Deep foundation pit monitoring parameter prediction method based on neural network

Abstract

The invention provides a deep foundation pit monitoring parameter prediction method based on a neural network, and relates to the technical field of deep foundation pit parameter prediction.The method comprises the steps that risk factors of a deep foundation pit are obtained, and deep foundation pit monitoring parameters are determined; setting a deep foundation pit monitoring frequency, acquiring deep foundation pit monitoring data according to the deep foundation pit monitoring parameters, and performing data preprocessing; acquiring a neural network algorithm of the deep foundation pit monitoring parameters, constructing a deep foundation pit monitoring parameter prediction model, performing model training according to the deep foundation pit monitoring data subjected to data preprocessing, acquiring deep foundation pit prediction parameters, and determining a matching algorithm of the deep foundation pit monitoring parameters according to the actual deep foundation pit parameters; obtaining deep foundation pit prediction parameters corresponding to the matching algorithm, determining the toughness grade of the deep foundation pit, and executing a deep foundation pit maintenance strategy; and accurate prediction of the monitoring parameters of the deep foundation pit is realized.

Description

Technical Field

[0001] The invention relates to the technical field of deep foundation pit parameter prediction, and in particular to a method for predicting deep foundation pit monitoring parameters based on a neural network. Background Art

[0002] With the rapid development of infrastructure construction in my country, the scale of foundation pit projects has become deeper and larger; deep foundation pits have complex environments, long construction periods, numerous risk factors, and high probability of accidents. Deep foundation pit accidents often lead to serious losses; existing prediction technologies have many limitations, making it difficult to accurately predict deep foundation pit monitoring parameters. In addition, there are many existing neural network algorithms, and different neural networks require different adaptation conditions.

[0003] Therefore, the present invention provides a method for predicting deep foundation pit monitoring parameters based on a neural network. Summary of the invention

[0004] The present invention provides a prediction method for deep foundation pit monitoring parameters based on a neural network, which determines the deep foundation pit monitoring parameters by acquiring risk factors of the deep foundation pit; sets the deep foundation pit monitoring frequency, acquires deep foundation pit monitoring data according to the deep foundation pit monitoring parameters and performs data preprocessing; acquires a neural network algorithm for the deep foundation pit monitoring parameters, constructs a deep foundation pit monitoring parameter prediction model and performs model training based on the deep foundation pit monitoring data after data preprocessing, acquires the deep foundation pit prediction parameters, and determines a matching algorithm for the deep foundation pit monitoring parameters according to the actual parameters of the deep foundation pit; acquires the deep foundation pit prediction parameters corresponding to the matching algorithm, determines the deep foundation pit toughness grade and executes the deep foundation pit maintenance strategy; and realizes accurate prediction of the deep foundation pit monitoring parameters.

[0005] The present invention provides a method for predicting deep foundation pit monitoring parameters based on a neural network, comprising: Step 1: Obtain risk factors of deep foundation pits and determine monitoring parameters of deep foundation pits; Step 2: Set the deep foundation pit monitoring frequency, obtain the deep foundation pit monitoring data according to the deep foundation pit monitoring parameters, and perform data preprocessing; Step 3: Obtain the neural network algorithm of deep foundation pit monitoring parameters, build a deep foundation pit monitoring parameter prediction model and perform model training based on the deep foundation pit monitoring data after data preprocessing, obtain the deep foundation pit prediction parameters, and determine the matching algorithm of the deep foundation pit monitoring parameters based on the actual parameters of the deep foundation pit; Step 4: Obtain the deep foundation pit prediction parameters corresponding to the matching algorithm, determine the deep foundation pit toughness grade and implement the deep foundation pit maintenance strategy.

[0006] According to the neural network-based deep foundation pit monitoring parameter prediction method provided by the present invention, the risk factors of the deep foundation pit are obtained and the deep foundation pit monitoring parameters are determined, including: Obtain risk factors for deep foundation pits and determine the monitorable parameters of deep foundation pits, including deformation parameters of retaining structures, stress characteristic parameters of supports, settlement parameters around foundation pits, groundwater level change parameters, foundation pit uplift parameters, and parameters of conditions around foundation pits; According to the actual monitoring equipment of the deep foundation pit, some monitoring parameters are selected from the monitorable parameters of the deep foundation pit and recorded as deep foundation pit monitoring parameters, wherein the deep foundation pit monitoring parameters include: deep foundation pit settlement monitoring parameters, deep foundation pit support end horizontal displacement monitoring parameters and deep foundation pit support axial force monitoring parameters.

[0007] According to the neural network-based deep foundation pit monitoring parameter prediction method provided by the present invention, the deep foundation pit monitoring frequency is set, the deep foundation pit monitoring data is obtained according to the deep foundation pit monitoring parameters, and the data is preprocessed, including: Obtain the deep foundation pit monitoring accuracy requirements, set the deep foundation pit monitoring frequency, and obtain the deep foundation pit monitoring data based on the deep foundation pit monitoring parameters and the actual deep foundation pit monitoring equipment; The deep foundation pit monitoring data is preprocessed, wherein the data preprocessing includes: data cleaning processing, data standardization processing and data formatting processing.

[0008] According to the neural network-based deep foundation pit monitoring parameter prediction method provided by the present invention, a neural network algorithm for deep foundation pit monitoring parameters is obtained, a deep foundation pit monitoring parameter prediction model is constructed, and model training is performed based on deep foundation pit monitoring data after data preprocessing, including: A neural network algorithm for obtaining deep foundation pit monitoring parameters, wherein the neural network algorithm includes: BP neural network algorithm, GA-BP neural network algorithm, NARX dynamic neural network algorithm and Elman neural network algorithm; A deep foundation pit monitoring parameter prediction model is constructed, and model parameters of the deep foundation pit monitoring parameter prediction model are determined based on the deep foundation pit monitoring data after data preprocessing and the neural network algorithm, and model training is performed on the deep foundation pit monitoring parameter prediction model; sorting the deep foundation pit monitoring data after data preprocessing according to the data acquisition time of the deep foundation pit monitoring data after data preprocessing; According to the data sorting results, the deep foundation pit monitoring data with a preset proportion of data preprocessing is selected as training samples; The deep foundation pit monitoring parameter prediction model is trained based on the training samples.

