A Method for Predicting the Performance of High-Performance Composite Materials of a Hub-Type Transmission Device

Abstract

The present invention relates to the technical field of composite material performance prediction, and specifically to a method for predicting the performance of high-performance composite materials of a hub-type transmission device, which includes the following steps: obtaining the mechanical properties and load information of the composite material, defining a preliminary prediction framework and generating preliminary dynamic response prediction values, collecting real-time data and denoising it, calculating the difference between the prediction value and the actual value, adjusting the input factors and updating the prediction value, obtaining strain data, analyzing the strain distribution, screening the severe regions, adjusting the mesh density, performing strength analysis, and generating local strength prediction results. The present invention optimizes the input factors through the analysis of the difference between real-time data and prediction, improves the prediction accuracy, the local strain analysis and mesh density adjustment reflect the strength distribution of the material under specific conditions, and the adaptive adjustment of the prediction framework can be optimized according to the actual working conditions, avoiding over-reliance on initial assumptions. The refined mesh and strain analysis provide local strength prediction, improving the flexibility and reliability of the prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of composite material performance prediction, and particularly to a method for predicting the performance of high-performance composite materials for a hub transmission device. Background Art

[0002] The technical field of composite material performance prediction involves the processes of modeling, analyzing, and evaluating the physical and mechanical properties of various composite materials under specific application scenarios. The core content of this technical field is to obtain the key performance indicators of composite materials under complex loading or service conditions, such as strength, stiffness, fatigue life, etc., through means such as theoretical analysis, experimental measurement, and data-driven modeling. Overall, the composite material performance prediction technology involves multiple links such as material microstructure modeling, performance parameter identification, experimental data analysis, and numerical simulation calculation, and is widely used in the design and optimization of high-performance structural components in aerospace, automotive, energy equipment, etc., especially suitable for scenarios where material behavior needs to be quickly obtained in the early design stage.

[0003] Among them, the method for predicting the performance of high-performance composite materials for a hub transmission device refers to, for high-performance composite material components applied in a hub transmission device, based on the known material composition and geometric characteristics, adopting a multi-scale mechanical modeling method, combining the thermo-mechanical coupling calculation theory and the structural response solution method, constructing a prediction model for the anisotropic performance of the composite material, and determining the key parameters in the model through experimental data regression to achieve the analysis and prediction of its strength distribution and thermal deformation trend. This method usually completes through means such as finite element modeling analysis, material lamination theory calculation, and stress-strain state analysis.

[0004] One of the deficiencies faced by the prior art in practical applications is that the prediction of the mechanical properties of composite materials usually relies too much on theoretical models and experimental data, while ignoring the dynamic feedback under actual working conditions. This method cannot fully handle the influence of working condition changes on material properties, and usually locks in the prediction parameters at the initial stage of design and cannot be adjusted according to the data obtained subsequently. Since these methods rely more on preliminary assumptions, during long-term use, the difference between the prediction results and the actual performance gradually expands, and it is impossible to effectively respond to the material behavior in a complex working environment. In addition, the prior art analyzes the stress-strain distribution rather roughly and cannot perform fine response prediction on local areas. Especially in the application of high-performance structural components, this deficiency may lead to design inaccuracies, affecting the long-term service life and safety of materials. Summary of the Invention

[0005] The purpose of the present invention is to solve the drawbacks existing in the prior art, and to propose a method for predicting the performance of high-performance composite materials for a hub transmission device.

[0006] To achieve the above object, the present invention adopts the following technical solutions: A method for predicting the performance of high-performance composite materials of a hub-type transmission device, comprising the following steps:

[0007] S1: Obtain the basic mechanical properties of the composite material, including elastic modulus, strength, and fracture toughness data and working load information, define a preliminary prediction framework based on the operating environment, establish a reference dynamic response value, and generate a preliminary dynamic response prediction value through the load data and environmental conditions;

[0008] S2: Based on the preliminary dynamic response prediction value, collect real-time data of the composite material under actual working conditions and perform noise filtering, compare with the current working load and environmental data, and generate a real-time monitoring data set;

[0009] S3: According to the real-time monitoring data set, calculate the difference between the dynamic response prediction value and the actual measurement value, adjust the input factors in the prediction framework, and generate an updated dynamic response prediction value;

[0010] S4: Based on the updated dynamic response prediction value, obtain the strain data of the composite material, analyze the regional strain distribution, and according to the relationship between the load intensity and the strain, screen the regions with drastic strain changes and reallocate the grid density to generate a local strain and load coupling region;

[0011] S5: Combine the local strain and load coupling region, adjust the grid density of the corresponding region, perform refined strength analysis, and generate a local strength prediction result based on the load and strain relationship of the region.

[0012] As a further solution of the present invention, the preliminary dynamic response prediction value includes a reference dynamic response value, load data, and environmental conditions; the real-time monitoring data set includes sensor data, noise-filtered data, working load information, and environmental data; the updated dynamic response prediction value includes a prediction framework after error analysis, adjusted input factors, and optimized correlation data; the local strain and load coupling region includes regions with drastic strain changes and reallocated grid density; the local strength prediction result includes a strength analysis result after adjusting the grid density and the load and strain relationship of the region.

