An intelligent detection method and system based on building foundation settlement

Abstract

The application discloses a kind of based on building ground settlement intelligent detection method and system, it is related to ground settlement detection technical field, by the pre-processing of original feature dataset RAW, obtain reliable standard feature dataset SFD, lay high-quality data foundation for subsequent accurate analysis.Based on average longitudinal wave velocity AVG_V p The dynamic elastic modulus Ed of the pile body is calculated, and the elastic compression amount S caused by the load of the pile body is calculated according to the dynamic elastic modulus Ed p , and the quantitative separation of the deformation of the pile body itself is realized. The soil settlement amount S t is separated from the average total ground settlement amount AVG_S o , and the multi-dimensional risk index F reflecting the comprehensive risk is calculated by combining the soil settlement rate T_S v in the current detection period, and the clear decision instruction Air is generated by comparing the risk index F with the preset risk threshold interval θ. Finally, the feedback calibration mechanism of man-machine combination is established, and the long-term reliability of the precise grading early warning and system judgment ability is realized.

Description

Technical Field

[0001] This invention relates to the field of foundation settlement detection technology, specifically to an intelligent detection method and system for building foundation settlement. Background Technology

[0002] In the field of civil engineering, ensuring the long-term stability and safety of buildings is a crucial core issue. All the loads of a building are ultimately transferred to the soil beneath through its foundation. Therefore, as the hub connecting the superstructure and the earth, the health of the foundation directly determines the safety of the entire building. Among the many indicators for evaluating the working condition of the foundation, its vertical displacement, i.e., foundation settlement, is the most critical and direct measurement. Minor settlement is a normal process of mutual adaptation between the building and the foundation; however, excessive settlement may indicate potential risks. Therefore, regular and accurate monitoring of building foundation settlement has become an indispensable part of modern engineering safety assurance systems. With the acceleration of urbanization, a large number of high-rise and super high-rise buildings are springing up. In these scenarios, whether it is the newly constructed building itself or existing buildings affected by surrounding construction, the foundation settlement exhibits unprecedented complexity. This poses a challenge to traditional monitoring methods and creates a need for a deeper analysis and understanding of the settlement process—that is, a more intelligent and systematic method and system for detecting building foundation settlement.

[0003] However, current methods and systems for detecting building foundation settlement still have significant shortcomings and deficiencies in addressing the aforementioned problems. First, existing automated monitoring technologies often only provide a macroscopic, general estimate of the total foundation settlement. This single value acts like a black box, failing to reveal the underlying mechanism of settlement, i.e., it cannot answer the core question of whether settlement is contributed by the elastic compression of the pile foundation itself or by the compaction deformation of the foundation soil, leading to ambiguity in the data's origin. Second, although engineers can use equipment such as multi-channel cross-hole ultrasonic instruments to detect the integrity of the pile concrete, this work is usually independent of settlement monitoring, creating data silos. Effective automated correlation analysis of these two types of data, which are vastly different in nature and have different periods, largely relies on the engineer's personal experience and post-hoc judgment, lacking real-time and quantitative standards. Finally, most existing automated early warning systems are based on a fixed, absolute threshold for the total settlement amount or its rate. This one-size-fits-all approach cannot effectively distinguish settlement risks of different causes and modes, and is prone to misjudgment of risks. It may generate unnecessary alarms for benign settlement, or it may be indifferent to the slow but serious accumulation of risks. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for intelligent detection of building foundation settlement, which solves the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent detection of building foundation settlement, comprising the following steps:

[0006] S1. By reading the construction data file and collecting monitoring data using different sensor devices deployed on the building foundation, a raw feature dataset RAW is constructed, and then the dataset is preprocessed to obtain a standard feature dataset SFD.

[0007] S2. Based on the standard feature dataset SFD, calculate the dynamic elastic modulus Ed of the pile body according to the theory of elasticity. Then, using the calculated dynamic elastic modulus Ed, calculate the load-induced elastic compression S of the pile body according to Hooke's law in mechanics of materials. p ;

[0008] S3, elastic compression based on pile load S p Using the standard feature dataset SFD, calculate the soil settlement S. o and the soil settlement rate T_S during the current monitoring period v And based on the soil settlement S o Elastic compression S caused by pile load p and the soil settlement rate T_S during the current monitoring period v Calculate the risk index F;

[0009] S4. Compare the risk index F with the preset risk threshold range θ, and generate a decision instruction Air based on the comparison results;

[0010] S5. Execute decision-making actions based on decision command Air, and conduct manual surveys periodically to obtain evaluation results J. Execute optimization actions based on the difference between evaluation results J and decision command Air. Store the original feature dataset RAW, risk index F, evaluation results J, and current risk threshold interval θ in the historical database HIS, and perform iterative optimization of the risk threshold interval θ based on the historical database HIS.

[0011] Preferably, S1 includes S11 and S12;

[0012] S11. Obtain the pile design length H, pile cross-sectional area A, pile concrete density ρ, Poisson's ratio ν, and building inspection pile equivalent load W by reading the construction data file;

[0013] Displacement sensors deployed on the building foundation platform are used to calculate the total foundation settlement S according to a preset observation time window Q. tCollect data to obtain a set of total foundation settlement data SS within a preset detection period of days t. t ;

[0014] By using a multi-channel trans-hole ultrasonic probe pre-embedded inside the building foundation, the total foundation settlement S is collected by a displacement sensor. t Simultaneous longitudinal wave velocity V p The data collection yields a set of P-wave velocity data SV within a preset detection period of days t. p ;

[0015] Combined with the total foundation settlement data set SS t Longitudinal wave velocity data set SV p The original feature dataset RAW is constructed by taking the pile design length H, pile cross-sectional area A, pile concrete density ρ, Poisson's ratio ν, and the equivalent load W of the building inspection pile.

