include the surrounding rock deformation vector , the sorted surrounding rock pressure data vector , the stress distribution data vector , and the microseismic activity data vector .

[0015] The surrounding rock deformation vector , the sorted surrounding rock pressure data vector , and the stress distribution data vector are arranged in a certain spatial order, and the microseismic activity data vector is arranged in a certain time order.

[0016] Construction phase data Including the current excavation depth, the position of the working face, the completed support steps, and converted into the corresponding feature vector;

[0017] Reinforcement measures parameters Including the length, spacing, grouting amount of anchor rods, and prestress parameters of anchor cables, and converted into the corresponding feature vector.

[0018] In the preferred scheme, in step S3, the model For the input sample The specific analysis steps are as follows:

[0019] S31, the LSTM layer performs convolution operation on the microseismic activity data vector Perform analysis and processing;

[0020] According to the occurrence order and time interval of the microseismic event, the model The pre-set 1-3 layers of LSTM layer extract the evolution trend features of the microseismic activity and output the feature vector of the evolution stage of the microseismic source parameters ;

[0021] Wherein, the LSTM layer contains 64-256 hidden units per layer;

[0022] S32, the CNN performs convolution operation on the spatial distribution data of the surrounding rock deformation and damage zone, extracts spatial features through different size convolution kernels, and outputs the spatial feature vector of the surrounding rock deformation and stress distribution And ;

[0023] Wherein, in order to identify the shape and position of the local concentrated area of the surrounding rock deformation and the stress concentration area, the surrounding rock deformation vector , the sorted surrounding rock pressure data vector And the stress distribution data vector Through 2-3 convolution layers for spatial feature extraction, the convolution kernel size is 3x3 to 5x5, the step is 1 or 2, and each convolution layer is followed by a max pooling layer;

[0024] S33, using feature engineering to extract key features from construction phase data And reinforcement measures parameters Into a vector , and as the input data of the SVM model, the feature vectors output by LSTM and CNN after their respective fully connected layers , And are calculated by weighted summation, and the feature vectors are fused The calculation formula is as follows:

[0025] ;

[0026] wherein the weights are respectively , , ;

[0027] The fusion feature vector is input into the SVM model, and a prediction is performed according to the fusion feature vector , and a SoftMax function is used to output the deformation prediction value , the stress prediction value , the damage zone distribution prediction probability , and the microseismic source parameter evolution stage prediction probability .

[0028] In the preferred scheme, in step S4, the standard surrounding rock data in the database is:

[0029] ;

[0030] wherein represents the deformation dimension of different monitoring positions, respectively corresponds to four deformation levels, and the standard surrounding rock data corresponds to the vector representation of the typical deformation under different deformation levels;

[0031] The Euclidean distance between the deformation prediction value of the current surrounding rock and each standard surrounding rock data is calculated, and the Euclidean distance calculation formula is:

[0032] ;

[0033] The deformation level is determined by comparing the size of the Euclidean distance :

[0034] If , it is determined to be first-level slight deformation, and the daily deformation amount is represented as ;

[0035] If , it is determined to be second-level moderate deformation, and the daily deformation amount is represented as ;

[0036] If , it is determined to be third-level intense deformation, and the daily deformation amount is represented as ;

[0037] If , it is determined to be fourth-level extremely intense deformation, and the daily deformation amount is represented as .

[0038] In the preferred solution, according to the Euclidean distance Determine the surrounding rock deformation level and combine it with the model Output of predicted deformation value of surrounding rock , stress prediction value , predicted probability of damage zone distribution and the predicted probability of the evolution stage of microseismic source parameters Conduct comprehensive early warning;

[0039] Defining comprehensive early warning indicators , which is a weighted combination of the following:

[0040] ;

[0041] in, 、 、 、 These are the weight coefficients of the four factors, namely, deformation level, stress, probability of distribution of damage zone, and evolution stage of microseismic source parameters, in the calculation of comprehensive early warning indicators;

[0042] Determine according to the actual situation and importance of the project 、 、 、 , and satisfies ;

[0043] get: ;

[0044] Among them, obtain the surrounding rock deformation grade , according to the surrounding rock deformation grade Substitute the value into the deformation level function Get the corresponding function value and use the stress prediction value and stress threshold 、 Compare and substitute the stress function Get the corresponding function value and predict the probability of damage zone distribution and the predicted probability of the evolution stage of microseismic source parameters , and then substituted into the probability function of the damage area distribution after comparing with their respective thresholds and the probability function of the evolution stage of microseismic source parameters Get the corresponding function value and multiply the corresponding function value by the corresponding weight coefficient 、 、 、 Then add them together to get the comprehensive early warning index .

[0045] In a preferred scheme, the comprehensive early warning index is calculated as follows:

[0046] Let the surrounding rock deformation level determined by the Euclidean distance be ;

[0047] wherein the surrounding rock deformation level takes the values 1, 2, 3, and 4, corresponding to first-level slight deformation, second-level moderate deformation, third-level intense deformation, and fourth-level extremely intense deformation, respectively, for reflecting the severity of surrounding rock deformation;

[0048] The deformation level function is determined according to the surrounding rock deformation level and is assigned different basic scores, so that the surrounding rock deformation level is quantified, i.e., according to the input surrounding rock deformation level value, a corresponding fixed score is directly output, and the formula is as follows:

[0049] ;

[0050] Let and be stress threshold values, and the function value of the stress function is determined according to the relationship between the stress prediction value and the threshold value:

[0051] ;

[0052] Let and be damage zone distribution probability threshold values, and , , then the damage zone distribution probability function is expressed as follows:

[0053] ;

[0054] Let and , be microseismic source parameter evolution stage probability threshold values, and , , then the microseismic source parameter evolution stage probability function is expressed as follows:

[0055] ;

[0056] Let , and be early warning threshold values, and the comprehensive early warning index Compare with the set warning threshold to determine the warning level;

[0057] Among them, , , , then the formula is expressed as:

[0058] ;

[0059] Displayed through a visual interface, according to The value of is represented by different colors in the animation simulation of the surrounding rock deformation. The color cloud map of stress distribution is rendered with the value of color, and the predicted probability of damage zone distribution is represented by different transparency and color. The probability of different damage zone categories in the stratum, as well as the predicted probability according to the evolution stage of microseismic source parameters High-risk microseismic hotspots are highlighted in vibrant colors.