[0013] As a further solution of the present invention, the specific steps of S1 are:

[0014] S101: Obtain the elastic modulus, strength, and fracture toughness of the composite material, synchronously collect the amplitude, direction, and frequency of the load during actual use, and extract the temperature, humidity, and corrosion medium type in the environment to obtain material performance data and environmental load data;

[0015] S102: Call the elastic modulus, strength, fracture toughness in the material property data and environmental load data, as well as the temperature, humidity, and corrosion medium type, convert the environmental data into material property adjustment values, calculate the standard frequency based on the elastic modulus, determine the material failure critical value by combining the strength and fracture toughness, and establish the reference response data;

[0016] S103: Call the standard frequency, critical stress, and crack growth rate in the reference response data, calculate the difference between the load frequency and the standard frequency, compare the critical stress with the load amplitude to obtain the remaining load-bearing range, adjust the growth rate in combination with the crack growth rate and environmental corrosion type, and integrate the difference value, remaining load-bearing range, and adjusted growth rate to generate the preliminary dynamic response prediction value.

[0017] As a further solution of the present invention, the specific steps of S2 are as follows:

[0018] S201: Based on the preliminary dynamic response prediction value, collect the real-time data of the sensor corresponding to the dynamic response prediction value of the composite material, smooth the fluctuation components in the sensor signal that exceed the prediction range, compare the difference between the original data and the truncation threshold range, screen out the abnormal fluctuation points, and generate the filtered sensor data set;

[0019] S202: Based on the filtered sensor data set, call the displacement gradient of the current working load and the environmental temperature data, calculate the linear relationship between the sensor strain value and the load displacement gradient, extract the compensation and correction term of the environmental temperature to the data, and perform a time series difference operation on the corrected data with the load and environmental reference to generate the load difference and environmental deviation;

[0020] S203: Integrate the load difference and environmental deviation with the filtered sensor data set, align the time series nodes according to the load cycle period, update the data association relationship at a fixed sampling interval, and generate the real-time monitoring data set.

[0021] As a further solution of the present invention, the specific calculation formula for comparing the difference between the original data and the truncation threshold range is:

[0022] ;

[0023] where, represents the dynamic fluctuation deviation value of the th sensor at time , represents the original monitoring data of the th sensor at time , represents the lower truncation threshold of the th sensor at time ; Representative Sensors at time The upper cutoff threshold at time, Representative The average value of all monitoring data of a sensor in a certain time window in the past. Represents the current time node The absolute value of the difference between the predicted and measured Represents the average absolute value of the difference of all time nodes in the current evaluation period.

[0024] As a further solution of the present invention, the specific steps of S3 are:

[0025] S301: obtaining the current dynamic response prediction value and the actual measurement value in the real-time monitoring data set, calculating the numerical difference between the two at the same time node, taking the absolute value of the difference as a reference, and generating a dynamic difference value;

[0026] S302: Based on the dynamic difference value, identify the time node where the difference value exceeds the preset error threshold, extract the original data of the input factor corresponding to the node, and perform reverse correction on the original impact ratio of the input factor according to the proportional relationship between the difference value and the error threshold to generate an input factor adjustment ratio;

[0027] S303: calling the input factor adjustment ratio, performing cumulative operation on the associated data in the prediction framework, performing product summation on the corrected impact degree and the original input factor, recalculating the full-time dynamic response prediction value, and generating an updated dynamic prediction value.

[0028] As a further solution of the present invention, the specific steps of S4 are:

[0029] S401: extracting strain data of the composite material under the updated dynamic response according to the updated dynamic response prediction value, collecting strain values of multiple regions based on the load intensity and strain relationship, calculating strain distribution of the region, and analyzing the strain intensity difference and change trend of the region to obtain regional strain distribution data;

[0030] S402: according to the regional strain distribution data, a threshold of strain change is set, regions with drastic strain changes are screened, key locations with stress concentration are identified, and marked as key regions, thereby obtaining data of drastic strain regions;

[0031] S403: According to the severe strain area data, the mesh density of the screening area is adjusted, the mesh division is refined, the strain response of the adjusted area is calculated, and a local strain and load coupling area is generated.

[0032] As a further solution of the present invention, the specific calculation formula for the strain distribution in the calculation area is:

[0033] ;

[0034] Calculate the regional strain distribution to obtain the strain intensity and the difference in regional strain intensity;

[0035] Among them, represents the strain value at the region , represents the load intensity received by this region, represents the stress area of this region, represents the material elastic coefficient of this region, represents the strain reaction radius of this region.

[0036] As a further solution of the present invention, the specific steps of S5 are as follows:

[0037] S501: According to the local strain and load coupling region, extract strain data and load information, establish a mesh division standard based on the relationship between regional strain and load, determine the mesh refinement region, obtain the initial mesh data of the refinement region, and perform mesh division optimization to generate a high-density mesh region;

[0038] S502: Based on the high-density mesh region, extract the local stress and strain information of the density mesh region, calculate the local strength through the stress-strain relationship to obtain the local strength prediction value;

[0039] S503: According to the local strength prediction value, compare the local strength prediction value with the set safety strength standard, determine the strength exceeding region, and generate a local strength prediction result by analyzing the relationship between load action and local strain.

[0040] As a further solution of the present invention, the specific calculation formula for establishing a mesh division standard based on the relationship between regional strain and load is:

[0041] ;

[0042] Among them, represents the refinement value of the mesh element, represents the strain value of the th grid point, represents the load corresponding to the grid point, represents the geometric region corresponding to the grid point, represents the maximum strain value within the strain region, represents the minimum strain value within the strain region, represents the total number of points within the mesh region.