[0016] Preferably, in step S12, the total settlement data set SS in the original feature dataset RAW is processed. t and P-wave velocity data set SV p Preprocessing is required;

[0017] Calculate the total foundation settlement data set SS separately t and P-wave velocity data set SV p The standard deviation of settlement σS t and the standard deviation of P-wave velocity σV p For the total foundation settlement data set SS t The moving standard deviation σS of the settlement exceeds 5 times. t The values ​​were discarded based on the total foundation settlement data set SS. t The remaining data is used to calculate the average total foundation settlement AVG_S t For the P-wave velocity data set SV p The P-wave velocity exceeds 5 times the standard deviation σV p The values ​​were discarded based on the P-wave velocity data set SV. p The average P-wave velocity AVG_V is calculated from the remaining data. p ;

[0018] Combined with the average total foundation settlement AVG_S t Average P-wave velocity AVG_V p The standard feature dataset SFD is constructed by considering the pile design length H, pile cross-sectional area A, pile concrete density ρ, Poisson's ratio ν, and the equivalent load W of the building inspection pile.

[0019] Preferably, S2 includes S21 and S22;

[0020] S21. Average P-wave velocity AVG_V based on the standard feature dataset SFD p The density ρ and Poisson's ratio ν of the pile concrete, according to the theory of elasticity, for a homogeneous, isotropic, infinite elastic body, the average longitudinal wave velocity AVG_V inside it. p The calculation formula is reversed to obtain the dynamic elastic modulus Ed of the pile body;

[0021] The formula for calculating the dynamic elastic modulus Ed of the pile is as follows:

[0022] ;

[0023] In the formula, ε represents the minimum value to prevent the denominator from being zero, and its specific value is 1*10. -8 .

[0024] Preferably, in step S22, based on the dynamic elastic modulus Ed of the pile, combined with the pile design length H, pile cross-sectional area A, and equivalent load W of the building inspection pile in the standard feature dataset SFD, the load-induced elastic compression S of the pile is calculated according to Hooke's law in mechanics of materials. p ;

[0025] Among them, the elastic compression S caused by pile load p The calculation formula is as follows:

[0026] .

[0027] Preferably, S3 includes S31 and S32;

[0028] S31, Elastic compression based on pile load S p The average total foundation settlement AVG_S in the standard feature dataset SFD t By calculating the average total foundation settlement AVG_S t Elastic compression S caused by pile load p The difference is used to obtain the soil settlement S. o By calculating the soil settlement S o The ratio of the soil settlement rate T_S to the preset detection period number t is used to obtain the soil settlement rate T_S within the current detection period. v , where T represents the timestamp of the current detection period.

[0029] Preferably, in step S32, based on the soil settlement So, the elastic compression Sp caused by the pile load, and the soil settlement rate T_Sv during the current monitoring period, the relative contribution rate factor reflecting the dominance of building settlement is obtained by dividing the soil settlement So by the average total foundation settlement AVG_St. The soil settlement rate T_Sv in the current detection cycle is combined with the hyperbolic tangent function and a preset soil settlement rate control factor λ is introduced to construct a risk index evaluation model and calculate the risk index F.

[0030] The risk index evaluation model expression is as follows:

[0031] ;

[0032] In the formula, tanh represents the hyperbolic tangent function, and λ represents the soil settlement rate control factor set by professionals in the field according to the building completion stage.

[0033] Preferably, S4 includes S41;

[0034] S41. Based on the risk index F, compare it with the preset risk threshold interval θ, and generate a decision instruction Air according to the comparison result, wherein the risk threshold interval θ includes the first-level risk threshold θ1 and the second-level risk threshold θ2.

[0035] If the risk index F < the first-level risk threshold θ1, then the decision instruction Air is generated to determine that the building foundation settlement is a natural phenomenon within a safe range, and the zero-level decision action A0 is generated to indicate that no intervention is required.

[0036] If the first-level risk threshold θ1 ≤ risk index F < second-level risk threshold θ2, then the decision instruction Air is generated to determine that although the building foundation settlement is within the safe range, it is an abnormal phenomenon, and the first-level decision action A1 is generated to automatically generate a report and send it to the central control management platform, suggesting that the management personnel increase the detection frequency;

[0037] If the risk index F ≥ the secondary risk threshold θ2, then the decision instruction Air is generated to determine that the building foundation settlement is within the dangerous range, and the secondary decision action A2 is generated to automatically generate an alarm to notify relevant personnel via APP and SMS, and send a file containing the original feature dataset RAW, the standard feature dataset SFD, the dynamic elastic modulus of the pile Ed, and the elastic compression S caused by the pile load to the central control management platform. p Soil settlement S o and the soil settlement rate T_S during the current monitoring period v The data report recommends manual verification.

[0038] Preferably, S5 includes S51;

[0039] S51. Execute decision-making actions based on decision command Air, conduct regular manual surveys and obtain evaluation results J, and determine whether to execute optimization actions based on the difference between evaluation results J and decision command Air.

[0040] If the decision action level generated in the decision instruction Air is a level 0 decision action A0 or a level 1 decision action A1, then according to the preset manual survey cycle, manual surveys will be carried out periodically and evaluation results J will be obtained.

[0041] If the decision action level generated in the decision instruction Air is a level 2 decision action A2, then immediately conduct a manual inspection according to the content of level 2 decision action A2 and obtain the evaluation result J;

[0042] If the assessment result J is consistent with the decision instruction Air, then relevant staff will be arranged to generate and implement a solution based on the building foundation settlement problem.

[0043] If the assessment result J is inconsistent with the decision instruction Air, check whether there are any faults in the displacement sensors and multi-channel trans-hole ultrasonic instruments deployed on the building foundation platform.

[0044] If the displacement sensor and multi-channel cross-hole ultrasonic instrument deployed on the building foundation platform are faulty, replace them with fault-free displacement sensor and multi-channel cross-hole ultrasonic instrument.

[0045] If the displacement sensor and multi-channel trans-hole ultrasonic instrument deployed on the building foundation are not faulty, the preset risk threshold range θ value will be adjusted based on the assessment result J and the historical database HIS.

