Intelligent optimization method and system for grouting parameters of fully weathered granite stratum

Abstract

The invention discloses a fully weathered granite stratum grouting parameter intelligent optimization method and system, and relates to the technical field of rock mass grouting. The invention discloses an intelligent optimization system for grouting parameters of a fully weathered granite stratum. The intelligent optimization system comprises an initial grouting parameter determination module and an intelligent grouting parameter optimization module, through cooperative work of the grouting reinforcement parameter pre-output model and the grouting reinforcement reaction effect prediction model, intelligent optimization of fully-weathered granite stratum grouting operation is achieved, the reaction influence features in the grouting process are accurately calculated through the steps of data preprocessing, parameter prediction, feature extraction, classified analysis and the like, and the grouting reinforcement effect prediction method is high in practicability. Grouting parameters are adjusted in real time, so that the grouting efficiency and the reinforcing effect are improved, and the safety and reliability of a project are ensured; and meanwhile, the time period of the grouting feedback parameter set X is determined through historical data optimization, and the accuracy and response speed of grouting operation are further improved.

Description

Technical Field

[0001] The invention relates to the technical field of rock mass grouting, and in particular to a method and system for intelligently optimizing grouting parameters in a fully weathered granite stratum. Background Art

[0002] Completely weathered granite strata have the characteristics of loose structure, unevenness, significantly reduced rock strength, increased permeability, low bearing capacity, and may show expansion or contraction. These characteristics make it easy to sink or crack when bearing loads, so grouting reinforcement is often required to improve its stability and bearing capacity; at the same time, during the grouting process, the slurry may diffuse unevenly and have poor permeability due to the uneven nature of the stratum, resulting in unstable reinforcement effect; in addition, chemical reactions will also directly affect the temperature, fluidity and solidification rate of the slurry. These factors determine the interaction between the slurry and the stratum and the uniformity, stability and construction safety of the reinforcement effect. Therefore, in view of the special properties of the completely weathered granite stratum, it is very important to optimize the grouting parameters. Scientific adjustment of the grouting parameters can significantly improve the reinforcement effect, enhance the long-term stability and bearing capacity of the foundation, and thus ensure the quality and safety of the project. Summary of the invention

[0003] The present invention aims to provide a method and system for intelligently optimizing grouting parameters in a fully weathered granite formation, so as to improve the grouting effect of the fully weathered granite formation.

[0004] An intelligent optimization method for grouting parameters in a fully weathered granite formation comprises the following steps: Obtain the basic parameters of the stratum to be grouted, analyze based on the basic parameters of the stratum to be grouted and the grouting reinforcement parameter pre-output model to obtain the initial grouting parameters; perform grouting operation on the fully weathered granite stratum based on the initial grouting parameters; the grouting reinforcement parameter pre-output model includes a data preprocessing layer, a grouting parameter prediction layer and a result output layer, and is constructed based on a confidence rule library; Get the grouting feedback parameter set X of the formation to be grouted, X={X i |i=1, 2, ..., I}, I is the total number of grouting feedback parameters for the stratum to be grouted in the grouting feedback parameter set X, X i Represents the grouting feedback parameter item of the stratum to be grouted; based on the grouting feedback parameter set X of the stratum to be grouted and the grouting reinforcement reaction effect prediction model, the grouting stratum reaction effect parameter is obtained by analysis; the time period for obtaining the grouting feedback parameter set X of the stratum to be grouted is optimized and determined based on historical data; The grouting reinforcement reaction effect prediction model includes a data feature extraction layer, a data feature classification analysis layer, a feature reaction prediction layer and a feedback parameter output layer, which are used to calculate the reaction influence characteristics during the grouting process and intelligently optimize the grouting parameters; The grouting formation reaction effect parameters are input into the grouting reinforcement parameter pre-output model for real-time analysis to obtain the grouting intelligent optimization parameters. The grouting operation of the fully weathered granite formation is continued based on the grouting intelligent optimization parameters.

[0005] As a preferred technical solution of the present invention, the grouting reinforcement parameter pre-output model includes a data pre-processing layer, a grouting parameter prediction layer and a result output layer; The data preprocessing layer is used to preprocess the basic parameters of the formation to be grouting, and obtain the preprocessed basic parameters of the formation to be grouting; The grouting parameter prediction layer is used to perform prediction analysis based on the basic parameters of the pre-processed grouting formation to obtain the initial grouting parameters; The grouting parameter prediction layer is constructed based on the grouting confidence rule base; The result output layer is used to output the initial grouting parameters.

