An AI-based data center seismic isolation performance testing method

Abstract

The present invention discloses a method for detecting the seismic isolation performance of a data center based on AI intelligence, which specifically relates to the technical field of seismic isolation performance detection. By collecting seismic vibration characteristic data in real time on data center buildings and equipment, and extracting seismic input characteristics in combination with a historical seismic wave database, a seismic isolation performance prediction model is constructed and trained, and the foundation shear wave velocity gradient anomalies and the dynamic displacement response deviation rate of the seismic isolation layer are further extracted. The deviation index is comprehensively calculated, and classification processing is performed according to the deviation level. In particular, in the case of medium deviation levels, the dynamic updating of foundation physical parameters and the optimization of the model are adopted, which significantly improves the accuracy of the prediction model and the system adaptability, and effectively guarantees the seismic isolation safety and continuous operation capability of the data center in a strong earthquake environment.

Description

Technical Field

[0001] The present invention relates to the technical field of seismic isolation performance detection technology, and in particular to a data center seismic isolation performance detection method based on AI intelligence. Background Art

[0002] Data center seismic isolation performance testing refers to the testing and evaluation of the seismic isolation system (such as seismic isolation devices, seismic isolation layers, etc.) of the data center building and its key equipment through professional technical means, to verify whether its shock absorption and seismic isolation effects in vibration environments such as earthquakes meet the design requirements and safety standards, to ensure that the data center can maintain structural safety and stable equipment operation when an earthquake occurs, thereby ensuring the continuity of data storage and services.

[0003] The existing technology has the following shortcomings:

[0004] Existing techniques for training DNN models for seismic isolation performance prediction based on simulation modeling often lead to systematic deviations due to inconsistencies between actual engineering boundary conditions and simulation assumptions. For example, a coastal data center, due to uneven settlement of its foundation caused by land reclamation, was mistakenly set as a homogeneous elastic body during simulation. This caused the model to severely underestimate the amplification effect of bottom displacement in a major earthquake scenario. Ultimately, the isolation device failed prematurely during a strong earthquake, causing equipment tilt and system failure, exposing the significant impact of deviations in simulation boundary conditions on the accuracy and safety of seismic isolation performance predictions. Summary of the Invention

[0005] The purpose of the present invention is to provide a data center seismic isolation performance detection method based on AI intelligence to address the shortcomings of the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a data center seismic isolation performance detection method based on AI intelligence, comprising:

[0007] Deploy multi-dimensional IoT sensors on data center building structures and equipment to collect real-time equipment vibration characteristic data during earthquake events and extract seismic wave characteristic parameters from a seismic database containing historical seismic wave records;

[0008] Based on the collected equipment vibration characteristic data and seismic wave characteristic data, a deep neural network model is constructed and trained to learn the nonlinear mapping relationship between seismic input and structural response, forming a seismic isolation performance prediction model;

[0009] For the actual monitored equipment vibration characteristic data and seismic wave characteristic data, the foundation shear wave velocity gradient anomaly and the isolation layer dynamic displacement response deviation rate are extracted respectively, and the comprehensive deviation index is calculated based on the preset weights;

[0010] According to the comprehensive deviation index, the systematic deviation degree of the isolation performance prediction model is divided into high deviation, medium deviation and low deviation levels, and corresponding treatment is performed respectively;

[0011] For situations with medium deviation levels, the foundation physical model parameters are dynamically updated, and the seismic isolation performance prediction model is retrained or optimized.

[0012] Preferably, the multi-dimensional Internet of Things sensor includes an acceleration sensor, a displacement sensor, a tilt sensor, a multi-axis gyroscope and an environmental sensor.

[0013] Preferably, the method for extracting the outlier value of the foundation shear wave velocity gradient is as follows: inputting a discrete data point set of the measured foundation shear wave velocity with depth: ;in, It's depth, is the corresponding shear wave velocity; the shear wave velocity versus depth data set obtained by simulation modeling is: ; N is the total number of discrete data points of the measured foundation shear wave velocity with depth, The total number of shear wave velocity versus depth data obtained for simulation modeling; The jth depth point in the simulation model refers to the depth position of a layer in the foundation profile set during the simulation modeling process. For simulation models in depth The shear wave velocity at represents the foundation shear wave velocity value set at this depth in the simulation modeling;