[0046] The original feature dataset RAW, risk index F, evaluation result J, and current risk threshold interval θ are stored in the historical database HIS. When the sample data of the currently detected building in the historical database HIS does not reach the preset number of samples, professionals in the field select other building sample data that can be used for the current detected building from the historical database HIS based on the original feature dataset RAW and the current risk threshold interval θ of the current detected building, and combine it with the sample data of the current detected building to perform logistic regression training. Using the risk index F as input and the evaluation result j as label, the risk level classification model is retrained. Based on the new risk level classification model, the values ​​of the first-level risk threshold θ1 and the second-level risk threshold θ2 are automatically calculated and updated, and professionals in the field manually verify the newly generated first-level risk threshold θ1 and second-level risk threshold θ2. When the sample data of the currently detected building in the historical database HIS reaches the preset number of samples, the logistic regression training is performed entirely using the sample data of the current detected building, and professionals in the field manually verify the newly generated first-level risk threshold θ1 and second-level risk threshold θ2.

[0047] A smart detection system for building foundation settlement includes a data acquisition and preprocessing module, a pile body analysis module, a risk index calculation module, an evaluation module, and an optimization module.

[0048] The data acquisition and preprocessing module reads construction data files and collects monitoring data using different sensor devices deployed on the building foundation to construct the raw feature dataset RAW. The dataset is then preprocessed to obtain the standard feature dataset SFD.

[0049] The pile analysis module calculates the dynamic elastic modulus Ed of the pile body based on the standard feature dataset SFD and the theory of elasticity. Then, using the calculated dynamic elastic modulus Ed, it calculates the load-induced elastic compression S of the pile body according to Hooke's law in mechanics of materials. p ;

[0050] The risk index calculation module calculates the risk index based on the elastic compression S induced by pile load. p Using the standard feature dataset SFD, calculate the soil settlement S. o and the soil settlement rate T_S during the current monitoring period v And based on the soil settlement S o Elastic compression S caused by pile load p and the soil settlement rate T_S during the current monitoring period v Calculate the risk index F;

[0051] The assessment module compares the risk index F with a preset risk threshold range θ and generates a decision instruction Air based on the comparison results.

[0052] The optimization module executes decision-making actions based on decision instructions Air, and periodically conducts manual surveys to obtain evaluation results J. Based on the difference between the evaluation results J and the decision instructions Air, it executes optimization actions, stores the original feature dataset RAW, risk index F, evaluation results J and the current risk threshold interval θ in the historical database HIS, and performs iterative optimization of the risk threshold interval θ based on the historical database HIS.

[0053] This invention provides a method and system for intelligent detection of building foundation settlement, which has the following advantages:

[0054] (1) By analyzing the original total foundation settlement S t With longitudinal wave velocity V pRigorous preprocessing lays a stable and reliable data foundation for the entire analysis process. Based on this, it innovatively fuses two types of monitoring data with different characteristics using algorithms, achieving quantitative separation of the causes of total settlement and further constructing a multi-dimensional risk index F that can comprehensively assess the magnitude and development trend of risk. The system makes intelligent, tiered decision responses based on this risk index F, and ultimately includes a human-machine collaborative feedback calibration closed loop to ensure its long-term accuracy and adaptability. This complete technology chain successfully solves the core technical problems of ambiguity in data causes and single-dimensional risk assessment in traditional monitoring methods, providing a new and more technologically advanced solution for the safety management of building foundations.

[0055] (2) By deploying displacement sensors at key locations in the building foundation and combining them with cross-hole ultrasonic probes embedded in the pile body, the total settlement S of the building foundation is measured. t The longitudinal wave velocity V of the pile concrete p Long-term, synchronous data acquisition was conducted. The large amount of raw feature datasets (RAW) acquired were subjected to rigorous 5-sigma outlier removal and periodic averaging to ensure that all data input into the core computational model were stable, reliable, and representative standard feature datasets (SFD), thus improving the diagnostic accuracy of the entire system from the source. Based on this high-quality data, the acquired average P-wave velocity (AVG_V) was utilized... p By using physical formulas to calculate the dynamic elastic modulus Ed of the pile body, which reflects the current true state of the pile, the elastic compression S caused by the pile load can be further quantitatively estimated. p This allows for the quantitative separation of the causes of total settlement. Once the system can clearly distinguish the contributions of piles and soil to the settlement, it can then calculate the average total foundation settlement AVG_S. t The soil settlement S was accurately separated from the soil. o And combined with the soil settlement rate T_S during the current testing period v Through an innovatively constructed multidimensional risk index evaluation model, the final risk index F is calculated. The significant benefit of this series of coherent calculation processes is that it successfully transforms a vague macroscopic sedimentation phenomenon into a specific, quantifiable, and single diagnostic indicator F that can profoundly reveal the source, magnitude, and speed of development of risk, providing a solid data basis for subsequent scientific decision-making.

[0056] (3) After obtaining the risk index F, it is compared with a preset multi-level risk threshold θ to achieve automated and high-precision graded early warning. This intelligent decision-making mechanism completely changes the drawback of the traditional one-size-fits-all alarm mode. It can provide timely and accurate alarms for truly dangerous signals, while remaining silent for benign and predictable settlement, greatly reducing the false alarm rate and providing managers with clear, explicit, and executable action instructions. Furthermore, to ensure the long-term reliability of the system throughout the entire building life cycle monitoring process, the system's automatic decision-making instructions Air are compared with the authoritative manual survey evaluation results J of domain experts. Once a deviation is found, feedback-based manual optimization can be performed on the data acquisition equipment or the risk threshold range θ. This mechanism ensures that the system is not a fixed and unchanging tool, but an intelligent detection technology that can be continuously corrected and improved with the help of human wisdom during use, and whose judgment ability is continuously improved, ultimately achieving a high degree of unity between monitoring effect and engineering practice. Attached Figure Description

[0057] Figure 1 This is a schematic diagram illustrating the steps of an intelligent detection method for building foundation settlement according to the present invention.

[0058] Figure 2 This is a schematic diagram of a block diagram of an intelligent detection system for building foundation settlement according to the present invention;

[0059] Figure 3 This is a graph showing the changes in the risk index F curve. Detailed Implementation

[0060] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0061] Example 1

[0062] This invention provides a method for intelligent detection of building foundation settlement. Please refer to [link / reference]. Figure 1 This includes the following steps:

[0063] S1. By reading the construction data file and collecting monitoring data using different sensor devices deployed on the building foundation, a raw feature dataset RAW is constructed, and then the dataset is preprocessed to obtain a standard feature dataset SFD.