[0006] As a preferred technical solution of the present invention, the basic steps of training the grouting parameter prediction layer include: Obtain the historical grouting data set of fully decomposed granite formation and the historical grouting verification set of fully decomposed granite formation; Based on the historical grouting data set of fully weathered granite strata, rule extraction is performed to obtain N stratum premise attributes S n and the corresponding grouting parameter fuzzy function P n , n=1,2,…,N; N stratum premise attributes S n and the corresponding grouting parameter fuzzy function P n Combine and obtain the grouting confidence rule base; An initial grouting parameter prediction layer is constructed based on the grouting confidence rule base; model training is performed based on the historical grouting verification set of fully weathered granite strata and the initial grouting parameter prediction layer, and the initial grouting parameter prediction layer is adjusted to obtain the initial grouting parameter prediction layer to be evaluated; model evaluation is performed on the initial grouting parameter prediction layer to be evaluated to obtain the model evaluation result of the initial grouting parameter prediction layer; if the model evaluation result of the initial grouting parameter prediction layer meets the preset output confidence level, the initial grouting parameter prediction layer to be evaluated is used as the grouting parameter prediction layer in the grouting reinforcement parameter pre-output model; otherwise, the grouting confidence rule base is reconstructed for model training.

[0007] As a preferred technical solution of the present invention, the grouting reinforcement reaction effect prediction model includes a data feature extraction layer, a data feature classification analysis layer, a feature reaction prediction layer and a feedback parameter output layer; The data feature extraction layer is used to extract the grouting feedback parameter item X of the grouting formation to be grouting i Perform feature extraction to obtain the grouting feedback parameter feature X of the grouting formation to be grouted i '; The data feature classification analysis layer is used to classify the grouting feedback parameter item feature X based on the grouting formation to be grouted. i 'Perform classification feature analysis to obtain the formation response feature vector set Y j , j = 1, 2, …, J, J is the total number of attributes affected by formation reaction characteristics; The characteristic reaction prediction layer is used to predict the formation reaction based on the characteristic vector set Y j Perform formation reaction prediction and obtain the predicted formation reaction feature vector set Y j '; The feedback parameter output layer is used to predict the formation response feature vector set Y based on all the predicted formation response feature vectors. j 'Output grouting formation reaction effect parameters.

[0008] As a preferred technical solution of the present invention, the specific steps of constructing the characteristic reaction prediction layer include: The feature response prediction layer consists of J attribute prediction layers F j , M attribute interaction prediction layers H m and prediction feature fusion layer; where m=1, 2, …, M; In the attribute prediction layer F j In the above example, the formation response feature vector set Y j Perform reaction prediction and obtain the initial formation reaction prediction characteristics T j ; In the attribute interaction prediction layer H m In the prediction of initial formation response characteristics T j and initial formation response prediction characteristics T j 'Perform interactive reaction prediction to obtain interactive formation reaction prediction characteristics W(j, j'); initial formation reaction prediction characteristics T j ' is the predicted characteristic T except for the initial formation reaction j The predicted characteristics of the initial formation response are random; the value of M is determined by the total number of (j, j'); In the prediction feature fusion layer, based on the formation reaction feature vector set Y j The corresponding interactive formation response prediction features W(j, j') are fused to obtain the predicted formation response feature vector set Y j '.

[0009] As a preferred technical solution of the present invention, the basic steps of the training data feature classification analysis layer include: The data feature classification analysis layer is constructed based on the BP neural network model and is trained using the formation feature response classification training set; the formation feature response classification training set contains several groups of formation feature response classification training samples; each group of formation feature response classification training samples contains formation parameter features and corresponding classification label values; The formation characteristic response classification training set is input into the BP neural network model for training to obtain the initial data feature classification analysis layer; the initial data feature classification analysis layer is evaluated, and if the initial data feature classification analysis layer passes the model evaluation, the initial data feature classification analysis layer is used as the data feature classification analysis layer in the grouting reinforcement reaction effect prediction model; otherwise, the formation characteristic response classification training set is used to continue model training.

[0010] As a preferred technical solution of the present invention, the specific steps of optimizing and determining the time period of the grouting feedback parameter set X of the grouting formation to be grouted based on historical data include: Construct K reaction time period individuals G k , each reaction time period individual G k Generate based on historical data range; response time period individual G k The fitness is S k ; K reaction time period individuals G k Combination, get the reaction time period iteration population; set the maximum number of iterations, Construct a numerical twin model of formation grouting response; use the reaction time period to iterate the population and obtain the fitness S k ; In the process of population iteration, a population variation mechanism is introduced; the population variation mechanism is specifically to calculate the individual G in each reaction time period in the population iteration. k The fitness of the opposing individuals, and retain the high fitness as the new reaction time cycle individual G k ; Using formula D k =L+US k Get the reaction time period individual G k The fitness of the opposing individual D k , where L is the reaction time period of individual G k The lower bound of the search space, U is the reaction time period of individual G k The upper bound of the search space; Calculate the fitness S k Specific steps: Based on the reaction time cycle individual G k Monitor the reaction changes of the numerical twin model of formation grouting reaction, and obtain several reaction change degrees; calculate the average values ​​of several reaction change degrees to obtain the reaction change mean; use the reaction change mean as the reaction time period individual G k The fitness S k ; When the maximum number of iterations is reached, the reaction time period individual corresponding to the maximum fitness is output, which is the optimal reaction time period individual; the time period for obtaining the grouting feedback parameter set X for the grouting formation to be grouted is determined based on the optimal reaction time period individual.