[0060] In the preferred solution, S5, monitoring frequency adjustment: according to the deformation prediction value of the surrounding rock , stress prediction value , predicted probability of damage zone distribution and the predicted probability of the evolution stage of microseismic source parameters and comprehensive early warning indicators Determine the current status of the underground powerhouse cavern complex. The specific steps are as follows:

[0061] Assume the deformation speed is , by calculating the two adjacent monitoring moments, let the time interval be The deformation difference , the deformation speed formula is as follows:

[0062] ;

[0063] when mm / day, determine the monitoring frequency , and is expressed by the following formula:

[0064] ;

[0065] in, and To further divide the speed interval threshold, that is mm / day, mm / day;

[0066] According to the determined monitoring frequency , adjust the data collection cycle of the monitoring equipment to ensure that sufficient surrounding rock status information is obtained in a timely manner.

[0067] In a preferred solution, S6, monitoring measurement data stage analysis and construction evaluation: through the week analysis and month analysis, determine the influence law between construction stage data , surrounding rock monitoring data and reinforcement measures parameters , the specific steps are as follows:

[0068] S61, data collection and arrangement: set the surrounding rock deformation data sequence obtained by each monitoring as , the surrounding rock pressure data sequence as , the steel frame stress data sequence as , and the concrete stress data sequence as ;

[0069] Wherein, represents the monitoring time, is the total number of monitoring times;

[0070] S62, stage analysis according to week and month

[0071] Week analysis:

[0072] The data obtained every week is analyzed respectively, that is, the average deformation of surrounding rock deformation data every week is calculated:

[0073] ;

[0074] Wherein, is the monitoring number of the th week, represents the monitoring time set corresponding to the th week;

[0075] Similarly, the average surrounding rock pressure every week , the average steel frame stress every week , the average concrete stress every week ;

[0076] Analyze the change trend of each week data, calculate the difference value of adjacent two weeks average deformation Observe the increase and decrease of deformation and the change of each stress data to determine the change trend of surrounding rock state;

[0077] Month analysis:

[0078] Calculate the average deformation every month:

[0079] ;

[0080] Wherein, is the monitoring number of the th month, represents the monitoring time set The monitoring time set corresponding to each month, and the monthly average surrounding rock pressure can be calculated in the same way The monthly average steel frame stress The monthly average concrete stress

[0081] Through the change trend of the weekly data and the monthly monitoring measurement data, the change law of each monitoring measurement data over time is determined .

[0082] In the preferred scheme, in step S6, the construction condition is evaluated according to the change law The specific steps are as follows:

[0083] Let be the surrounding rock deformation increment threshold value, be the surrounding rock pressure safety threshold interval, be the steel frame stress safety threshold interval, be the concrete stress safety threshold interval;

[0084] If there is a week k, then the week risk condition determination formula is as follows:

[0085] ;

[0086] For two consecutive months and , the month risk condition determination formula is as follows:

[0087] ;

[0088] If the construction condition has a risk in a week or during two consecutive months, in-depth assessment is performed and corresponding measures are taken.

[0089] The application provides a three-dimensional calculation and analysis method for excavation with or without support, and has the following beneficial effects through cooperation between the structures:

[0090] First, the surrounding rock deformation law of a high-stress underground powerhouse cavern group in an excavation state with or without support can be accurately analyzed, not only the current deformation amount can be accurately mastered, but also the development trend of the deformation can be deduced according to the data change curve and related calculation, thereby providing a key basis for taking targeted deformation control measures in advance;

[0091] Second, the accuracy and reliability of stress analysis are significantly improved, the stress data such as surrounding rock pressure, steel frame stress, concrete stress and the like are comprehensively considered by means of a three-dimensional calculation model, the stress distribution state is accurately calculated, and the stress change condition can be tracked in real time, when facing a complex high-stress environment, whether the stress exceeds the safety threshold value can be accurately judged, and a potential stress concentration risk area can be found in time;​

[0092] Thirdly, the distribution law of the damage zone is accurately determined, based on comprehensive monitoring of the surrounding rock state and in-depth analysis of the three-dimensional model, combined with the damage zone distribution prediction probability and other indicators, the possible position, range and development trend of the damage zone can be accurately depicted;

[0093] Fourthly, the evolution law of the microseismic source parameters is effectively monitored and analyzed, through the use of microseismic source parameter evolution stage prediction probability and other data, combined with professional analysis methods and models, the law and trend of microseismic activity can be clearly grasped. BRIEF DESCRIPTION OF DRAWINGS

[0094] The present application will be further described below in combination with the drawings and embodiments:

[0095] Figure 1 is the main structure diagram of the process of the present application;

[0096] Figure 2 is the system device structure diagram in embodiment 1 of the present application;

[0097] Figure 3 is the electronic equipment schematic diagram in embodiment 2 of the present application. DETAILED DESCRIPTION

[0098] In order to better understand the purpose, structure and function of the present application, the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0099] Embodiment 1

[0100] As shown in Figure 1 ,a three-dimensional calculation and analysis method for excavation with or without support, comprising the following steps: 2 S1, data acquisition and preprocessing: obtaining the geological information of the field monitoring equipment and the construction area;

[0101] According to the geological information of the field monitoring equipment and the construction area of the underground powerhouse cavern group, multi-source data in the current construction process are collected, including construction stage data

[0102] , surrounding rock monitoring data and reinforcement measure parameters ;

[0103] S2, data cleaning and feature extraction: performing preprocessing operation on the collected multi-source data, and extracting feature vectors according to the preprocessed multi-source data;

[0104] The preprocessing operation includes clearing, removing outliers and error data, and performing integration and standardization processing on multi-source data to make the data have unified format and dimension;

[0105] Specifically, the original data in the collected multi-source data is cleaned, and the obviously abnormal or error data points are removed. For the surrounding rock deformation data, if the deformation value of a certain monitoring point suddenly deviates seriously from the surrounding monitoring points and historical data, the data point may be caused by sensor failure or other interference factors, and is removed or corrected.