[0009] The neural network-based deep foundation pit monitoring parameter prediction method provided by the present invention further includes: The deep foundation pit monitoring data after data preprocessing of the remaining training samples in the data sorting results are selected as verification samples, and the deep foundation pit monitoring parameter prediction model is verified to obtain the prediction accuracy of the deep foundation pit monitoring parameter prediction model; When the prediction accuracy of the deep foundation pit monitoring parameter prediction model is not lower than the set accuracy, it is determined that the model training of the deep foundation pit monitoring parameter prediction model is completed.

[0010] According to the neural network-based deep foundation pit monitoring parameter prediction method provided by the present invention, the deep foundation pit prediction parameters are obtained, and a matching algorithm for determining the deep foundation pit monitoring parameters according to the deep foundation pit actual parameters includes: The deep foundation pit monitoring parameters are predicted according to the deep foundation pit monitoring parameter prediction model after model training, so as to obtain the deep foundation pit prediction parameters; The actual parameters of the deep foundation pit are obtained according to the training samples and the verification samples, the error value between the actual parameters of the deep foundation pit and the predicted parameters of the deep foundation pit is determined, and the neural network algorithm corresponding to the smallest error value is selected, which is recorded as the matching algorithm of the deep foundation pit monitoring parameters.

[0011] According to the neural network-based deep foundation pit monitoring parameter prediction method provided by the present invention, the deep foundation pit prediction parameters corresponding to the matching algorithm are obtained, the deep foundation pit toughness level is determined and the deep foundation pit maintenance strategy is implemented, including: Obtain the deep foundation pit prediction parameters corresponding to the matching algorithm; Obtain the parameter increment of the actual parameters of the deep foundation pit, record the parameter increment as an independent parameter according to the parameter type of the deep foundation pit monitoring parameters, and perform a normality test; Obtain the parameter test value of the normality test. When the parameter test value is not lower than the preset test value, it is determined that the corresponding independent parameter conforms to the normal distribution; The corresponding independent parameters that conform to the normal distribution are divided into intervals for analysis to obtain the toughness influencing factors of the deep foundation pit, among which the toughness influencing factors include: robustness influencing factors, redundancy influencing factors, recovery influencing factors, wisdom influencing factors and adaptability influencing factors; The predicted expected value and predicted standard deviation of deep foundation pit prediction parameters are obtained according to the normality test; The interval level is divided according to the predicted expected value and the predicted standard deviation, and the interval corresponding to plus or minus 1 times the predicted standard deviation near the predicted expected value is recorded as the level 1 interval; The intervals corresponding to plus or minus 2 times the predicted standard deviation near the predicted expected value are recorded as level 2 intervals; The intervals corresponding to plus or minus three times the predicted standard deviation around the predicted expected value are recorded as level 3 intervals; The corresponding intervals outside the three intervals are recorded as level 4 intervals; According to the influencing factors of deep foundation pit toughness and the prediction parameters of deep foundation pit, the deep foundation pit toughness grade is determined and the deep foundation pit maintenance strategy is implemented.

[0012] The neural network-based deep foundation pit monitoring parameter prediction method provided by the present invention further includes: According to the monitorable parameters of deep foundation pit, the deep foundation pit toughness assessment index is determined, and the assessment index matrix is ​​established to conduct a comprehensive assessment of the toughness of the deep foundation pit. The deep foundation pit toughness assessment index includes: the deformation assessment index of the retaining structure, the support force characteristic assessment index, the settlement assessment index around the foundation pit, the groundwater level change assessment index, the foundation pit uplift assessment index and the foundation pit surrounding condition assessment index; The evaluation index matrix is ​​dimensionless; According to the experts' weighted scoring of deep foundation pit toughness assessment indicators, the subjective weights of deep foundation pit toughness assessment indicators are determined; Obtain the coefficient of variation value based on the predicted expected value and the predicted standard deviation, and determine the objective weight of the deep foundation pit toughness evaluation index; The weight coefficient of the subjective weight to the indicator weight and the weight coefficient of the objective weight to the indicator weight are set to determine the indicator weight of the deep foundation pit toughness assessment indicator; According to the deep foundation pit toughness assessment index and the index weight of the deep foundation pit toughness assessment index, the toughness assessment distance is obtained and dimensionless processed to determine the final toughness assessment distance; ; Where d represents the final toughness assessment distance; a1 represents the weight coefficient of subjective weight to indicator weight; w1i1 represents the subjective weight of the i1th deep foundation pit toughness assessment indicator; represents the objective weight of the i1th deep foundation pit toughness assessment index; si1 represents the average value of the i1th deep foundation pit toughness assessment index; ui1 represents the variance of the i1th deep foundation pit toughness assessment index; xi1j1 represents the j1th level quantitative value of the i1th deep foundation pit toughness assessment index; j1 represents the interval level, with values ​​of 1, 2, 3, and 4; u1 represents the predicted expected value; represents the prediction standard deviation; n1 represents the number of deep foundation pit toughness evaluation indicators; According to the actual parameters of the deep foundation pit, the final toughness assessment distance is matched with the deep foundation pit toughness grade, and a mapping table of the final toughness assessment distance and the deep foundation pit toughness grade is determined; The predicted final toughness assessment distance and the predicted deep foundation pit toughness grade corresponding to the deep foundation pit prediction parameters are obtained, and a corresponding judgment is made according to the final toughness assessment distance-deep foundation pit toughness grade mapping table. If it is determined that the predicted final toughness assessment distance corresponds to the predicted deep foundation pit toughness grade, then the deep foundation pit toughness grade determined according to the deep foundation pit prediction parameters is accurate.