[0011] An intelligent optimization system for grouting parameters in fully weathered granite formations, comprising: The initial grouting parameter determination module includes a data acquisition unit and a parameter analysis unit; the data acquisition unit is used to obtain the basic parameters of the stratum to be grouted; the parameter analysis unit is used to analyze based on the basic parameters of the stratum to be grouted and the grouting reinforcement parameter pre-output model to obtain the initial grouting parameters; the grouting operation of the fully weathered granite stratum is performed based on the initial grouting parameters; the grouting reinforcement parameter pre-output model includes a data preprocessing layer, a grouting parameter prediction layer and a result output layer, and is constructed based on a confidence rule library; The grouting parameter intelligent optimization module includes a data monitoring unit, a reaction prediction unit and an intelligent optimization unit; the data monitoring unit is used to obtain the grouting feedback parameter set X of the formation to be grouted, X={X i |i=1, 2, ..., I}, I is the total number of grouting feedback parameters for the stratum to be grouted in the grouting feedback parameter set X, X i represents the grouting feedback parameter item of the stratum to be grouted; the reaction prediction unit is used to analyze based on the grouting feedback parameter set X of the stratum to be grouted and the grouting reinforcement reaction effect prediction model to obtain the grouting stratum reaction effect parameters; the time period for obtaining the grouting feedback parameter set X of the stratum to be grouted is optimized and determined based on historical data; the grouting reinforcement reaction effect prediction model includes a data feature extraction layer, a data feature classification analysis layer, a feature reaction prediction layer and a feedback parameter output layer, which are used to calculate the reaction influence characteristics during the grouting process and perform intelligent optimization of the grouting parameters; the intelligent optimization unit is used to input the grouting stratum reaction effect parameters into the grouting reinforcement parameter pre-output model for real-time analysis to obtain the grouting intelligent optimization parameters, and continue the grouting operation of the fully weathered granite stratum based on the grouting intelligent optimization parameters.

[0012] The present invention has the following advantages: 1. The present invention realizes the intelligent optimization of grouting operation in fully weathered granite strata through the collaborative work of the grouting reinforcement parameter pre-output model and the grouting reinforcement reaction effect prediction model. The process accurately calculates the reaction influence characteristics in the grouting process through the steps of data preprocessing, parameter prediction, feature extraction and classification analysis, and adjusts the grouting parameters in real time, thereby improving the grouting efficiency and reinforcement effect, and ensuring the safety and reliability of the project; at the same time, the time period of the grouting feedback parameter set X is determined by optimizing historical data, which further improves the accuracy and response speed of the grouting operation.