[0106] Meanwhile, meaningful features are extracted from the original data in the multi-source data. For the surrounding rock deformation data, in addition to the direct deformation value, the deformation rate can also be calculated, such as the deformation difference between adjacent two monitoring times divided by the time interval, and the deformation rate is recorded as , and the deformation acceleration is calculated by further differentiating the deformation rate , and is recorded as Feature; for the microseismic activity data, the magnitude , frequency , and energy features of the microseismic event are extracted; for the surrounding rock pressure data, the pressure gradient feature is calculated.

[0107] The feature vectors extracted from the surrounding rock monitoring data include the surrounding rock deformation vector , the sorted surrounding rock pressure data vector , the stress distribution data vector , and the microseismic activity data vector ;

[0108] The surrounding rock deformation vector , the sorted surrounding rock pressure data vector , and the stress distribution data vector are arranged in a certain spatial order, and the microseismic activity data vector is arranged in a certain time order.

[0109] The construction phase data include the current excavation depth, the position of the working face, and the completed support steps, and are converted into corresponding feature vectors;

[0110] The reinforcement measure parameters include the length and spacing of the anchor rod, the grouting amount, and the prestress parameters of the anchor cable, and are converted into corresponding feature vectors.

[0111] S3, model prediction and analysis: the multi-source data after preprocessing is integrated to form a complete input sample , and is input into the model analysis is performed;

[0112] wherein the model is a pre-trained ensemble model integrating long short-term memory network (LSTM), convolutional neural network (CNN) and support vector machine (SVM), which is used to input the sample The specific analysis steps are as follows:

[0113] The pre-processed construction phase data , surrounding rock monitoring data and reinforcement measures parameters are integrated to form a complete input sample ;

[0114] The constructed input sample is sent to the trained model combining long short-term memory network (LSTM), convolutional neural network (CNN), support vector machine (SVM) and Transformer ;

[0115] In this embodiment, the model aims to comprehensively and accurately predict the surrounding rock state of underground powerhouse cavern group by integrating the advantages of LSTM, CNN and SVM three models. First, the LSTM layer focuses on processing the time series features of microseismic activity data, and mining the evolution trend of microseismic events over time. At the same time, CNN performs convolution operation on the spatial distribution data of surrounding rock deformation and stress, and extracts key spatial features. Then, key features are extracted from construction phase data and reinforcement measures parameters using feature engineering, and weighted fusion is performed with the feature vectors output by LSTM and CNN. Finally, the fused feature vectors are input into the SVM model for final prediction, and key information such as surrounding rock deformation, stress, damage zone distribution and microseismic source parameter evolution stage is output, providing scientific basis and decision support for underground engineering construction.

[0116] The LSTM layer performs analysis and processing on the microseismic activity data vector ;

[0117] According to the occurrence order and time interval of microseismic events, the model pre-sets 1-3 layers of LSTM layers to extract the evolution trend features of microseismic activity and output the feature vector of microseismic source parameter evolution stage ;

[0118] wherein each layer of the LSTM layer contains 64-256 hidden units;

[0119] Specifically, according to the complexity of engineering data and the training effect of the model, after multiple tests and optimizations, it is found that 64-256 hidden units can capture the time series characteristics of microseismic activity while avoiding overfitting and underfitting problems, thereby better extracting the evolution trend characteristics of microseismic activity. Further, for the case of small data volume and relatively simple microseismic activity, 64 hidden units are sufficient to learn key information; when the data volume is large and the microseismic activity is complex, increasing the number of hidden units to 256 helps improve the expression ability and accuracy of the model.

[0120] The CNN performs convolution operations on the spatial distribution data of the surrounding rock deformation and the damage zone, extracts spatial features through different sizes of convolution kernels, and outputs the spatial feature vectors of the surrounding rock deformation and stress distribution and ;

[0121] Among them, the shape and position of the local concentrated area of the surrounding rock deformation and the stress concentration area are identified by the surrounding rock deformation vector and the sorted surrounding rock pressure data vector The spatial features are extracted through 2-3 convolution layers, the convolution kernel size is 3x3 to 5x5, the step size is 1 or 2, and each convolution layer is followed by a max pooling layer;

[0122] Specifically, 2-3 convolution layers are used to gradually extract the spatial features of surrounding rock deformation and stress distribution, from local features at a shallow level to more abstract and global features at a deep level. Smaller convolution kernels (3x3 and 5x5) can focus on local area features such as subtle changes in surrounding rock deformation and the starting point of stress concentration, while larger step sizes can expand the receptive field while appropriately reducing computational complexity, helping to capture larger spatial relationships. The max pooling layer after each convolution layer is used to reduce the resolution of the feature map, further extract main features, and reduce the number of model parameters and computational complexity, allowing the model to converge faster and generalize better to new data.

[0123] Key features are extracted from construction phase data and reinforcement measures parameters using feature engineering and composed into a vector , which is used as input data for the SVM model. The feature vectors , and output by LSTM and CNN after their respective fully connected layers are calculated using weighted summation, with the following steps:

[0124] Let the weights be , , ;

[0125] Specifically, the determination of the weights , , uses a grid search method based on cross-validation, which tries different combinations of weights on the training set to find the weight values that make the model have the best prediction performance on the validation set. For example, the initial weight range is set to 0 to 1 with a step of 0.1, and through the evaluation of different combinations, the current weight values are finally determined, so that the features extracted by LSTM and CNN can reasonably contribute information in the decision-making process of the SVM model according to their importance to the prediction of surrounding rock state, thereby improving the prediction accuracy of the overall model.

[0126] ;

[0127] The fusion feature vector is obtained, the fusion feature vector is input into the SVM model, and the final prediction is performed according to the fusion feature vector , and the SoftMax function is used to output the deformation prediction value , stress prediction value , damage zone distribution prediction probability and microseismic source parameter evolution stage prediction probability .