[0014] The shear wave velocity data is transformed into a one-dimensional discrete wavelet transform and decomposed into multiple scales, retaining only the low-frequency approximate components. The retained low-frequency wavelet is reconstructed into a smooth shear wave velocity trend curve, which is recorded as , differentiate the trend curve by depth and calculate the gradient , the expression is: ; is the i-th measured depth point, The measured shear wave velocity trend curve at depth The value at The measured shear wave velocity trend curve at depth The simulated shear wave velocity is simply linearly interpolated and smoothed to calculate the simulated gradient , the expression is: ;

[0015] Indicates the simulated shear wave velocity at depth The value at To simulate the shear wave velocity at depth The value at ; perform depth-weighted averaging on GGI(z) to obtain the overall foundation shear wave velocity gradient anomaly value , the expression is: Where, is the maximum depth.

[0016] Preferably, the method for extracting the deviation rate of the dynamic displacement response of the seismic isolation layer is as follows: obtaining the displacement time history of the seismic isolation layer actually monitored, marked as ; The displacement history of the isolation layer predicted by simulation is marked as ; The actual displacement time Simulated displacement time history Perform optimal alignment on the time axis and calculate the Euclidean distance matrix for each point between the two time course curves , the expression is: ; Search for a minimum cumulative cost path, connecting the start and end, representing the optimal alignment path, and output a set of aligned point pair sequences ; Indicates the actual displacement Corresponding simulation displacement ; Indicates the displacement time history of the simulation Aligned time points, calculate the instantaneous error for each pair of aligned points , the expression is: ; For simulations predicting displacement in time Calculate the average error of all alignment points : ;W is the number of point pairs on the alignment path, and the maximum predicted displacement peak value of the simulated displacement history is taken : The final isolation layer dynamic displacement response deviation rate IDRΔ is defined as: .

[0017] Preferably, the weight of the foundation shear wave velocity anomaly is set to 50%; the weight of the dynamic displacement deviation rate of the seismic isolation layer is set to 50%; the GGI is normalized and the absolute value of the deviation from 1 is used as the metric: |GGI−1|; the comprehensive deviation index CPI is calculated using the following formula: .

[0018] Preferably, the systematic deviation degree of the seismic isolation performance prediction model is divided into high deviation, medium deviation and low deviation levels according to the comprehensive deviation index, and corresponding processing is performed respectively, specifically including:

[0019] If CPI ≥ 0.40, the systematic deviation degree of the isolation performance prediction model is classified as high deviation, indicating that the model is seriously mismatched and needs to be fully revised;

[0020] If 0.20≤CPI<0.40, the systematic deviation of the isolation performance prediction model is classified as medium deviation, indicating that the model has mismatch and needs local optimization.

[0021] If CPI < 0.20, the systematic deviation degree of the isolation performance prediction model is classified as low deviation, indicating that the model is consistent with the actual situation and no adjustment is required.

[0022] Preferably, for the case of medium deviation level, Gaussian process regression is used to establish a proxy relationship between foundation parameters and CPI values. The proxy model can predict the corresponding CPI after each foundation parameter update. The proxy model expression is: ; is the total square error between the CPI predicted by the proxy model and the actual CPI; r is the number of samples; is the predicted CPI of the i-th sample, is the actual CPI of the i-th sample;

[0023] Randomly sample several combinations of foundation physical parameters to actually evaluate CPI;

[0024] Predict the parameter group that will cause CPI to fall in the next step based on the proxy model;

[0025] Update foundation physical parameters and adjust simulation output;

[0026] Recalculate CPI and feed it back into the proxy model;

[0027] The termination condition is to find a parameter combination that reduces the CPI to a low deviation level or to reach the maximum number of iterations.

[0028] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0029] 1. This invention deploys multi-dimensional IoT sensors on data center structures and equipment to collect real-time vibration signature data from earthquake events. This data is then combined with a historical seismic wave database to extract seismic characteristics. A deep neural network is then used to establish a nonlinear mapping relationship between seismic input and structural response, enabling intelligent prediction of seismic isolation performance. This method innovatively extracts outliers in the foundation shear wave velocity gradient and the deviation rate of the dynamic displacement response of the isolation layer, calculates a comprehensive deviation index, and classifies systematic deviations according to the CPI. This method implements a graded approach based on different degrees of deviation, effectively addressing the inaccurate seismic isolation performance predictions in existing technologies caused by mismatches between simulation assumptions and actual boundary conditions.