[0013] Compared with the prior art, the beneficial effects of this application are as follows: The deep foundation pit monitoring parameters are determined by obtaining the risk factors of the deep foundation pit; the deep foundation pit monitoring frequency is set, and the deep foundation pit monitoring data is obtained and data preprocessing is performed according to the deep foundation pit monitoring parameters; the neural network algorithm of the deep foundation pit monitoring parameters is obtained, and a deep foundation pit monitoring parameter prediction model is constructed and the model training is performed based on the deep foundation pit monitoring data after data preprocessing to obtain the deep foundation pit prediction parameters, and the matching algorithm of the deep foundation pit monitoring parameters is determined according to the actual parameters of the deep foundation pit; the deep foundation pit prediction parameters corresponding to the matching algorithm are obtained, the deep foundation pit toughness grade is determined and the deep foundation pit maintenance strategy is implemented; and accurate prediction of the deep foundation pit monitoring parameters is achieved.

[0014] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0015] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 It is a flow chart of a method for predicting deep foundation pit monitoring parameters based on a neural network provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0018] Embodiment 1: The embodiment of the present invention provides a prediction method for deep foundation pit monitoring parameters based on neural network, such as Figure 1 As shown, including: Step 1: Obtain risk factors of deep foundation pits and determine monitoring parameters of deep foundation pits; Step 2: Set the deep foundation pit monitoring frequency, obtain the deep foundation pit monitoring data according to the deep foundation pit monitoring parameters, and perform data preprocessing; Step 3: Obtain the neural network algorithm of deep foundation pit monitoring parameters, build a deep foundation pit monitoring parameter prediction model and perform model training based on the deep foundation pit monitoring data after data preprocessing, obtain the deep foundation pit prediction parameters, and determine the matching algorithm of the deep foundation pit monitoring parameters based on the actual parameters of the deep foundation pit; Step 4: Obtain the deep foundation pit prediction parameters corresponding to the matching algorithm, determine the deep foundation pit toughness grade and implement the deep foundation pit maintenance strategy.

[0019] In this embodiment, the risk factors refer to factors that affect the toughness of the deep foundation pit.

[0020] In this embodiment, the deep foundation pit monitoring parameters refer to monitorable parameters for analyzing the toughness of the deep foundation pit, and are parameters obtained by selecting some monitoring parameters from the monitorable parameters of the deep foundation pit.

[0021] In this embodiment, the deep foundation pit monitoring frequency refers to the frequency of monitoring the deep foundation pit monitoring parameters.

[0022] In this embodiment, the deep foundation pit monitoring data refers to data acquired based on the deep foundation pit monitoring parameters and actual deep foundation pit monitoring equipment.

[0023] In this embodiment, the neural network algorithm refers to an algorithm used for predicting deep foundation pit monitoring parameters.

[0024] In this embodiment, the deep foundation pit monitoring parameter prediction model refers to a model used for predicting deep foundation pit monitoring parameters.

[0025] In this embodiment, the deep foundation pit prediction parameter refers to the prediction data obtained by predicting the deep foundation pit monitoring parameter according to the deep foundation pit monitoring parameter prediction model after model training.

[0026] In this embodiment, the matching algorithm refers to an optimal neural network algorithm for predicting deep foundation pit monitoring parameters.

[0027] In this embodiment, the deep foundation pit toughness level refers to the level of stable operation of the deep foundation pit.

[0028] In this embodiment, the deep foundation pit maintenance strategy refers to a strategy for ensuring stable operation of the deep foundation pit.

[0029] The working principle and beneficial effects of the above technical solution are: determining the monitoring parameters of the deep foundation pit by acquiring the risk factors of the deep foundation pit; setting the monitoring frequency of the deep foundation pit, acquiring the deep foundation pit monitoring data according to the deep foundation pit monitoring parameters and performing data preprocessing; acquiring the neural network algorithm of the deep foundation pit monitoring parameters, constructing a deep foundation pit monitoring parameter prediction model and performing model training based on the deep foundation pit monitoring data after data preprocessing, acquiring the deep foundation pit prediction parameters, and determining the matching algorithm of the deep foundation pit monitoring parameters according to the actual parameters of the deep foundation pit; acquiring the deep foundation pit prediction parameters corresponding to the matching algorithm, determining the deep foundation pit toughness grade and implementing the deep foundation pit maintenance strategy; and realizing the accurate prediction of the deep foundation pit monitoring parameters.

[0030] Embodiment 2: The embodiment of the present invention provides a prediction method for deep foundation pit monitoring parameters based on a neural network, obtains risk factors of the deep foundation pit, and determines the deep foundation pit monitoring parameters, including: Obtain risk factors for deep foundation pits and determine the monitorable parameters of deep foundation pits, including deformation parameters of retaining structures, stress characteristic parameters of supports, settlement parameters around foundation pits, groundwater level change parameters, foundation pit uplift parameters, and parameters of conditions around foundation pits; According to the actual monitoring equipment of the deep foundation pit, some monitoring parameters are selected from the monitorable parameters of the deep foundation pit and recorded as deep foundation pit monitoring parameters, wherein the deep foundation pit monitoring parameters include: deep foundation pit settlement monitoring parameters, deep foundation pit support end horizontal displacement monitoring parameters and deep foundation pit support axial force monitoring parameters.

[0031] In this embodiment, the risk factors refer to factors that affect the toughness of the deep foundation pit.

[0032] In this embodiment, the monitorable parameters refer to the monitorable parameters determined according to the risk factors, that is, the specific risk factors of the deep foundation pit are determined by the monitorable parameters.

[0033] In this embodiment, the deep foundation pit monitoring parameters refer to monitorable parameters for analyzing the toughness of the deep foundation pit, and are parameters obtained by selecting some monitoring parameters from the monitorable parameters of the deep foundation pit.

[0034] In this embodiment, the deep foundation pit settlement monitoring parameters belong to the settlement parameters around the foundation pit; the deep foundation pit support end horizontal displacement monitoring parameters belong to the surrounding structure deformation parameters; the deep foundation pit support axial force monitoring parameters belong to the support force characteristic parameters.