[0064] S2. Based on the standard feature dataset SFD, calculate the dynamic elastic modulus Ed of the pile body according to the theory of elasticity. Then, using the calculated dynamic elastic modulus Ed, calculate the load-induced elastic compression S of the pile body according to Hooke's law in mechanics of materials. p ;

[0065] S3, elastic compression based on pile load S p Using the standard feature dataset SFD, calculate the soil settlement S. o and the soil settlement rate T_S during the current monitoring period v And based on the soil settlement S o Elastic compression S caused by pile load p and the soil settlement rate T_S during the current monitoring period v Calculate the risk index F;

[0066] S4. Compare the risk index F with the preset risk threshold range θ, and generate a decision instruction Air based on the comparison results;

[0067] S5. Execute decision-making actions based on decision command Air, and conduct manual surveys periodically to obtain evaluation results J. Execute optimization actions based on the difference between evaluation results J and decision command Air. Store the original feature dataset RAW, risk index F, evaluation results J, and current risk threshold interval θ in the historical database HIS, and perform iterative optimization of the risk threshold interval θ based on the historical database HIS.

[0068] In this embodiment, a composite foundation consisting of natural soil and artificial piles is used as an example. The standard feature dataset SFD is obtained by preprocessing the original feature dataset RAW, and the elastic compression S caused by pile load is calculated based on this dataset. p This method enables precise quantitative separation of settlement sources, forming a more comprehensive and in-depth diagnostic system than traditional single-source settlement monitoring. Furthermore, the calculated multidimensional risk index F reflects the nature and development trend of foundation risks in real time and comprehensively, ensuring that safety assessments can make scientific judgments based on actual risk changes. Finally, by comparing the risk index F with the risk threshold range θ to generate clear decision instructions Air, and establishing a feedback calibration mechanism based on manual survey and assessment results J, accurate graded early warnings can be achieved, and the accuracy of system judgments can be optimized in a timely manner. The overall scheme not only enhances the diagnostic depth of the settlement monitoring process but also allows for effective logical optimization based on real-world feedback, effectively addressing the shortcomings of traditional monitoring schemes, such as ambiguity in causes and misjudgment of risks. Traditional methods often only focus on a general total foundation settlement S. t This method measures the total foundation settlement S t With longitudinal wave velocity V pDeep integration and analysis provide a more flexible and intelligent solution for the safety management of building foundations, avoiding the limitations of traditional solutions that cannot accurately assess risks, and ensuring the scientific nature and reliability of engineering monitoring.

[0069] Example 2

[0070] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically: S1 includes S11 and S12;

[0071] S11. Obtain the pile design length H, pile cross-sectional area A, pile concrete density ρ, Poisson's ratio ν, and building inspection pile equivalent load W by reading the construction data file;

[0072] Displacement sensors deployed on the building foundation platform are used to calculate the total foundation settlement S according to a preset observation time window Q. t Collect data to obtain a set of total foundation settlement data SS within a preset detection period of days t. t ;

[0073] By using a multi-channel trans-hole ultrasonic probe pre-embedded inside the building foundation, the total foundation settlement S is collected by a displacement sensor. t Simultaneous longitudinal wave velocity V p The data collection yields a set of P-wave velocity data SV within a preset detection period of days t. p ;

[0074] Combined with the total foundation settlement data set SS t Longitudinal wave velocity data set SV p The original feature dataset RAW is constructed by taking the pile design length H, pile cross-sectional area A, pile concrete density ρ, Poisson's ratio ν, and the equivalent load W of the building inspection pile.

[0075] S12. The total settlement data set SS in the original feature dataset RAW t and P-wave velocity data set SV p Preprocessing is required;

[0076] Calculate the total foundation settlement data set SS separately t and P-wave velocity data set SV p The standard deviation of settlement σS t and the standard deviation of P-wave velocity σV p For the total foundation settlement data set SS t The moving standard deviation σS of the settlement exceeds 5 times. t The values ​​were discarded based on the total foundation settlement data set SS. t The remaining data is used to calculate the average total foundation settlement AVG_St For the P-wave velocity data set SV p The P-wave velocity exceeds 5 times the standard deviation σV p The values ​​were discarded based on the P-wave velocity data set SV. p The average P-wave velocity AVG_V is calculated from the remaining data. p ;

[0077] Combined with the average total foundation settlement AVG_S t Average P-wave velocity AVG_V p The standard feature dataset SFD is constructed by considering the pile design length H, pile cross-sectional area A, pile concrete density ρ, Poisson's ratio ν, and the equivalent load W of the building inspection pile.

[0078] S2 includes S21 and S22;

[0079] S21. Average P-wave velocity AVG_V based on the standard feature dataset SFD p The density ρ and Poisson's ratio ν of the pile concrete, according to the theory of elasticity, for a homogeneous, isotropic, infinite elastic body, the average longitudinal wave velocity AVG_V inside it. p The calculation formula is reversed to obtain the dynamic elastic modulus Ed of the pile body;

[0080] The formula for calculating the dynamic elastic modulus Ed of the pile is as follows:

[0081] ;

[0082] In the formula, ε represents the minimum value to prevent the denominator from being zero, and its specific value is 1*10. -8 ;

[0083] S22. Based on the dynamic elastic modulus Ed of the pile, combined with the pile design length H, pile cross-sectional area A, and equivalent load W of the building inspection pile in the standard feature dataset SFD, the load-induced elastic compression S of the pile is calculated according to Hooke's law in mechanics of materials. p ;

[0084] Among them, the elastic compression S caused by pile load p The calculation formula is as follows:

[0085] .