[0030] 2. This invention can use Gaussian process regression to establish a proxy relationship between foundation parameters and system deviations at medium deviation levels. It dynamically updates foundation physical parameters through a Bayesian optimization strategy, and combines incremental learning to optimize the isolation performance prediction model, quickly correcting the model to a low deviation level. This avoids the high cost and low efficiency of full reconstruction in traditional methods. Overall, this invention significantly improves the prediction accuracy and adaptability of the isolation system of data centers in complex seismic environments, enhancing the resilience and security of critical infrastructure in earthquake disasters. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0032] Figure 1 This is a mind map of the method of the present invention. DETAILED DESCRIPTION

[0033] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0034] For examples, see Figure 1 As shown, the AI-based data center seismic isolation performance detection method described in this embodiment includes:

[0035] Deploy multi-dimensional IoT sensors on data center building structures and equipment to collect real-time equipment vibration characteristic data during earthquake events and extract seismic wave characteristic parameters from a seismic database containing historical seismic wave records;

[0036] Based on the collected equipment vibration characteristic data and seismic wave characteristic data, a deep neural network model is constructed and trained to learn the nonlinear mapping relationship between seismic input and structural response, forming a seismic isolation performance prediction model;

[0037] For the actual monitored equipment vibration characteristic data and seismic wave characteristic data, the foundation shear wave velocity gradient anomaly and the isolation layer dynamic displacement response deviation rate are extracted respectively, and the comprehensive deviation index is calculated based on the preset weights;

[0038] According to the comprehensive deviation index, the systematic deviation degree of the isolation performance prediction model is divided into high deviation, medium deviation and low deviation levels, and corresponding treatment is performed respectively;

[0039] For situations with medium deviation levels, the foundation physical model parameters are dynamically updated, and the seismic isolation performance prediction model is retrained or optimized.

[0040] A sensor network is deployed on the building structure of the data center and key equipment inside the computer room (such as cabinets, racks, and server arrays) to monitor in real time the dynamic response characteristics of the overall structure and internal equipment when an earthquake occurs, including but not limited to key indicators such as acceleration, displacement, angular velocity, and inclination.

[0041] Accelerometer (MEMS type, high sensitivity): used to capture absolute acceleration changes under seismic excitation, with a frequency range of 0.1–200 Hz and a measurement range of ±2g to ±5g.

[0042] Displacement sensor (LVDT or laser displacement meter): monitors the horizontal and vertical displacement changes of the isolation device layer or cabinet with an accuracy of up to 0.1mm.

[0043] Tilt sensor: Captures changes in the tilt angle of equipment or local structures, with a range of ±30° and a resolution of 0.001°.

[0044] Multi-axis gyroscope: Captures three-dimensional angular velocity changes and is used to identify changes in device torsion and vibration modes.

[0045] Environmental sensors: temperature, humidity, etc., used to calibrate sensor drift caused by environmental changes.

[0046] Typical deployment locations include: Foundations and isolation layers: Focus on monitoring support displacement and energy dissipation. Frame beam-column joints: Monitor overall frame vibration characteristics. Floor slab centers and edges: Assess in-plane deformation and floor slab torsional effects. Server cabinet bottoms, middles, and tops: Capture cabinet vibration amplification effects. Rack connection points: Monitor changes in equipment connection stiffness. UPS power supply equipment and large cooling equipment foundations: Focus on the inertial response of large-mass equipment.

[0047] Wired (PoE Ethernet) or wireless (LoRa, Wi-Fi 6) communication is preferred to ensure stable communication during earthquakes. All sensors are synchronized to the nanosecond level using GPS time synchronization or IEEE 1588 PTP (Precision Time Protocol), ensuring precise and consistent data timelines for subsequent vibration pattern analysis.

[0048] The data collection content of equipment vibration characteristic data includes:

[0049] Acquisition time window: continuous recording from 5 seconds before earthquake warning to 300 seconds after the earthquake.

[0050] Acquisition data types: three-axis acceleration, three-axis displacement, tilt angle change, and spectrum characteristics.

[0051] Data sampling rate: acceleration data ≥ 200Hz (meeting high-frequency vibration sampling requirements); displacement data ≥ 50Hz; tilt angle data ≥ 10Hz.

[0052] Real-time data preprocessing includes: De-noising: using wavelet denoising or bandpass filter to filter out high-frequency noise from the sensor.

