Flange connection tightness calculation modeling method, calculation method, system, equipment and medium

Abstract

The invention discloses a flange connection tightness calculation modeling method, a calculation method, a system, equipment and a medium. The modeling method comprises the following steps: acquiring flange joint surface historical gaps under different temperatures and loads; constructing a basic gap physical model according to the historical gap of the flange joint surface under different temperatures and loads; according to the basic gap physical model and the historical gap of the flange joint surface under different temperatures and loads, calculating gap deviations under different temperatures and loads; constructing a sample data set according to the gap deviations under different temperatures and loads; and deploying a deep learning model, and training the deep learning model by using the sample data set to obtain a gap deviation prediction model. According to the method, the calculation precision of the flange connection tightness is improved, and real-time and continuous calculation of the flange connection tightness can be realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of flange connection safety, and in particular relates to a flange connection tightness calculation modeling method, a calculation method, a system, a device and a medium. Background Art

[0002] Flange connection is one of the most common connection forms in industrial equipment, and is widely used in important fields such as wind power equipment, aerospace equipment, and petrochemical pipelines. The tightness of flange connection is directly related to the safety, reliability, and service life of the equipment. However, in actual operation, the tightness of flange connection may decrease due to factors such as temperature changes, load fluctuations, and loose bolts, and even cause serious problems such as leakage, vibration, and structural damage.

[0003] At present, flange connection tightness monitoring mainly relies on the following methods: First, regular offline inspections, such as using a torque wrench to detect bolt preload and an ultrasonic thickness gauge to detect bolt length changes, etc. The second is simple sensor monitoring, such as pressure sensors monitoring medium pressure changes, bolt strain monitoring, etc.; The third is the theoretical calculation method, which predicts the flange connection status based on elasticity theory and finite element analysis.

[0004] However, these methods have obvious shortcomings: Offline inspection cannot achieve real-time monitoring, and equipment shutdown inspection increases operation and maintenance costs; simple sensor monitoring is difficult to fully reflect the actual status of flange connections under complex working conditions; theoretical calculation methods are usually based on ideal assumptions, ignoring actual influencing factors such as temperature changes and dynamic loads, and the calculation accuracy is limited.

[0005] Therefore, it is urgent to propose a method that can accurately, efficiently and real-time evaluate the tightness status of flange connections. Summary of the invention

[0006] The purpose of the present invention is to provide a flange connection tightness calculation modeling method, calculation method, system, equipment and medium to solve the problems of insufficient monitoring accuracy and poor real-time performance of traditional methods.

[0007] The present invention solves the above technical problems through the following technical solutions: a flange connection tightness calculation modeling method, comprising: Obtain the historical clearance of flange joint surfaces under different temperatures and loads; The basic gap physical model is constructed based on the historical gap of the flange joint surface under different temperatures and loads; Calculate the gap deviation under different temperatures and loads according to the basic gap physical model and the historical gaps of the flange joint surfaces under different temperatures and loads; Construct a sample data set based on gap deviations at different temperatures and loads; Deploy a deep learning model, and use the sample data set to train the deep learning model to obtain a gap deviation prediction model.

[0008] Furthermore, a basic gap physical model is constructed according to the historical gap of the flange joint surface under different temperatures and loads, including: Constructing a flange finite element model; wherein the flange finite element model includes an upper flange, a lower flange, and bolts connecting the upper flange and the lower flange, and the gap between the upper flange and the lower flange is a flange joint surface gap; Based on the flange finite element model, simulation calculations are performed at different temperatures to obtain the simulated clearances of the flange joint surfaces at different temperatures; the simulated clearances of the flange joint surfaces at different temperatures are fitted to obtain a temperature-joint surface clearance fitting function; Based on the flange finite element model, simulation calculations of different loads are performed to obtain the simulated clearance of the flange joint surface under different loads; the simulated clearance of the flange joint surface under different loads is fitted to obtain a load-joint surface clearance fitting function; The flange joint surface gap caused by temperature and the flange joint surface gap caused by load are calculated according to the historical gap of the flange joint surface under different temperatures and loads, as well as the temperature-joint surface gap fitting function and the load-joint surface gap fitting function; According to the flange joint surface gap caused by temperature, the flange joint surface gap caused by load and the flange joint surface historical gap, the basic gap physical model is obtained.