[0043] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0044] In the present invention, the input factors are optimized through real-time data and prediction difference analysis, improving the prediction accuracy. The local strain analysis and mesh density adjustment reflect the strength distribution of the material under specific conditions. The adaptive adjustment of the prediction framework can be optimized according to the actual working conditions, avoiding over-reliance on initial assumptions. The refined mesh and strain analysis provide local strength prediction, enhancing the flexibility and reliability of the prediction. Brief Description of the Drawings

[0045] Figure 1 It is a schematic diagram of the step flow of the present invention. Detailed Embodiments

[0046] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0047] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as limiting the present invention. In addition, in the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0048] Please refer to Figure 1 , a method for predicting the performance of high-performance composite materials of a hub-type transmission device, comprising the following steps:

[0049] S1: Obtain the basic mechanical properties of the composite material, including elastic modulus, strength and fracture toughness data, collect the working load information, define a preliminary prediction framework based on the operating environment, establish a reference dynamic response value, and generate a preliminary dynamic response prediction value by inputting the load data and environmental conditions;

[0050] S2: Based on the preliminary dynamic response prediction value, collect the real-time data of the composite material under the actual working conditions, filter the noise of the collected sensor data, and compare the obtained data with the current working load and environmental data to generate a real-time monitoring data set;

[0051] S3: According to the real-time monitoring data set, calculate the difference between the current dynamic response prediction value and the actual measured value, adjust the input factors in the prediction framework through error analysis, and adjust the associated data using an optimization method to generate an updated dynamic response prediction value;

[0052] S4: Based on the updated dynamic response prediction values, obtain the strain data of the composite material, analyze the regional strain distribution, and according to the relationship between the load intensity and strain, screen out the regions with drastic strain changes, and reassign the mesh density to the regions to generate a local strain and load coupling region;

[0053] S5: Combine the local strain and load coupling region, adjust the mesh density of the corresponding region, perform a refined strength analysis, and generate local strength prediction results based on the load and strain relationship of the region.

[0054] The preliminary dynamic response prediction values include the benchmark dynamic response value, load data, and environmental conditions; the real-time monitoring data set includes sensor data, data after noise filtering, working load information, and environmental data; the updated dynamic response prediction values include the prediction framework after error analysis, adjusted input factors, and optimized associated data; the local strain and load coupling region includes the regions with drastic strain changes and the reassigned mesh density; the local strength prediction results include the strength analysis results after adjusting the mesh density and the load and strain relationship of the region.

[0055] The specific steps of S1 are as follows:

[0056] S101: Obtain the elastic modulus, strength, and fracture toughness of the composite material, synchronously collect the amplitude, direction, and frequency of the load during actual use, and extract the temperature, humidity, and corrosion medium type in the environment to obtain material property data and environmental load data;

[0057] Perform a tensile test on the composite material, apply tensile force at a constant rate until fracture, and record the slope of the linear segment of the stress-strain curve as the elastic modulus. For example, when the measured stress is 700 MPa and the corresponding strain is 0.2%, the elastic modulus is obtained by dividing the stress by the strain as 350 GPa. Load the specimen to fracture through a three-point bending test, record the maximum load of 800 N, and calculate the bending strength as 300 MPa according to the span, specimen width, and thickness. Use a compact tension specimen to prefabricate a crack and load it to the critical state, and calculate the fracture toughness as 35 MPa· , install sensors at the root of the wind turbine blade to monitor that the load amplitude fluctuates in the range of 50 - 200 kN, the direction changes periodically between 0° and 90° as the blade rotates, the vibration frequency is determined by spectral analysis with the main frequency being 0.5 - 2 Hz, the temperature in the coastal environment fluctuates in the range of 15 - 30 °C, the humidity is 60 - 95%, and the corrosion medium is confirmed by chemical detection to contain 3.5% sodium chloride solution. Integrate the elastic modulus, strength, and fracture toughness as material property data, and integrate the load amplitude, direction, frequency, and environmental parameters as environmental load data to generate material property data and environmental load data.

[0058] S102: Call the elastic modulus, strength, fracture toughness in the material property data and environmental load data, and temperature, humidity, and corrosion medium type. Convert the environmental data into material property adjustment values, calculate the standard frequency based on the elastic modulus, and determine the material failure critical value by combining strength and fracture toughness to establish the reference response data;

[0059] According to the elastic modulus of 350 GPa and the material density of 1600 kg / m³, combined with the structural characteristic length of 5 m, calculate the standard frequency as the square root of the ratio of the elastic modulus to the density divided by twice the product of the characteristic length and pi, and the result is 47.1 Hz. For a temperature of 30 °C, calculate the temperature attenuation coefficient as 0.99 according to the rule that the elastic modulus decreases by 0.1% per degree Celsius. For a humidity of 95%, calculate the humidity attenuation coefficient as 0.825 according to the rule that the strength decreases by 0.05% per percentage of humidity. For the sodium chloride solution corrosion environment, select the fracture toughness attenuation coefficient of 0.7. After comprehensive adjustment, the critical stress is the original strength of 300 MPa multiplied by the temperature, humidity, and corrosion coefficients in turn, resulting in 171.6 MPa. The crack growth rate is calculated based on an initial crack length of 2 mm and a load fluctuation amplitude of 100 MPa, and the growth rate is 2.1×10^-3 mm per cycle according to the empirical model. Integrate the standard frequency, critical stress, and growth rate as the reference response data.