[0128] S4, construction monitoring and early warning: the deformation prediction value , stress prediction value , damage zone distribution prediction probability and microseismic source parameter evolution stage prediction probability output by the model are integrated and encoded as input to the Transformer model;

[0129] In the input layer of the Transformer, the input data is first positionally encoded, which adds position information to each input feature through a specific function, so that the model can capture the sequence order of the data, which is crucial for processing time series related prediction results (such as time series prediction values of deformation and stress) and spatial information of different monitoring locations (such as damage zone distribution probability in different areas);

[0130] Subsequently, the data enters the core architecture composed of multiple encoder layers and decoder layers. In each encoder layer, there are two key parts: multi-head self-attention mechanism (Multi-Head Self-Attention) and feed-forward neural network (Feed-Forward Network);

[0131] The multi-head self-attention mechanism allows the model to simultaneously focus on different parts of the input data in different representation subspaces, enabling comprehensive capture of complex dependencies and interactions between different prediction results. For example, it can simultaneously focus on the correlation between the trend of deformation values and stress concentration areas, as well as the potential link between microseismic activity and the development of damage zones. Through the calculation of the multi-head attention mechanism, a set of reweighted and integrated feature representations is output.

[0132] Next, this set of feature representations enters the feedforward neural network, which is composed of two linear transformation layers and a ReLU activation function. The purpose of this network is to further perform nonlinear transformation and feature extraction on the features processed by the attention mechanism, enhancing the model's expressive power and enabling it to learn more complex feature combinations and patterns.

[0133] After sequential processing by multiple encoder layers, the data undergoes continuous feature refinement and information fusion. Finally, in the output layer, a linear projection layer maps the last feature representation to a single comprehensive early warning indicator . The weights of this linear projection layer are continuously optimized during the model training process using the backpropagation algorithm, with the goal of finding the best mapping relationship between the input prediction results and the comprehensive early warning indicator , so that the comprehensive early warning indicator can accurately reflect the overall safety state of the surrounding rock, allowing for comparison with a pre-set threshold to achieve effective engineering risk early warning and reasonable construction decision-making, providing strong support for the construction safety of underground powerhouse cavern groups.

[0134] In the preferred embodiment, in step S4, the deformation grade of the surrounding rock and the early warning are determined based on the Euclidean distance formula, with the specific steps as follows:

[0135] Let the standard surrounding rock data in the database be:

[0136] ;

[0137] where represents the deformation dimension at different monitoring locations, such as the deformation values at different measuring points on the vault, side wall, etc., corresponding to four deformation grades, respectively, and the standard surrounding rock data correspond to the vector representation of the typical deformation under different deformation grades.

[0138] Calculate the Euclidean distance between the deformation prediction value of the current surrounding rock and each standard surrounding rock data , with the Euclidean distance calculation formula being:​

[0139] ;

[0140] By comparing the size of the deformation level:

[0141] If , it is determined to be a first-level slight deformation, and the daily deformation amount is represented as ;

[0142] wherein the average value of the deformation amount of each monitoring position is taken to measure the daily deformation amount, denoted as , then

[0143] and the duration is long and the duration is set as , , is the threshold value of the slight deformation duration determined according to engineering experience, accompanied by local spalling phenomenon, judged by on-site monitoring equipment, and the local spalling judgment index is set as , satisfies , is the threshold value for determining local spalling.

[0144] If , it is determined to be a second-level moderate deformation, and the daily deformation amount is represented as , and initial support cracking occurs, and the initial support cracking judgment index is set as , satisfies , is the threshold value for determining initial support cracking, spalling, also satisfies and steel frame distortion phenomenon, and the steel frame distortion judgment index is set as , satisfies , is the threshold value for determining steel frame distortion.

[0145] If , it is determined to be a third-level strong deformation, and the daily deformation amount is represented as , and the support is severely cracked, and the support cracking judgment index is set as , satisfies , is the threshold value for determining support cracking, steel frame distortion and fracture, and the steel frame distortion and fracture judgment index is set as , satisfies , is the threshold value for determining steel frame distortion and fracture, and is accompanied by obvious new line damage, and the obvious new line damage judgment index is set as , satisfies , is the threshold value for determining obvious new line damage.​

[0146] like , it is determined to be extremely severe deformation of level 4, and the daily deformation is expressed as , and the supporting structure is seriously damaged, the judgment index of serious damage of the supporting structure is ,satisfy , To determine the threshold of serious damage to the support structure, the steel frame is twisted and broken, and meets the , the clearing line damage is obvious, and the clearing line damage obvious judgment index is set as ,satisfy , In order to determine the threshold of obvious line damage, even 2-inch cracks may appear locally, the crack size judgment index is set as ,satisfy inch corresponding length threshold .

[0147] In the preferred solution, when the Euclidean distance After determining the deformation level of the surrounding rock, combined with the model Output of predicted deformation value of surrounding rock , stress prediction value , predicted probability of damage zone distribution and the predicted probability of the evolution stage of microseismic source parameters Conduct comprehensive early warning.

[0148] Defining comprehensive early warning indicators , which is a weighted combination of the following:

[0149] ;

[0150] in, 、 、 、 The weight coefficients of the four factors, namely deformation level, stress, probability of distribution of damage zone, and evolution stage of microseismic source parameters, in the calculation of comprehensive early warning indicators reflect the relative importance of each factor to the overall risk assessment. The weight coefficient determines the relative contribution of each factor in the calculation of comprehensive early warning indicators.

[0151] Determine according to the actual situation and importance of the project 、 、 、 , and satisfies ;

[0152] get: ;

[0153] Among them, first obtain the surrounding rock deformation level , according to the surrounding rock deformation grade Substitute the value into the deformation level function Get the corresponding function value, then, the stress prediction value and stress threshold 、 Compare and substitute the stress function Get the corresponding function value and predict the probability of damage zone distribution and the predicted probability of the evolution stage of microseismic source parameters , and then substituted into the probability function of the damage area distribution after comparing with their respective thresholds and the probability function of the evolution stage of microseismic source parameters Get the corresponding function value, and finally multiply the corresponding function value by the corresponding weight coefficient 、 、 、 Then add them together to get the comprehensive early warning index .