[0013] 2. The present invention can optimize the grouting parameters in real time and improve the grouting effect by obtaining the grouting feedback parameter set X of the formation to be grouted and analyzing it based on the grouting reinforcement reaction effect prediction model; wherein, the data feature extraction layer extracts key features from the formation grouting feedback parameters, the data feature classification analysis layer uses the BP neural network model to classify the formation reaction characteristics, the characteristic reaction prediction layer predicts the formation reaction, and the feedback parameter output layer outputs the grouting formation reaction effect parameters, which are then input into the grouting reinforcement parameter pre-output model for real-time analysis to obtain the grouting intelligent optimization parameters; this process not only improves the optimization accuracy and construction efficiency of the grouting parameters, but also enhances the construction safety, reduces resource waste, and reduces costs; at the same time, the time period for obtaining the grouting feedback parameters is optimized through historical data, which further improves the accuracy and reliability of the model and provides a scientific basis for grouting engineering decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The present invention is a schematic diagram of the structure of a fully weathered granite formation grouting parameter intelligent optimization system used in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0016] Embodiment 1, a method for intelligent optimization of grouting parameters in a fully weathered granite formation, comprising the following steps: Obtain the basic parameters of the stratum to be grouted, analyze based on the basic parameters of the stratum to be grouted and the grouting reinforcement parameter pre-output model to obtain the initial grouting parameters; perform grouting operation on the fully weathered granite stratum based on the initial grouting parameters; the grouting reinforcement parameter pre-output model includes a data preprocessing layer, a grouting parameter prediction layer and a result output layer, and is constructed based on a confidence rule library; The fully weathered granite strata have poor stability and are prone to collapse when exposed to water. The effect of using conventional materials for anti-seepage reinforcement grouting in engineering is poor. Therefore, before grouting, it is necessary to conduct a detailed analysis of the basic parameters of the strata, including the physical and mechanical properties of the strata, the development of cracks, and the self-stabilization capacity. The data preprocessing layer is used to preprocess the basic parameters of the formation to be grouting, and obtain the preprocessed basic parameters of the formation to be grouting; The grouting parameter prediction layer is used to perform prediction analysis based on the basic parameters of the pre-processed grouting formation to obtain the initial grouting parameters; The grouting parameter prediction layer is constructed based on the grouting confidence rule base; The result output layer is used to output the initial grouting parameters; The basic steps of training the grouting parameter prediction layer include: Obtain the historical grouting data set of fully decomposed granite formation and the historical grouting verification set of fully decomposed granite formation; Based on the historical grouting data set of fully weathered granite strata, rule extraction is performed to obtain N stratum premise attributes S n and the corresponding grouting parameter fuzzy function P n , n=1,2,…,N; N stratum premise attributes S n and the corresponding grouting parameter fuzzy function P n Combine and obtain the grouting confidence rule base; An initial grouting parameter prediction layer is constructed based on the grouting confidence rule base; model training is performed based on the historical grouting verification set of the fully weathered granite stratum and the initial grouting parameter prediction layer, and the initial grouting parameter prediction layer is adjusted to obtain the initial grouting parameter prediction layer to be evaluated; a model evaluation is performed on the initial grouting parameter prediction layer to be evaluated to obtain the model evaluation result of the initial grouting parameter prediction layer; if the model evaluation result of the initial grouting parameter prediction layer meets the preset output confidence level, the initial grouting parameter prediction layer to be evaluated is used as the grouting parameter prediction layer in the grouting reinforcement parameter pre-output model; otherwise, the grouting confidence rule base is reconstructed for model training; By preprocessing the basic parameters of the formation through the data preprocessing layer, key information can be extracted more accurately, providing more reliable input data for grouting parameter prediction; the grouting parameter prediction layer is constructed based on the grouting confidence rule library. This rule-based method can combine expert experience and historical data to improve the accuracy of grouting parameter prediction; the result output layer directly outputs the initial grouting parameters, which simplifies the decision-making process and improves the efficiency of grouting operations; For example, adjusting the viscosity of cement slurry in the grouting parameters can improve its diffusion law in fully weathered granite, thereby improving the grouting reinforcement effect. Different grouting materials, such as ordinary cement slurry, cement clay slurry, etc., have different application effects in fully weathered granite formations. Model prediction can optimize material selection and improve anti-seepage effects. After grouting reinforcement, the uniaxial compressive strength and shear strength of fully weathered granite formations are significantly improved, and the cohesion and internal friction angle are also improved, thereby enhancing the stability of the formation. Get the grouting feedback parameter set X of the formation to be grouted, X={X i |i=1, 2, ..., I}, I is the total number of grouting feedback parameters for the stratum to be grouted in the grouting feedback parameter set X, X i Represents the grouting feedback parameter item of the formation to be grouted; By collecting and analyzing different grouting feedback parameters, we can better understand the characteristics of the formation and its changing laws of response to grouting, so as to optimize the grouting design scheme; for example, the speed of reinforcement diffusion, the exothermic reaction parameters during the grouting process, because during the grouting process of granite formations, the grouting materials (such as cement slurry, chemical slurry, etc.) react chemically with the mineral components in the granite formation, thereby producing a series of physical and chemical changes, thereby enhancing the strength, density and impermeability of the formation; or because the chain reaction of temperature and material changes produces negative effects, the chemical reactions occurring during the grouting process (such as endothermic or exothermic reactions) will have a significant impact on the grouting process, especially in terms of temperature, grouting material properties, slurry fluidity and grouting effect; for example, gypsum grouting fluid will undergo endothermic reactions in some cases, absorbing surrounding heat. Although this helps to alleviate the temperature increase caused by the exothermic reaction, excessive heat absorption may cause the slurry to solidify more slowly, and the reaction heat absorption will reduce the temperature of the slurry, which may cause the slurry fluidity to change and affect the permeability of the grouting; Based on the grouting feedback parameter set X of the stratum to be grouted and the grouting reinforcement reaction effect prediction model, the grouting stratum reaction effect parameters are obtained; the time period for obtaining the grouting feedback parameter set X of the stratum to be grouted is optimized and determined based on historical data; The grouting reinforcement reaction effect prediction model includes a data feature extraction layer, a data feature classification analysis layer, a feature reaction prediction layer and a feedback parameter output layer, which are used to calculate the reaction influence characteristics during the grouting process and intelligently optimize the grouting parameters; The data feature extraction layer is used to extract the grouting feedback parameter item X of the grouting formation to be grouting i Perform feature extraction to obtain the grouting feedback parameter feature X of the grouting formation to be grouted i '; The data feature classification analysis layer is used to classify the grouting feedback parameter item feature X based on the grouting formation to be grouted. i 'Perform classification feature analysis to obtain the formation response feature vector set Y j , j = 1, 2, …, J, J is the total number of attributes affected by formation reaction characteristics; The characteristic reaction prediction layer is used to predict the formation reaction based on the characteristic vector set Y j Perform formation reaction prediction and obtain the predicted formation reaction feature vector set Y j '; The feedback parameter output layer is used to predict the formation response feature vector set Y based on all the predicted formation response feature vectors. j 'Output grouting formation reaction effect parameters; The data feature extraction layer can extract key features from a large number of original grouting feedback parameters, which are the basis for subsequent analysis; the data feature classification analysis layer further refines the response characteristics of the formation through classification feature analysis, and improves the pertinence and accuracy of the prediction model; the characteristic response prediction layer uses the formation response feature vector set for prediction, making the prediction results closer to the actual engineering situation; the model can dynamically adjust the grouting parameters, such as grouting pressure, grouting rate, etc., according to the actual response of the formation, to achieve the best reinforcement effect; through the grouting formation response effect parameters output by the feedback parameter output layer, engineers can monitor the grouting process in real time, adjust the strategy in time, and avoid invalid or excessive grouting; accurate prediction and timely parameter adjustment can prevent potential risks in the construction process, such as ground subsidence and cracks; The specific steps of constructing the feature response prediction layer include: The feature response prediction layer consists of J attribute prediction layers F j , M attribute interaction prediction layers H m and prediction feature fusion layer; where m=1, 2, …, M; In the attribute prediction layer F j In the above example, the formation response feature vector set Y j Perform reaction prediction and obtain the initial formation reaction prediction characteristics T j ; In the attribute interaction prediction layer H m In the prediction of initial formation response characteristics T j and initial formation response prediction characteristics T j 'Perform interactive reaction prediction to obtain interactive formation reaction prediction characteristics W(j, j'); initial formation reaction prediction characteristics T j ' is the predicted characteristic T except for the initial formation reaction j The predicted characteristics of the initial formation response are random and random; the value of M is determined by the total number of (j, j') and the total number of all possible combinations; In the prediction feature fusion layer, based on the formation reaction feature vector set Y j The corresponding interactive formation response prediction features W(j, j') are fused to obtain the predicted formation response feature vector set Y j '; The attribute prediction layer independently predicts each formation response feature, reducing the error accumulation of a single model; the attribute interaction prediction layer considers the interaction between features, captures the complexity of formation response, and further improves the accuracy of prediction; through multi-layer prediction and interactive prediction, the model can better adapt to different types of formations and different grouting conditions, enhancing the robustness and generalization ability of the model; The basic steps of the training data feature classification analysis layer include: The data feature classification analysis layer is constructed based on the BP neural network model and is trained using the formation feature response classification training set; the formation feature response classification training set contains several groups of formation feature response classification training samples; each group of formation feature response classification training samples contains formation parameter features and corresponding classification label values; Input the formation characteristic response classification training set into the BP neural network model for training to obtain the initial data characteristic classification analysis layer; perform model evaluation on the initial data characteristic classification analysis layer. If the initial data characteristic classification analysis layer passes the model evaluation, the initial data characteristic classification analysis layer is used as the data characteristic classification analysis layer in the grouting reinforcement reaction effect prediction model; otherwise, continue model training using the formation characteristic response classification training set; Through multiple iterations of training, the BP neural network model can effectively learn the complex relationship between formation characteristics and classification labels, and improve classification accuracy; the model can automatically extract and utilize key information in formation parameter characteristics, reducing the subjectivity and uncertainty of manual feature selection; fast and accurate classification results reduce the time of trial and error and speed up the construction progress; for example, dividing the classification attributes into physical attributes and chemical attributes can effectively distinguish the impact categories of various reactions in the grouting process; The specific steps of optimizing and determining the time period of the grouting feedback parameter set X of the formation to be grouted based on historical data include: Construct K reaction time period individuals G k , each reaction time period individual G k Generate based on historical data range; response time period individual G k The fitness is S k ; K reaction time period individuals G k Combination, get the reaction time period iteration population; set the maximum number of iterations, Construct a numerical twin model of formation grouting response; use the reaction time period to iterate the population and obtain the fitness S k ; In the process of population iteration, a population variation mechanism is introduced; the population variation mechanism is specifically to calculate the individual G in each reaction time period in the population iteration. k The fitness of the opposing individuals, and retain the high fitness as the new reaction time cycle individual G k ; Using formula D k =L+US k Get the reaction time period individual G k The fitness of the opposing individual D k , where L is the reaction time period of individual G k The lower bound of the search space, U is the reaction time period of individual G k The upper bound of the search space; Calculate the fitness S k Specific steps: Based on the reaction time cycle individual G k Monitor the reaction changes of the numerical twin model of formation grouting reaction, and obtain several reaction change degrees; calculate the average values ​​of several reaction change degrees to obtain the reaction change mean; use the reaction change mean as the reaction time period individual G k The fitness S k ; When the maximum number of iterations is reached, the reaction time period individual corresponding to the maximum fitness is output, which is the optimal reaction time period individual; based on the optimal reaction time period individual, the time period for obtaining the grouting feedback parameter set X for the grouting formation to be grouted is determined; By constructing the reaction time cycle individual G k Based on the historical data range, the time period for obtaining the grouting feedback parameter set X of the stratum to be grouted can be determined more accurately, thereby optimizing the grouting process. The numerical twin model of stratum grouting response, combined with population iteration and fitness calculation, can more accurately predict the grouting effect and improve the success rate of grouting reinforcement. The population variation mechanism is introduced to calculate the individual G in each reaction time period. k The fitness of the opposing individuals, and retain the high fitness as the new reaction time cycle individual G k , enhancing the adaptability and robustness of the model; by optimizing the determined time period, it can reduce the unnecessary use of grouting materials, reduce resource waste, and reduce the negative impact on the environment; the population variation mechanism increases the diversity of the population by introducing new individuals, which helps to avoid the algorithm from falling into the local optimal solution during the search process.