[0060] S103: Call the standard frequency, critical stress, and crack growth rate in the reference response data, calculate the difference between the load frequency and the standard frequency, compare the critical stress with the load amplitude to obtain the remaining load-bearing range, adjust the growth rate by combining the crack growth rate and the environmental corrosion type, and integrate the difference value, remaining load-bearing range, and adjusted growth rate to generate the preliminary dynamic response prediction value;

[0061] Calculate the difference between the actual load frequency of 2 Hz and the standard frequency of 47.1 Hz, and the difference value is 45.1 Hz. The difference ratio is the proportion of the actual difference to the standard frequency, which is 95.7%. Convert the load amplitude of 200 kN into the stress per unit area. Assuming the load-bearing area is 0.02 m², the actual stress is 10 MPa. Compare it with the critical stress of 171.6 MPa, and the remaining load-bearing range is the difference between the two, which is 161.6 MPa. According to the sodium chloride solution corrosion environment, increase the crack growth rate by 1.5 times on the original rate, and the adjusted rate is 3.15×10^-3 mm per cycle. According to the weight distribution rule of 60% for the difference ratio, 30% for the remaining load-bearing range, and 10% for the growth rate, calculate the dynamic response prediction value as the weighted sum of each item, which is 0.919, and generate the dynamic response prediction value.

[0062] The specific steps of S2 are as follows:

[0063] S201: Based on the preliminary dynamic response prediction value, collect the real-time data of the sensors corresponding to the dynamic response prediction value of the composite material. Through the dynamic fluctuation threshold truncation method, smooth the fluctuation components in the sensor signal that exceed the prediction range, compare the difference between the original data and the truncation threshold range, screen out the abnormal fluctuation points, and generate a filtered sensor data set;

[0064] The specific calculation formula for comparing the difference between the original data and the truncation threshold range is:

[0065] ;

[0066] Where, represents the dynamic fluctuation deviation value of the th sensor at time , represents the original monitoring data of the th sensor at time , represents the lower truncation threshold of the th sensor at time , represents the upper truncation threshold of the th sensor at time , represents the average value of all monitoring data of the th sensor within a certain past time window, represents the absolute value of the difference between the prediction and the actual measurement at the current time node , represents the average value of the absolute values of the differences at all time nodes within the current evaluation time period;

[0067] In a monitoring task of a certain composite material structure, the data of the y = 3 sensor at time t = 14:00 is collected and enters the analysis process. The original monitoring data collected by the sensor is mm, which is sourced from the real-time data upload interface of the fiber Bragg grating displacement sensor. The sampling frequency is 1 Hz, and the data is included in the calculation after being normalized and filtered to remove noise.

[0068] In the statistical data of the past 60 minutes, the upper and lower dynamic thresholds generated by the fluctuation anomaly detection module are: mm, sourced from the historical minimum fluctuation boundary set by the sliding window method; mm, sourced from the maximum stable fluctuation boundary in the past 60 minutes. The two are screened by the range method of historical data at the sampling frequency and a median filtering process is introduced to form a fluctuation interval.

[0069] The average value of the sensor numbered 3 in the past 1 hour is: mm, which is obtained by averaging the monitoring data of the past 60 time points. The data comes from the monitoring upload cache table within a 60-minute cycle, and the calculation method is the sum of all historical values divided by the number of time points.

[0070] In the previous execution step, the predicted value at this moment obtained from the prediction model is 6.4 mm, and the measured value is 5.6 mm; the absolute value of the difference between the two is: mm; and the average value of the predicted and measured differences at all time points within this monitoring cycle is: ; which is the average value of the predicted difference sequence from the past 60 time points.

[0071] Weight term V t rises as the prediction error increases. To prevent its excessive influence, the denominator is set to to ensure the pulling coefficient of the balance offset term and maintain the dynamic adjustment ability of the weight within the control range.

[0072] Substituting all data into the formula, the calculation process is as follows:

[0073] The first term is the absolute value of the difference at the midpoint of the truncation interval, calculated as:

[0074] ;

[0075] The second term is the fluctuation deviation adjustment term, which includes the square root term of the historical mean deviation and the normalization term of the predicted difference, calculated as:

[0076] ; The product of the two terms is:

[0077] ;

[0078] Finally, the results of the two parts are combined:

[0079] ;

[0080] This result shows that the dynamic fluctuation deviation value of sensor No. 3 at 14:00 is 0.506 mm. This value is used to compare with the fluctuation identification threshold. If it is greater than the set screening line value of 0.45 mm, it is determined as an abnormal fluctuation point and written as an abnormal mark into the filtered sensor dataset for subsequent data cleaning processes and error feedback model calculations.