[0053] Baseline correction: Eliminate low-frequency drift to ensure the physical accuracy of displacement data.

[0054] Anomaly detection: Eliminate error packets by setting threshold detection (such as abnormal acceleration jumps) or AI-assisted judgment.

[0055] Compression coding: Use lossless compression (such as FLAC coding) to compress data in real time to reduce communication pressure.

[0056] Obtain earthquake data through authoritative earthquake database: seismic wave types include: surface records; near-fault records; bedrock records.

[0057] All seismic waves are normalized using displacement, velocity, and acceleration amplitude scales. Feature extraction includes: peak acceleration (PGA); peak velocity (PGV); peak displacement (PGD); effective duration (e.g., the time period between 5% and 95% of accumulated energy); Fourier spectrum characteristics (dominant frequency, characteristic period); and Arias intensity (an indicator of seismic input energy). Each seismic record is converted into a set of standardized feature vectors, which serve as the input feature set for the DNN model.

[0058] Based on the collected equipment vibration characteristic data and seismic wave characteristic data, a deep neural network model is constructed and trained to learn the nonlinear mapping relationship between seismic input and structural response, forming a seismic isolation performance prediction model. Specifically:

[0059] Eliminate incomplete data samples, data samples with obvious outliers, or data samples with drift failures. Unify the unit system (e.g., acceleration m / s², displacement m). Extract equipment vibration characteristic data (e.g., maximum acceleration, maximum displacement, main vibration frequency, vibration duration) as a feature set. Normalize seismic wave characteristic parameters (e.g., PGA, PGV, PGD, effective duration, characteristic period, etc.) to form a standard input vector. Optional feature enhancement: Add frequency domain features (e.g., the main peak frequency of the Fourier spectrum); add time domain statistics (e.g., root mean square acceleration RMS). Normalize all input and output data (e.g., to a mean of 0 and a standard deviation of 1) to improve network convergence speed and stability. Divide the data into training, validation, and test sets in an 8:1:1 ratio, ensuring that the seismic waves in the test set do not overlap with those in the training set to assess the model's generalization capabilities.

[0060] The input vector includes: seismic wave characteristic parameter vector (such as 10-20 dimensions); equipment vibration characteristic vector (such as 10 dimensions); the total input feature dimension is usually controlled within 20-50 dimensions.

[0061] Build a multi-layer perceptron (MLP) architecture with 3-5 hidden layers, 64-256 neurons per layer, ReLU activation function (to avoid vanishing gradients), and a batch normalization layer to stabilize the training process.

[0062] The output node corresponds to the isolation performance prediction target, such as: the maximum horizontal displacement of the isolation device; the maximum acceleration response of the equipment; the energy dissipation rate of the isolation system; the activation function is selected according to the output target, such as linear activation for regression problems.

[0063] The regression task uses the mean square error loss function;

[0064] Optimizer selection: Adam Optimizer (the learning rate is initially set to 0.001);

[0065] Batch size: between 32 and 128, adjusted according to the amount of data and GPU memory.

[0066] Training rounds: Dynamically adjusted according to the error convergence of the validation set, usually 100-500 rounds.

[0067] Introducing the Dropout mechanism (Dropout rate between 0.2-0.5)

[0068] Use the early stopping strategy to terminate training when the validation set error does not decrease for several consecutive rounds.

[0069] R² (coefficient of determination), MAE (mean absolute error), and RMSE (root mean square error) are used as the main evaluation indicators. A scatter plot of the predicted results and the actual observed values ​​is also drawn to verify the fitting effect.

[0070] The model's stability is tested using extreme seismic wave data not seen in the test set (e.g., near-fault pulse earthquakes). The model's performance is evaluated under earthquakes of varying intensities and types.

[0071] Solidify and package the trained and verified DNN model (for example, save it in ONNX or TensorFlow SavedModel format).

[0072] The model is deployed in the data center's intelligent monitoring system to enable real-time input of collected seismic wave characteristics and preliminary equipment response data when an earthquake occurs; and to quickly predict the response of the isolation system and potential performance degradation risks.

[0073] Regularly collect new actual earthquake response data, expand the data set, conduct incremental learning, and continuously improve the accuracy of model predictions.