[0017] Example 2, a fully weathered granite formation grouting parameter intelligent optimization system, see Figure 1 As shown, including: The initial grouting parameter determination module includes a data acquisition unit and a parameter analysis unit; the data acquisition unit is used to obtain the basic parameters of the stratum to be grouted; the parameter analysis unit is used to analyze based on the basic parameters of the stratum to be grouted and the grouting reinforcement parameter pre-output model to obtain the initial grouting parameters; the grouting operation of the fully weathered granite stratum is performed based on the initial grouting parameters; the grouting reinforcement parameter pre-output model includes a data preprocessing layer, a grouting parameter prediction layer and a result output layer, and is constructed based on a confidence rule library; The grouting parameter intelligent optimization module includes a data monitoring unit, a reaction prediction unit and an intelligent optimization unit; the data monitoring unit is used to obtain the grouting feedback parameter set X of the formation to be grouted, X={X i |i=1, 2, ..., I}, I is the total number of grouting feedback parameters for the stratum to be grouted in the grouting feedback parameter set X, X irepresents the grouting feedback parameter item of the stratum to be grouted; the reaction prediction unit is used to analyze based on the grouting feedback parameter set X of the stratum to be grouted and the grouting reinforcement reaction effect prediction model to obtain the grouting stratum reaction effect parameters; the time period for obtaining the grouting feedback parameter set X of the stratum to be grouted is optimized and determined based on historical data; the grouting reinforcement reaction effect prediction model includes a data feature extraction layer, a data feature classification analysis layer, a feature reaction prediction layer and a feedback parameter output layer, which are used to calculate the reaction influence characteristics during the grouting process and perform intelligent optimization of the grouting parameters; the intelligent optimization unit is used to input the grouting stratum reaction effect parameters into the grouting reinforcement parameter pre-output model for real-time analysis to obtain the grouting intelligent optimization parameters, and continue the grouting operation of the fully weathered granite stratum based on the grouting intelligent optimization parameters.