[0086] In this embodiment, displacement sensors deployed on the building foundation platform are used to monitor the total foundation settlement S according to a preset observation time window Q. t Data collection is performed to obtain a dataset SS of the total foundation settlement within a preset detection period of days t. tMeanwhile, the total foundation settlement S was collected by using a multi-channel trans-hole ultrasonic transducer probe pre-embedded inside the building foundation, along with a displacement sensor. t Simultaneous synchronization of longitudinal wave velocity V p The data set SV of the longitudinal wave velocity within the same detection period (day t) is obtained. p In addition, static parameters such as pile design length H, pile cross-sectional area A, pile concrete density ρ, Poisson's ratio ν, and the current total building load W are obtained from the construction data files. Finally, all the above dynamic data sets are combined with the static parameters to construct the original feature dataset RAW. The original feature dataset RAW is preprocessed, and the total foundation settlement data set SS is calculated separately. t The standard deviation of settlement σS t and P-wave velocity data set SV p Standard deviation of longitudinal wave velocity σV p Based on the 5-sigma principle, outliers exceeding 5 times their respective standard deviations were removed from both datasets. Then, based on the remaining valid data after the removal, the average total foundation settlement AVG_S was calculated. t and average longitudinal wave velocity AVG_V p Finally, these two highly reliable average values ​​are integrated with other static parameters to construct the final standard feature dataset SFD used for computation. The average P-wave velocity AVG_V from the standard feature dataset SFD is then used. p The pile concrete density ρ and Poisson's ratio ν are used as the basis for inverse calculations based on the classical formulas describing the relationship between these three factors and the elastic modulus in elasticity theory, thus obtaining the dynamic elastic modulus Ed of the pile. Based on the obtained dynamic elastic modulus Ed, and combined with the pile design length H, pile cross-sectional area A, and current total building load W from the standard feature dataset SFD, the load-induced elastic compression S of the pile is calculated according to Hooke's law in mechanics of materials. p .

[0087] Example 3

[0088] This embodiment is an explanation based on Embodiment 2. Please refer to it. Figure 1 Specifically: S3 includes S31 and S32;

[0089] S31, Elastic compression based on pile load S p The average total foundation settlement AVG_S in the standard feature dataset SFD t By calculating the average total foundation settlement AVG_S t Elastic compression S caused by pile load p The difference is used to obtain the soil settlement S. o By calculating the soil settlement S oThe ratio of the soil settlement rate T_S to the preset detection period number t is used to obtain the soil settlement rate T_S within the current detection period. v , where T represents the timestamp of the current detection period;

[0090] Among them, soil settlement S o The calculation expression is as follows:

[0091] ;

[0092] Soil settlement rate T_S during the current monitoring period v The calculation expression is as follows:

[0093] ;

[0094] S32, based on the soil settlement S o Elastic compression S caused by pile load p And the soil settlement rate T_Sv during the current monitoring period, by expressing the soil settlement amount S o With average total foundation settlement AVG_S t By dividing the two, we obtain the relative contribution rate factor, which reflects the degree to which building settlement is dominant. The soil settlement rate T_S during the current monitoring period. v By combining the hyperbolic tangent function with a preset soil settlement rate control factor λ, a risk index evaluation model is constructed, and the risk index F is calculated.

[0095] The risk index evaluation model expression is as follows:

[0096] ;

[0097] In the formula, tanh represents the hyperbolic tangent function, and λ represents the soil settlement rate control factor set by professionals in the field according to the building completion stage.

[0098] In this embodiment, based on the elastic compression Sp of the pile load and the average total foundation settlement AVG_St in the standard feature dataset SFD, the crucial diagnostic indicator—soil settlement S—is obtained by calculating the difference between the two. o This calculation clarifies the absolute amount contributed by foundation soil deformation to the total settlement. Next, by calculating the ratio of this soil settlement So to the preset monitoring period number of days t, the soil settlement rate T_S within the current monitoring period timestamp T, characterizing the rate of settlement development, is obtained. v By measuring the soil settlement S o With average total foundation settlement AVG_S t By dividing, we obtain a dimensionless relative contribution rate factor that reflects the dominance of building settlement. At the same time, the soil settlement rate T_S during the current monitoring period will be... v By combining the hyperbolic tangent function with a soil settlement rate adjustment factor λ, set by professionals in the field according to the building completion stage, an evaluation term that nonlinearly reflects the dynamic trend of settlement is constructed. Finally, the relative contribution rate factor is multiplied by this dynamic trend evaluation term to construct a complete risk index evaluation model and calculate the final risk index F. The risk index F does not view any single indicator in isolation, but creatively outputs a new diagnostic quantity that simultaneously reflects where the problem is and how fast it is developing. Compared to any single parameter in traditional monitoring, this risk index F more comprehensively and three-dimensionally reveals the true risk state of the building foundation, providing unprecedented, high-quality judgment basis for subsequent scientific decision-making.

[0099] Example 4

[0100] This embodiment is an explanation based on Embodiment 3. Please refer to it. Figure 1 and Figure 3 Specifically: S4 includes S41;

[0101] S41. Based on the risk index F, compare it with the preset risk threshold interval θ, and generate a decision instruction Air according to the comparison result, wherein the risk threshold interval θ includes the first-level risk threshold θ1 and the second-level risk threshold θ2.

[0102] If the risk index F < the first-level risk threshold θ1, then the decision instruction Air is generated to determine that the building foundation settlement is a natural phenomenon within a safe range, and the zero-level decision action A0 is generated to indicate that no intervention is required.

[0103] If the first-level risk threshold θ1 ≤ risk index F < second-level risk threshold θ2, then the decision instruction Air is generated to determine that although the building foundation settlement is within the safe range, it is an abnormal phenomenon, and the first-level decision action A1 is generated to automatically generate a report and send it to the central control management platform, suggesting that the management personnel increase the detection frequency;

[0104] If the risk index F ≥ the secondary risk threshold θ2, then the decision instruction Air is generated to determine that the building foundation settlement is within the dangerous range, and the secondary decision action A2 is generated to automatically generate an alarm to notify relevant personnel via APP and SMS, and send a file containing the original feature dataset RAW, the standard feature dataset SFD, the dynamic elastic modulus of the pile Ed, and the elastic compression S caused by the pile load to the central control management platform. p Soil settlement S o and the soil settlement rate T_S during the current monitoring period v The data report recommends manual verification.

[0105] S5 includes S51;

[0106] S51. Execute decision-making actions based on decision command Air, conduct regular manual surveys and obtain evaluation results J, and determine whether to execute optimization actions based on the difference between evaluation results J and decision command Air.

[0107] If the decision action level generated in the decision instruction Air is a level 0 decision action A0 or a level 1 decision action A1, then according to the preset manual survey cycle, manual surveys will be carried out periodically and evaluation results J will be obtained.