[0074] For the actual monitored equipment vibration characteristic data and seismic wave characteristic data, the foundation shear wave velocity gradient anomaly and the isolation layer dynamic displacement response deviation rate are extracted respectively, and the comprehensive deviation index is calculated based on the preset weights, specifically including:

[0075] The method for extracting the outlier value of foundation shear wave velocity gradient is as follows: input the discrete data point set of measured foundation shear wave velocity with depth: ;in, It's depth, is the corresponding shear wave velocity. The shear wave velocity versus depth data set obtained by simulation modeling is: ; N is the total number of discrete data points of the measured foundation shear wave velocity with depth, The total number of shear wave velocity versus depth data obtained for simulation modeling; The jth depth point in the simulation model refers to the depth position of a layer in the foundation profile set during the simulation modeling process. For simulation models in depth The shear wave velocity at represents the foundation shear wave velocity value set at this depth in the simulation modeling.

[0076] Select a wavelet basis suitable for smooth trend extraction, such as Daubechies wavelet (db4) or Symlets wavelet (sym5). Perform a one-dimensional discrete wavelet transform on the shear wave velocity data and decompose it into multiple scales (e.g., layer 3 or layer 4). Only the low-frequency approximate components are retained, and the high-frequency detail components are discarded. The retained low-frequency wavelet is used to reconstruct a smooth shear wave velocity trend curve, which is recorded as . Differentiate the trend curve by depth and calculate the gradient (shear wave velocity change rate) , the expression is: ; is the i-th measured depth point, The measured shear wave velocity trend curve at depth The value at The measured shear wave velocity trend curve at depth The simulated shear wave velocity is simply linearly interpolated and smoothed (wavelet processing is not required if the simulated data is ideally layered) and the gradient of the simulation is calculated. , the expression is: ; Indicates the simulated shear wave velocity at depth The value at To simulate the shear wave velocity at depth The value at

[0077] In each depth segment, the ratio of the actual gradient to the simulated gradient is calculated : If GGI(z)≈1, it indicates that the actual situation is consistent with the simulation. If GGI(z)>1, it indicates that the actual foundation gradient is steeper (hardening trend). If GGI(z)<1, it indicates that the actual foundation is softened or layered.

[0078] Perform depth-weighted averaging on GGI(z) to obtain the overall foundation shear wave velocity gradient anomaly , the expression is: Where, is the maximum depth.

[0079] The extraction method of the dynamic displacement response deviation rate of the isolation layer is as follows: obtain the actual monitored isolation layer displacement time history, marked as ; The displacement history of the isolation layer predicted by simulation is marked as ; The two time histories can be of different lengths (different earthquake durations or slightly different sampling frequencies). The sampling frequencies should be unified, or a simple linear interpolation process should be performed.

[0080] The actual displacement time Simulated displacement time history Perform optimal alignment on the time axis. Allow local time stretching / compression to make the overall response trend as close as possible. Calculate the Euclidean distance matrix for each point between the two time history curves. , the expression is: ; Search for a minimum cumulative cost path, connecting the start and end, representing the optimal alignment path, and output a set of aligned point pair sequences ; Indicates the actual displacement Corresponding simulation displacement ; Indicates the displacement time history of the simulation Aligned time points, calculate the instantaneous error for each pair of aligned points , the expression is: ; For simulations predicting displacement in time Calculate the average error of all alignment points : ;W is the number of point pairs on the alignment path, and the maximum predicted displacement peak value of the simulated displacement history is taken : The final isolation layer dynamic displacement response deviation rate IDRΔ is defined as: .

[0081] The weight of the ground shear wave velocity anomaly (GGI) is set to 50%; the weight of the isolation layer dynamic displacement deviation rate (IDRΔ) is set to 50%; the weights can be flexibly adjusted according to project requirements (for example, more emphasis on displacement deviation for high-rise buildings and more emphasis on soft soil foundations). gradient anomaly).

[0082] GGI is normalized and the absolute value of deviation from 1 is used as the metric: |GGI−1|; the comprehensive deviation index CPI is calculated using the following formula: .

[0083] According to the comprehensive deviation index, the systematic deviation degree of the isolation performance prediction model is divided into high deviation, medium deviation and low deviation levels, and corresponding treatment is performed respectively, including:

[0084] If CPI ≥ 0.40, the systematic deviation degree of the isolation performance prediction model is classified as high deviation, indicating that the model is seriously mismatched and needs to be fully revised;

[0085] If 0.20≤CPI<0.40, the systematic deviation of the isolation performance prediction model is classified as medium deviation, indicating that there is a certain mismatch and local optimization is required;

[0086] If CPI < 0.20, the systematic deviation degree of the isolation performance prediction model is classified as low deviation, indicating that the model is consistent with the actual situation and no adjustment is required.