[0081] S202: Based on the filtered sensor dataset, call the displacement gradient of the current workload and the ambient temperature data, calculate the linear relationship between the sensor strain value and the load displacement gradient, extract the compensation and correction term of the ambient temperature for the data, perform a time series difference operation on the corrected data with the load and ambient benchmarks to generate the load difference and ambient deviation;

[0082] Based on the time series data of displacement gradient and ambient temperature, taking a bridge health monitoring scenario as an example, the original strain data under load cycles is collected. Assuming that the displacement gradient of the current working load is 0.25 mm per N and the ambient temperature is 25 °C, the relationship between strain and displacement gradient is established through linear regression analysis. For example, the displacement gradients at three sampling points are 0.2, 0.25, and 0.3 mm per N respectively, and the corresponding strain values are 50, 62, and 75 microstrains. When calculating, the average value of the displacement gradient is 0.25 mm per N, and the average value of the strain is 62.33 microstrains. The linear coefficient is obtained as 250 microstrains per mm per N through the ratio of covariance to variance. The temperature compensation term is calculated based on the difference between the ambient temperature and the reference temperature of 20 °C, and the temperature coefficient is 0.5 microstrains per °C. When the ambient temperature is 25 °C, the compensation value is 2.5 microstrains. The corrected strain value is the original strain minus the compensation value. For example, the original strain of 62 microstrains is corrected to 59.5 microstrains. The load difference is calculated by the difference between the current load and the reference load of 1000 N. For example, when the current load is 1050 N, the difference is 50 N. The environmental deviation is the difference between the ambient temperature and the reference temperature of 5 °C. Finally, the time series of load difference and environmental deviation is output.

[0083] S203: Integrate the load difference and environmental deviation with the filtered sensor dataset, align the time series nodes according to the load cycle period, update the data association relationship at a fixed sampling interval, establish a multi-dimensional structure including strain, load, and temperature, and generate a real-time monitoring dataset;

[0084] Taking the monitoring of a wind turbine tower as an example, the load cycle period is 10 minutes. Align the time nodes of the load difference, environmental deviation, and corrected strain data. The sampling interval is 5 seconds. Assume that the load peak in a certain period is 1200 N, and the ambient temperature fluctuates between 18 and 22 °C. The corrected strain data is divided into 120 groups according to the time window. Each group contains a strain value, a load difference, and a temperature deviation. For example, the strain value at the 5th second is 58.3 microstrains, the load difference is 48 N, and the temperature deviation is 4.2 °C, which is stored as the first group of data. The data at the 10th second is 59.1 microstrains, 52 N, and 4.5 °C, which is stored as the second group of data. The missing time points are filled by interpolating the data before and after. For example, when the 7.5th second is missing, take the intermediate value of the data at the 5th second and the 10th second. The strain value is 58.7 microstrains, the load difference is 50 N, and the temperature deviation is 4.35 °C. After updating the data association, a real-time monitoring dataset containing 120 groups of three-dimensional data points is formed.

[0085] The specific steps of S3 are as follows:

[0086] S301: Obtain the current dynamic response prediction value and the actual measurement value in the real-time monitoring data set, calculate the numerical difference between the two at the same time node, and use the absolute value of the difference as the benchmark to generate a dynamic difference value;

[0087] First, data is collected in real time through sensor devices installed on the target object. For example, in a building, accelerometers, inclinometers and other monitoring devices are deployed to obtain the actual measurement values of the structural response. At the same time, the future structural response is predicted through a data-driven model (such as a long short-term memory network LSTM or a finite element model based on physical rules). The two types of data are aligned with timestamps to ensure the comparability of data at each time node. In actual operation, the time node is set to once every 5 minutes. The data will enter the main server through the data middleware for parsing, and then the predicted value and the actual value at each time point are extracted and the difference is calculated to obtain the dynamic difference value at that moment. The difference is absolutized to form a difference index sequence. The system will establish the association between this sequence and the sensor number, location, and data source to facilitate the subsequent tracking of abnormal points. For example, in a subway tunnel construction project, the monitoring system shows a predicted deformation of 3.6 mm at a certain moment, while the actual measurement is 2.9 mm. Then the difference at this time point is 0.7 mm. This difference value will be immediately marked and uploaded to become the dynamic difference value. This operation is completed in the server with streaming data processing. The data is sent into the operation module in real time through the pipeline, and the difference calculation is completed one by one for all time periods and all sensor data, and the dynamic difference value sequence is output as the input for subsequent processing.

[0088] S302: Based on the dynamic difference value, identify the time nodes where the difference value exceeds the preset error threshold, extract the original input factor data corresponding to these nodes, and perform an inverse correction on the original influence ratio of the input factor according to the proportional relationship between the difference value and the error threshold to generate an input factor adjustment ratio;

[0089] After generating the dynamic difference value sequence, the platform needs to determine whether there are time nodes exceeding the error tolerance range. The set error threshold is usually determined by historical data analysis and safety requirements. For example, in deformation monitoring, it is usually set in the range of 1.0 mm to 2.0 mm. If the difference value at the current time point is higher than this threshold, the system identifies this time point as an abnormal node. Subsequently, all the original input factor data corresponding to this moment is retrieved from the database, including environmental parameters (such as wind speed, humidity), load factors, and structural characteristic parameters (such as stiffness, damping coefficient). Then, according to the ratio of the current difference value to the error threshold, the reverse correction of the influence weight is performed. That is, if the difference value is much larger than the threshold, it indicates that the original influence ratio of the current input factor is overestimated, and the system will lower its weight. Conversely, if the difference value is close to the threshold, the correction amplitude is smaller. The weights of the input factors here usually come from the preliminary model fitting stage and are obtained through the fitting training results. For example, in building deformation prediction, it is found that the wind speed input value is 6 m / s, but the model prediction result is too high, indicating that the contribution of wind speed to the predicted value is over-amplified. Then the originally set influence ratio of wind speed is reduced to half of the original. Similarly, other parameters are adjusted according to the same ratio. Finally, a new adjusted ratio data set containing all input factors is output for subsequent prediction use.

