A geotechnical engineering model optimization training method and system based on a multi-source knowledge base

Abstract

The application discloses a kind of geotechnical engineering model optimization training method and system based on multi-source knowledge base, comprising: constructing geotechnical engineering multi-source knowledge base;Real-time acquisition of field sensor monitoring data and preprocessing obtain standardized time series monitoring data;Adopt Mamba model to extract time series characteristics to generate time series state feature dataset for standardized data;Based on knowledge base, generate initial computable prior information and parameter constraint condition of geotechnical engineering model and establish initial particle state set;Particle flow filtering algorithm is used to iteratively update particle state, obtain geotechnical engineering model parameter posterior distribution;Carry out model numerical simulation and calculate simulation error evaluation index;According to error evaluation index, judge whether posterior distribution reaches predetermined optimization convergence condition and output optimized training geotechnical engineering model.The application effectively improves the accuracy and convergence stability of geotechnical engineering model parameter optimization.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of model optimization training, and particularly relates to a geotechnical engineering model optimization training method and system based on a multi-source knowledge base. BACKGROUND

[0002] Geotechnical engineering numerical modeling and parameter optimization technology is one of the important research contents in the field of modern geotechnical engineering. The geotechnical engineering model is usually based on field monitoring data, geological survey data and historical engineering cases, and the mechanical properties and evolution process of complex rock-soil mass are analyzed and predicted by numerical simulation method to support the design and decision of actual engineering. With the development of monitoring technology and data analysis method, in recent years, the geotechnical engineering model optimization training based on multi-source knowledge fusion and data driving has been widely applied.

[0003] At present, the mainstream technical route of geotechnical engineering model parameter optimization includes inversion analysis method and data assimilation method. The inversion analysis method depends on the explicit mathematical model, and the engineering parameters are inversely calculated by minimizing the error between the monitoring data and the numerical model. The implementation process needs to be highly dependent on the model and data accuracy, and the parameter convergence performance is difficult to guarantee, which is easy to fall into local optimum. The particle filter algorithm, which is a representative of the data assimilation method, can estimate the probability distribution of parameters through the particle set in the state space, and iteratively optimize in real time by fusing observation data. However, in the application process of the existing particle filter algorithm in the field of geotechnical engineering, the multi-source geological data, geophysical data, historical engineering experience and specification requirements and other knowledge information have not been fully and effectively utilized as prior constraints of model parameters, resulting in insufficient data utilization and poor convergence stability in the parameter optimization process. At the same time, in the complex geotechnical environment, there are missing and abnormal situations in the monitoring data, and the current data processing method cannot effectively balance the data quality and model parameter updating accuracy requirements, so that the accuracy and reliability of the parameter optimization result are difficult to guarantee.

[0004] Therefore, how to provide a geotechnical engineering model optimization training method and system based on a multi-source knowledge base is a problem that needs to be solved by those skilled in the art. SUMMARY

[0005] One object of the present application is to provide a geotechnical engineering model optimization training method and system based on a multi-source knowledge base. In view of the problems of difficult multi-source knowledge unified modeling, insufficient parameter optimization accuracy and convergence stability in the existing geotechnical engineering model optimization process, the technical scheme of constructing a multi-source knowledge base of geotechnical engineering, real-time acquisition and preprocessing of sensor monitoring data, extracting time series features by Mamba model, and iterative optimization by particle flow filtering algorithm is proposed. The present application has the advantages of high model parameter optimization accuracy and strong convergence stability.

[0006] According to the geotechnical engineering model optimization training method based on the multi-source knowledge base, the method comprises the following steps:

[0007] Constructing a multi-source knowledge base of geotechnical engineering;

[0008] Real-time collection and preprocessing of sensor monitoring data of a geotechnical engineering site to obtain standardized time series monitoring data;

[0009] Time series feature extraction of the standardized time series monitoring data by using the Mamba model to generate a time series state feature data set;

[0010] Generating initial computable prior information and parameter constraints of the geotechnical engineering model according to the multi-source knowledge base of geotechnical engineering, and combining the time series state feature data set to establish an initial particle state set;

[0011] Iterative updating of the initial particle state set by using a particle flow filtering algorithm to obtain an updated posterior distribution of the geotechnical engineering model parameters;

[0012] Numerical simulation of the geotechnical engineering model based on the updated posterior distribution of the geotechnical engineering model parameters, and calculation of an error evaluation index;

[0013] Judging whether the posterior distribution of the geotechnical engineering model parameters reaches a predetermined optimization convergence condition based on the error evaluation index, and obtaining a final optimized and trained geotechnical engineering model when the predetermined optimization convergence condition is met.

[0014] Optionally, the multi-source knowledge base of geotechnical engineering is constructed, specifically as follows:

[0015] Dynamic confidence evaluation of geological survey data, geophysical survey data, historical engineering case data, construction log data, test data and specification experience data, and automatic elimination of low-confidence data according to a confidence threshold to obtain a high-confidence filtered data set;

[0016] Semantic analysis of the high-confidence filtered data set to obtain a semantic constraint mapping table;

[0017] Dynamic standardization processing of the high-confidence filtered data set according to the semantic constraint mapping table to generate a standardized data set;

[0018] Construction of a space-time topology index library by using the spatial position, time stamp and semantic label of each data record in the standardized data set;

[0019] Multi-layer space-time association processing of the historical engineering case data and the construction log data according to the space-time topology index library to obtain a multi-source dynamic association data set;

[0020] Unified structure integration of the standardized data set and the multi-source dynamic association data set to obtain the multi-source knowledge base of geotechnical engineering.

[0021] Optionally, the real-time acquisition of the geotechnical engineering site sensor monitoring data and preprocessing, obtain standardized time series monitoring data, specifically:

[0022] The data integrity of the real-time acquisition of the sensor monitoring data is checked, and the statistical distribution characteristics of the time series data before and after the data are used to interpolate and complete the missing data, and the sensor monitoring data after the missing data completion is obtained;

[0023] The sensor monitoring data after the missing data completion is subjected to abnormal data identification and elimination, and the sensor monitoring data after the abnormal data elimination is obtained;

[0024] The semantic constraint mapping table is used to map each data record of the sensor monitoring data after the abnormal data elimination to the predetermined standard data field, and the semantic standardized sensor monitoring data is obtained;

[0025] The semantic standardized sensor monitoring data is subjected to data smoothing processing, and the denoised sensor monitoring data is obtained;

[0026] The denoised sensor monitoring data is subjected to time series data resampling, and the sensor monitoring data with uniform time series sampling frequency is obtained;

[0027] The sensor monitoring data with uniform time series sampling frequency is subjected to numerical range standardization operation, and each value in the data is converted to a uniform standard numerical interval, and the standardized time series monitoring data is generated.

[0028] Optionally, the Mamba model is used to extract the time series characteristics of the standardized time series monitoring data, and the time series state feature data set is generated, specifically:

[0029] Based on the standardized time series monitoring data, a multi-scale time series sequence group is constructed, and the adaptive scale learning mechanism built-in the Mamba model is used to determine the optimal convolution scale of each time series sequence, and the adaptive scale feature data containing the optimal scale information is generated;

[0030] The adaptive scale feature data is subjected to dynamic state space mapping using the state space conversion mechanism in the Mamba model, and the time series feature representation is converted into dynamic state space feature data;

[0031] According to the historical case time series feature mode library in the geotechnical engineering multi-source knowledge base, the dynamic state space feature data is subjected to time series feature mode similarity matching, the matching degree between the current monitoring data feature and the historical feature mode is calculated, and the state feature data is obtained;

[0032] The time sequence dynamic sparse attention mechanism is adopted for the state feature data, key state features with discriminability in the geotechnical engineering evolution process are highlighted, redundant and low correlation features are suppressed, and sparse optimized key state feature data is generated;

[0033] The importance of the sparse optimized key state feature data is scored by using the interpretable time sequence state feature screening mechanism, a few key feature dimensions with high importance in both historical time sequence pattern matching and dynamic state space mapping are selected for reservation, and time sequence state principal component feature data is obtained;

[0034] The geotechnical engineering time sequence state feature data set is constructed based on the time sequence state principal component feature data.