[0087] When the Comprehensive Deviation Index (CPI) is at a medium deviation level (0.20 ≤ CPI < 0.40), the isolation performance prediction model, while generally performing well, deviates from actual foundation physical parameters (such as shear wave velocity distribution, damping ratio, and isolation device stiffness). Targeted dynamic updates are necessary to avoid systematic error accumulation. The goal is to dynamically correct foundation model parameters and locally optimize the DNN prediction model performance at the lowest possible cost, rather than completely rebuilding the model.

[0088] Set the range of foundation parameters to be adjusted, for example:

[0089] Foundation shear wave velocity adjustment coefficient αVs∈[0.8,1.2];

[0090] Foundation damping ratio adjustment coefficient αξ∈[0.9,1.1];

[0091] Initial stiffness adjustment coefficient of the isolation device αk∈[0.85,1.15];

[0092] Optimization goal: Minimize the comprehensive deviation index CPI.

[0093] Gaussian process regression is used to establish a proxy relationship between foundation parameters and CPI values. The proxy model can predict the corresponding CPI after each foundation parameter update. The proxy model expression is: ; is the total square error between the CPI predicted by the proxy model and the actual CPI, which is used to measure the fitting accuracy of the proxy model; r is the number of samples; is the predicted CPI of the i-th sample, is the actual CPI of the i-th sample.

[0094] Randomly sample several combinations of foundation physical parameters to actually evaluate CPI.

[0095] Based on the proxy model, the parameter group that may cause the CPI to fall in the next step is predicted (using the upper confidence limit algorithm UCB or the expected improvement EI criterion).

[0096] Updated foundation physics parameters and slightly adjusted simulation output.

[0097] Recalculate CPI and feed it back into the agent model.

[0098] The termination condition is to find a parameter combination that reduces the CPI to a low deviation level (CPI < 0.20) or to reach the maximum number of iterations (for example, 50).

[0099] Instead of retraining the entire deep neural network DNN, only the last few layers (such as the output layer or the penultimate layer) are trained with small step increments.

[0100] Incremental training using new corrected simulation data (mini-batches).

[0101] Maintain the model's previous knowledge while improving prediction accuracy for new ground conditions.

[0102] Independent validation seismic wave samples are used to evaluate the accuracy of the fine-tuned model's seismic isolation performance predictions under the new foundation parameters. Once the model is confirmed to meet the new low-bias rating, the updated version is solidified.