[0035] In this embodiment, some monitoring parameters are selected from the monitorable parameters of the deep foundation pit, among which the deep foundation pit settlement monitoring parameters, the deep foundation pit support end horizontal displacement monitoring parameters and the deep foundation pit support axial force monitoring parameters are key parameters affecting the toughness of the deep foundation pit. Only the above parameters are selected to simplify the analysis, and monitoring parameters can be added later according to the analysis requirements.

[0036] The working principle and beneficial effects of the above technical solution are: by obtaining the risk factors of deep foundation pits and determining the deep foundation pit monitoring parameters, a data foundation is laid for the subsequent prediction of deep foundation pit monitoring parameters.

[0037] Embodiment 3: The embodiment of the present invention provides a prediction method for deep foundation pit monitoring parameters based on a neural network, sets a deep foundation pit monitoring frequency, obtains deep foundation pit monitoring data according to the deep foundation pit monitoring parameters, and performs data preprocessing, including: Obtain the deep foundation pit monitoring accuracy requirements, set the deep foundation pit monitoring frequency, and obtain the deep foundation pit monitoring data based on the deep foundation pit monitoring parameters and the actual deep foundation pit monitoring equipment; The deep foundation pit monitoring data is preprocessed, wherein the data preprocessing includes: data cleaning processing, data standardization processing and data formatting processing.

[0038] In this embodiment, the deep foundation pit monitoring accuracy requirement refers to the accuracy requirement for monitoring the deep foundation pit monitoring parameters.

[0039] In this embodiment, the deep foundation pit monitoring frequency refers to the frequency of monitoring the deep foundation pit monitoring parameters, which is determined according to the deep foundation pit monitoring accuracy. For example, the deep foundation pit monitoring frequency is once every two days.

[0040] In this embodiment, the actual monitoring equipment for the deep foundation pit refers to equipment used to monitor monitoring parameters of the deep foundation pit, such as infrared monitoring equipment.

[0041] In this embodiment, the deep foundation pit monitoring data refers to data acquired based on deep foundation pit monitoring parameters and actual deep foundation pit monitoring equipment, and usually one year of monitoring data is acquired.

[0042] The working principle and beneficial effects of the above technical solution are: by setting the deep foundation pit monitoring frequency, acquiring deep foundation pit monitoring data according to deep foundation pit monitoring parameters and performing data preprocessing, the accuracy of the deep foundation pit monitoring data is ensured.

[0043] Embodiment 4: The embodiment of the present invention provides a prediction method for deep foundation pit monitoring parameters based on a neural network, obtains a neural network algorithm for deep foundation pit monitoring parameters, constructs a deep foundation pit monitoring parameter prediction model, and performs model training based on deep foundation pit monitoring data after data preprocessing, including: A neural network algorithm for obtaining deep foundation pit monitoring parameters, wherein the neural network algorithm includes: BP neural network algorithm, GA-BP neural network algorithm, NARX dynamic neural network algorithm and Elman neural network algorithm; A deep foundation pit monitoring parameter prediction model is constructed, and model parameters of the deep foundation pit monitoring parameter prediction model are determined based on the deep foundation pit monitoring data after data preprocessing and the neural network algorithm, and model training is performed on the deep foundation pit monitoring parameter prediction model; sorting the deep foundation pit monitoring data after data preprocessing according to the data acquisition time of the deep foundation pit monitoring data after data preprocessing; According to the data sorting results, the deep foundation pit monitoring data with a preset proportion of data preprocessing is selected as training samples; The deep foundation pit monitoring parameter prediction model is trained based on the training samples.

[0044] In this embodiment, different neural networks require different adaptation conditions for obtaining the neural network algorithms for deep foundation pit monitoring parameters. Therefore, four neural network algorithms are selected to determine the optimal neural network algorithm for predicting deep foundation pit monitoring parameters.

[0045] In this embodiment, the deep foundation pit monitoring parameter prediction model refers to a model used for predicting deep foundation pit monitoring parameters.

[0046] In this embodiment, the model parameters of the deep foundation pit monitoring parameter prediction model are determined based on the neural network algorithm, and the deep foundation pit monitoring parameter prediction model is trained respectively. Different neural network algorithms correspond to different model parameters, and the model training results are also different.

[0047] In this embodiment, the model parameters refer to the parameters of the deep foundation pit monitoring parameter prediction model, such as input layer parameters, hidden layer parameters and output layer parameters.

[0048] In this embodiment, data sorting refers to sorting based on the time at which the data is acquired. The older the data acquisition time is, the higher the data is ranked.

[0049] In this embodiment, the training samples refer to samples used for model training of the deep foundation pit monitoring parameter prediction model.

[0050] In this embodiment, a preset proportion of deep foundation pit monitoring data that has been preprocessed is selected as training samples according to the data sorting results. For example, the first 0.7 proportion of the deep foundation pit monitoring data in one year, that is, the first 245 days of deep foundation pit monitoring data, is selected as training samples.

[0051] The working principle and beneficial effects of the above technical solution are: by obtaining the neural network algorithm of deep foundation pit monitoring parameters, a deep foundation pit monitoring parameter prediction model is constructed and the model training is carried out according to the deep foundation pit monitoring data after data preprocessing, which lays the model foundation for the subsequent deep foundation pit monitoring parameter prediction.

[0052] Embodiment 5: The embodiment of the present invention provides a method for predicting deep foundation pit monitoring parameters based on a neural network, further comprising: The deep foundation pit monitoring data after data preprocessing of the remaining training samples in the data sorting results are selected as verification samples, and the deep foundation pit monitoring parameter prediction model is verified to obtain the prediction accuracy of the deep foundation pit monitoring parameter prediction model; When the prediction accuracy of the deep foundation pit monitoring parameter prediction model is not lower than the set accuracy, it is determined that the model training of the deep foundation pit monitoring parameter prediction model is completed.

[0053] In this embodiment, the verification sample refers to a sample used to perform model verification on the deep foundation pit monitoring parameter prediction model.