[0035] Optionally, the initial computable prior information and parameter constraint conditions of the geotechnical engineering model are generated according to the geotechnical engineering multi-source knowledge base, and an initial particle state set is established in combination with the time sequence state feature data set, and specifically:

[0036] The experience value range of the initial structure and parameters of the geotechnical engineering model is determined according to the historical engineering case data and test data in the geotechnical engineering multi-source knowledge base, and an experience constraint set of parameters is formed;

[0037] The spatial parameter partition of the geotechnical engineering model is established according to the spatial distribution characteristics and geological attribute classification information in the geophysical exploration data and geological survey data, and initial parameter spatial partition information is obtained;

[0038] The initial physical and mechanical constraint conditions of the geotechnical engineering model are extracted according to the specification experience data and construction log data, and a physical constraint set containing initial stress boundary conditions and mechanical parameter constraints is obtained;

[0039] The experience constraint set of parameters, the initial parameter spatial partition information and the physical constraint set are fused to form the initial computable prior information and the parameter constraint conditions;

[0040] The principal component feature dimensions in the geotechnical engineering time sequence state feature data set are classified and mapped according to the spatial position and parameter type, and a time sequence feature mapping set is obtained;

[0041] According to the initial computable prior information and parameter constraint conditions of the unified structure of the geotechnical engineering model and in combination with the time sequence feature mapping set, an initial particle state set is generated by using random sampling.

[0042] Optionally, the initial particle state set is iteratively updated by using a particle flow filtering algorithm to obtain an updated geotechnical engineering model parameter posterior distribution, and specifically:

[0043] The initial particle state distribution of the particle flow filtering algorithm is determined according to the initial particle state set, and the statistical characteristic parameters of the initial particle state distribution are calculated.

[0044] Collecting the latest standardized time series monitoring data, and determining the mapping relationship between the real-time observation data and the particle state parameters according to the time series feature mapping set;

[0045] Calculating the particle flow path between the particle state parameters from the initial state distribution to the posterior state distribution under the condition of real-time observation data by using the particle flow filtering algorithm, and determining the change of the weight of each particle on the particle flow path;

[0046] Updating the numerical value of each particle state parameter through the particle flow path and the change of the particle weight, and performing boundary check and constraint adjustment on the updated particle state parameter according to the parameter constraint condition to generate the corrected particle state parameter meeting the constraint condition;

[0047] Calculating the statistical characteristic parameters of the updated particle state set by using the corrected particle state parameter, and estimating the geotechnical engineering model parameter posterior distribution according to the statistical characteristic parameters;

[0048] Judging whether the geotechnical engineering model parameter posterior distribution is stable and convergent, if not, re-calculating the particle flow path to continue iteration and updating until the stable and convergent geotechnical engineering model parameter posterior distribution is obtained.

[0049] Optionally, the geotechnical engineering model numerical simulation is performed based on the updated geotechnical engineering model parameter posterior distribution, and the error evaluation index is calculated, specifically:

[0050] According to the updated geotechnical engineering model parameter posterior distribution, the parameters of the geotechnical engineering model are valued, the parameters of the geotechnical engineering model are one by one corresponding to the geotechnical engineering model space grid, and the geotechnical engineering model space grid data after parameter valuation is obtained;

[0051] According to the initial physical and mechanical constraint conditions of the geotechnical engineering model, the initial boundary conditions of the geotechnical engineering model are determined, and the initial boundary conditions are applied to the geotechnical engineering model space grid data after parameter valuation;

[0052] The geotechnical engineering model space grid data after parameter valuation and application of the initial boundary conditions is subjected to numerical simulation calculation, and the model response result is obtained;

[0053] According to the model response result, the simulation calculation values at the corresponding position and time are extracted, and the simulation data sequence which can be directly compared with the actual monitoring data is formed;

[0054] The numerical difference between the simulation data sequence and the actual monitoring data at each monitoring point is calculated, and the difference sequence at the corresponding time of each monitoring point is obtained;

[0055] Calculate the arithmetic mean of the absolute value of the difference, the variance of the difference sequence, and the root mean square error of the difference sequence for the difference sequence of all monitoring points, and record them as error evaluation indicators of the model simulation results.

[0056] Optionally, the method further comprises:

[0057] According to the error evaluation indicators of the numerical simulation results of the geotechnical engineering model, the error absolute value, variance, and root mean square error of each monitoring point and time sequence are comprehensively scored, and an error comprehensive score sequence corresponding to each iteration is generated.

[0058] In the error comprehensive score sequence, the change rate of the error comprehensive scores of adjacent two iterations is calculated according to the time sequence index, and an error convergence trend curve is obtained.

[0059] According to the preset optimization convergence threshold and the error convergence trend curve, the difference between the current iteration error comprehensive score and the threshold is calculated, and it is judged whether the error convergence trend curve remains below the threshold fluctuation range at consecutive multiple time indexes.

[0060] When the error convergence trend curve remains below the threshold fluctuation range for a continuous preset number of iterations, the current geotechnical engineering model parameter posterior distribution is determined to satisfy the predetermined optimization convergence condition, and the geotechnical engineering model parameter posterior distribution of the current iteration is output as the final optimized and trained geotechnical engineering model parameter posterior distribution.

[0061] When the error convergence trend curve does not reach the predetermined optimization convergence condition, the error comprehensive score of the current iteration is recorded and the particle flow filtering algorithm is executed to continue the iteration update of the parameter posterior distribution until the predetermined optimization convergence condition is satisfied.

[0062] A geotechnical engineering model optimization training system based on a multi-source knowledge base, comprising:

[0063] A multi-source knowledge base construction module for constructing a multi-source knowledge base of geotechnical engineering.

[0064] A sensor data acquisition and preprocessing module for acquiring sensor monitoring data of a geotechnical engineering site and preprocessing the data to obtain standardized time series monitoring data.

[0065] A time series feature extraction module for extracting time series features from the standardized time series monitoring data using a Mamba model to generate a time series state feature dataset.

[0066] The initial particle state set generation module is used to generate the initial computable prior information and parameter constraints of the geotechnical engineering model based on the geotechnical engineering multi-source knowledge base, and to establish the initial particle state set.

[0067] The particle flow filtering iterative update module is used to iteratively update the initial particle state set using the particle flow filtering algorithm to obtain the updated posterior distribution of geotechnical engineering model parameters.

[0068] The geotechnical engineering model numerical simulation module is used to perform numerical simulations of geotechnical engineering models and calculate error evaluation indicators.

[0069] The optimization convergence judgment module is used to determine whether the posterior distribution of geotechnical engineering model parameters has reached the predetermined optimization convergence condition. If the predetermined optimization convergence condition is met, the final optimized training posterior distribution of geotechnical engineering model parameters is output.