[0103] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0104] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0105] It should be understood that the term "and / or" herein is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent three situations: A exists alone, A and B exist at the same time, and B exists alone, where A and B may be singular or plural. In addition, the character " / " herein generally indicates that the objects associated with each other are in an "or" relationship, but it may also indicate an "and / or" relationship, which can be understood by referring to the context. A person of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0106] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A data center seismic isolation performance testing method based on AI intelligence, characterized by: include: Deploy multi-dimensional IoT sensors on data center building structures and equipment to collect real-time equipment vibration characteristic data during earthquake events and extract seismic wave characteristic parameters from a seismic database containing historical seismic wave records; Based on the collected equipment vibration characteristic data and seismic wave characteristic data, a deep neural network model is constructed and trained to learn the nonlinear mapping relationship between seismic input and structural response, forming a seismic isolation performance prediction model; For the actual monitored equipment vibration characteristic data and seismic wave characteristic data, the foundation shear wave velocity gradient anomaly and the isolation layer dynamic displacement response deviation rate are extracted respectively, and the comprehensive deviation index is calculated based on the preset weights; The method for extracting the outlier value of foundation shear wave velocity gradient is as follows: input the discrete data point set of measured foundation shear wave velocity with depth: ;in, It's depth, is the corresponding shear wave velocity; the shear wave velocity versus depth data set obtained by simulation modeling is: ; N is the total number of discrete data points of the measured foundation shear wave velocity with depth, The total number of shear wave velocity versus depth data obtained for simulation modeling; is the j-th depth point in the simulation model, For simulation models in depth The shear wave velocity at represents the foundation shear wave velocity value set at this depth in the simulation modeling. The shear wave velocity data is subjected to a one-dimensional discrete wavelet transform and decomposed into multiple scales, retaining only the low-frequency approximate component. The retained low-frequency wavelet is reconstructed back into a smooth shear wave velocity trend curve, which is recorded as , differentiate the trend curve by depth and calculate the gradient , the expression is: ; For the Measured depth points, The measured shear wave velocity trend curve at depth The value at The measured shear wave velocity trend curve at depth The simulated shear wave velocity is simply linearly interpolated and smoothed to calculate the simulated gradient , the expression is: ; Indicates the simulated shear wave velocity at depth The value at To simulate the shear wave velocity at depth The value at In each depth segment, the ratio of the actual gradient to the simulated gradient is calculated : ; Perform depth-weighted averaging on GGI(z) to obtain the overall foundation shear wave velocity gradient anomaly , the expression is: Where, is the maximum depth; The extraction method of the dynamic displacement response deviation rate of the isolation layer is as follows: obtain the actual monitored isolation layer displacement time history, marked as ; The displacement history of the isolation layer predicted by simulation is marked as ; The actual displacement time Simulated displacement time history Perform optimal alignment on the time axis and calculate the Euclidean distance matrix for each point between the two time course curves , the expression is: ; Search for a minimum cumulative cost path, connecting the start and end, representing the optimal alignment path, and output a set of aligned point pair sequences ; Indicates the actual displacement Corresponding simulation displacement ; Indicates the displacement time history of the simulation Aligned time points, calculate the instantaneous error for each pair of aligned points , the expression is: ; For simulations predicting displacement in time Calculate the average error of all alignment points : ;W is the number of point pairs on the alignment path, and the maximum predicted displacement peak value of the simulated displacement history is taken : The final isolation layer dynamic displacement response deviation rate IDRΔ is defined as: ; According to the comprehensive deviation index, the systematic deviation degree of the isolation performance prediction model is divided into high deviation, medium deviation and low deviation levels, and corresponding treatment is performed respectively; For situations with medium deviation levels, the foundation physical model parameters are dynamically updated, and the seismic isolation performance prediction model is retrained or optimized.

2. The AI-based data center seismic isolation performance detection method according to claim 1 is characterized by: The multi-dimensional Internet of Things sensor includes an acceleration sensor, a displacement sensor, a tilt sensor, a multi-axis gyroscope and an environmental sensor.

3. The AI-based data center seismic isolation performance detection method according to claim 1 is characterized by: The weight of the foundation shear wave velocity anomaly is set to 50%; the weight of the dynamic displacement deviation rate of the isolation layer is set to 50%; the GGI is normalized and the absolute value of the deviation from 1 is used as the metric: |GGI−1|; Calculate the comprehensive deviation index CPI, the calculation formula is: .

4. The AI-based data center seismic isolation performance testing method according to claim 1 is characterized by: According to the comprehensive deviation index, the systematic deviation degree of the isolation performance prediction model is divided into high deviation, medium deviation and low deviation levels, and corresponding treatment is performed respectively, including: If CPI ≥ 0.40, the systematic deviation degree of the isolation performance prediction model is classified as high deviation, indicating that the model is seriously mismatched and needs to be fully revised; If 0.20≤CPI < 0.40, the systematic deviation of the isolation performance prediction model is classified as medium deviation, indicating that the model has mismatch and requires local optimization. If CPI < 0.20, the systematic deviation degree of the isolation performance prediction model is classified as low deviation, indicating that the model is consistent with the actual situation and no adjustment is required.

5. The AI-based data center seismic isolation performance detection method according to claim 4 is characterized by: For the case of medium deviation level, Gaussian process regression is used to establish a proxy relationship between foundation parameters and CPI values. The proxy model can predict the corresponding CPI after each foundation parameter update. The proxy model expression is: ; is the total square error between the CPI predicted by the proxy model and the actual CPI; r is the number of samples; is the predicted CPI of the i-th sample, is the actual CPI of the i-th sample; Randomly sample several combinations of foundation physical parameters to actually evaluate CPI; Predict the parameter group that will cause CPI to fall in the next step based on the proxy model; Update foundation physical parameters and adjust simulation output; Recalculate CPI and feed it back into the proxy model; The termination condition is to find a parameter combination that reduces the CPI to a low deviation level or to reach the maximum number of iterations.

Images