[0054] In this embodiment, the deep foundation pit monitoring data that has been preprocessed after the remaining data of the training samples are selected from the data sorting results as verification samples. For example, the last 0.3 proportion of the data of the deep foundation pit monitoring data of one year, that is, the last 105 days of deep foundation pit monitoring data, are selected as verification samples.

[0055] In this embodiment, the prediction accuracy refers to the accuracy of the deep foundation pit monitoring parameter prediction model in predicting the deep foundation pit monitoring parameters, which is determined by the data ratio of the output value of the deep foundation pit monitoring parameter prediction model to the verification sample, that is, the closer the output value is to the verification sample, the higher the prediction accuracy.

[0056] In this embodiment, the setting accuracy refers to a setting value used to determine whether the deep foundation pit monitoring parameter prediction model has been trained.

[0057] The working principle and beneficial effect of the above technical solution are: by obtaining verification samples, the deep foundation pit monitoring parameter prediction model is verified, thereby ensuring the model training effectiveness and accuracy of the deep foundation pit monitoring parameter prediction model.

[0058] Embodiment 6: The embodiment of the present invention provides a prediction method for deep foundation pit monitoring parameters based on a neural network, obtains deep foundation pit prediction parameters, and determines a matching algorithm for deep foundation pit monitoring parameters according to actual parameters of the deep foundation pit, including: The deep foundation pit monitoring parameters are predicted according to the deep foundation pit monitoring parameter prediction model after model training, so as to obtain the deep foundation pit prediction parameters; The actual parameters of the deep foundation pit are obtained according to the training samples and the verification samples, the error value between the actual parameters of the deep foundation pit and the predicted parameters of the deep foundation pit is determined, and the neural network algorithm corresponding to the smallest error value is selected, which is recorded as the matching algorithm of the deep foundation pit monitoring parameters.

[0059] In this embodiment, the deep foundation pit prediction parameter refers to the prediction data obtained by predicting the deep foundation pit monitoring parameter according to the deep foundation pit monitoring parameter prediction model after model training.

[0060] In this embodiment, different neural network algorithms obtain different deep foundation pit prediction parameters.

[0061] In this embodiment, the actual parameters of the deep foundation pit refer to monitoring parameters during the actual operation of the deep foundation pit, which can be obtained based on training samples and verification samples.

[0062] In this embodiment, the matching algorithm refers to the optimal neural network algorithm for predicting deep foundation pit monitoring parameters. For example, when the error value between the deep foundation pit prediction parameters corresponding to the GA-BP neural network algorithm and the actual parameters of the deep foundation pit is the smallest, the GA-BP neural network algorithm is recorded as the matching algorithm for the deep foundation pit monitoring parameters.

[0063] The working principle and beneficial effect of the above technical solution are: by obtaining the deep foundation pit prediction parameters, the matching algorithm of the deep foundation pit monitoring parameters is determined according to the actual parameters of the deep foundation pit, which lays the foundation for the neural network algorithm for the subsequent prediction of the deep foundation pit monitoring parameters.

[0064] Embodiment 7: The embodiment of the present invention provides a prediction method for deep foundation pit monitoring parameters based on a neural network, obtains deep foundation pit prediction parameters corresponding to a matching algorithm, determines the deep foundation pit toughness level and executes a deep foundation pit maintenance strategy, including: Obtain the deep foundation pit prediction parameters corresponding to the matching algorithm; Obtain the parameter increment of the actual parameters of the deep foundation pit, record the parameter increment as an independent parameter according to the parameter type of the deep foundation pit monitoring parameters, and perform a normality test; Obtain the parameter test value of the normality test. When the parameter test value is not lower than the preset test value, it is determined that the corresponding independent parameter conforms to the normal distribution; The corresponding independent parameters that conform to the normal distribution are divided into intervals for analysis to obtain the toughness influencing factors of the deep foundation pit, among which the toughness influencing factors include: robustness influencing factors, redundancy influencing factors, recovery influencing factors, wisdom influencing factors and adaptability influencing factors; The predicted expected value and predicted standard deviation of deep foundation pit prediction parameters are obtained according to the normality test; The interval level is divided according to the predicted expected value and the predicted standard deviation, and the interval corresponding to plus or minus 1 times the predicted standard deviation near the predicted expected value is recorded as the level 1 interval; The intervals corresponding to plus or minus 2 times the predicted standard deviation near the predicted expected value are recorded as level 2 intervals; The intervals corresponding to plus or minus three times the predicted standard deviation around the predicted expected value are recorded as level 3 intervals; The corresponding intervals outside the three intervals are recorded as level 4 intervals; According to the influencing factors of deep foundation pit toughness and the prediction parameters of deep foundation pit, the deep foundation pit toughness grade is determined and the deep foundation pit maintenance strategy is implemented.

[0065] In this embodiment, the parameter type refers to the type of deep foundation pit monitoring parameters, such as the deep foundation pit settlement monitoring parameter type, the deep foundation pit support end horizontal displacement monitoring parameter type and the deep foundation pit support axial force monitoring parameter type.

[0066] In this embodiment, the parameter increment refers to a parameter value determined according to the difference between the predicted parameter of the deep foundation pit and the actual parameter of the deep foundation pit.

[0067] In this embodiment, the independent parameters refer to the parameter increments of the actual parameters of the deep foundation pit determined by the parameter type of the deep foundation pit monitoring parameters, such as the independent parameters for monitoring the settlement of the deep foundation pit, the independent parameters for monitoring the horizontal displacement of the support ends of the deep foundation pit, and the independent parameters for monitoring the axial force of the support of the deep foundation pit.

[0068] In this embodiment, the normality test refers to testing whether the independent parameters conform to the normal distribution, and the normality test is performed according to the expected value and standard deviation of the independent parameters.

[0069] In this embodiment, the parameter test value refers to a value obtained based on the expected value and standard deviation of the independent parameter for determining whether the independent parameter conforms to the normal distribution. When the parameter test value is not lower than the preset test value, for example, the parameter test value is not lower than 0.05, it is determined that the corresponding independent parameter conforms to the normal distribution.