[0090] S303: Invoke the adjustment ratio of the input factors, perform an accumulation operation on the associated data within the prediction framework, sum the product of the corrected influence degree and the original input factors, recalculate the dynamic response prediction value for the entire time period, and generate the updated dynamic prediction value;

[0091] After completing the correction ratio of the input factors, it is necessary to apply it to the prediction model again to update the overall prediction value. For the original prediction model, the system first applies the correction ratio to each input factor and performs the adjustment item by item according to the data correspondence. Then, the corrected data is input into the original prediction framework again, and the model performs re-inference to generate the dynamic prediction value sequence for the entire time period. For example, in structural displacement prediction, the original input factors such as wind speed, ground humidity, and traffic load are 5, 0.3, and 200 respectively, and their corresponding influence ratios are 0.2, 0.4, and 0.4. After correction, the ratios become 0.1, 0.6, and 0.5. The original data is matched with the new ratios item by item to generate the corrected input data, which is then input into the prediction module again to recalculate the prediction result at this time point. This process is carried out one by one for each time point, so that the prediction results for the entire time period are all the new corrected values. If the system is a deep neural network structure, it performs re-inference after dynamically adjusting the model weights, or uses a patch layer to append processing logic in front of the original model to form a new prediction sequence by correcting the influence of the input factors in real time. Finally, an updated dynamic response prediction value data set is formed for model result display and data release.

[0092] The specific steps of S4 are as follows:

[0093] S401: Extract the strain data of the composite material under the updated dynamic response according to the updated dynamic response prediction value. Based on the relationship between load intensity and strain, collect the strain values of multiple regions, calculate the strain distribution of the regions, analyze the strain intensity differences and change trends of the regions, and obtain the regional strain distribution data;

[0094] The specific calculation formula for calculating the strain distribution of the region is:

[0095] ;

[0096] Calculate the strain distribution of the region to obtain the strain intensity and the regional strain intensity difference;

[0097] Where, represents the strain value at region , represents the load intensity received by this region, represents the stress area of this region, represents the material elastic coefficient of this region, represents the strain reaction radius of this region;

[0098] Suppose in a certain composite material structure, region is selected for analysis. The following parameter values are obtained through experiments and calculations:

[0099] ;

[0100] ;

[0101] ;

[0102] ;

[0103] Substitute the above numerical values into the formula:

[0104] ;

[0105] The calculation steps are as follows:

[0106] Calculate :

[0107] ;

[0108] Calculate :

[0109] ;

[0110] Substitute the above results into the formula:

[0111] ;

[0112] Calculate the final result:

[0113] ;

[0114] This result indicates that in the region the strain value is .

[0115] S402: According to the regional strain distribution data, set the threshold of strain change, screen the regions with drastic strain changes, identify the key parts with stress concentration, and mark them as key regions to obtain the data of regions with drastic strain changes;

[0116] Based on the regional strain distribution data, the threshold of strain change can be set, and the regions with drastic strain changes can be screened out through the threshold. Regions with drastic strain changes usually show different strain values from the surrounding regions or have large fluctuations in their strain values. In actual operation, first, an initial threshold can be set according to historical data. For example, regions with a strain change value greater than 0.003 are considered regions with drastic strain changes. Then, calculate the strain change rate of each region, and regions with a larger change rate will be determined as regions with drastic strain changes. For example, if the strain change rate of a certain region between two time steps is 0.008, exceeding the set threshold of 0.003, then this region will be identified as a region with drastic strain changes. In this way, the key parts with stress concentration can be effectively identified. These parts may have the risk of crack initiation or material failure, so they should be marked as key monitoring regions to ensure continuous monitoring of material properties.

[0117] S403: According to the data of regions with drastic strain changes, adjust the mesh density of the screened regions, refine the mesh division, calculate the strain response of the adjusted regions, and generate the local strain and load coupling regions;

[0118] Based on the data of the severely strained area screened out, the grid density of this area is adjusted. In actual operation, the density of grid division is a key factor. The denser the grid, the more accurate the calculation result. Usually, the adjustment amount of grid density can be judged according to the severity of strain change. For areas with large strain changes, a higher grid density can be adopted. Suppose the grid size of a severely strained area is 10mm, and through analysis, it is found that the strain change in this area is large. Therefore, the grid size is reduced to 5mm to improve the calculation accuracy. For the area after adjusting the grid density, the strain response of this area can be recalculated. When recalculating, the finite element analysis method can be used, combined with the geometric shape, material properties and load conditions of the area to obtain a more accurate strain response result. In this process, the accuracy of the strain response is improved by refining the grid, which can better describe the load distribution and strain relationship in the local area, and finally obtain the local strain and load coupling area, further providing a basis for subsequent design optimization.

[0119] The specific steps of S5 are as follows:

[0120] S501: According to the local strain and load coupling area, extract strain data and load information, establish a grid division standard based on the relationship between regional strain and load, determine the grid refinement area, obtain the initial grid data of the refinement area, and perform grid division optimization to generate a high-density grid area;

[0121] The specific calculation formula for establishing the grid division standard based on the relationship between regional strain and load is:

[0122] ;

[0123] Among them, represents the refinement value of the grid element, represents the strain value of the th grid point, represents the load corresponding to the grid point, represents the geometric area corresponding to the grid point, represents the maximum strain value within the strain area, represents the minimum strain value within the strain area, represents the total number of points within the grid area;

[0124] Formula derivation process:

[0125] Calculate the weight factor of each grid point

[0126] ;

[0127] This weight factor comprehensively considers the influence of strain, load and geometric size, and is used to reflect the importance of each grid point in grid division.