[0070] The beneficial effects of this invention are:

[0071] (1) By constructing a multi-source knowledge base for geotechnical engineering and using the Mamba model to extract features from time-series monitoring data, this invention effectively realizes the unified modeling and full utilization of multi-source knowledge in the field of geotechnical engineering, effectively improves the accuracy and stability of parameter estimation in the process of geotechnical engineering model optimization, and enhances the reliability and accuracy of model optimization training.

[0072] (2) This invention achieves refined optimization of geotechnical engineering model parameters by iteratively updating particle state parameters and adjusting constraints through particle flow filtering algorithm, which significantly improves the parameter convergence and prediction accuracy of geotechnical engineering model and shows better adaptability in complex geotechnical engineering practical application scenarios.

[0073] (3) In terms of geotechnical engineering model parameter optimization and data processing, this invention effectively solves the shortcomings of low data quality and insufficient convergence in the prior art through real-time data preprocessing, comprehensive evaluation of error trends and optimization convergence judgment mechanism. It breaks through the stability and accuracy bottleneck of particle filter algorithm in geotechnical engineering application, realizes specific and significant improvement of model optimization reliability and stability, and effectively improves the engineering applicability of geotechnical engineering model optimization training method. Attached Figure Description

[0074] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0075] Figure 1 This is a flowchart of a geotechnical engineering model optimization training method based on a multi-source knowledge base proposed in this invention. Detailed Implementation

[0076] The application will now be described in further detail with reference to the drawings. These drawings are simplified schematic illustrations of the basic structure of the application and therefore only show the components relevant to the application.

[0077] Reference Figure 1 A geotechnical engineering model optimization training method based on a multi-source knowledge base, comprising:

[0078] Constructing a geotechnical engineering multi-source knowledge base;

[0079] Real-time collection of sensor monitoring data of a geotechnical engineering site and preprocessing to obtain standardized time series monitoring data;

[0080] Using a Mamba model to extract time series features from the standardized time series monitoring data to generate a time series state feature dataset;

[0081] Generating initial computable prior information and parameter constraints of the geotechnical engineering model according to the geotechnical engineering multi-source knowledge base, and combining the time series state feature dataset to establish an initial particle state set for optimization training of the geotechnical engineering model;

[0082] Using a particle flow filtering algorithm to iteratively update the initial particle state set to obtain an updated geotechnical engineering model parameter posterior distribution;

[0083] Based on the updated geotechnical engineering model parameter posterior distribution, performing numerical simulation of the geotechnical engineering model to obtain a numerical simulation result of the geotechnical engineering model, and calculating an error evaluation index between the numerical simulation result of the geotechnical engineering model and the actual monitoring data;

[0084] Based on the error evaluation index, determining whether the geotechnical engineering model parameter posterior distribution meets a predetermined optimization convergence condition, and obtaining a final optimized and trained geotechnical engineering model when the predetermined optimization convergence condition is met.

[0085] In this embodiment, the geotechnical engineering multi-source knowledge base is constructed, specifically:

[0086] Performing dynamic confidence evaluation on geological survey data, geophysical survey data, historical engineering case data, construction log data, test data, and specification experience data, and automatically eliminating low-confidence data according to a confidence threshold to obtain a high-confidence filtered data set;

[0087] The dynamic confidence evaluation is specifically:

[0088] After obtaining geological survey data, geophysical survey data, historical engineering case data, construction log data, test data and specification experience data, multi-dimensional quality analysis is performed on the collection time, collection frequency, collection location, collection instrument type, historical reliability record, spatial correlation with similar data and time continuity of each data source, and a comprehensive confidence score is calculated for each data record according to a preset confidence evaluation rule, including a time freshness factor, a source reliability factor, a spatial consistency factor and a cross-validation factor, wherein the time freshness factor is used to reflect the closeness of the data to the current time, the source reliability factor is used to reflect the reliability of the collection instrument, the collection unit and the data transmission process, the spatial consistency factor is used to measure the matching degree of the data in space with adjacent data sets, and the cross-validation factor is used to evaluate the mutual consistency between the data and other data sources;

[0089] The factors are combined according to the predetermined weight to form a comprehensive confidence score matrix, the data records below the confidence threshold are marked and removed, and a high-confidence filtered data set containing only data with a confidence score higher than the threshold is obtained;

[0090] Semantic analysis is performed on the high-confidence filtered data set to obtain a semantic constraint mapping table;

[0091] The high-confidence filtered data set is dynamically standardized according to the semantic constraint mapping table to generate a standardized data set;

[0092] The dynamic standardization processing is specifically:

[0093] According to the field name, dimension definition and value range in the semantic constraint mapping table, each data record in the high-confidence filtered data set is converted by numerical standardization one by one, the mapping relationship between the original value of the data record and the predetermined standard value range of the corresponding field in the semantic constraint mapping table is calculated, the standardization mapping coefficient is dynamically adjusted based on the confidence score attached to the data record, and the original value of each data record is dynamically standardized and mapped into a unified standard value interval to obtain a standardized data set;

[0094] A spatio-temporal topology index library is constructed using the spatial position, timestamp and semantic label of each data record in the standardized data set;

[0095] According to the spatio-temporal topology index library, multi-layer spatio-temporal association processing is performed on the historical engineering case data and the construction log data to obtain a multi-source dynamic association data set;

[0096] The multi-layer spatio-temporal association processing is specifically:

[0097] According to the spatial position, the time stamp and the semantic label information recorded in the historical engineering case data and the construction log data in the space-time topology index library, the historical engineering case data and the construction log data are associated and matched in the spatial dimension according to the spatial position, and data records in the same or adjacent spatial units are determined;

[0098] The data records matched in the space are associated in the time dimension, time sequence association links are established according to the collection time of each data record, and a double-layer association structure with the spatial position as the first layer and the time sequence as the second layer is formed;

[0099] The data in the association structure is further subdivided in the semantic dimension according to the semantic label of the data record, a third layer of association in the semantic dimension is established, and a multi-source dynamic association data set is formed;

[0100] The standardized data set and the multi-source dynamic association data set are integrated in a unified structure, and a geotechnical engineering multi-source knowledge base is obtained.

[0101] In the embodiment, the sensor monitoring data of the geotechnical engineering site is collected in real time and preprocessed to obtain standardized time sequence monitoring data, specifically:

[0102] The real-time collected sensor monitoring data is subjected to data integrity check, and the missing data is interpolated and completed by using the statistical distribution characteristics of the data before and after the time sequence, to obtain sensor monitoring data after completion of missing data;

[0103] The sensor monitoring data after completion of missing data is subjected to abnormal data identification and elimination, to obtain sensor monitoring data after elimination of abnormal data;

[0104] The sensor monitoring data after elimination of abnormal data is mapped to the predetermined standard data field by using a semantic constraint mapping table, to obtain sensor monitoring data after semantic standardization;

[0105] The sensor monitoring data after semantic standardization is subjected to data smoothing processing, to obtain denoised sensor monitoring data;

[0106] The denoised sensor monitoring data is subjected to time sequence data resampling, to obtain sensor monitoring data with uniform time sequence sampling frequency;

[0107] The sensor monitoring data with uniform time sequence sampling frequency is subjected to numerical range standardization operation, each value in the data is converted in the numerical range according to a uniform standard numerical interval, and standardized time sequence monitoring data is generated.