[0108] If the decision action level generated in the decision instruction Air is a level 2 decision action A2, then immediately conduct a manual inspection according to the content of level 2 decision action A2 and obtain the evaluation result J;

[0109] If the assessment result J is consistent with the decision instruction Air, then relevant staff will be arranged to generate and implement a solution based on the building foundation settlement problem.

[0110] If the assessment result J is inconsistent with the decision instruction Air, check whether there are any faults in the displacement sensors and multi-channel trans-hole ultrasonic instruments deployed on the building foundation platform.

[0111] If the displacement sensor and multi-channel cross-hole ultrasonic instrument deployed on the building foundation platform are faulty, replace them with fault-free displacement sensor and multi-channel cross-hole ultrasonic instrument.

[0112] If the displacement sensor and multi-channel trans-hole ultrasonic instrument deployed on the building foundation are not faulty, the preset risk threshold range θ value will be adjusted based on the assessment result J and the historical database HIS.

[0113] The original feature dataset RAW, risk index F, evaluation result J, and current risk threshold interval θ are stored in the historical database HIS. When the sample data of the currently detected building in the historical database HIS does not reach the preset number of samples, professionals in the field select other building sample data that can be used for the current detected building from the historical database HIS based on the original feature dataset RAW and the current risk threshold interval θ of the current detected building, and combine it with the sample data of the current detected building to perform logistic regression training. Using the risk index F as input and the evaluation result j as label, the risk level classification model is retrained. Based on the new risk level classification model, the values ​​of the first-level risk threshold θ1 and the second-level risk threshold θ2 are automatically calculated and updated, and professionals in the field manually verify the newly generated first-level risk threshold θ1 and second-level risk threshold θ2. When the sample data of the currently detected building in the historical database HIS reaches the preset number of samples, the logistic regression training is performed entirely using the sample data of the current detected building, and professionals in the field manually verify the newly generated first-level risk threshold θ1 and second-level risk threshold θ2.

[0114] The following is a specific example of how the decision instruction Air is generated:

[0115] The example of generating the decision instruction Air uses a composite foundation consisting of natural soil layers and artificial piles as an example:

[0116] The background of the building foundation is in the early stage of completion. The composite ground load has not yet stabilized and a rapid settlement stage has appeared. According to the building safety code, the allowable settlement rate in this stage is 5-10 mm / year. The preset observation time window Q is 12 hours for data collection, and the preset detection cycle t is 30 days. The building foundation settlement is detected once based on the collected data.

[0117] Standard Feature Dataset SFD=[AVG_S t 2.75mm, AVG_V p :4100m / s, W: 800000N, H: 30m,

[0118] A: 0.283m 2 ρ: 2500 kg / m 3 ,ν:0.2];

[0119] Based on the background of the building under inspection, the median of the allowable settlement rate of 5-10 mm / year, i.e. 7 to 8 mm / year, is taken as the reference settlement rate. Based on the number of days required for each mm of settlement, the soil settlement rate control factor λ is set to 50 days / mm.

[0120] Level 1 risk threshold θ1: 0.5; Level 2 risk threshold θ2: 1.0;

[0121] The dynamic elastic modulus Ed of the pile is calculated as follows:

[0122] ;

[0123] Elastic compression S caused by pile load p The specific calculations are as follows:

[0124] ;

[0125] Soil settlement S o The specific calculations are as follows:

[0126] ;

[0127] Soil settlement rate T_S during the current monitoring period v The specific calculations are as follows:

[0128] ;

[0129] The risk index F is calculated as follows:

[0130]

[0131] Since the risk index F < the first-level risk threshold θ1, the generated decision instruction Air is a natural phenomenon that determines that the building foundation settlement is within a safe range, and the generated zero-level decision action A0 is that no intervention is required.

[0132] An analysis of a specific calculation example reveals the average total foundation settlement AVG_S obtained over a 30-day monitoring period. t Numerical value for the average total foundation settlement AVG_S t The main reason is that the elastic compression S of the pile body is caused by the load on the pile foundation not yet being stable in the early stage after completion. p As the main contributor, it is not the soil settlement S o As the main contributor, combined with the soil settlement rate T_S during the current monitoring cycle v The risk index F obtained after co-calculation with the set soil settlement rate control factor λ is within the safe range, which demonstrates the superiority of this invention in avoiding misjudgment caused solely by changes in the average total foundation settlement AVG_St.

[0133] In this embodiment, by comparing the risk index F with a scientifically divided multi-level risk threshold interval θ, this method directly maps the risk assessment result into a specific decision instruction Air. This unique advantage lies in its transformation of the traditional black-and-white alarm mode of monitoring, replacing it with a graded response that provides clear guidance, such as no intervention required, increased frequency, and suggested reconnaissance. This makes automated decision-making not only accurate but also highly operable, effectively avoiding the decision-making confusion faced by managers when dealing with complex data. More importantly, by establishing an optimized closed loop based on the consistency comparison between the manual reconnaissance assessment result J and the decision instruction Air, a fundamental guarantee is provided for the long-term reliability of the detection method. The unique advantage of this innovative logic is that it introduces an authoritative arbitration mechanism of human experts into the automation mechanism. When the judgment is inconsistent with the facts, this mechanism can trigger a feedback adjustment of the core decision-making basis—sensor fault detection and the risk threshold interval θ. This ensures that this method is not a fixed, unchanging tool, but a reliable intelligent detection technology that can continuously "learn" from engineering practice and constantly calibrate itself, maintaining a high level of judgment accuracy throughout the entire life cycle of the building.

[0134] Example 5

[0135] A smart detection system for building foundation settlement, please refer to... Figure 2 Specifically, it includes a data acquisition and preprocessing module, a pile body analysis module, a risk index calculation module, an evaluation module, and an optimization module;

[0136] The data acquisition and preprocessing module reads construction data files and collects monitoring data using different sensor devices deployed on the building foundation to construct the raw feature dataset RAW. The dataset is then preprocessed to obtain the standard feature dataset SFD.

