A method for evaluating, predicting and dynamically controlling the stability of roadway surrounding rock

Abstract

The present invention provides a method for real-time identification and dynamic regulation of the stability state of roadway surrounding rock. This method realizes the real-time evaluation and prediction of the stability of surrounding rock by constructing an evaluation model and a prediction model considering time series. The static evaluation result and the dynamic evaluation result are weighted and added to realize the evaluation of the stability state of surrounding rock at the current moment. The historical data of dynamic evaluation indexes and the advanced detection data of roadway excavation such as joint density and formation fissure water content are used to predict the values of dynamic evaluation indexes of the stability of surrounding rock at future moments. The dynamic evaluation result obtained by substituting the predicted values of dynamic evaluation indexes into the dynamic evaluation model is weighted and added to the static evaluation result to realize the evaluation of the stability state of surrounding rock at future moments. Further combined with the golden eagle optimization algorithm, the dynamic regulation of the operation state at the current moment is realized, ensuring that the surrounding rock is in a stable state during the roadway excavation process at the next moment.

Description

Technical Field

[0001] The present invention relates to the field of underground space tunneling safety, and particularly to a method for evaluating, predicting and dynamically controlling the stability of roadway surrounding rock. Background Technique

[0002] At present, with the continuous reduction of shallow resources in China, the development of deep mineral resources has gradually become the mining trend of China's mineral resources. And constructing deep roadways is the key to realizing the development of deep underground resources. However, compared with shallow roadways, deep roadways are more likely to be deformed under the influence of mining, stoping operations, and blasting operations, which will then lead to the cracking and failure of the roadway surrounding rock until it becomes unstable.

[0003] Roof and rock burst accidents caused by the instability of the rock mass around deep roadways are one of the major disasters occurring during roadway tunneling. Roof and rock burst accidents often lead to direct injuries such as personnel being buried and crushed, or indirect injuries such as asphyxiation, and are also likely to cause roadway blockages and trap personnel. Taking effective measures to reduce the occurrence probability of roadway surrounding rock instability can effectively reduce the occurrence of roof and rock burst accidents during tunneling. Conducting real-time stability evaluation of roadway surrounding rock has important engineering practical value.

[0004] At present, most studies use artificial neural networks to evaluate the stability of roadway surrounding rock. Most of them start from a single influencing factor for evaluation and prediction, and rarely consider the relationships between various factors and their impacts on the surrounding rock mass during roadway tunneling. The factors affecting roadway stability are numerous and complex, and they influence and restrict each other. Making a pre-evaluation of the stability of roadway surrounding rock in a timely and accurate manner is of great significance for discovering and taking measures to control the instability of the surrounding rock. In addition, the rock layer conditions of roadways are complex and changeable, and the in-situ stress and the dynamic evolution of the physical and mechanical parameters of the rock layer lead to a complex variety of stability types of roadway surrounding rock, making it a major problem during roadway tunneling to judge its stability. With the increase of mining depth, driving a series of complex roadways is likely to cause the surrounding rock to lose its original equilibrium state, resulting in roadway instability, threatening the lives of personnel and causing significant property losses.

[0005] Therefore, there is an urgent need to develop a method for evaluating, predicting and dynamically controlling the stability of roadway surrounding rock. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for real-time identification and dynamic control of roadway surrounding rock stability to solve the problems existing in the prior art.

[0007] The technical solution adopted to achieve the purpose of the present invention is as follows: A method for evaluating, predicting and dynamically controlling the stability of roadway surrounding rock includes the following steps:

[0008] 1) Obtain the data of the discriminant index for the stability of the roadway surrounding rock under the on-site operation conditions of the roadway. Among them, the discriminant index for the stability of the roadway surrounding rock includes static evaluation indexes and dynamic evaluation indexes. The static evaluation indexes include the dip angle of coal and rock strata, rock stratum structure, natural stress state of rock mass, roadway burial depth, designed size of roadway section, roadway construction technology, roadway driving method, and designed length of roadway. The dynamic evaluation indexes include temperature, humidity, pressure, surrounding rock deformation amount, influence of underground water inflow, opening degree of surrounding rock fissures, actual roadway width, and actual roadway height.