[0154] Comprehensive early warning indicators The specific calculation steps are as follows:

[0155] Assume that the surrounding rock deformation grade determined by Euclidean distance is ;

[0156] Among them, the surrounding rock deformation grade The values ​​are 1, 2, 3, and 4, corresponding to the first-level slight deformation, the second-level moderate deformation, the third-level strong deformation, and the fourth-level extremely strong deformation, respectively, and are used to reflect the severity of the surrounding rock deformation;

[0157] Deformation level function According to the surrounding rock deformation grade Assign different basic scores and adjust the surrounding rock deformation level Factor quantification, the higher the level, the greater the score, to reflect the importance of deformation level in comprehensive early warning, that is, according to the input surrounding rock deformation level value, directly output the corresponding fixed score, the formula is as follows:

[0158] ;

[0159] set up and is the stress threshold, according to the stress prediction value Determine the stress function in relation to the threshold Function value of :

[0160] ;

[0161] When the stress prediction value When the function value is 1, it indicates that the stress is at a low level, and the contribution to the comprehensive early warning index is relatively small;

[0162] When the stress prediction value is between and , the function value increases linearly, reflecting the gradually increasing influence of stress on risk in this interval;

[0163] When the stress prediction value is greater than , the function value increases with the growth of stress, and the growth rate is the ratio of to

[0164] , reflecting the rapid increase of risk when the stress exceeds the higher threshold; Let and be the damage zone distribution probability thresholds, and , then the damage zone distribution probability function

[0165] is expressed as follows:

[0166] According to the relationship between the damage zone distribution prediction probability and the thresholds , , the contribution of the function value to the comprehensive early warning index is quantified by a piecewise function, so that the function value can more reasonably reflect the influence degree of the damage zone probability.

[0167] Let and , be the microseismic source parameter evolution stage probability thresholds, and , then the microseismic source parameter evolution stage probability function is expressed as follows:

[0168] ;

[0169] This function determines the function value in a piecewise form according to the comparison between the prediction probability of the microseismic source parameter evolution stage and the thresholds , , so as to include the microseismic factor into the comprehensive early warning system and reflect the relationship between the influence degree of the microseismic factor on the early warning and the probability.

[0170] Let , and be the early warning thresholds, and compare the calculated comprehensive early warning index with the set early warning thresholds to determine the early warning level;

[0171] wherein, let , , The formula is expressed as:

[0172] ;

[0173] Specifically, when , that is, , the safety state can maintain normal construction and monitoring frequency without special measures;

[0174] When , that is, , the low-risk state, although the risk has increased, but still within the acceptable range, can maintain normal construction and monitoring frequency, but may need to take some simple special measures, such as strengthening the frequency of data monitoring, etc.

[0175] When , that is, , the medium-risk state, a warning signal is issued to prompt the construction personnel to strengthen the site inspection, check the integrity and stability of the supporting structure, and prepare possible emergency measures, such as preparing a certain amount of supporting materials and emergency equipment, etc.

[0176] When , that is, , the high-risk state, immediately issue an alarm, require suspension of construction, organize personnel to evacuate the dangerous area, take emergency rescue measures, such as rapid reinforcement of surrounding rock, start spraying concrete within hours, install temporary support, etc., while the project is fully evaluated and redesigned, a safer and more reliable construction plan is developed, including adjusting the excavation method, support parameters, etc., and before implementing the new plan, the surrounding rock state is continuously monitored.

[0177] Through the visual interface display, according to the value of , the animation simulation of the surrounding rock deformation size is represented by different colors, according to the value of , the color rendering of stress distribution color cloud is carried out, the distribution prediction probability of the damage zone is represented by different transparency and color probability of different damage zone categories, and according to the microseismic source parameter evolution stage prediction probability The high-risk microseismic hotspot area is highlighted in bright color.

[0178] For the surrounding rock deformation, according to the vector Using the values ​​in the image, the system uses preset color mapping rules to assign different deformations to different colors, which are then presented through animation simulations. As time progresses and construction progresses, the animation dynamically displays the spatial distribution of surrounding rock deformation and its changing trends, allowing engineers to clearly identify areas of increase and decrease in deformation, as well as the rate and direction of deformation, and thus gain an intuitive understanding of the surrounding rock's stability.

[0179] The stress distribution is presented in the form of a color cloud map, based on the stress prediction value Using color rendering technology, low-stress areas are represented by cool blue, high-stress areas by warm red, and intermediate stress values ​​correspond to varying degrees of transitional colors. This allows engineers to understand the stress distribution within the surrounding rock at a glance, similar to how they would understand air pressure or temperature distribution on a weather cloud map. This allows them to quickly locate areas of stress concentration and take timely measures, such as strengthening support structures or adjusting construction techniques, to prevent damage to the surrounding rock caused by stress concentration.

[0180] The probability map of the damage area distribution is obtained by predicting the probability of the damage area distribution. Using different levels of transparency and color to distinguish the probability of damage zones, low-probability damage zones are displayed in more transparent colors. As the probability increases, the color's transparency decreases and its saturation increases. This allows engineers to intuitively identify areas with a higher likelihood of damage, as well as the relative likelihood of damage in different areas. This provides a strong basis for proactively implementing reinforcement measures or adjusting the construction schedule, ensuring the safe progress of the project.

[0181] Microseismic activity hotspot map predicts the probability of microseismic source parameter evolution stage The map is created by highlighting areas with frequent and high-probability microseismic activity with bright colors, such as red or yellow, to create easily identifiable hotspots, while low-probability microseismic areas are represented by darker or less conspicuous colors. This allows engineering personnel to quickly determine the distribution and intensity trends of dangerous microseismic activity, rationally plan construction personnel's work areas, avoid dangerous operations in high-risk microseismic areas, and strengthen monitoring and early warning efforts in these areas, effectively preventing safety accidents such as surrounding rock instability caused by microseisms, and ensuring the safety and smooth construction of the underground powerhouse cavern complex.