[0108] In the embodiment, the Mamba model is used to extract time sequence characteristics from the standardized time sequence monitoring data, to generate a time sequence state feature data set, specifically:

[0109] Based on the standardized time series monitoring data, a multi-scale time series sequence group is constructed, and the adaptive scale learning mechanism built-in the Mamba model is used to determine the optimal convolution scale of each time series sequence, and adaptive scale feature data containing optimal scale information is generated;

[0110] After constructing a multi-scale time series sequence group based on standardized time series monitoring data, each time series sequence is input into the pre-defined multi-scale convolution kernel group in the Mamba model, different convolution kernel groups correspond to different receptive fields and time windows, and trainable scale weight parameters are established for each convolution kernel group in the adaptive scale learning unit of the Mamba model;

[0111] The convolution response of each convolution kernel group to the input time series sequence is calculated using the forward propagation process, and the scale weight parameters are dynamically adjusted based on the response through the back propagation algorithm and gradient optimization method, so that each time series sequence automatically selects the convolution kernel group with the highest feature extraction score in the multi-scale convolution kernel group;

[0112] The convolution kernel group parameters corresponding to the highest feature extraction score obtained by each time series sequence in the training iteration process are marked as the optimal convolution scale of the time series sequence, and finally the optimal convolution scale index corresponding to the convolution feature response is output for each time series sequence, forming adaptive scale feature data containing optimal scale information;

[0113] The adaptive scale feature data is mapped to the dynamic state space using the state space conversion mechanism in the Mamba model, and the time series feature representation is converted to dynamic state space feature data that can reflect the dynamic change characteristics of the geotechnical engineering physical process;

[0114] After obtaining adaptive scale feature data containing optimal convolution scale information, the feature data is input into the state space conversion unit built-in the Mamba model, and each time series feature vector is jointly encoded with its time index, spatial index and historical feature mode parameters according to the pre-set dynamic state variable set, forming an initial state representation containing input state, output state and state transition factor;

[0115] The Mamba model dynamically learns the transition rules between state variables, input response rules and output observation rules in the training stage through trainable state transition matrix, input mapping matrix and output mapping matrix, and obtains state space parameters describing the evolution of time series features over time;

[0116] The learned state transition matrix and input-output mapping matrix are applied to the adaptive scale feature data, the state of each time step is updated, and the state output is calculated to generate dynamic state space feature data that can reflect the dynamic change characteristics of the geotechnical engineering physical process, and the corresponding state variable, state transition parameter and time index are recorded in the output to form a complete and traceable dynamic state space feature sequence;

[0117] According to the historical case time sequence feature mode library in the geotechnical engineering multi-source knowledge base, the dynamic state space feature data is matched in time sequence feature mode, the matching degree between the current monitoring data feature and the historical feature mode is calculated, and the state feature data with historical time sequence feature mode matching weight is obtained;

[0118] After obtaining the dynamic state space feature data, the state space feature vectors of each time step are combined into fixed-length state sequence fragments in chronological order, and the key state variables, state transition parameters and time indexes of each state sequence fragment are extracted to form a matching sequence;

[0119] The historical time sequence feature modes stored in the historical case time sequence feature mode library in the geotechnical engineering multi-source knowledge base are constructed into standard state sequence templates according to the same state variables and state transition parameters;

[0120] By calculating the Euclidean distance, dynamic time warping distance and state transition matrix difference degree of the matching sequence and the standard state sequence template in the state variable value, state transition parameter and time index, the multi-index similarity score between each matching sequence and all standard state sequence templates is obtained;

[0121] The multi-index similarity score is integrated according to the preset weight to form a comprehensive matching degree score matrix, and the optimal matching historical time sequence feature mode template number and the corresponding comprehensive matching degree score are labeled for each time step in the score matrix, and the state feature data with historical time sequence feature mode matching weight is output;

[0122] The state feature data with historical time sequence feature mode matching weight is processed by a time sequence dynamic sparse attention mechanism to highlight the key state features that are discriminative in the geotechnical engineering evolution process and suppress redundant and low-correlation features, thereby generating sparse optimized key state feature data;

[0123] After obtaining the state feature data with historical time sequence feature mode matching weight, the feature tensors are constructed according to the time steps and state variable dimensions, and the initial attention weight matrix is generated based on the historical time sequence feature mode matching weight;

[0124] The initial attention weight matrix is dynamically updated by using a time sequence dynamic sparse attention mechanism, and the importance of each state variable in the feature tensor in the time sequence is weighted by introducing a sparse regularization constraint in the attention calculation process, to obtain a sparse attention weight distribution;

[0125] The sparse attention weight distribution is multiplied element by element with the feature tensor to obtain a weighted feature tensor after attention sparsification;

[0126] All state variable features in the weighted feature tensor are accumulated and summed along the time axis direction, and a preset discriminative threshold is used to screen out key state variable features that have a greater contribution to state changes in the geotechnical engineering evolution process, and to suppress redundant or low-correlation state variable features with a weight lower than the discriminative threshold, to generate sparse optimized key state feature data;

[0127] The sparse optimized key state feature data is scored for importance by using an interpretable time sequence state feature screening mechanism, a few key feature dimensions with high importance in both historical time sequence pattern matching and dynamic state space mapping are selected for retention, to obtain time sequence state principal component feature data;

[0128] After obtaining the sparse optimized key state feature data, a feature importance evaluation function is established according to the historical time sequence feature pattern matching weight and the state transition matrix parameters of the state variable in the dynamic state space mapping process;

[0129] The contribution score of each key state feature variable in the historical time sequence pattern matching and the sensitivity score in the dynamic state space feature mapping are calculated by using the feature importance evaluation function respectively;

[0130] The contribution score and the sensitivity score are weighted and fused according to a predetermined weight to form a comprehensive importance score index;

[0131] All key state feature variables are sorted according to the comprehensive importance score index, and a few feature variables with a higher comprehensive importance score are selected as interpretable key feature dimensions according to an importance score preset threshold, and time sequence state principal component feature data containing the key feature dimensions and corresponding importance scores are outputted;

[0132] A geotechnical engineering time sequence state feature data set is constructed based on the time sequence state principal component feature data;

[0133] After obtaining the time sequence state principal component feature data, the feature values of each key feature dimension in the data are restructured in a unified data storage format, and corresponding space-time indexes are generated based on the time stamp and spatial position information of the feature data;

[0134] The reorganized feature data is mapped into the spatial coordinate system and the time coordinate system of the geotechnical engineering model by using the space-time index, and the correspondence between the feature data and the actual monitoring points and monitoring time of the geotechnical engineering site is established;

[0135] According to the established space-time mapping relationship, the data of all key feature dimensions in each space-time unit are integrated to generate a time sequence state feature record with the monitoring point as the basic unit and the time sequence as the sorting;

[0136] All the time sequence state feature records are uniformly numbered and labeled to form a geotechnical engineering time sequence state feature dataset with a unified data structure and a clear space-time index.