[0018] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. An intelligent optimization method for grouting parameters in fully weathered granite formations, characterized in that: The following steps are involved: Obtaining basic parameters of the stratum to be grouting, analyzing based on the basic parameters of the stratum to be grouting and the grouting reinforcement parameter pre-output model, and obtaining initial grouting parameters; The grouting operation of the fully weathered granite formation is carried out based on the initial grouting parameters; the grouting reinforcement parameter pre-output model includes a data pre-processing layer, a grouting parameter prediction layer and a result output layer, which is constructed based on a confidence rule library; Get the grouting feedback parameter set X of the formation to be grouted, X={X i |i=1, 2, ..., I}, I is the total number of grouting feedback parameters for the stratum to be grouted in the grouting feedback parameter set X, X i Represents the grouting feedback parameter item of the formation to be grouted; Based on the grouting feedback parameter set X of the stratum to be grouted and the grouting reinforcement reaction effect prediction model, the grouting stratum reaction effect parameters are obtained; the time period for obtaining the grouting feedback parameter set X of the stratum to be grouted is optimized and determined based on historical data; The grouting reinforcement reaction effect prediction model includes a data feature extraction layer, a data feature classification analysis layer, a feature reaction prediction layer and a feedback parameter output layer, which are used to calculate the reaction influence characteristics during the grouting process and intelligently optimize the grouting parameters; The grouting formation reaction effect parameters are input into the grouting reinforcement parameter pre-output model for real-time analysis to obtain the grouting intelligent optimization parameters. The grouting operation of the fully weathered granite formation is continued based on the grouting intelligent optimization parameters.