[0128] Calculate the sum of squares of the weight factors

[0129] ;

[0130] By summing the squares of the weight factors at all grid points, a comprehensive index is obtained to guide the accuracy requirements of grid division.

[0131] Calculate the grid refinement value

[0132] ;

[0133] Grid refinement value It reflects the accuracy level that the grid division needs to achieve under the current load and strain distribution. The smaller the value, the finer the grid division required.

[0134] Calculation example:

[0135] Suppose in a certain structural analysis, the following are the actual data of each grid point:

[0136] Grid point 1: Strain , Load N, Area size m 2 .

[0137] Grid point 2: Strain , Load N, Area size m 2 .

[0138] Grid point 3: Strain , Load N, Area size m 2 .

[0139] Grid point 4: Strain , Load N, Area size m 2 .

[0140] Grid point 5: Strain , Load N, Area size m 2 .

[0141] Among them, the maximum strain value , the minimum strain value , so .

[0142] Calculate the weight factor of each grid point

[0143] ;

[0144] ;

[0145] ;

[0146] ;

[0147] ;

[0148] Calculate the sum of squares of the weight factors

[0149] ;

[0150] ;

[0151] Calculate the mesh refinement value

[0152] ;

[0153] Result analysis: The calculated mesh refinement value is 4.863, indicating that under the current strain and load distribution conditions, the mesh refinement degree should be controlled below this value to ensure the accuracy and precision of the calculation results.

[0154] S502: Based on the high-density mesh region, extract the local stress-strain information of the density mesh region, calculate the local strength through the stress-strain relationship, and obtain the local strength prediction value;

[0155] Based on the already optimized high-density mesh region, extract the local stress-strain data from each mesh node. The relationship between stress and strain is described by the known material constitutive relationship (such as the linear elastic material model). The constitutive model of the material can help predict the stress borne by the material at different strain levels. In this way, the corresponding local stress can be calculated in each mesh region. Then, use the mechanical model to deduce the strength of the local region, which can be used to predict the strength through, for example, the von Mises stress criterion to ensure accurate strength prediction in the local region under different load conditions. The key to this process is to be able to calculate the corresponding local strength prediction value according to the specific load and strain information of each region, further providing data support for the subsequent structural safety assessment.

[0156] S503: According to the local strength prediction value, compare the local strength prediction value with the set safety strength standard, determine the strength-exceeding region, and generate the local strength prediction result by analyzing the relationship between the load action and the local strain;

[0157] First, a safety strength standard is set, which can be determined by the tensile strength of the material or the safety factor specified in the engineering design. For example, if the set safety strength standard is 300 MPa and the predicted strength value in a certain local area is 350 MPa, then this area is determined as an area with excessive strength. Next, by comparing the local strength prediction value with the set safety standard, all areas with excessive strength are found. In this process, the relationship between the load action and the local strain also needs to be carefully analyzed because the change of local strain under different loads will affect the strength prediction. For example, when the wing surface bears a large load, the local strain increases, resulting in the local stress and strength possibly reaching or exceeding the safety strength standard. By analyzing the interaction relationship between the load and the strain, the areas that may exceed the standard can be accurately determined, and the local strength prediction results can be generated.

[0158] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still belong to the protection scope of the technical solution of the present invention.

Claims

1. A method for predicting the performance of high-performance composite materials of a hub-type transmission device, characterized in that, The following steps are involved: S1: Obtain the basic mechanical properties of composite materials, including elastic modulus, strength and fracture toughness data and working load information, define a preliminary prediction framework based on the operating environment, establish a benchmark dynamic response value, and generate preliminary dynamic response prediction values through load data and environmental conditions; S2: Based on the preliminary dynamic response prediction value, collect real-time data of the composite material under actual working conditions and perform noise filtering, compare it with the current working load and environmental data, and generate a real-time monitoring data set; the specific steps of S2 are: S201: Based on the preliminary dynamic response prediction value, collect the real-time sensor data corresponding to the dynamic response prediction value of the composite material, smooth the fluctuation component in the sensor signal that exceeds the prediction range, compare the difference between the original data and the truncation threshold range, filter out abnormal fluctuation points, and generate a filtered sensor data set; the specific calculation formula for comparing the difference between the original data and the truncation threshold range is: ； Among them, represents the dynamic fluctuation deviation value of the th sensor at time moment, represents the original monitoring data of the th sensor at time moment, represents the lower truncation threshold of the th sensor at time moment, represents the upper truncation threshold of the th sensor at time moment, represents the average value of all monitoring data of the th sensor within a certain past time window, represents the absolute value of the difference between the prediction and the actual measurement at the current time node , represents the average value of the absolute values of the differences at all time nodes within the current evaluation time period; S202: Based on the filtered sensor data set, the displacement gradient of the current working load and the ambient temperature data are called, the linear relationship between the sensor strain value and the load displacement gradient is calculated, the compensation correction item of the ambient temperature on the data is extracted, and the correction data is subjected to time series difference calculation with the load and environmental reference to generate the load difference and environmental deviation; S203: Integrate the load difference and environmental deviation with the filtered sensor data set, align time series nodes according to the load cycle period, update the data association relationship at a fixed sampling interval, and generate a real-time monitoring data set; S3: Calculate the difference between the dynamic response prediction value and the actual measurement value according to the real-time monitoring data set, adjust the input factors in the prediction framework, and generate an updated dynamic response prediction value; S4: Based on the updated dynamic response prediction value, obtain the composite material strain data, analyze the regional strain distribution, screen the area with drastic strain changes and redistribute the grid density according to the relationship between load intensity and strain, and generate the local strain and load coupling area; S5: In combination with the local strain and load coupling area, the mesh density of the corresponding area is adjusted to perform a refined strength analysis, and a local strength prediction result is generated according to the load and strain relationship of the area.