[0182] S5. Monitoring frequency adjustment

[0183] Assume the deformation speed is , by calculating the two adjacent monitoring moments, let the time interval be The deformation difference , then the deformation speed ;

[0184] When mm / day, determine the monitoring frequency :

[0185] ;

[0186] wherein, and are further divided speed interval thresholds, i.e. mm / day, mm / day;

[0187] According to the determined monitoring frequency , the data acquisition period of the monitoring device is adjusted to ensure timely acquisition of sufficient surrounding rock state information.

[0188] S6, monitoring and measuring data stage analysis and construction evaluation

[0189] S61, data collection and arrangement

[0190] Let the surrounding rock deformation data sequence obtained each time be , the surrounding rock pressure data sequence be , the steel frame stress data sequence be , and the concrete stress data sequence be ;

[0191] wherein, denotes the monitoring time, is the total number of monitoring times;

[0192] S62, stage analysis according to weeks and months

[0193] Week analysis:

[0194] The data obtained each week is arranged and analyzed respectively, for example, for the surrounding rock deformation data, the average deformation amount each week is calculated:

[0195] ;

[0196] wherein, is the monitoring time of the th week, denotes the monitoring time set corresponding to the th week;

[0197] Similarly, the average surrounding rock pressure each week , the average steel frame stress each week , and the average concrete stress each week can be calculated;

[0198] The change trend of the data each week is analyzed, and the difference between the average deformation amounts of adjacent two weeks is calculated Observe the increase and decrease of deformation and the changes of various stress data to determine the changing trend of surrounding rock conditions;

[0199] Monthly analysis:

[0200] Calculate the average monthly deformation:

[0201] ;

[0202] in, For the Monthly monitoring times, Indicates the The monitoring time set corresponding to the month is also used to calculate the average surrounding rock pressure per month. , Monthly average steel frame stress , monthly average concrete stress ;

[0203] By collating and analyzing the monthly monitoring and measurement data, drawing accurate data change curves, and using rigorous mathematical methods and professional data analysis methods to determine the change patterns of each monitoring and measurement data over time .

[0204] Specifically, statistical analysis trend fitting techniques, such as linear regression, polynomial fitting, or seasonal decomposition models, are used to conduct in-depth data analysis. For data exhibiting clear cyclical characteristics, the cycle length and fluctuation amplitude are precisely determined through methods such as Fourier transforms. For data with long-term growth or decline trends, linear regression models are used to accurately calculate the slope and intercept, allowing for a detailed description of the changing trend using precise mathematical expressions. If the data changes are complex and cannot be summarized using simple mathematical expressions, the characteristics of the changes are carefully described in detail, including but not limited to periodic trends, the impact of special events on the data, and the correlation changes between different monitored variables, to ensure a precise and unambiguous understanding of the data's changing patterns.

[0205] S63. Construction status evaluation

[0206] According to the summary of the monitoring measurement data change rules Evaluate the construction situation. If the surrounding rock deformation continues to increase rapidly, or If the pressure is continuously greater than a certain growth threshold, and the surrounding rock pressure, steel frame stress, concrete stress, etc. also show abnormal changes, that is, they exceed the corresponding safety threshold or the change trend does not conform to normal expectations, then the current construction situation may be risky and it is necessary to adjust the construction process and strengthen support measures in a timely manner. If the changes in various monitoring measurement data are stable and within the safe range, then the construction situation is good and construction can continue as planned, as shown below:

[0207] set up is the incremental threshold of surrounding rock deformation, is the safety threshold range of surrounding rock pressure, is the safety threshold range of steel frame stress, is the safety threshold range of concrete stress.

[0208] If there is a certain week k, the formula for determining the weekly risk situation is as follows:

[0209] ;

[0210] If it is determined that there is a risk in the construction situation in a certain week, corresponding measures need to be taken, such as adjusting construction parameters and strengthening on-site inspections;

[0211] For two consecutive months and , the monthly risk determination formula is as follows:

[0212] ;

[0213] If it is determined that there is a risk in the construction situation for two consecutive months, an in-depth assessment will be carried out and measures such as suspending construction and optimizing support plans will be implemented.

[0214] For continuous Week or Months are determined by the following week determination formula:

[0215] ;

[0216] Alternatively, it can be determined by the following month determination formula:

[0217] ;

[0218] The construction is judged to be in good condition and can continue as planned, while the normal monitoring frequency and data analysis process will be maintained.

[0219] Example 2

[0220] Further illustrate with reference to Example 1, Figure 3 The structure shown, Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0221] A memory, a processor, and a computer program stored in the memory and executable on the processor.

[0222] When the processor executes the program, the three-dimensional calculation and analysis method for excavation with or without support provided in the above embodiment is implemented.

[0223] Furthermore, the electronic device further includes:

[0224] Communication interface, used for communication between memory and processor.

[0225] Memory is used to store computer programs that can be run on the processor.

[0226] The memory may include a high-speed RAM memory and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0227] If the memory, processor, and communication interface are implemented independently, the communication interface, memory, and processor can be interconnected via a bus to facilitate communication. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be categorized as address buses, data buses, and control buses.

[0228] The processor can include one or more processing units, for example: the processor can include an application processor (AP), an application specific integrated circuit (ASIC), a modem processor, a central processing unit (CPU), an image signal processor (ISP), a controller, a memory, a video code, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Different processing units can be independent devices or integrated in one or more processors. The controller can be the nerve center and command center. The controller can generate operation control signals according to instruction operation codes and timing signals, and complete the control of fetching and executing instructions. The processor can also be provided with a memory for storing instructions and data. In some embodiments, the memory in the processor is a cache memory. The memory can save instructions or data that the processor has just used or repeatedly uses. If the processor needs to use the instructions or data again, it can directly call them from the memory. This avoids repeated access and reduces the waiting time of the processor, thereby improving the efficiency of the system.

[0229] The visualization module is configured to display images, videos, and the like. The visualization module can include a display panel, which can be a liquid crystal display (LCD), an organic light-emitting diode (OLED), an active-matrix organic light-emitting diode (AMOLED), a flexible light-emitting diode (FLED), a Miniled, a MicroLed, a Micro-oLed, a quantum dot light emitting diode (QLED), or the like.