[0009] 2) Standardize the data of the discriminant index for the stability of the roadway surrounding rock.

[0010] 3) Based on the standardized discriminant index for the stability of the roadway surrounding rock, realize the grade division of the stability of the surrounding rock. The stability of the roadway surrounding rock is divided into four grades: stable, moderately stable, unstable, and extremely unstable.

[0011] 4) Construct a real-time evaluation model for the stability of the roadway surrounding rock. The real-time evaluation model for the stability of the roadway surrounding rock includes a static evaluation model and a dynamic evaluation model.

[0012] 5) Use the real-time evaluation model for the stability of the roadway surrounding rock to realize the real-time evaluation of the stability of the roadway surrounding rock at the current moment.

[0013] 5.1) Substitute the static evaluation index values at the current moment into the static evaluation model to obtain the static evaluation result P j at the current moment. Substitute the dynamic evaluation index values at the current moment into the dynamic evaluation model to obtain the current dynamic evaluation result P dt at the current moment.

[0014] 5.2) Weight the obtained static evaluation index P j at the current moment and the dynamic evaluation index P dt at the current moment in the following way to obtain the evaluation result P t of the stability of the roadway surrounding rock at the current moment.

[0015] 6) Construct a prediction model for the dynamic evaluation index values of the stability of the roadway surrounding rock. The prediction model for the dynamic evaluation index values of the stability of the surrounding rock selects the LSTM model. The LSTM model includes an input gate, an output gate, a forget gate, a cell state, and a hidden state.

[0016] 7) Use the prediction model for the dynamic evaluation index values of the stability of the roadway surrounding rock to realize the prediction of the dynamic evaluation index values of the stability of the surrounding rock at future moments. Substitute the historical data and advanced detection data of the dynamic evaluation index into the prediction model for the dynamic evaluation index values of the stability of the surrounding rock to obtain the dynamic evaluation index values at future t+1, t+2, t+3, and t+4 moments. The advanced detection data includes the range of soft surrounding rock, joint density, and content of formation fissure water, etc.

[0017] 8) Substitute the dynamic evaluation index values obtained in step 7) into the real-time evaluation model of roadway surrounding rock stability to obtain the dynamic evaluation results of the surrounding rock stability at future times t+1, t+2, t+3, and t+4. Weightedly add the obtained dynamic evaluation results of the surrounding rock stability at future times t+1, t+2, t+3, and t+4 to the static evaluation results to obtain the evaluation results of the surrounding rock stability at future times t+1, t+2, t+3, and t+4.

[0018] 9) Based on the grade division of the evaluation results of the surrounding rock stability at times t+1, t+2, t+3, and t+4, obtain the future surrounding rock stability state. When the surrounding rock stability state at future times t+1, t+2, t+3, or t+4 is moderately stable, unstable, or extremely unstable, formulate a dynamic regulation plan.

[0019] Furthermore, the real-time evaluation model of roadway surrounding rock stability selects the Bayesian network evaluation model, and substitute the obtained surrounding rock stability discrimination indexes at future times t+1, t+2, t+3, and t+4 into the Bayesian network evaluation model to obtain the surrounding rock stability states at future times t+1, t+2, t+3, and t+4.

[0020] Furthermore, the construction process of the real-time evaluation model of roadway surrounding rock stability includes the following sub-steps:

[0021] a) Use the roadway surrounding rock stability discrimination indexes to construct the Bayesian network nodes.

[0022] b) Construct a directed acyclic graph to obtain the roadway surrounding rock stability model structure.

[0023] c) Conduct parameter learning of the Bayesian network model for roadway surrounding rock stability to obtain the conditional probabilities between the roadway surrounding rock stability indexes. Combine the roadway surrounding rock stability Bayesian network structure obtained in step b) with the conditional probabilities to obtain the roadway surrounding rock stability Bayesian static evaluation network.