[0137] The pile analysis module calculates the dynamic elastic modulus Ed of the pile body based on the standard feature dataset SFD and the theory of elasticity. Then, using the calculated dynamic elastic modulus Ed, it calculates the load-induced elastic compression S of the pile body according to Hooke's law in mechanics of materials. p ;

[0138] The risk index calculation module calculates the risk index based on the elastic compression S induced by pile load. p Using the standard feature dataset SFD, calculate the soil settlement S. o and the soil settlement rate T_S during the current monitoring period v And based on the soil settlement S o Elastic compression S caused by pile load p and the soil settlement rate T_S during the current monitoring period v Calculate the risk index F;

[0139] The assessment module compares the risk index F with a preset risk threshold range θ and generates a decision instruction Air based on the comparison results.

[0140] The optimization module executes decision-making actions based on decision instructions Air, and periodically conducts manual surveys to obtain evaluation results J. Based on the difference between the evaluation results J and the decision instructions Air, it executes optimization actions, stores the original feature dataset RAW, risk index F, evaluation results J and the current risk threshold interval θ in the historical database HIS, and performs iterative optimization of the risk threshold interval θ based on the historical database HIS.

[0141] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent detection method based on building foundation settlement, characterized in that: The method comprises the following steps: S1, constructing a raw feature data set RAW by reading a construction data file and collecting monitoring data by using different sensor devices deployed on the building foundation, and pre-processing the data set to obtain a standard feature data set SFD; S2, based on the standard feature data set SFD, calculating the pile body dynamic elastic modulus Ed according to the theory of elasticity, and combining the calculated pile body dynamic elastic modulus Ed, calculating the elastic compression amount S caused by the load of the pile body according to Hooke's law in material mechanics p ; S3, the elastic compression amount S of the pile body based on the load p and the standard feature data set SFD, the soil settlement amount S o and the soil settlement rate T_S in the current detection period v , and according to the soil settlement amount S o , the elastic compression amount S of the pile body based on the load p and the soil settlement rate T_S in the current detection period v calculate the risk index F; S4, comparing the risk index F with a preset risk threshold interval θ, and generating a decision instruction Air according to the comparison result; S5, executing a decision action based on the decision instruction Air, periodically performing manual surveying and obtaining an evaluation result J, executing an optimization action according to the difference between the evaluation result J and the decision instruction Air, storing the raw feature data set RAW, the risk index F, the evaluation result J and the current risk threshold interval θ in a historical database HIS, and realizing iterative optimization of the risk threshold interval θ according to the historical database HIS. 2.The building foundation settlement intelligent detection method according to claim 1, characterized in that: S1 comprises S11 and S12; S11, obtaining the pile body design length H, the pile body cross-sectional area A, the pile body concrete density ρ, the Poisson's ratio v and the equivalent load W of the building detection pile by reading the construction data file; The total foundation settlement S is obtained according to a preset observation time window Q through a displacement sensor arranged on a building foundation slab t The total foundation settlement data set SS in a preset detection period of days t is collected t ; The total settlement S of the foundation is collected by the displacement sensor through the probe of the multi-channel cross-hole ultrasonic instrument embedded in the building foundation t The longitudinal wave velocity V is synchronously collected p The longitudinal wave velocity data set SV in the preset detection period of days t is collected p ; a set of data of total settlement of foundation SS t a set of data of longitudinal wave velocity SV p a pile body design length H, a pile body cross-sectional area A, a pile body concrete density p, a Poisson's ratio v, and a building test pile equivalent load W, to construct a raw feature data set RAW.

3. The method according to claim 2, wherein: S12, pre-processing the total settlement data set SS in the raw feature data set RAW t and the longitudinal wave velocity data set SV p S12, pre-processing the total settlement data set SS in the raw feature data set RAW Calculate the total foundation settlement data set SS separately t and P-wave velocity data set SV p The standard deviation of settlement σS t and the standard deviation of P-wave velocity σV p For the total foundation settlement data set SS t The moving standard deviation σS of the settlement exceeds 5 times. t The values ​​were discarded based on the total foundation settlement data set SS. t The remaining data is used to calculate the average total foundation settlement AVG_S t For the P-wave velocity data set SV p The P-wave velocity exceeds 5 times the standard deviation σV p The values ​​were discarded based on the P-wave velocity data set SV. p The average P-wave velocity AVG_V is calculated from the remaining data. p ; AVG_S t , AVG_V p , H, A, ρ, ν, and W, to construct a standard feature dataset SFD.

4. The method according to claim 3, characterized in that: S2 comprises S21 and S22; S21, based on the average longitudinal wave velocity AVG V in the standard feature dataset SFD p , the pile body concrete density p and the Poisson's ratio v, according to the calculation formula of the average longitudinal wave velocity AVG V in the internal homogeneous, isotropic, infinite elastic body in the theory of elasticity, the pile body dynamic elastic modulus Ed is obtained by reverse calculation; p , the pile body concrete density p and the Poisson's ratio v, according to the calculation formula of the average longitudinal wave velocity AVG V in the internal homogeneous, isotropic, infinite elastic body in the theory of elasticity, the pile body dynamic elastic modulus Ed is obtained by reverse calculation; The calculation formula of the pile body dynamic elastic modulus Ed is as follows: ； In the formula, ε represents a minimum value for preventing the denominator from being zero, and specifically takes a value of 1*10 -8 .