[0137] In the embodiment, the initial computable prior information and parameter constraint conditions of the geotechnical engineering model are generated according to the geotechnical engineering multi-source knowledge base, and an initial particle state set is established in combination with the time sequence state feature dataset, specifically:

[0138] The experience value range of the initial structure and parameters of the geotechnical engineering model is determined according to the historical engineering case data and test data in the geotechnical engineering multi-source knowledge base to form a parameter experience constraint set;

[0139] The spatial parameter partition of the geotechnical engineering model is established according to the spatial distribution characteristics and geological attribute classification information in the geophysical exploration data and geological survey data to obtain initial parameter spatial partition information;

[0140] The initial physical and mechanical constraint conditions of the geotechnical engineering model are extracted according to the specification experience data and construction log data to obtain a physical constraint set containing initial stress boundary conditions and mechanical parameter constraints;

[0141] The parameter experience constraint set, the initial parameter spatial partition information and the physical constraint set are fused to form the initial computable prior information and the parameter constraint conditions;

[0142] After obtaining the parameter experience constraint set, the initial parameter spatial partition information and the physical constraint set, the parameter experience value range in the parameter experience constraint set is corresponded to a specific spatial partition according to the initial parameter spatial partition information to form the parameter experience constraint corresponding to the spatial partition;

[0143] The physical constraint condition set of each spatial partition is determined according to the initial stress boundary conditions and the mechanical parameter constraints in the physical constraint set;

[0144] The parameter experience constraint corresponding to each spatial partition and the physical constraint condition set are checked for compatibility, and the consistency between the parameter experience value range and the physical and mechanical constraints in each partition is determined through the checking, the conflicting constraint conditions are excluded, and the constraint conditions with high consistency are retained;

[0145] The parameter empirical constraint, the spatial partition information and the physical constraint after the compatibility check are uniformly expressed and structurally integrated to form unified initial computable prior information and parameter constraint conditions;

[0146] The principal component feature dimensions in the geotechnical engineering time sequence state feature dataset are classified and mapped according to spatial positions and parameter types to obtain a time sequence feature mapping set;

[0147] After obtaining the geotechnical engineering time sequence state feature dataset, all the principal component feature dimensions are preliminarily classified according to spatial coordinate systems based on the spatial position information recorded by each principal component feature dimension in the dataset to obtain feature dimension spatial groups based on spatial positions;

[0148] According to the parameter type definition in the geotechnical engineering model, the principal component feature dimensions in each spatial group are secondarily classified according to specific parameter types of the geotechnical engineering model to determine the explicit geotechnical parameter type corresponding to each feature dimension and generate a one-to-one mapping relationship between the feature dimensions and the geotechnical parameter types;

[0149] After the classification is completed, an index mapping table among the spatial positions, the feature dimensions and the parameter types is established, and each mapping relationship is labeled to obtain a time sequence feature mapping set with spatial positions as primary indexes and parameter types as secondary indexes;

[0150] According to the initial computable prior information and the parameter constraint conditions of the unified structure of the geotechnical engineering model and in combination with the time sequence feature mapping set, an initial particle state set is generated by using random sampling.

[0151] In the embodiment, the initial particle state set is iteratively updated by using a particle flow filtering algorithm to obtain an updated geotechnical engineering model parameter posterior distribution, and specifically:

[0152] The initial particle state distribution of the particle flow filtering algorithm is determined according to the initial particle state set, and statistical characteristic parameters of the initial particle state distribution are calculated;

[0153] The calculation process of the statistical characteristic parameters is as follows:

[0154] After the initial particle state distribution of the particle flow filtering algorithm is determined according to the initial particle state set, each particle state parameter in the initial particle state set is vectorized and arranged according to parameter dimensions to form a parameter dimension matrix;

[0155] The arithmetic mean value of all particle state values of each parameter dimension in the parameter dimension matrix is calculated to obtain an initial mean value vector of the parameter dimension;

[0156] The sum of squares of all particle state values of each parameter dimension minus the corresponding initial mean vector is calculated, and then divided by the number of particles minus one to obtain the variance vector of each parameter dimension;

[0157] The sum of the product of the deviations between different parameter dimensions is divided by the number of particles minus one to obtain the covariance matrix between the parameter dimensions;

[0158] The initial mean vector, variance vector and covariance matrix are combined to form the statistical characteristic parameters of the initial particle state distribution;

[0159] Collect the latest standardized time series monitoring data, and determine the mapping relationship between the real-time observation data and the particle state parameters according to the time series feature mapping set;

[0160] The particle flow path between the posterior state distribution of the particle state parameters from the initial state distribution to the real-time observation data is calculated using the particle flow filtering algorithm, and the change of the weight of each particle on the particle flow path is determined;

[0161] The calculation process of the particle flow path is:

[0162] After obtaining the initial particle state distribution and its statistical characteristic parameters, the conditional likelihood function is established between each particle state parameter in the initial particle state set and the real-time observation data, and the particle flow continuous transformation equation is constructed, and the probability density transformation between the initial state distribution and the posterior state distribution is described as a continuous evolution process of the time variable from 0 to 1;

[0163] The state transition matrix, observation matrix and control matrix parameters are set in the particle flow continuous transformation equation, and the drift term and diffusion term of the transformation equation are dynamically updated according to the real-time observation data;

[0164] Differential solution is performed on each particle state parameter in the time variable, the path trajectory point of the particle state in the continuous evolution process is calculated, and the intermediate value of each particle state parameter on the path trajectory point is recorded;

[0165] The particle weight change rate is calculated according to the conditional likelihood function and the current drift term and diffusion term at each path trajectory point, and the particle weight is updated, forming a particle flow path containing the continuous evolution trajectory of the particle state parameter and the particle weight change curve;

[0166] The values of each particle state parameter are updated through the particle flow path and the particle weight change, and the updated particle state parameters are subjected to boundary check and constraint adjustment according to the parameter constraint conditions, to generate corrected particle state parameters meeting the constraint conditions;

[0167] After updating the particle state parameter values using the particle flow path and particle weight change, the updated particle state parameters are compared with the parameter upper and lower limit values in the unified structure of the initial computable prior information and parameter constraints of the geotechnical engineering model in the parameter dimension, and it is checked whether each particle state parameter exceeds the corresponding parameter allowable range;

[0168] The particle state parameters exceeding the upper limit value are adjusted to the corresponding upper limit value using the truncation correction method, and the particle state parameters below the lower limit value are adjusted to the corresponding lower limit value using the truncation correction method;

[0169] The particle state parameters within the allowable range but violating the physical and mechanical constraint relationship are projected into the feasible region satisfying the physical and mechanical constraint condition using the constraint projection algorithm;

[0170] After completing the truncation correction and constraint projection, the statistical consistency of all corrected particle state parameters is recalculated, and the particle state set is updated to generate corrected particle state parameters that satisfy all parameter constraints;

[0171] The statistical characteristic parameters of the updated particle state set are calculated using the corrected particle state parameters, and the geotechnical engineering model parameter posterior distribution is estimated according to the statistical characteristic parameters;

[0172] It is judged whether the geotechnical engineering model parameter posterior distribution is stable and convergent, and if not, the particle flow path is recalculated for iterative updating until a stable and convergent geotechnical engineering model parameter posterior distribution is obtained.