[0024] d) Statistically analyze the grades of the roadway surrounding rock stability states to obtain the transition probability values of the roadway surrounding rock Bayesian transfer network. Combine the roadway surrounding rock stability static Bayesian static evaluation network obtained in step c) with the transition probability values to obtain the roadway surrounding rock stability dynamic Bayesian network.

[0025] Furthermore, in step 9), the Golden Eagle intelligent optimization algorithm is used to search for the optimal dynamic regulation value.

[0026] The technical effects of the present invention are beyond doubt:

[0027] A. Aiming at the technical problem that it is difficult to understand the stability of roadway surrounding rock in real time, an evaluation model and a prediction model considering time series are proposed to realize the real-time evaluation and prediction of the stability of surrounding rock, thus making up for the existing technical problems.

[0028] B. The static evaluation results and the dynamic evaluation results are weighted and added together to realize the evaluation of the surrounding rock stability state at the current moment; the historical data of the dynamic evaluation indexes and the advanced detection data of roadway excavation such as joint density and formation fissure water content are brought into the prediction model of the dynamic evaluation index value of surrounding rock stability to realize the prediction of the dynamic evaluation index value of surrounding rock stability at the future moment; finally, the dynamic evaluation results obtained by substituting the predicted dynamic evaluation index values into the dynamic evaluation model are weighted and added to the static evaluation results to realize the evaluation of the surrounding rock stability state at the future moment;

[0029] C. Based on the evaluation and prediction results of roadway surrounding rock stability, combined with the Golden Eagle optimization algorithm, the dynamic regulation of the operation state at the current moment is realized, ensuring that the surrounding rock is in a stable state during the roadway excavation at the next moment. Description of the Drawings

[0030] Figure 1 is the flow chart of the method for evaluating, predicting and dynamically regulating the stability of roadway surrounding rock;

[0031] Figure 2 is the structure diagram of the dynamic Bayesian network;

[0032] Figure 3 is the structure diagram of LSTM. Specific Embodiments

[0033] The present invention will be further described below in conjunction with embodiments, but it should not be understood that the above-mentioned subject scope of the present invention is limited to the following embodiments. Without departing from the above-mentioned technical idea of the present invention, various substitutions and changes should be included in the protection scope of the present invention according to the common general knowledge and customary means in the art.

[0034] Embodiment 1:

[0035] The conditions of roadway surrounding rock are complex and changeable, and its stability directly affects its safety. Accurately judging the type of surrounding rock stability has become one of the effective ways to ensure the safe and efficient mining of mineral resources. In addition, the historical operation data during roadway excavation will affect the operation at the current moment.

[0036] Refer to Figures 1 to 3 , this embodiment provides a method for evaluating, predicting and dynamically regulating the stability of roadway surrounding rock, including the following steps:

[0037] 1) Obtain the data of the discriminant index for the stability of the roadway surrounding rock under the on-site operation conditions of the roadway. Among them, the discriminant index for the stability of the roadway surrounding rock includes static evaluation indexes and dynamic evaluation indexes. The static evaluation indexes include natural factors and human factors. The natural factors include the dip angle of coal and rock strata, rock stratum structure, natural stress state of rock mass, and roadway burial depth. The human factors include the designed size of the roadway section, roadway construction technology, roadway driving method, and designed length of the roadway. The dynamic evaluation indexes include environmental factors, geological factors, and human factors. The environmental factors include temperature, humidity, and pressure. The geological factors include the deformation amount of the surrounding rock, influence of underground water inflow, and aperture of surrounding rock fissures. The human factors include the actual roadway width and actual roadway height.

[0038] 2) Standardize the data of the discriminant index for the stability of the roadway surrounding rock.