5. The intelligent detection method for building foundation settlement according to claim 4, characterized in that: S22, based on the dynamic elastic modulus Ed of the pile body, combined with the pile body design length H, the pile body cross-sectional area A and the building detection pile equivalent load W in the standard feature data set SFD, the elastic compression amount S caused by the load of the pile body is calculated according to Hooke's law in material mechanics p ; Wherein, the pile body load causes elastic compression amount S p The calculation formula is as follows: 。 6. The method for intelligent detection of building foundation settlement according to claim 5, characterized in that: S3 comprises S31 and S32; S31, the elastic compression amount S caused by the pile load p and the average total settlement amount of the foundation AVG_S in the standard feature data set SFD t , the soil settlement amount S is obtained by calculating the difference between the average total settlement amount of the foundation AVG_S t and the elastic compression amount S caused by the pile load p o , the soil settlement rate T_S in the current detection period is obtained by calculating the ratio of the soil settlement amount S o to the preset detection period t v , where T represents the current detection period timestamp.​ 7. The method according to claim 6, wherein: S32, based on the soil settlement S o Elastic compression S caused by pile load p and the soil settlement rate T_S during the current monitoring period v By measuring the soil settlement S o With average total foundation settlement AVG_S t By dividing the two, we obtain the relative contribution rate factor, which reflects the degree to which building settlement is dominant. The soil settlement rate T_S during the current monitoring period. v By combining the hyperbolic tangent function with a preset soil settlement rate control factor λ, a risk index evaluation model is constructed, and the risk index F is calculated. The expression of the risk index evaluation model is as follows: ； In the formula, tanh represents the hyperbolic tangent function, and λ represents the soil settlement rate control factor set by a person skilled in the art according to the building completion stage. 8.The method for intelligent detection of building foundation settlement according to claim 7, characterized in that: S4 comprises S41; S41, comparing the risk index F with a preset risk threshold interval θ, and generating a decision instruction Air according to the comparison result, wherein the risk threshold interval θ comprises a first risk threshold θ1 and a second risk threshold θ2; If the risk index F is less than the first risk threshold θ1, the decision instruction Air generated is a natural phenomenon that the building foundation settlement is within a safe range, and a zero-level decision action A0 is generated, which is that no intervention is needed; If the first risk threshold θ1 is less than or equal to the risk index F and less than the second risk threshold θ2, the decision instruction Air generated is that the building foundation settlement is within a safe range but is an abnormal phenomenon, and a first-level decision action A1 is generated to automatically generate a report and send it to a central control management platform, and it is suggested that the management personnel increase the detection frequency; If the risk index F is greater than or equal to the secondary risk threshold θ2, a decision instruction Air is generated to determine that the building foundation settlement is in a dangerous range, and a secondary decision action A2 is generated to automatically generate an alarm to notify the relevant responsible personnel through an APP and a short message, and send a data report including the original feature data set RAW, the standard feature data set SFD, the dynamic elastic modulus Ed of the pile body, the elastic compression amount S p of the pile body caused by the load o , and the soil settlement rate T_S in the current detection period v to the central control management platform, and manual survey verification is suggested. 9.The method for intelligent detection of building foundation settlement according to claim 8, characterized in that: S5 comprises S51; S51, executing a decision action based on the decision instruction Air, periodically performing manual surveying and obtaining an evaluation result J, and judging whether to execute an optimization according to the difference between the evaluation result J and the decision instruction Air; If the decision action level generated in the decision instruction Air is a zero-level decision action A0 or a first-level decision action A1, manual surveying is periodically performed according to a preset manual surveying period, and an evaluation result J is obtained; If the decision action level generated in the decision instruction Air is a second-level decision action A2, manual surveying is immediately performed according to the content of the second-level decision action A2, and an evaluation result J is obtained; If the evaluation result J is consistent with the decision instruction Air, relevant personnel are arranged to generate a solution based on the building foundation settlement problem and execute it; If the evaluation result J is not consistent with the decision instruction Air, it is checked whether there is a fault problem in the displacement sensor and the multi-channel cross-hole ultrasonic instrument deployed on the building foundation pile cap. If the displacement sensor and the multi-channel cross-hole ultrasonic instrument deployed on the building foundation platform have a failure problem, the displacement sensor and the multi-channel cross-hole ultrasonic instrument without failure are replaced; If the displacement sensor and the multi-channel cross-hole ultrasonic instrument deployed on the building foundation platform have no failure problem, the preset risk threshold interval θ value is adjusted according to the evaluation result J in combination with the historical database HIS; The original feature data set RAW, the risk index F, the evaluation result J and the current risk threshold interval θ are stored in the historical database HIS, when the sample data of the current detection building in the historical database HIS does not reach the preset sample quantity, other building sample data that can be used for the current detection building are selected from the historical database HIS by the person skilled in the art according to the original feature data set RAW of the current detection building and the current risk threshold interval θ, and the logical regression training is performed in combination with the sample data of the current detection building, the risk index F is taken as the input, the evaluation result J is taken as the label, the classification model of the risk level is retrained, the values of the first risk threshold θ1 and the second risk threshold θ2 are automatically calculated and updated based on the new classification model of the risk level, and the first risk threshold θ1 and the second risk threshold θ2 generated newly are manually checked by the person skilled in the art, when the sample data of the current detection building in the historical database HIS reaches the preset sample quantity, the logical regression training is completely performed using the sample data of the current detection building, and the first risk threshold θ1 and the second risk threshold θ2 generated newly are manually checked by the person skilled in the art. 10.A building foundation settlement intelligent detection system based on any one of claims 1-9, applied to a building foundation settlement intelligent detection method. The system comprises a data acquisition and preprocessing module, a pile body analysis module, a risk index calculation module, an evaluation module and an optimization module; The data acquisition and preprocessing module reads the construction data file, acquires monitoring data by using different sensor devices deployed on the building foundation, constructs an original feature data set RAW, and pre-processes the data set to obtain a standard feature data set SFD; The pile body analysis module calculates the dynamic elastic modulus Ed of the pile body based on the standard feature data set SFD according to the theory of elasticity, and calculates the elastic compression amount S caused by the load of the pile body according to Hooke's law in material mechanics in combination with the calculated dynamic elastic modulus Ed of the pile body p ; The risk index calculation module calculates the risk index based on the elastic compression S induced by pile load. p Using the standard feature dataset SFD, calculate the soil settlement S. o and the soil settlement rate T_S during the current monitoring period v And based on the soil settlement S o Elastic compression S caused by pile load p and the soil settlement rate T_S during the current monitoring period v Calculate the risk index F; The evaluation module compares the risk index F with the preset risk threshold interval θ, and generates a decision instruction Air according to the comparison result; The optimization module executes a decision action based on the decision instruction Air, regularly performs manual surveying and obtains an evaluation result J, executes an optimization action according to the difference between the evaluation result J and the decision instruction Air, stores the original feature data set RAW, the risk index F, the evaluation result J and the current risk threshold interval θ in the historical database HIS, and realizes iterative optimization of the risk threshold interval θ according to the historical database HIS.

Images