2. The intelligent optimization method for grouting parameters in fully weathered granite strata according to claim 1 is characterized in that: The grouting reinforcement parameter pre-output model includes a data pre-processing layer, a grouting parameter prediction layer and a result output layer; The data preprocessing layer is used to preprocess the basic parameters of the formation to be grouting, and obtain the preprocessed basic parameters of the formation to be grouting; The grouting parameter prediction layer is used to perform prediction analysis based on the basic parameters of the pre-processed grouting formation to obtain the initial grouting parameters; The grouting parameter prediction layer is constructed based on the grouting confidence rule base; The result output layer is used to output the initial grouting parameters.

3. The intelligent optimization method for grouting parameters in fully weathered granite strata according to claim 2 is characterized in that: The basic steps of training the grouting parameter prediction layer include: Obtain the historical grouting data set of fully decomposed granite formation and the historical grouting verification set of fully decomposed granite formation; Based on the historical grouting data set of fully weathered granite strata, rule extraction is performed to obtain N stratum premise attributes S n and the corresponding grouting parameter fuzzy function P n , n=1,2,…,N; N stratum premise attributes S n and the corresponding grouting parameter fuzzy function P n Combine and obtain the grouting confidence rule base; An initial grouting parameter prediction layer is constructed based on the grouting confidence rule base; model training is performed based on the historical grouting verification set of fully weathered granite strata and the initial grouting parameter prediction layer, and the initial grouting parameter prediction layer is adjusted to obtain the initial grouting parameter prediction layer to be evaluated; model evaluation is performed on the initial grouting parameter prediction layer to be evaluated to obtain the model evaluation result of the initial grouting parameter prediction layer; if the model evaluation result of the initial grouting parameter prediction layer meets the preset output confidence level, the initial grouting parameter prediction layer to be evaluated is used as the grouting parameter prediction layer in the grouting reinforcement parameter pre-output model; otherwise, the grouting confidence rule base is reconstructed for model training.

4. The intelligent optimization method for grouting parameters in fully weathered granite strata according to claim 3 is characterized in that: The grouting reinforcement reaction effect prediction model includes a data feature extraction layer, a data feature classification analysis layer, a feature reaction prediction layer and a feedback parameter output layer; The data feature extraction layer is used to extract the grouting feedback parameter item X of the grouting formation to be grouting i Perform feature extraction to obtain the grouting feedback parameter feature X of the grouting formation to be grouted i '; The data feature classification analysis layer is used to classify the grouting feedback parameter item feature X based on the grouting formation to be grouted. i 'Perform classification feature analysis to obtain the formation response feature vector set Y j , j = 1, 2, …, J, J is the total number of attributes affected by formation reaction characteristics; The characteristic reaction prediction layer is used to predict the formation reaction based on the characteristic vector set Y j Perform formation reaction prediction and obtain the predicted formation reaction feature vector set Y j '; The feedback parameter output layer is used to predict the formation response feature vector set Y based on all the predicted formation response feature vectors. j 'Output grouting formation reaction effect parameters.

5. The intelligent optimization method for grouting parameters in fully weathered granite strata according to claim 4 is characterized in that: The specific steps of constructing the feature response prediction layer include: The feature response prediction layer consists of J attribute prediction layers F j , M attribute interaction prediction layers H m and prediction feature fusion layer; where m=1, 2, …, M; In the attribute prediction layer F j In the above example, the formation response feature vector set Y j Perform reaction prediction and obtain the initial formation reaction prediction characteristics T j ; In the attribute interaction prediction layer H m In the prediction of initial formation response characteristics T j and initial formation response prediction characteristics T j 'Perform interactive reaction prediction to obtain interactive formation reaction prediction characteristics W(j, j'); initial formation reaction prediction characteristics T j ' is the predicted characteristic T except for the initial formation reaction j The predicted characteristics of the initial formation response are random; the value of M is determined by the total number of (j, j'); In the prediction feature fusion layer, based on the formation reaction feature vector set Y j The corresponding interactive formation response prediction features W(j, j') are fused to obtain the predicted formation response feature vector set Y j '.