[0039] 3) Based on the standardized discriminant index for the stability of the roadway surrounding rock, realize the classification of the stability level of the surrounding rock. The stability of the roadway surrounding rock is divided into four levels: extremely unstable, unstable, moderately stable, and stable. Denote the data to be classified as the initial data t. When there is a risk of collapse of the roadway surrounding rock in the next column of data (t + 1), severely restricting the driving of the roadway, the stability of the roadway surrounding rock is classified as extremely unstable. The extremely unstable state means that the rock mass is in a granular structure, the tectonic influence is very serious, the joint planes and their combinations are disorderly, forming a large number of fragmented bodies, and most of the blocks are filled with mud, even in a state of stone sandwiched with soil or soil sandwiched with stone. When there is a risk of collapse of the roadway surrounding rock in the third column of data (t + 2), restricting the driving of the roadway, the stability of the roadway surrounding rock is classified as unstable. The unstable state means that the rock mass is in a fractured structure, the biting force is weak, forming more unstable blocks, and the structure of thin layers, medium-thick layers, and soft rock interbeds with poor interlayer bonding. When there is a risk of collapse of the roadway surrounding rock in the fourth column of data (t + 3), affecting the driving of the roadway, the stability of the roadway surrounding rock is classified as moderately stable. The moderately stable state means that the structure of thin layers and hard-soft rock interbeds with good interlayer bonding has a slight influence on the driving process of the roadway. When there is a risk of collapse of the surrounding rock in the fifth column of data (t + 4) and subsequent columns, affecting the driving of the roadway, the stability of the roadway surrounding rock is classified as stable. The stable state means that the rock mass is in a massive and thick-layered structure with good interlayer bonding, mainly composed of primary and tectonic joints, generally no unstable rock blocks will appear, and it does not affect the driving of the roadway.

[0040] 4) Construct a real-time evaluation model for the stability of the roadway surrounding rock. The real-time evaluation model for the stability of the roadway surrounding rock includes a static evaluation model and a dynamic evaluation model. Use the method of combining static and dynamic to realize the real-time evaluation of the stability of the roadway surrounding rock. The real-time evaluation model for the stability of the roadway surrounding rock selects the Bayesian network evaluation model. The construction process of the model includes the following sub-steps:

[0041] 4.1) Use the evaluation indexes of roadway surrounding rock stability to construct the nodes of the Bayesian network.

[0042] 4.2) Construct a directed acyclic graph to obtain the model structure of roadway surrounding rock stability.

[0043] 4.3) Conduct parameter learning of the Bayesian network model for roadway surrounding rock stability to obtain the conditional probabilities among the evaluation indexes of roadway surrounding rock stability. Combine the Bayesian network structure of roadway surrounding rock stability obtained in step 4.2) with the conditional probabilities to obtain the Bayesian static evaluation network of roadway surrounding rock stability.

[0044] 4.4) Statistically analyze the stability state grades of roadway surrounding rock to obtain the transition probability values of the Bayesian transition network of roadway surrounding rock. Combine the static Bayesian evaluation network of roadway surrounding rock stability obtained in step 4.3) with the transition probability values to obtain the dynamic Bayesian network of roadway surrounding rock stability.

[0045] 5) Use the real-time evaluation model of roadway surrounding rock stability to realize the real-time evaluation of the stability of roadway surrounding rock at the current moment. Substitute the dynamic evaluation index values of the surrounding rock stability at the future times t + 1, t + 2, t + 3, and t + 4 into the dynamic evaluation model to obtain the dynamic evaluation results of the surrounding rock stability at the future times t + 1, t + 2, t + 3, and t + 4. Weight and add the obtained dynamic evaluation results and the static evaluation results to obtain the evaluation results of the surrounding rock stability at the future times t + 1, t + 2, t + 3, and t + 4, so as to realize the real-time evaluation of the stability of roadway surrounding rock.

[0046] 5.1) Use the method combining static and dynamic approaches to realize the real-time evaluation of roadway surrounding rock stability. Substitute the static evaluation index value and the dynamic evaluation index value at the current moment into the static evaluation model and the dynamic evaluation model respectively to obtain the static evaluation result P j and the dynamic evaluation result P dt .

[0047] 5.2) Weight the obtained static evaluation result P j and the dynamic evaluation result P dt in the following way to obtain the evaluation result P t of the roadway surrounding rock stability at the current moment.