[0070] In this embodiment, performing interval division analysis on the corresponding independent parameters that conform to the normal distribution refers to dividing the interval levels and analyzing the interval levels of the independent parameters.

[0071] In this embodiment, the toughness influencing factor refers to the influencing factor that affects the toughness grade of the deep foundation pit.

[0072] In this embodiment, the predicted expected value refers to the expected value of the deep foundation pit prediction parameter obtained according to the normality test; the predicted standard deviation refers to the standard deviation of the deep foundation pit prediction parameter obtained according to the normality test.

[0073] In this embodiment, the interval grade refers to the grade determined according to the predicted expected value and the predicted standard deviation, which is used for the subsequent determination of the deep foundation pit toughness grade.

[0074] In this embodiment, the deep foundation pit toughness grade refers to the grade of stable operation of the deep foundation pit. The higher the deep foundation pit toughness grade, the safer and more stable the operation of the deep foundation pit.

[0075] In this embodiment, the level 1 interval refers to the interval corresponding to the poor operation of the deep foundation pit; the level 2 interval refers to the interval corresponding to the poor operation of the deep foundation pit; the level 3 interval refers to the interval corresponding to the good operation of the deep foundation pit; and the level 4 interval refers to the interval corresponding to the excellent operation of the deep foundation pit.

[0076] In this embodiment, the toughness grade of the deep foundation pit is determined based on the influencing factors of the toughness of the deep foundation pit and the prediction parameters of the deep foundation pit. For example, the corresponding deep foundation pit prediction parameters are obtained based on the influencing factors of the toughness of the deep foundation pit. If the deep foundation pit prediction parameters are in the level 1 interval, the deep foundation pit toughness grade is level 1; if the deep foundation pit prediction parameters are in the level 2 interval, the deep foundation pit toughness grade is level 2.

[0077] In this embodiment, the deep foundation pit maintenance strategy refers to a strategy for ensuring stable operation of the deep foundation pit.

[0078] The working principle and beneficial effects of the above technical solution are: by obtaining the deep foundation pit prediction parameters corresponding to the matching algorithm, determining the deep foundation pit toughness grade and implementing the deep foundation pit maintenance strategy, accurate prediction of the deep foundation pit monitoring parameters is achieved, ensuring the safe operation of the deep foundation pit.

[0079] Embodiment 8: The embodiment of the present invention provides a method for predicting deep foundation pit monitoring parameters based on a neural network, further comprising: According to the monitorable parameters of deep foundation pit, the deep foundation pit toughness assessment index is determined, and the assessment index matrix is ​​established to conduct a comprehensive assessment of the toughness of the deep foundation pit. The deep foundation pit toughness assessment index includes: the deformation assessment index of the retaining structure, the support force characteristic assessment index, the settlement assessment index around the foundation pit, the groundwater level change assessment index, the foundation pit uplift assessment index and the foundation pit surrounding condition assessment index; The evaluation index matrix is ​​dimensionless; According to the experts' weighted scoring of deep foundation pit toughness assessment indicators, the subjective weights of deep foundation pit toughness assessment indicators are determined; Obtain the coefficient of variation value based on the predicted expected value and the predicted standard deviation, and determine the objective weight of the deep foundation pit toughness evaluation index; The weight coefficient of the subjective weight to the indicator weight and the weight coefficient of the objective weight to the indicator weight are set to determine the indicator weight of the deep foundation pit toughness assessment indicator; According to the deep foundation pit toughness assessment index and the index weight of the deep foundation pit toughness assessment index, the toughness assessment distance is obtained and dimensionless processed to determine the final toughness assessment distance; ; Where d represents the final toughness assessment distance; a1 represents the weight coefficient of subjective weight to indicator weight; w1i1 represents the subjective weight of the i1th deep foundation pit toughness assessment indicator; represents the objective weight of the i1th deep foundation pit toughness assessment index; si1 represents the average value of the i1th deep foundation pit toughness assessment index; ui1 represents the variance of the i1th deep foundation pit toughness assessment index; xi1j1 represents the j1th level quantitative value of the i1th deep foundation pit toughness assessment index; j1 represents the interval level, with values ​​of 1, 2, 3, and 4; u1 represents the predicted expected value; represents the prediction standard deviation; n1 represents the number of deep foundation pit toughness evaluation indicators; According to the actual parameters of the deep foundation pit, the final toughness assessment distance is matched with the deep foundation pit toughness grade, and a mapping table of the final toughness assessment distance and the deep foundation pit toughness grade is determined; The predicted final toughness assessment distance and the predicted deep foundation pit toughness grade corresponding to the deep foundation pit prediction parameters are obtained, and a corresponding judgment is made according to the final toughness assessment distance-deep foundation pit toughness grade mapping table. If it is determined that the predicted final toughness assessment distance corresponds to the predicted deep foundation pit toughness grade, then the deep foundation pit toughness grade determined according to the deep foundation pit prediction parameters is accurate.

[0080] In this embodiment, the deep foundation pit toughness assessment index refers to an index used to assess the toughness of the deep foundation pit, which is determined based on the monitorable parameters of the deep foundation pit, and includes: an assessment index for deformation of the retaining structure, an assessment index for support force characteristics, an assessment index for settlement around the foundation pit, an assessment index for groundwater level change, an assessment index for foundation pit uplift, and an assessment index for the condition around the foundation pit.

[0081] In this embodiment, the evaluation index matrix refers to a matrix determined according to the deep foundation pit toughness evaluation index, which is used for subsequent calculations.

[0082] In this embodiment, the evaluation index matrix is ​​dimensionless to simplify calculation.

[0083] In this embodiment, the subjective weight refers to the indicator weight value determined by experts' weighted scoring of the deep foundation pit toughness assessment indicator.

[0084] In this embodiment, the objective weight refers to the indicator weight value determined according to the coefficient of variation value.