[0173] In the embodiment, the geotechnical engineering model numerical simulation is performed based on the updated geotechnical engineering model parameter posterior distribution, and the error evaluation index is calculated, specifically:

[0174] According to the updated geotechnical engineering model parameter posterior distribution, the parameters of the geotechnical engineering model are assigned, and the parameters of the geotechnical engineering model are corresponded to the geotechnical engineering model spatial grid one by one to obtain the geotechnical engineering model spatial grid data after parameter assignment;

[0175] According to the initial physical and mechanical constraint conditions of the geotechnical engineering model, the initial boundary conditions of the geotechnical engineering model are determined, and the initial boundary conditions are applied to the geotechnical engineering model spatial grid data after parameter assignment;

[0176] After obtaining the initial physical and mechanical constraint conditions of the geotechnical engineering model, the various physical quantity constraint parameters included therein are divided into stress constraint, displacement constraint and load constraint according to the mechanical type;

[0177] The geometric structure information of the space grid of the geotechnical engineering model is used to match the spatial positions of the constraint parameters of each physical quantity, the stress constraint parameters are corresponded to the boundary elements in the space grid which are the same as the actual stress areas, the displacement constraint parameters are corresponded to the boundary elements in the space grid which are the same as the actual fixed supports or displacement control areas, and the load constraint parameters are corresponded to the boundary elements in the space grid which are the same as the external load action areas;

[0178] The corresponding physical quantity constraint parameters in each boundary element are numerically assigned and the boundary condition types are labeled, and after the constraint parameter mapping and assignment of all the boundary elements are completed, the initial boundary conditions including the stress, displacement and load boundary conditions are formed;

[0179] The geotechnical engineering model space grid data after the parameter assignment and the initial boundary condition application are subjected to numerical simulation calculation, and the model response results are obtained;

[0180] According to the model response results, the simulation calculation values at the corresponding positions and time points are extracted, and the simulation data sequence which can be directly compared with the actual monitoring data is formed;

[0181] The numerical difference values of the simulation data sequence and the actual monitoring data at each monitoring point are calculated, and the difference sequence of each monitoring point at the corresponding time is obtained;

[0182] The calculation process of the difference sequence is as follows:

[0183] After the simulation data sequence and the actual monitoring data are obtained, the simulation data and the actual monitoring data are one-to-one corresponded according to the spatial coordinate index and the time sequence index of the monitoring points, and the simulation calculation values and the actual monitoring values of the same monitoring point at the same time index are paired;

[0184] The difference value of each paired data is calculated, the difference value is the simulation calculation value minus the actual monitoring value, and the single-point difference value of each monitoring point at each time index is obtained;

[0185] All the single-point difference values of the same monitoring point are sequentially grouped into the difference sequence according to the time sequence, and the time index and the spatial coordinate information corresponding to each difference value are recorded, and the difference sequence of each monitoring point is formed;

[0186] The arithmetic mean value of the difference absolute values, the variance of the difference sequence and the root mean square error of all the monitoring point difference sequences are calculated, and are recorded as the error evaluation indexes of the model simulation results.

[0187] In the embodiment, after the error evaluation indexes are used to judge whether the geotechnical engineering model parameter posterior distribution reaches the predetermined optimization convergence condition, when the predetermined optimization convergence condition is met, the final optimized and trained geotechnical engineering model is obtained, and specifically:

[0188] According to the error evaluation index of the numerical simulation result of the geotechnical engineering model, the absolute value, variance and root mean square error of each monitoring point and time sequence are comprehensively scored by multiple indexes to generate an error comprehensive score sequence corresponding to each iteration;

[0189] In the error comprehensive score sequence, the change rate of the error comprehensive scores of the adjacent two iterations is calculated according to the time sequence index to obtain an error convergence trend curve;

[0190] The calculation process of the change rate of the error comprehensive scores of the adjacent two iterations is as follows:

[0191] After obtaining the error comprehensive score sequence corresponding to each iteration, the error comprehensive score sequence is sequentially arranged according to the time sequence index to form a time sequence matrix containing the iteration time index and the corresponding error comprehensive score;

[0192] The error comprehensive scores of the adjacent two iterations in the time sequence matrix are subjected to difference operation, the difference is the error comprehensive score of the latter iteration minus the error comprehensive score of the former iteration, and the difference is divided by the absolute value of the error comprehensive score of the former iteration to obtain the change rate of the error comprehensive scores of the adjacent two iterations;

[0193] According to the preset optimization convergence threshold and the error convergence trend curve, the difference between the current iteration error comprehensive score and the threshold is calculated, and it is judged whether the error convergence trend curve remains below the threshold fluctuation range at consecutive multiple time indexes;

[0194] The calculation method of the difference between the current iteration error comprehensive score and the threshold is as follows:

[0195] After obtaining the error convergence trend curve, a one-to-one correspondence is established between the error comprehensive score corresponding to each iteration time index in the trend curve and the preset optimization convergence threshold to construct a control matrix containing the iteration time index, the error comprehensive score and the convergence threshold;

[0196] The error comprehensive score and the corresponding convergence threshold of each row in the control matrix are subjected to difference calculation, the difference is the error comprehensive score minus the convergence threshold, and the positive and negative signs and the absolute values of the difference are recorded;

[0197] The calculated difference, positive and negative signs and absolute values are sequentially arranged into a difference sequence according to the iteration time index, and the deviation degree of the error comprehensive score from the threshold at each iteration time index is marked in the difference sequence to form a complete error comprehensive score and threshold difference dataset;

[0198] When the error convergence trend curve remains below the threshold fluctuation range for a continuous preset number of iterations, the current geotechnical engineering model parameter posterior distribution is determined to satisfy the predetermined optimization convergence condition, and the geotechnical engineering model parameter posterior distribution of the current iteration is output as the final optimized geotechnical engineering model parameter posterior distribution.

[0199] When the error convergence trend curve does not reach the predetermined optimization convergence condition, the error comprehensive score of the current iteration is recorded, and the particle flow filtering algorithm is executed to continue the iterative update of the parameter posterior distribution until the predetermined optimization convergence condition is met.

[0200] A geotechnical engineering model optimization training system based on a multi-source knowledge base, comprising:

[0201] A multi-source knowledge base construction module for constructing a geotechnical engineering multi-source knowledge base;

[0202] A sensor data acquisition and preprocessing module for acquiring sensor monitoring data in a geotechnical engineering site and preprocessing to obtain standardized time series monitoring data;

[0203] A time series feature extraction module for extracting time series features from standardized time series monitoring data using a Mamba model to generate a time series state feature dataset;

[0204] An initial particle state set generation module for generating initial computable prior information and parameter constraints of a geotechnical engineering model based on a geotechnical engineering multi-source knowledge base, and establishing an initial particle state set;

[0205] A particle flow filtering iterative update module for iteratively updating the initial particle state set using a particle flow filtering algorithm to obtain an updated geotechnical engineering model parameter posterior distribution;

[0206] A geotechnical engineering model numerical simulation module for numerical simulation of a geotechnical engineering model and calculation of error evaluation indicators;

[0207] An optimization convergence judgment module for determining whether the geotechnical engineering model parameter posterior distribution meets the predetermined optimization convergence condition, and outputting the final optimized geotechnical engineering model parameter posterior distribution when the predetermined optimization convergence condition is met.

[0208] Embodiment: In order to verify the feasibility of the application in implementation, the application is applied to the safety risk monitoring scene of a certain large geotechnical engineering foundation pit construction. In the actual construction process, various sensor devices are buried on site to monitor key data such as foundation pit displacement, settlement, stress, and groundwater level changes in real time. Generally, only single-source monitoring data is used for model parameter optimization, and multi-source data resources are not fully utilized, resulting in low data utilization, insufficient model accuracy, and poor convergence stability, leading to large fluctuations in risk assessment and prediction results and affecting the accuracy of construction safety management decisions.

[0209] In the implementation process, first, the geological survey data, geophysical survey data, historical engineering case data, construction log data, test data, and specification experience data of the foundation pit project are collected to construct a multi-source knowledge base of geotechnical engineering. The system dynamically evaluates the confidence of multi-source data, automatically filters out data with low confidence, and further constructs a standardized data set through semantic analysis and dynamic standardization processing. Then, through the spatio-temporal correlation between data, a multi-layer spatio-temporal index library and multi-source dynamic correlation data set are constructed, and a unified structure of the multi-source knowledge base of geotechnical engineering is formed.