[0048] P t = α′P j + β′P dt

[0049]

[0050]

[0051] It should be noted that the static evaluation index refers to the influence of various factors on the stability of surrounding rock at the current moment without considering the impact of time series, including but not limited to indicators such as the dip angle of coal and rock strata, the cross-sectional size of the roadway, the depth of the roadway, the roadway construction technology, the rock stratum structure, the natural stress state of the rock mass, the roadway driving method, and the length of the roadway. The index value of the static index described in the present invention is less affected by the driving process, but the static index has a certain impact on roadway driving and its stability. Since the impact it generates has been determined during the driving process and does not change with the change of the driving position, it is a static evaluation index. The static evaluation achieved through static indexes can only obtain a rough evaluation of the stability of surrounding rock and is difficult to reflect the influence degree of real-time operation data on the stability of surrounding rock at the current stage. That is, the influence of the static evaluation index on the stability of surrounding rock is determined before roadway driving and does not change with time during the driving process. The dynamic evaluation index refers to the factors that appear during roadway driving and affect the stability of surrounding rock, including but not limited to temperature, humidity, pressure, the deformation amount of surrounding rock, roadway width, roadway height, the influence of underground water inflow, and the opening degree of surrounding rock fissures. The index value of the dynamic evaluation index described in the present invention changes continuously with the stability of surrounding rock during the driving process. That is, the dynamic evaluation index needs to be monitored in real time. Through dynamic evaluation, the influence of real-time operation data on the current operation can be reflected. However, the influence of operation technology on it cannot be well reflected in the dynamic evaluation index. Relying solely on dynamic evaluation to obtain the influence at the current moment is relatively weak, and static evaluation cannot consider the influence of various factors on the operation in combination with real-time conditions. Therefore, a method of combining static and dynamic is adopted, and the static evaluation result and the dynamic evaluation result are weighted and added to achieve real-time evaluation of the stability of roadway surrounding rock.

[0052] 6) Construct a prediction model for the dynamic evaluation index value of roadway surrounding rock stability. The prediction model for the dynamic evaluation index value of surrounding rock stability selects the LSTM model. The LSTM model includes a series of repeated LSTM cells. Each LSTM cell includes a forget gate, an input gate, and an output gate. At time t, the forget gate, input gate, and output gate are represented by ft, it, and ot respectively. Each LSTM cell first uses the forget gate to filter out the information to be discarded, then combines the valid information in the input gate, and finally obtains the output hidden layer from the output gate. The LSTM cell passes the cell state vector to the next LSTM cell for continuous write, read, and reset operations. The LSTM model includes an input gate, an output gate, a forget gate, a cell state, and a hidden state.

[0053] 7) The prediction model of the dynamic evaluation index value of the roadway surrounding rock stability is used to predict the dynamic evaluation index value of the surrounding rock stability at future time. The historical data of the dynamic evaluation index and the advanced detection data are substituted into the prediction model of the dynamic evaluation index value of the roadway surrounding rock stability to obtain the dynamic evaluation index values at future times t+1, t+2, t+3, and t+4. The advanced detection data includes the range of soft surrounding rock, joint density, and formation fissure water content, etc.

[0054] 8) The dynamic evaluation index values obtained in step 7) are substituted into the real-time evaluation model of the roadway surrounding rock stability to obtain the dynamic evaluation results of the surrounding rock stability at future times t+1, t+2, t+3, and t+4. The dynamic evaluation results of the surrounding rock stability at future times t+1, t+2, t+3, and t+4 are weighted and added to the static evaluation results to obtain the evaluation results of the surrounding rock stability at future times t+1, t+2, t+3, and t+4.

[0055] 9) Based on the grade division of the evaluation results of the surrounding rock stability at times t+1, t+2, t+3, and t+4, the future surrounding rock stability state is obtained. When the surrounding rock stability state at future time t+1, t+2, t+3, or t+4 is moderately stable, unstable, or extremely unstable, a dynamic regulation plan is formulated.