[0085] In this embodiment, the coefficient of variation value refers to the absolute value of the data dispersion degree of the deep foundation pit prediction parameter, which is determined according to the prediction expected value and the prediction standard deviation.

[0086] In this embodiment, the weight coefficient refers to the influence of the subjective weight and the objective weight on the indicator weight.

[0087] In this embodiment, the indicator weight refers to the influence weight of the deep foundation pit toughness assessment indicator.

[0088] In this embodiment, the toughness assessment distance refers to the distance value used for deep foundation pit toughness assessment obtained according to the deep foundation pit toughness assessment index and the indicator weight of the deep foundation pit toughness assessment index. It refers to the relative distance between the assessment value and the optimal value. The toughness of the deep foundation pit is reflected according to the toughness assessment distance, that is, the smaller the toughness assessment distance, the higher the corresponding deep foundation pit toughness.

[0089] In this embodiment, the final toughness assessment distance refers to data determined by performing dimensionless processing on the toughness assessment distance.

[0090] In this embodiment, the final toughness assessment distance-deep foundation pit toughness grade mapping table refers to a mapping table obtained by corresponding the final toughness assessment distance with the deep foundation pit toughness grade according to the actual parameters of the deep foundation pit, and is used to determine whether the deep foundation pit toughness grade determined according to the deep foundation pit prediction parameters is accurate.

[0091] In this embodiment, the predicted final toughness assessment distance refers to the final toughness assessment distance obtained according to the deep foundation pit prediction parameters; the predicted deep foundation pit toughness grade refers to the deep foundation pit toughness grade obtained according to the deep foundation pit prediction parameters.

[0092] In this embodiment, a corresponding judgment is made according to the final toughness assessment distance-deep foundation pit toughness grade mapping table. For example, the final toughness assessment distance d1 is predicted, and a corresponding judgment is made according to the final toughness assessment distance-deep foundation pit toughness grade mapping table to determine that the corresponding deep foundation pit toughness grade is level 1. If the predicted deep foundation pit toughness grade is level 1, it is determined that the predicted final toughness assessment distance corresponds to the predicted deep foundation pit toughness grade, and the deep foundation pit toughness grade determined according to the deep foundation pit prediction parameters is accurate; if the predicted deep foundation pit toughness grade is not level 1, it is determined that the predicted final toughness assessment distance does not correspond to the predicted deep foundation pit toughness grade, and the deep foundation pit toughness grade determined according to the deep foundation pit prediction parameters is inaccurate. The corresponding deep foundation pit toughness grade is determined according to the predicted final toughness assessment distance as the deep foundation pit toughness grade determined by the deep foundation pit prediction parameters.

[0093] The working principle and beneficial effects of the above technical solution are: by determining the deep foundation pit toughness evaluation index, establishing the evaluation index matrix, and determining the final toughness evaluation distance, the accuracy of the deep foundation pit toughness grade determined according to the deep foundation pit prediction parameters is ensured.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A prediction method for deep foundation pit monitoring parameters based on neural network, characterized in that: include: Step 1: Obtain risk factors of deep foundation pits and determine monitoring parameters of deep foundation pits; Step 2: Set the deep foundation pit monitoring frequency, obtain the deep foundation pit monitoring data according to the deep foundation pit monitoring parameters, and perform data preprocessing; Step 3: Obtain the neural network algorithm of deep foundation pit monitoring parameters, build a deep foundation pit monitoring parameter prediction model and perform model training based on the deep foundation pit monitoring data after data preprocessing, obtain the deep foundation pit prediction parameters, and determine the matching algorithm of the deep foundation pit monitoring parameters based on the actual parameters of the deep foundation pit; Step 4: Obtain the deep foundation pit prediction parameters corresponding to the matching algorithm, determine the deep foundation pit toughness grade and implement the deep foundation pit maintenance strategy.

2. The method for predicting deep foundation pit monitoring parameters based on neural network according to claim 1 is characterized in that: Obtain risk factors for deep foundation pits and determine monitoring parameters for deep foundation pits, including: Obtain risk factors for deep foundation pits and determine the monitorable parameters of deep foundation pits, including deformation parameters of retaining structures, stress characteristic parameters of supports, settlement parameters around foundation pits, groundwater level change parameters, foundation pit uplift parameters, and parameters of conditions around foundation pits; According to the actual monitoring equipment of the deep foundation pit, some monitoring parameters are selected from the monitorable parameters of the deep foundation pit and recorded as deep foundation pit monitoring parameters, wherein the deep foundation pit monitoring parameters include: deep foundation pit settlement monitoring parameters, deep foundation pit support end horizontal displacement monitoring parameters and deep foundation pit support axial force monitoring parameters.

3. The method for predicting deep foundation pit monitoring parameters based on neural network according to claim 1 is characterized in that: Set the deep foundation pit monitoring frequency, obtain the deep foundation pit monitoring data according to the deep foundation pit monitoring parameters and perform data preprocessing, including: Obtain the deep foundation pit monitoring accuracy requirements, set the deep foundation pit monitoring frequency, and obtain the deep foundation pit monitoring data based on the deep foundation pit monitoring parameters and the actual deep foundation pit monitoring equipment; The deep foundation pit monitoring data is preprocessed, wherein the data preprocessing includes: data cleaning processing, data standardization processing and data formatting processing.

4. The method for predicting deep foundation pit monitoring parameters based on neural network according to claim 1 is characterized in that: Obtain the neural network algorithm of deep foundation pit monitoring parameters, build a deep foundation pit monitoring parameter prediction model and perform model training based on the deep foundation pit monitoring data after data preprocessing, including: A neural network algorithm for obtaining deep foundation pit monitoring parameters, wherein the neural network algorithm includes: BP neural network algorithm, GA-BP neural network algorithm, NARX dynamic neural network algorithm and Elman neural network algorithm; A deep foundation pit monitoring parameter prediction model is constructed, and model parameters of the deep foundation pit monitoring parameter prediction model are determined based on the deep foundation pit monitoring data after data preprocessing and the neural network algorithm, and model training is performed on the deep foundation pit monitoring parameter prediction model; sorting the deep foundation pit monitoring data after data preprocessing according to the data acquisition time of the deep foundation pit monitoring data after data preprocessing; According to the data sorting results, the deep foundation pit monitoring data with a preset proportion of data preprocessing is selected as training samples; The deep foundation pit monitoring parameter prediction model is trained based on the training samples.