[0210] At the same time, the sensors installed on site continuously collect real-time monitoring data. For the collected data, the system first performs integrity check, uses statistical interpolation method to complete the missing data, and then identifies and removes abnormal data. Subsequently, semantic standardization processing, data smoothing processing, resampling and numerical range standardization are performed to generate high-quality standardized time series monitoring data.

[0211] The system then uses the Mamba model to extract time series features from the standardized time series monitoring data. First, a multi-scale time series sequence group is constructed, and the built-in adaptive scale learning mechanism of the model is used to determine the optimal scale. Then, the state space is converted to generate dynamic state feature data, and by similarity matching with the historical time series feature patterns in the multi-source knowledge base, the matching weight is obtained. Further, the dynamic sparse attention mechanism is used to highlight the key features, and finally the principal component features with higher importance are selected to form the geotechnical engineering time series state feature data set.

[0212] In the model optimization phase, the system uses the multi-source knowledge base of geotechnical engineering to determine the initial structure and parameter experience range of the model, establishes a spatial parameter partition, and combines with the physical and mechanical constraint conditions for constraint fusion to form the initial computable prior information and parameter constraint conditions. Then, according to the above conditions and combined with the time series state feature data set, the initial particle state set required by the particle filter algorithm is generated by random sampling.

[0213] After using the particle flow filter algorithm, the system determines the initial particle state distribution, updates the particle flow path and particle weight in real time, and iteratively updates the particle state through boundary check and constraint adjustment until the model parameter posterior distribution converges stably.

[0214] After convergence, the system assigns the model parameters to the geotechnical engineering model space grid, applies clear initial boundary conditions, performs detailed numerical simulation calculation, extracts the simulation data sequence, and compares it with the actual monitoring data to obtain the error evaluation index set of the model.

[0215] To ensure the stability and accuracy of the model parameter optimization, the system forms an error convergence trend curve through multi-index comprehensive scoring to determine whether the predetermined optimization convergence condition is reached. When the error trend curves of continuous multiple iterations are all within the fluctuation threshold range, the system determines convergence and outputs the final optimized training geotechnical engineering model parameter posterior distribution.

[0216] The following table is the comparison data of displacement prediction results and actual monitoring values of part of the monitoring points in the specific construction scene:

[0217] Table 1 Comparison table of geotechnical engineering model predicted displacement and actual monitoring displacement

[0218]

[0219] According to the data in the above table 1, after using the optimization training method of the present application, the absolute error between the predicted displacement value and the actual measured displacement value of each monitoring point is controlled within 0.16mm, and the relative error is less than 1%, which shows high prediction accuracy and reliability. Specifically, the actual monitoring displacement value of measuring point 01 is 12.53mm, the predicted displacement value is 12.42mm, and the relative error is only 0.88%, which effectively overcomes the common large fluctuation problem in the prediction of this position by the single source data method. In addition, the relative errors of measuring point 03 and measuring point 05 are 0.74% and 0.72% respectively, which reflects the high stability of the system in the small displacement prediction scene.

[0220] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can make equivalent replacement or change according to the technical solution and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for optimizing and training geotechnical engineering models based on a multi-source knowledge base, characterized in that, include: Constructing a multi-source knowledge base for geotechnical engineering; Real-time acquisition and preprocessing of sensor monitoring data from geotechnical engineering sites yields standardized time-series monitoring data. The Mamba model is used to extract time-series features from standardized time-series monitoring data to generate a time-series state feature dataset. The initial computable prior information and parameter constraints of the geotechnical engineering model are generated based on the multi-source knowledge base of geotechnical engineering, and the initial particle state set is established by combining the temporal state feature dataset. The particle flow filtering algorithm is used to iteratively update the initial particle state set to obtain the updated posterior distribution of geotechnical engineering model parameters. Numerical simulation of the geotechnical engineering model is performed based on the updated posterior distribution of geotechnical engineering model parameters, and error evaluation index is calculated. Based on the error evaluation index, it is determined whether the posterior distribution of the parameters of the geotechnical engineering model has reached the predetermined optimization convergence condition. When the predetermined optimization convergence condition is met, the final optimized geotechnical engineering model is obtained.

2. The geotechnical engineering model optimization training method based on a multi-source knowledge base according to claim 1, characterized in that, The construction of the multi-source knowledge base for geotechnical engineering specifically involves: Dynamic confidence assessment is performed on geological exploration data, geophysical exploration data, historical engineering case data, construction log data, test data, and standard experience data. Low-confidence data is automatically eliminated based on confidence thresholds to obtain a high-confidence filtered data set. Semantic parsing is performed on the high-confidence filtered data set to obtain a semantic constraint mapping table; The high-confidence filtered dataset is dynamically standardized based on the semantic constraint mapping table to generate a standardized dataset. A spatiotemporal topology index library is constructed using the spatial location, timestamp, and semantic tags of each data record in a standardized dataset. Based on the spatiotemporal topology index library, multi-layer spatiotemporal correlation processing is performed on historical engineering case data and construction log data to obtain a multi-source dynamic correlation dataset. A unified structure is used to integrate standardized datasets and multi-source dynamically associated datasets to obtain a multi-source knowledge base for geotechnical engineering.

3. The geotechnical engineering model optimization training method based on a multi-source knowledge base according to claim 1, characterized in that, The real-time acquisition and preprocessing of sensor monitoring data from the geotechnical engineering site to obtain standardized time-series monitoring data specifically involves: The sensor monitoring data collected in real time is checked for data integrity, and the missing data is interpolated and filled in using the statistical distribution characteristics of the data before and after the time series, so as to obtain the sensor monitoring data after the missing data is filled in. The sensor monitoring data after missing data is filled in is subjected to abnormal data identification and removal to obtain sensor monitoring data after abnormal data removal. Using a semantic constraint mapping table, each data record of the sensor monitoring data after removing abnormal data is mapped to a predetermined standard data field to obtain semantically standardized sensor monitoring data. The semantically standardized sensor monitoring data is smoothed to obtain denoised sensor monitoring data. The denoised sensor monitoring data is resampled in time series to obtain sensor monitoring data with a uniform time series sampling frequency; Perform numerical range standardization on sensor monitoring data with a uniform time-series sampling frequency, converting each value in the data into a uniform standard numerical range to generate standardized time-series monitoring data.

4. The geotechnical engineering model optimization training method based on a multi-source knowledge base according to claim 1, characterized in that, The process of extracting time-series features from standardized time-series monitoring data using the Mamba model to generate a time-series state feature dataset is as follows: A multi-scale time series sequence group is constructed based on standardized time series monitoring data, and the optimal convolution scale for each time series sequence is determined by the adaptive scale learning mechanism built into the Mamba model, generating adaptive scale feature data containing optimal scale information. The dynamic state space mapping is performed on the adaptive scaling feature data using the state space transformation mechanism in the Mamba model, which transforms the temporal feature representation into dynamic state space feature data. Based on the historical case time-series feature pattern library in the geotechnical engineering multi-source knowledge base, time-series feature pattern similarity matching is performed on the dynamic state space feature data, and the matching degree between the current monitoring data features and historical feature patterns is calculated to obtain state feature data; A time-series dynamic sparse attention mechanism is adopted for the state feature data to highlight the discriminative key state features in the geotechnical engineering evolution process, suppress redundant and low-correlation features, and generate sparsely optimized key state feature data. An interpretable temporal state feature screening mechanism is used to score the importance of key state feature data in sparse optimization. A few key feature dimensions with high importance in both historical temporal pattern matching and dynamic state space mapping are selected and retained to obtain temporal state principal component feature data. A geotechnical engineering time-series state feature dataset was constructed based on the time-series state principal component feature data.