6. The method for intelligent optimization of grouting parameters in fully weathered granite strata according to claim 5, characterized in that: The basic steps of the training data feature classification analysis layer include: The data feature classification analysis layer is constructed based on the BP neural network model and is trained using the formation feature response classification training set; the formation feature response classification training set contains several groups of formation feature response classification training samples; each group of formation feature response classification training samples contains formation parameter features and corresponding classification label values; The formation characteristic response classification training set is input into the BP neural network model for training to obtain the initial data feature classification analysis layer; the initial data feature classification analysis layer is evaluated, and if the initial data feature classification analysis layer passes the model evaluation, the initial data feature classification analysis layer is used as the data feature classification analysis layer in the grouting reinforcement reaction effect prediction model; otherwise, the formation characteristic response classification training set is used to continue model training.

7. The method for intelligent optimization of grouting parameters in fully weathered granite strata according to claim 6, characterized in that: The specific steps of optimizing and determining the time period of the grouting feedback parameter set X of the formation to be grouted based on historical data include: Construct K reaction time period individuals G k , each reaction time period individual G k Generate based on historical data range; response time period individual G k The fitness is S k ; K reaction time period individuals G k Combination, get the reaction time period iteration population; set the maximum number of iterations, Construct a numerical twin model of formation grouting response; use the reaction time period to iterate the population and obtain the fitness S k ; In the process of population iteration, a population variation mechanism is introduced; the population variation mechanism is specifically to calculate the individual G in each reaction time period in the population iteration. k The fitness of the opposing individuals, and retain the high fitness as the new reaction time cycle individual G k ; Using formula D k =L+US k Get the reaction time period individual G k The fitness of the opposing individual D k , where L is the reaction time period of individual G k The lower bound of the search space, U is the reaction time period of individual G k The upper bound of the search space; Calculate the fitness S k Specific steps: Based on the reaction time cycle individual G k Monitor the reaction changes of the numerical twin model of formation grouting reaction, and obtain several reaction change degrees; calculate the average values ​​of several reaction change degrees to obtain the reaction change mean; use the reaction change mean as the reaction time period individual G k The fitness S k ; When the maximum number of iterations is reached, the reaction time period individual corresponding to the maximum fitness is output, which is the optimal reaction time period individual; the time period for obtaining the grouting feedback parameter set X for the grouting formation to be grouted is determined based on the optimal reaction time period individual.

8. An intelligent optimization system for grouting parameters in fully weathered granite formations, characterized in that: The system applies the intelligent optimization method for grouting parameters in fully weathered granite formations as described in any one of claims 1 to 7, including: The initial grouting parameter determination module includes a data acquisition unit and a parameter analysis unit; the data acquisition unit is used to obtain the basic parameters of the stratum to be grouted; the parameter analysis unit is used to analyze based on the basic parameters of the stratum to be grouted and the grouting reinforcement parameter pre-output model to obtain the initial grouting parameters; the grouting operation of the fully weathered granite stratum is performed based on the initial grouting parameters; the grouting reinforcement parameter pre-output model includes a data preprocessing layer, a grouting parameter prediction layer and a result output layer, and is constructed based on a confidence rule library; The grouting parameter intelligent optimization module includes a data monitoring unit, a reaction prediction unit and an intelligent optimization unit; the data monitoring unit is used to obtain the grouting feedback parameter set X of the formation to be grouted, X={X i |i=1, 2, ..., I}, I is the total number of grouting feedback parameters for the stratum to be grouted in the grouting feedback parameter set X, X i represents the grouting feedback parameter item of the stratum to be grouted; the reaction prediction unit is used to analyze based on the grouting feedback parameter set X of the stratum to be grouted and the grouting reinforcement reaction effect prediction model to obtain the grouting stratum reaction effect parameters; the time period for obtaining the grouting feedback parameter set X of the stratum to be grouted is optimized and determined based on historical data; the grouting reinforcement reaction effect prediction model includes a data feature extraction layer, a data feature classification analysis layer, a feature reaction prediction layer and a feedback parameter output layer, which are used to calculate the reaction influence characteristics during the grouting process and perform intelligent optimization of the grouting parameters; the intelligent optimization unit is used to input the grouting stratum reaction effect parameters into the grouting reinforcement parameter pre-output model for real-time analysis to obtain the grouting intelligent optimization parameters, and continue the grouting operation of the fully weathered granite stratum based on the grouting intelligent optimization parameters.

Images