5. The method for predicting deep foundation pit monitoring parameters based on neural network according to claim 4 is characterized in that: Also includes: The deep foundation pit monitoring data after data preprocessing of the remaining training samples in the data sorting results are selected as verification samples, and the deep foundation pit monitoring parameter prediction model is verified to obtain the prediction accuracy of the deep foundation pit monitoring parameter prediction model; When the prediction accuracy of the deep foundation pit monitoring parameter prediction model is not lower than the set accuracy, it is determined that the model training of the deep foundation pit monitoring parameter prediction model is completed.

6. The method for predicting deep foundation pit monitoring parameters based on neural network according to claim 5 is characterized in that: Obtain the predicted parameters of deep foundation pits and determine the matching algorithm of deep foundation pit monitoring parameters according to the actual parameters of deep foundation pits, including: The deep foundation pit monitoring parameters are predicted according to the deep foundation pit monitoring parameter prediction model after model training, so as to obtain the deep foundation pit prediction parameters; The actual parameters of the deep foundation pit are obtained according to the training samples and the verification samples, the error value between the actual parameters of the deep foundation pit and the predicted parameters of the deep foundation pit is determined, and the neural network algorithm corresponding to the smallest error value is selected, which is recorded as the matching algorithm of the deep foundation pit monitoring parameters.

7. The method for predicting deep foundation pit monitoring parameters based on neural network according to claim 1 is characterized in that: Obtain the deep foundation pit prediction parameters corresponding to the matching algorithm, determine the deep foundation pit toughness level and implement the deep foundation pit maintenance strategy, including: Obtain the deep foundation pit prediction parameters corresponding to the matching algorithm; Obtain the parameter increment of the actual parameters of the deep foundation pit, record the parameter increment as an independent parameter according to the parameter type of the deep foundation pit monitoring parameters, and perform a normality test; Obtain the parameter test value of the normality test. When the parameter test value is not lower than the preset test value, it is determined that the corresponding independent parameter conforms to the normal distribution; The corresponding independent parameters that conform to the normal distribution are divided into intervals for analysis to obtain the toughness influencing factors of the deep foundation pit, among which the toughness influencing factors include: robustness influencing factors, redundancy influencing factors, recovery influencing factors, wisdom influencing factors and adaptability influencing factors; The predicted expected value and predicted standard deviation of deep foundation pit prediction parameters are obtained according to the normality test; The interval level is divided according to the predicted expected value and the predicted standard deviation, and the interval corresponding to plus or minus 1 times the predicted standard deviation near the predicted expected value is recorded as the level 1 interval; The intervals corresponding to plus or minus 2 times the predicted standard deviation near the predicted expected value are recorded as level 2 intervals; The intervals corresponding to plus or minus three times the predicted standard deviation around the predicted expected value are recorded as level 3 intervals; The corresponding intervals outside the three intervals are recorded as level 4 intervals; According to the influencing factors of deep foundation pit toughness and the prediction parameters of deep foundation pit, the deep foundation pit toughness grade is determined and the deep foundation pit maintenance strategy is implemented.

8. The method for predicting deep foundation pit monitoring parameters based on neural network according to claim 7 is characterized in that: Also includes: According to the monitorable parameters of deep foundation pit, the deep foundation pit toughness assessment index is determined, and the assessment index matrix is ​​established to conduct a comprehensive assessment of the toughness of the deep foundation pit. The deep foundation pit toughness assessment index includes: the deformation assessment index of the retaining structure, the support force characteristic assessment index, the settlement assessment index around the foundation pit, the groundwater level change assessment index, the foundation pit uplift assessment index and the foundation pit surrounding condition assessment index; The evaluation index matrix is ​​dimensionless; According to the experts' weighted scoring of deep foundation pit toughness assessment indicators, the subjective weights of deep foundation pit toughness assessment indicators are determined; Obtain the coefficient of variation value based on the predicted expected value and the predicted standard deviation, and determine the objective weight of the deep foundation pit toughness evaluation index; The weight coefficient of the subjective weight to the indicator weight and the weight coefficient of the objective weight to the indicator weight are set to determine the indicator weight of the deep foundation pit toughness assessment indicator; According to the deep foundation pit toughness assessment index and the index weight of the deep foundation pit toughness assessment index, the toughness assessment distance is obtained and dimensionless processed to determine the final toughness assessment distance; ; Where d represents the final toughness assessment distance; a1 represents the weight coefficient of the subjective weight to the indicator weight; w1i1 represents the subjective weight of the i1th deep foundation pit toughness assessment indicator; represents the objective weight of the i1th deep foundation pit toughness assessment index; si1 represents the average value of the i1th deep foundation pit toughness assessment index; ui1 represents the variance of the i1th deep foundation pit toughness assessment index; xi1j1 represents the j1th level quantitative value of the i1th deep foundation pit toughness assessment index; j1 represents the interval level, with values ​​of 1, 2, 3, and 4; u1 represents the predicted expected value; represents the prediction standard deviation; n1 represents the number of deep foundation pit toughness evaluation indicators; According to the actual parameters of the deep foundation pit, the final toughness assessment distance is matched with the deep foundation pit toughness grade, and a mapping table of the final toughness assessment distance and the deep foundation pit toughness grade is determined; The predicted final toughness assessment distance and the predicted deep foundation pit toughness grade corresponding to the deep foundation pit prediction parameters are obtained, and a corresponding judgment is made according to the final toughness assessment distance-deep foundation pit toughness grade mapping table. If it is determined that the predicted final toughness assessment distance corresponds to the predicted deep foundation pit toughness grade, then the deep foundation pit toughness grade determined according to the deep foundation pit prediction parameters is accurate.

Images