[0056] In this embodiment, the real-time evaluation and prediction of the surrounding rock stability are realized by constructing an evaluation model and a prediction model considering time series. The static evaluation results and the dynamic evaluation results are weighted and added to realize the evaluation of the surrounding rock stability state at the current time. The historical data of the dynamic evaluation index and the advanced detection data of the roadway excavation such as joint density and formation fissure water content are used to predict the dynamic evaluation index value of the surrounding rock stability at future time. The dynamic evaluation results obtained by substituting the predicted dynamic evaluation index values into the dynamic evaluation model and the static evaluation results are weighted and added to realize the evaluation of the surrounding rock stability state at future time. Further, the dynamic regulation of the current operation state is realized to ensure that the surrounding rock is in a stable state during the roadway excavation at the next time.

[0057] Example 2:

[0058] The main steps of this embodiment are the same as those of Example 1. Among them, to realize the prediction of the dynamic evaluation index value of the roadway surrounding rock stability at future time, LSTM (Long Short-Term Memory) is selected to predict the dynamic evaluation index value. The long short-term memory network model includes an input gate, an output gate, and a forget gate, and also includes a cell state and a hidden state.

[0059] In step 6), the construction process of the prediction model includes the following sub-steps:

[0060] a) Input the data related to the surrounding rock state during the roadway driving process into the LSTM algorithm, including the historical data of dynamic evaluation indicators and the advanced drilling data such as joint density and formation fissure water content. First, through the forget gate, the role of the forget gate is to control whether to forget the hidden cell state of the previous layer with a certain probability, and the probability value range is (0, 1). When the value is 1, it means "retain all previous information"; when the value is 0, it means "forget all previous information". The calculation formula is as follows:

[0061] F(t) = σ(W f h t-1 +U f X t +b f )

[0062] Among them, W f and U f correspond to the weight matrices of h (t-1) and x (t) respectively. x (t) is the data input of the roadway surrounding rock at time t, h (t-1) is the hidden state of the previous moment, and b f is the corresponding threshold value.

[0063] b) Use the weight matrix and the sigmod activation function to obtain the output of the roadway surrounding rock state at time t, and this output is a probability obtained by the forget gate.

[0064] c) The data related to the roadway surrounding rock state after passing through the forget gate will be passed into the input gate. The role of the input gate is to be responsible for processing the input of the surrounding rock state at the current sequence position. The mathematical expression is as follows:

[0065] i (t) = σ(W i h (t-1) +U i X (t) +b i )

[0066] C (t) = tanh(W c h (t-1) +U c X (t) +b c )

[0067] Both parts are related to the input gate. Among them, W i 、W C and U i 、U c correspond to the weight matrices of h (t-1) and x (t) respectively. x (t)is the data input of the roadway surrounding rock at time t, h (t-1) is the hidden state at the previous moment, b f is the corresponding threshold value; C (t) is the output h at the previous moment (t-1) and the current input data x (t) A temporary neuron obtained by calculating the neuron state using the hyperbolic tangent function.

[0068] d) Output the dynamic evaluation index value of the roadway surrounding rock at the current moment through the output gate. The role of the output gate is to output the hidden state h at time t - 1 (t-1) and the cell state c at time t (t) , and the mathematical expression is as follows:

[0069] O t = σ(W o h (t-1) + U o X (t) + b o )

[0070] h (t) = o t × tanh(C (t) )

[0071] Among them, W o and U o correspond to the weight matrices of h (t-1) and x (t) respectively, b o is the corresponding threshold value, Ot is the output of the output gate, and h(t) is the predicted value of the dynamic evaluation index of the roadway surrounding rock output by the neuron at time t.

[0072] Substitute the historical data of the dynamic evaluation index and the advanced detection data such as the soft surrounding rock range, joint density, and formation fissure water content into the prediction model of the dynamic evaluation index value of the surrounding rock stability, and based on this, predict the dynamic evaluation index values at future times t + 1, t + 2, t + 3, and t + 4.

[0073] Example 3:

[0074] The main steps of this example are the same as those of Example 1. Among them, in step 9), based on the on-site operation conditions of the roadway tunneling, select the controllable indicators to construct a multi-objective optimization model, find the optimal parameters by establishing the objective function, and further obtain the optimal values of the regulation of each index variable, so as to obtain the dynamic regulation scheme.