5. The geotechnical engineering model optimization training method based on a multi-source knowledge base according to claim 1, characterized in that, The process involves generating initial computable prior information and parameter constraints for the geotechnical engineering model based on a multi-source knowledge base for geotechnical engineering, and establishing an initial particle state set by combining it with a time-series state feature dataset. Specifically: Based on historical engineering case data and experimental data in the geotechnical engineering multi-source knowledge base, the empirical range of initial structure and parameters of the geotechnical engineering model is determined, and a set of empirical parameter constraints is formed. Based on the spatial distribution characteristics and geological attribute classification information in geophysical exploration data and geological survey data, spatial parameter partitioning of geotechnical engineering models is established to obtain initial parameter spatial partitioning information; Based on standard experience data and construction log data, the initial physical and mechanical constraints of the geotechnical engineering model are extracted to obtain a set of physical constraints including initial stress boundary conditions and mechanical parameter constraints. The set of empirical constraints, the initial parameter space partitioning information, and the set of physical constraints are fused together to form the initial computable prior information and parameter constraints. The principal component feature dimensions in the geotechnical engineering time-series state feature dataset are classified and mapped according to spatial location and parameter type to obtain a time-series feature mapping set; Based on the initial computable prior information and parameter constraints of the geotechnical engineering model with a unified structure, and combined with the temporal feature mapping set, an initial particle state set is generated by random sampling.

6. The geotechnical engineering model optimization training method based on a multi-source knowledge base according to claim 1, characterized in that, The process of iteratively updating the initial particle state set using a particle flow filtering algorithm to obtain the updated posterior distribution of geotechnical engineering model parameters is as follows: The initial particle state distribution of the particle flow filtering algorithm is determined based on the initial particle state set, and the statistical characteristic parameters of the initial particle state distribution are calculated. Collect the latest standardized time-series monitoring data, and determine the mapping relationship between real-time observation data and particle state parameters based on the time-series feature mapping set; The particle flow filtering algorithm is used to calculate the particle flow path between the initial state distribution and the posterior state distribution under real-time observation data conditions, and to determine the weight changes of each particle on the particle flow path. The values ​​of each particle state parameter are updated by the particle flow path and particle weight changes, and the updated particle state parameters are checked for boundaries and adjusted for constraints according to the parameter constraints to generate corrected particle state parameters that meet the constraints. The statistical characteristic parameters of the updated particle state set are calculated using the corrected particle state parameters, and the posterior distribution of geotechnical engineering model parameters is estimated based on the statistical characteristic parameters. Determine whether the posterior distribution of geotechnical engineering model parameters is stable and convergent. If it is not stable and convergent, recalculate the particle flow path and continue iterative updates until a stable and convergent posterior distribution of geotechnical engineering model parameters is obtained.

7. The geotechnical engineering model optimization training method based on a multi-source knowledge base according to claim 1, characterized in that, The numerical simulation of the geotechnical engineering model is performed based on the updated posterior distribution of the geotechnical engineering model parameters, and the error evaluation index is calculated, specifically as follows: Based on the updated posterior distribution of geotechnical engineering model parameters, parameter values ​​are assigned to the geotechnical engineering model. The parameters of the geotechnical engineering model are mapped one-to-one with the spatial grid of the geotechnical engineering model to obtain the spatial grid data of the geotechnical engineering model after parameter assignment. The initial boundary conditions of the geotechnical engineering model are determined based on the initial physical and mechanical constraints of the geotechnical engineering model, and the initial boundary conditions are applied to the spatial grid data of the geotechnical engineering model after parameter assignment. Numerical simulation calculations were performed on the spatial grid data of the geotechnical engineering model after parameter assignment and initial boundary conditions were applied to obtain the model response results; Based on the model response results, the simulation calculation values ​​at the corresponding locations and times are extracted to form a simulation data sequence that can be directly compared with the actual monitoring data; Calculate the numerical difference between the simulated data sequence and the actual monitoring data at each monitoring point to obtain the difference sequence at each monitoring point at the corresponding time. For all monitoring points, calculate the arithmetic mean of the absolute values ​​of the differences, the variance of the difference sequence, and the root mean square error, and record them as error evaluation indicators for the model simulation results.

8. The method for optimizing and training geotechnical engineering models based on a multi-source knowledge base according to claim 1, characterized in that, The step of determining whether the posterior distribution of the geotechnical engineering model parameters meets the predetermined optimization convergence condition based on the error evaluation index, and obtaining the final optimized geotechnical engineering model after meeting the predetermined optimization convergence condition, is as follows: Based on the error evaluation index of the numerical simulation results of the geotechnical engineering model, the absolute value, variance and root mean square error of each monitoring point and time series are comprehensively scored by multiple indicators, and the error comprehensive score sequence corresponding to each iteration is generated. The rate of change of the comprehensive error score between two adjacent iterations is calculated by indexing the time series in the comprehensive error score sequence to obtain the error convergence trend curve. Based on the preset optimization convergence threshold and error convergence trend curve, calculate the difference between the current iteration error comprehensive score and the threshold, and determine whether the error convergence trend curve remains below the threshold fluctuation range across multiple consecutive time indices. When the error convergence trend curve remains below the threshold fluctuation range within a consecutive preset number of iterations, the posterior distribution of the current geotechnical engineering model parameters is determined to meet the predetermined optimization convergence condition, and the posterior distribution of the geotechnical engineering model parameters in the current iteration is output as the posterior distribution of the geotechnical engineering model parameters after the final optimization training. When the error convergence trend curve fails to meet the predetermined optimization convergence condition, record the comprehensive error score of the current iteration and return to execute the particle flow filtering algorithm to continue iteratively updating the posterior distribution of parameters until the predetermined optimization convergence condition is met.

9. A geotechnical engineering model optimization training system based on a multi-source knowledge base, comprising executing the geotechnical engineering model optimization training method based on a multi-source knowledge base as described in any one of claims 1 to 8, characterized in that, include: The multi-source knowledge base construction module is used to build a multi-source knowledge base for geotechnical engineering. The sensor data acquisition and preprocessing module is used to acquire sensor monitoring data at the geotechnical engineering site and preprocess it to obtain standardized time-series monitoring data. The temporal feature extraction module is used to extract temporal features from standardized time-series monitoring data using the Mamba model, and generate a time-series state feature dataset. The initial particle state set generation module is used to generate the initial computable prior information and parameter constraints of the geotechnical engineering model based on the geotechnical engineering multi-source knowledge base, and to establish the initial particle state set. The particle flow filtering iterative update module is used to iteratively update the initial particle state set using the particle flow filtering algorithm to obtain the updated posterior distribution of geotechnical engineering model parameters. The geotechnical engineering model numerical simulation module is used to perform numerical simulations of geotechnical engineering models and calculate error evaluation indicators. The optimization convergence judgment module is used to determine whether the posterior distribution of geotechnical engineering model parameters has reached the predetermined optimization convergence condition. If the predetermined optimization convergence condition is met, the final optimized training posterior distribution of geotechnical engineering model parameters is output.

Images