2. The performance prediction method of the high-performance composite material for the hub-type transmission device according to claim 1, characterized in that The preliminary dynamic response prediction value includes the benchmark dynamic response value, load data, and environmental conditions; the real-time monitoring data set includes sensor data, noise-filtered data, workload information, and environmental data; the updated dynamic response prediction value includes the prediction framework after error analysis, adjusted input factors, and optimized associated data; the local strain and load coupling area includes the area with drastic strain changes and the redistributed grid density; the local strength prediction result includes the strength analysis result after adjusting the grid density and the regional load-strain relationship.

3. The high-performance composite material performance prediction method for the hub-type transmission device according to claim 1, characterized in that The specific steps of S1 are: S101: Obtain the elastic modulus, strength and fracture toughness of the composite material, simultaneously collect the amplitude, direction and frequency of the load in actual use, extract the temperature, humidity and corrosive medium type in the environment, and obtain material performance data and environmental load data; S102: calling the elastic modulus, strength, fracture toughness, temperature, humidity, and corrosive medium type in the material performance data and the environmental load data, converting the environmental data into a material performance adjustment value, calculating the standard frequency based on the elastic modulus, determining the material failure critical value in combination with the strength and fracture toughness, and establishing the benchmark response data; S103: calling the standard frequency, critical stress, and crack growth rate in the benchmark response data, calculating the difference between the load frequency and the standard frequency, comparing the critical stress and the load amplitude to obtain the remaining load range, adjusting the growth rate in combination with the crack growth rate and the environmental corrosion type, integrating the difference value, the remaining load range, and the adjusted growth rate, and generating a preliminary dynamic response prediction value.

4. The method for predicting the performance of high-performance composite materials of the hub-type transmission device according to claim 1, wherein The specific steps of S3 are: S301: obtaining the current dynamic response prediction value and the actual measurement value in the real-time monitoring data set, calculating the numerical difference between the two at the same time node, taking the absolute value of the difference as a reference, and generating a dynamic difference value; S302: Based on the dynamic difference value, identify the time node where the difference value exceeds the preset error threshold, extract the original data of the input factor corresponding to the node, and perform reverse correction on the original impact ratio of the input factor according to the proportional relationship between the difference value and the error threshold to generate an input factor adjustment ratio; S303: calling the input factor adjustment ratio, performing cumulative operation on the associated data in the prediction framework, performing product summation on the corrected impact degree and the original input factor, recalculating the full-time dynamic response prediction value, and generating an updated dynamic prediction value.

5. The high-performance composite material property prediction method for the hub-type transmission device according to claim 4, characterized in that, The specific steps of S4 are: S401: extracting strain data of the composite material under the updated dynamic response according to the updated dynamic response prediction value, collecting strain values of multiple regions based on the load intensity and strain relationship, calculating strain distribution of the region, and analyzing the strain intensity difference and change trend of the region to obtain regional strain distribution data; S402: according to the regional strain distribution data, a strain change threshold is set, the region with drastic strain change is screened, key locations with stress concentration are identified, and marked as key areas, thereby obtaining data of drastic strain areas; S403: According to the severe strain area data, the mesh density of the screening area is adjusted, the mesh division is refined, the strain response of the adjusted area is calculated, and a local strain and load coupling area is generated.

6. The high-performance composite material property prediction method for the hub-type transmission device according to claim 5, characterized in that The specific calculation formula for the strain distribution in the calculation area is: ； Calculate the regional strain distribution and obtain the strain intensity and regional strain intensity difference; Among them, represents the strain value at the representative area, represents the load intensity received by this area, represents the stress area of this area, represents the material elastic coefficient of this area, represents the strain reaction radius of this area.

7. The performance prediction method of the high-performance composite material for the hub-type transmission device according to claim 5, characterized in that The specific steps of S5 are: S501: extracting strain data and load information according to the local strain and load coupling area, establishing a meshing standard based on the regional strain and load relationship, determining a mesh refinement area, obtaining initial mesh data of the refinement area, and performing meshing optimization to generate a high-density mesh area; S502: extracting local stress-strain information of the high-density grid area based on the high-density grid area, calculating the local strength through the stress-strain relationship, and obtaining a local strength prediction value; S503: According to the local strength prediction value, compare the local strength prediction value with the set safety strength standard to determine the area where the strength exceeds the standard, and generate a local strength prediction result by analyzing the relationship between the load action and the local strain.

8. The method for predicting the performance of the high-performance composite material of the hub-type transmission device according to claim 7, characterized in that, The specific calculation formula for establishing the mesh division standard based on the relationship between regional strain and load is as follows: ； Among them, represents the refinement value of the grid cell, represents the th strain value of the grid point, represents the load corresponding to the grid point, represents the geometric area corresponding to the grid point, represents the maximum strain value within the strain area, represents the minimum strain value within the strain area, represents the total number of points within the grid area.

Images