[0075] According to the real-time evaluation results, screen out the controllable indicators T1, T2, T3... T n .

[0076] Define the dynamic regulation values △d1, △d2, △d3... △d for each index n Add the dynamic regulation values to the selected adjustable index values to obtain the regulated index values K1, K2, K3... K n .

[0077] K i = T i + Δd i i = 1, 2, 3,... n

[0078] Substitute the index data at the current moment after dynamic regulation into the surrounding rock stability prediction model to obtain the stability of the surrounding rock at future times t+1, t+2, t+3, t+4 after regulation

[0079] Use the Golden Eagle Optimizer (GEO) to search for the optimal dynamic regulation values △d1, △d2, △d3... △d n . This part includes formulating the objective function and finding the optimal solution

[0080] The formulation of the objective function needs to be determined considering the on-site operation conditions of the roadway. During the roadway excavation process, dynamic regulation will slow down the excavation speed. Therefore, it is necessary to minimize the regulation amplitude to the greatest extent. In addition, it is also necessary to ensure that the regulated values can still guarantee the stability of the roadway surrounding rock at future times t+1, t+2, t+3, t+4 and do not affect the operation. Thus, the dynamic regulation objective function can be obtained

[0081]

[0082] The first stage is the selection stage. At this stage, find the approximate location of the optimal target. Each golden eagle must select a target to perform cruise and attack operations, and then calculate the attack and cruise vectors of each golden eagle relative to the selected target. If the new target (calculated through the attack and cruise vectors) is better than the previous target in memory, update the memory

[0083] The second stage is the search stage, which can be simulated by a vector starting from the current position of the golden eagle and ending at the target position in the golden eagle's memory, that is, search near the optimal dynamic regulation value of the roadway surrounding rock stability

[0084] The third stage is the hunting stage, that is, find the dynamic regulation values of each regulation index. Obtain the optimal search position in the objective function and output the optimal dynamic regulation values of each regulation index

[0085] Optimize the search to obtain the dynamic regulation values △d1, △d2, △d3... △d of the roadway surrounding rock stability n , and compare with the selected adjustable index values T1, T2, T3... T nAdd them up to obtain the regulated index values K1, K2, K3... K n And use the regulated operation data to guide the roadway tunneling operation at the next moment, thereby realizing the dynamic regulation of the stability of the surrounding rock, and at the same time ensuring that the surrounding rock is in a stable state during the roadway tunneling and does not affect the tunneling progress of the roadway.