[0230] Optionally, in a specific implementation, if the memory, the processor, and the communication interface are integrated on a chip, the memory, the processor, and the communication interface can communicate with each other through an internal interface.

[0231] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the three-dimensional calculation analysis method for excavation with or without support.

[0232] The embodiment of the present application further provides a computer program product, which can run computer instructions, and the computer instructions are executed by a processor to realize the three-dimensional calculation analysis method for excavation with or without support.

[0233] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered a list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- based system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions, or in conjunction with which the instructions maybe executed. In the context of this specification, a "computer-readable medium" can be any means that can contain, store, communicate, propagate or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electronic connection (an electronic device with one or N wires), a portable computer diskette (a magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or other suitable medium on which the program can be printed, because the program can be electronically obtained from the paper or other medium by optical scanning of the paper or other medium, followed by editing, interpreting or otherwise processing the electronically obtained program with an appropriate medium, and then storing the program in a computer memory if necessary.

[0234] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A three-dimensional calculation and analysis method for excavation with or without support, characterized in that: The following steps are involved: S1. Data collection and preprocessing: Obtain geological information of on-site monitoring equipment and construction areas; Collect multi-source data during the current construction process based on the on-site monitoring equipment of the underground powerhouse caverns and the geological information of the construction area, including construction stage data , surrounding rock monitoring data and reinforcement measures parameters ; S2. Data cleaning and feature extraction: preprocessing the collected multi-source data and extracting feature vectors based on the preprocessed multi-source data; In step S2, the preprocessing operation includes clearing and removing outliers and erroneous data, and is used to integrate and standardize multi-source data so that each data has a unified format and dimension. From the surrounding rock monitoring data The extracted feature vectors include the surrounding rock deformation vector , the sorted surrounding rock pressure data vector , stress distribution data vector and microseismicity data vector ; surrounding rock deformation vector , the sorted surrounding rock pressure data vector , stress distribution data vector Arranged in a certain spatial order, the microseismic activity data vector Arrange in a certain chronological order; Construction phase data Including the current excavation depth, tunnel face position, and completed support steps, and converted into corresponding feature vectors; Reinforcement measures parameters Including the length, spacing, grouting amount of anchor rods, prestressing parameters of anchor cables, and converting them into corresponding eigenvectors; S3. Model prediction and analysis: Integrate preprocessed multi-source data to form a complete input sample , and input into the model Conduct analysis and obtain prediction results; Model This is a pre-trained ensemble model that integrates a long short-term memory network (LSTM), a convolutional neural network (CNN), a support vector machine (SVM), and a Transformer. It also includes the following steps: S4. Construction monitoring and early warning: The prediction results are integrated and feature encoded, and the prediction results are input into the Transformer. The Transformer's input layer encodes the position of the data in the prediction results to capture the sequence order. The data then enters multiple encoder layers and captures the correlation between different prediction results. The output is a re-weighted integrated feature representation, which then enters a feedforward neural network consisting of two linear transformation layers and a ReLU activation function for nonlinear transformation and feature extraction. After processing by multiple encoder layers, the data is mapped to a comprehensive early warning indicator through a linear projection layer at the output layer. , whose weights are optimized by the back propagation algorithm during training to determine the input of each prediction result and the comprehensive warning index The best mapping relationship between them, and based on the comprehensive early warning indicators Compare with the threshold to determine the warning level; Among them, each encoder layer includes a multi-head self-attention mechanism and a feedforward neural network.

2. The three-dimensional calculation and analysis method for excavation with or without support according to claim 1 is characterized in that: In step S3, the model For input samples The specific analysis steps are as follows: S31, LSTM layer for microseismic activity data vector Perform analytical processing; According to the sequence and time interval of microseismic events, the model is executed The pre-set 1-3 LSTM layers extract the evolution trend characteristics of microseismic activity and output the characteristic vector of the evolution stage of microseismic source parameters ; Among them, each LSTM layer contains 64-256 hidden units; S32 and CNN perform convolution operations on the spatial distribution data of surrounding rock deformation and damage zones, extract spatial features through convolution kernels of different scales, and output spatial feature vectors of surrounding rock deformation and stress distribution. and ; Among them, in order to identify the shape and position of the local concentrated area of ​​surrounding rock deformation and stress concentration area, the surrounding rock deformation vector , the sorted surrounding rock pressure data vector and stress distribution data vector Spatial feature extraction is performed through 2-3 convolutional layers with kernel sizes ranging from 3×3 to 5×5 and strides of 1 or 2. Each convolutional layer is followed by a maximum pooling layer. S33. Using feature engineering to extract data from the construction phase and reinforcement measures parameters Extract key features and form vectors , and used as the input data of the SVM model, the feature vectors output by LSTM and CNN after passing through their respective fully connected layers 、 and And use weighted summation to calculate and fuse the feature vector The calculation formula is as follows: ； The weights are 、 、 ; The fused feature vector Input into the SVM model and according to the fusion feature vector Execute the prediction and use the SoftMax function to output the deformation prediction value of the surrounding rock , stress prediction value , predicted probability of damage zone distribution and the predicted probability of the evolution stage of microseismic source parameters .

3. The three-dimensional calculation and analysis method for excavation with or without support according to claim 2 is characterized in that: In step S4, the standard surrounding rock data in the database is for: ； in, Indicates the deformation dimension of different monitoring positions, Corresponding to four deformation levels, standard surrounding rock data Vector representation of typical deformation conditions corresponding to different deformation levels; Calculate the deformation prediction value of the current surrounding rock Compared with various standard surrounding rock data Euclidean distance , the Euclidean distance calculation formula is: ； By comparing the Euclidean distance The size of determines the deformation level: like , it is determined to be a slight deformation of the first level, and the daily deformation amount is expressed as ; like , it is determined to be a secondary medium deformation, and the daily deformation is expressed as ; like , it is determined to be a grade 3 severe deformation, and the daily deformation is expressed as ; like , it is determined to be a level 4 extremely strong deformation, and the daily deformation is expressed as .