Claims

1. A method for evaluating, predicting and dynamically controlling the stability of roadway surrounding rock, characterized in that, Including the following steps: 1) Obtain the data of the roadway surrounding rock stability discrimination index under the on-site operation conditions of the roadway; wherein, the roadway surrounding rock stability discrimination index includes static evaluation indexes and dynamic evaluation indexes; the static evaluation indexes include the dip angle of coal and rock strata, rock stratum structure, natural stress state of rock mass, roadway burial depth, designed size of roadway section, roadway construction technology, roadway driving method, and designed length of roadway; the dynamic evaluation indexes include temperature, humidity, pressure, surrounding rock deformation amount, influence of underground water inflow, opening degree of surrounding rock fissures, actual roadway width, and actual roadway height; 2) Standardize the data of the roadway surrounding rock stability discrimination index; 3) Based on the standardized data of the roadway surrounding rock stability discrimination index, realize the grade division of the surrounding rock stability; divide the roadway surrounding rock stability into four grades: stable, moderately stable, unstable, and extremely unstable; 4) Construct a real-time evaluation model for roadway surrounding rock stability; the real-time evaluation model for roadway surrounding rock stability includes a static evaluation model and a dynamic evaluation model; 5) Use the real-time evaluation model for roadway surrounding rock stability to realize the real-time evaluation of the roadway surrounding rock stability at the current moment; 5.1) Substitute the static evaluation index value at the current moment into the static evaluation model to obtain the static evaluation result P at the current moment j ; Substitute the dynamic evaluation index value at the current moment into the dynamic evaluation model to obtain the current dynamic evaluation result P dt ; 5.2) The obtained static evaluation index P at the current moment j and the dynamic evaluation index P dt are weighted in the following manner to obtain the evaluation result P of the roadway surrounding rock stability at the current moment t ; P t = α′P j + β′P dt 6) Build a prediction model for the dynamic evaluation index value of the roadway surrounding rock stability; the prediction model for the dynamic evaluation index value of the surrounding rock stability selects the LSTM model; the LSTM model includes a series of repeated LSTM cells; each LSTM cell includes a forget gate, an input gate, and an output gate; at time t, the forget gate, the input gate, and the output gate are represented by f t , i t , and o t respectively; each LSTM cell first uses the forget gate to filter out the information to be discarded, then merges the valid information in the input gate, and finally obtains the output hidden layer from the output gate; the LSTM cell passes the cell state vector to the next LSTM cell for continuous write, read, and reset operations; 7) Use the prediction model for the dynamic evaluation index value of roadway surrounding rock stability to predict the dynamic evaluation index value of surrounding rock stability at a future moment, substitute the historical data of the dynamic evaluation index value and the advanced detection data into the prediction model for the dynamic evaluation index value of surrounding rock stability, and predict the dynamic evaluation index values at future times t + 1, t + 2, t + 3, and t + 4; the advanced detection data includes the range of soft surrounding rock, joint density, and formation fissure water content; 8) Substitute the dynamic evaluation index values obtained in step 7) into the real-time evaluation model for roadway surrounding rock stability to obtain the dynamic evaluation results of the surrounding rock stability at future times t + 1, t + 2, t + 3, and t + 4; perform weighted addition on the obtained dynamic evaluation results of the surrounding rock stability at future times t + 1, t + 2, t + 3, and t + 4 and the static evaluation results to obtain the evaluation results of the surrounding rock stability at future times t + 1, t + 2, t + 3, and t + 4; 9) Based on the grade division of the evaluation results of the surrounding rock stability at times t + 1, t + 2, t + 3, and t + 4, obtain the future surrounding rock stability state; when the surrounding rock stability state at future time t + 1, t + 2, t + 3, or t + 4 is moderately stable, unstable, or extremely unstable, formulate a dynamic regulation plan.

2. The method for evaluating, predicting and dynamically regulating the stability of roadway surrounding rock according to claim 1, wherein: The real-time evaluation model for roadway surrounding rock stability selects a Bayesian network evaluation model, and substitute the obtained surrounding rock stability discrimination indexes at future times t + 1, t + 2, t + 3, and t + 4 into the Bayesian network evaluation model to obtain the surrounding rock stability states at future times t + 1, t + 2, t + 3, and t + 4.

3. A method for evaluating, predicting and dynamically regulating the stability of roadway surrounding rock according to claim 2, characterized in that, The construction process of the real-time evaluation model for roadway surrounding rock stability includes the following sub-steps: a) Use the roadway surrounding rock stability discrimination index to construct the Bayesian network nodes; b) Construct a directed acyclic graph to obtain the roadway surrounding rock stability model structure; c) Conduct parameter learning of the Bayesian network model for the stability of roadway surrounding rock to obtain the conditional probabilities among the stability indexes of the roadway surrounding rock; combine the Bayesian network structure and conditional probabilities of the roadway surrounding rock stability obtained in step b) to obtain the static Bayesian evaluation network for the stability of the roadway surrounding rock. d) Statistically analyze the stability state levels of the roadway surrounding rock to obtain the transition probability values of the Bayesian transfer network of the roadway surrounding rock; combine the static Bayesian evaluation network of the roadway surrounding rock stability obtained in step c) and the transition probability values to obtain the dynamic Bayesian network for the stability of the roadway surrounding rock.

4. A method for evaluating, predicting and dynamically regulating the stability of roadway surrounding rock according to claim 1, characterized in that In step 9), the Golden Eagle intelligent optimization algorithm is used to search for the optimal dynamic regulation value.

Images