4. The three-dimensional calculation and analysis method for excavation with or without support according to claim 3 is characterized in that: According to Euclidean distance Determine the surrounding rock deformation level and combine it with the model Output of predicted deformation value of surrounding rock , stress prediction value , predicted probability of damage zone distribution and the predicted probability of the evolution stage of microseismic source parameters Conduct comprehensive early warning; Defining comprehensive early warning indicators , which is a weighted combination of the following: ； in, 、 、 、 These are the weight coefficients of the four factors, namely, deformation level, stress, probability of distribution of damage zone, and evolution stage of microseismic source parameters, in the calculation of comprehensive early warning indicators; Determine according to the actual situation and importance of the project 、 、 、 , and satisfies ; get: ; Among them, obtain the surrounding rock deformation grade , according to the surrounding rock deformation grade Substitute the value into the deformation level function Get the corresponding function value and use the stress prediction value and stress threshold 、 Compare and substitute the stress function Get the corresponding function value and predict the probability of damage zone distribution and the predicted probability of the evolution stage of microseismic source parameters , and then substituted into the probability function of the damage area distribution after comparing with their respective thresholds and the probability function of the evolution stage of microseismic source parameters Get the corresponding function value and multiply the corresponding function value by the corresponding weight coefficient 、 、 、 Then add them together to get the comprehensive early warning index .

5. The three-dimensional calculation and analysis method for excavation with or without support according to claim 4 is characterized in that: The comprehensive early warning indicators The specific calculation steps are as follows: Assume that the surrounding rock deformation grade determined by Euclidean distance is ; Among them, the surrounding rock deformation grade The values ​​are 1, 2, 3, and 4, corresponding to the first-level slight deformation, the second-level moderate deformation, the third-level strong deformation, and the fourth-level extremely strong deformation, respectively, and are used to reflect the severity of the surrounding rock deformation; Deformation level function According to the surrounding rock deformation grade Assign different basic scores and adjust the surrounding rock deformation level Factor quantification, that is, according to the input surrounding rock deformation level value, directly output the corresponding fixed score, the formula is as follows: ； set up and is the stress threshold, according to the stress prediction value Determine the stress function in relation to the threshold Function value of : ； set up and is the probability threshold of the damage zone distribution, and , , then the probability function of the damage zone distribution is The formula is as follows: ； set up and , is the probability threshold of the microseismic source parameter evolution stage, and , , then the probability function of the microseismic source parameter evolution stage is The formula is as follows: ； set up 、 and is the warning threshold, and the comprehensive warning index calculated based on Compare with the set warning threshold to determine the warning level; Among them, , , , then the formula is expressed as: ； Displayed through a visual interface, according to The value of is represented by different colors in the animation simulation of the surrounding rock deformation. The color cloud map of stress distribution is rendered with the value of color, and the predicted probability of damage zone distribution is represented by different transparency and color. The probability of different damage zone categories in the stratum, as well as the predicted probability according to the evolution stage of microseismic source parameters High-risk microseismic hotspots are highlighted in vibrant colors.

6. The three-dimensional calculation and analysis method for excavation with or without support according to claim 4 is characterized in that: S5. Monitoring frequency adjustment: According to the deformation prediction value of surrounding rock , stress prediction value , predicted probability of damage zone distribution and the predicted probability of the evolution stage of microseismic source parameters and comprehensive early warning indicators Determine the current status of the underground powerhouse cavern complex. The specific steps are as follows: Assume the deformation speed is , by calculating the two adjacent monitoring moments, let the time interval be The deformation difference , the deformation speed formula is as follows: ； when mm / day, determine the monitoring frequency , and is expressed by the following formula: ； in, and To further divide the speed interval threshold, that is mm / day, mm / day; According to the determined monitoring frequency , adjust the data collection cycle of the monitoring equipment to ensure that sufficient surrounding rock status information is obtained in a timely manner.

7. The three-dimensional calculation and analysis method for excavation with or without support according to claim 6 is characterized in that: S6. Monitoring and measurement data stage analysis and construction evaluation: Through weekly and monthly analysis, determine the construction stage data , surrounding rock monitoring data and reinforcement measures parameters The specific steps are as follows: S61. Data collection and collation: Assume that the surrounding rock deformation data sequence obtained in each monitoring is , the surrounding rock pressure data series is , the steel frame stress data series is , the concrete stress data series is ; in, Indicates the monitoring time, is the total number of monitoring times; S62. Conduct stage analysis by week and month Weekly Analysis: The data obtained each week are sorted and analyzed, that is, the weekly average deformation of the surrounding rock deformation data is calculated: ； in, For the Weekly monitoring times, Indicates the The set of monitoring times corresponding to the week; Similarly, the weekly average surrounding rock pressure can be calculated , Weekly average steel frame stress , weekly average concrete stress ; Analyze the trend of weekly data by calculating the difference between the average deformation of two consecutive weeks Observe the increase and decrease of deformation and the changes of various stress data to determine the changing trend of surrounding rock conditions; Monthly analysis: Calculate the average monthly deformation: ； in, For the Monthly monitoring times, Indicates the The monitoring time set corresponding to the month can be used to calculate the average surrounding rock pressure per month. , Monthly average steel frame stress , monthly average concrete stress ; By sorting and analyzing the changing trends of weekly data and monthly monitoring measurement data, we can determine the changing patterns of each monitoring measurement data over time. .

8. The three-dimensional calculation and analysis method for excavation with or without support according to claim 7 is characterized in that: In step S6, according to the change rule The specific steps for carrying out the construction status evaluation are as follows: set up is the incremental threshold of surrounding rock deformation, is the safety threshold range of surrounding rock pressure, is the safety threshold range of steel frame stress, is the safety threshold range of concrete stress; If there is a certain week k, the formula for determining the weekly risk situation is as follows: ； For two consecutive months and , the monthly risk determination formula is as follows: ； If a construction situation is identified as risky for a particular week or two consecutive months, an in-depth assessment is performed and appropriate measures are taken.

Images