[0009] Furthermore, a sample data set is constructed based on the gap deviation under different temperatures and loads, including: Constructing a characteristic vector according to each set of temperature and load; wherein the characteristic vector includes temperature, load and interaction value between temperature and load; Performing standardization on the feature vector; The standardized feature vector and its gap deviation constitute a sample, and then a sample data set is obtained.

[0010] Furthermore, there are multiple deep learning models, each of which has the same structure and different initial weights, and each of the multiple deep learning models is trained using the sample data set; When each gap deviation prediction model is applied to perform real-time calculation, the final real-time gap deviation is equal to the weighted average of the real-time gap deviations output by each gap deviation prediction model.

[0011] Furthermore, the deep learning model is trained using the sample data set, specifically including: Inputting the feature vectors in the sample data set into the deep learning model; The deep learning model extracts and predicts the feature vector to obtain a predicted gap deviation; The improved LASSO regression loss function is used to calculate the loss error between the gap deviation corresponding to the feature vector and the predicted gap deviation, and the parameters of the deep learning model are adjusted according to the loss error to realize the training of the deep learning model. Among them, the improved LASSO regression loss function introduces L2 regularization on the basis of the original LASSO regression loss function.

[0012] Based on the same concept, the present invention also provides a method for calculating the tightness of flange connection, comprising: Calling a basic gap physical model and a gap deviation prediction model; wherein the basic gap physical model and the gap deviation prediction model are constructed using the flange connection tightness calculation modeling method as described above; Get the real-time temperature and load of the flange; Inputting the real-time temperature and the real-time load into a foundation gap physical model to obtain a real-time foundation gap; Preprocessing the real-time temperature and the real-time load to obtain a real-time feature vector; Inputting the real-time feature vector into a gap deviation prediction model to obtain a real-time gap deviation; The real-time gap of the flange joint surface is calculated according to the real-time basic gap and the real-time gap deviation.

[0013] Furthermore, the calculation method also includes providing an early warning based on the calculated real-time clearance of the flange joint surface, specifically including: When the real-time gap between the flange joint surfaces is less than the first gap threshold, the tightness of the flange connection is normal; When the first gap threshold ≤ the real-time gap of the flange joint surface < the second gap threshold, a flange connection tightness attention prompt is issued; When the second gap threshold ≤ the real-time gap of the flange joint surface < the third gap threshold, a flange connection tightness warning prompt is issued; When the real-time gap between the flange joint surfaces is ≥ the third gap threshold, a flange connection tightness danger warning is issued.

[0014] Based on the same concept, the present invention also provides a flange connection tightness calculation system, including a data access layer, a calculation layer and a result output layer; The data access layer is used to obtain the real-time temperature and real-time load of the flange, and transmit the real-time temperature and real-time load to the calculation layer; The computing layer is deployed with a basic gap physical model and a gap deviation prediction model, and is used to input the real-time temperature and real-time load into the basic gap physical model to obtain a real-time basic gap; pre-process the real-time temperature and real-time load to obtain a real-time feature vector; input the real-time feature vector into the gap deviation prediction model to obtain a real-time gap deviation; calculate the real-time gap of the flange joint surface according to the real-time basic gap and the real-time gap deviation; wherein the basic gap physical model and the gap deviation prediction model are constructed using the flange connection tightness calculation modeling method as described above; The result output layer is used to display the real-time temperature and real-time load of the flange and the calculated real-time gap of the flange joint surface.

[0015] Based on the same concept, the present invention also provides an electronic device, including a memory, a processor, and a computer program / instructions stored in the memory, wherein the processor executes the computer program / instructions to implement the flange connection tightness calculation modeling method or the flange connection tightness calculation method as described above.

[0016] Based on the same concept, the present invention also provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the flange connection tightness calculation modeling method or flange connection tightness calculation method as described above.

[0017] Compared with the prior art, the advantages of the present invention are: The present invention constructs a basic clearance physical model according to the historical clearance of the flange joint surface under different temperatures and loads, takes into account the influence of temperature and load changes on the flange joint surface clearance, and improves the calculation accuracy of the flange connection tightness through the coupling analysis of temperature field and load field; at the same time, a gap deviation prediction model is constructed based on the deep learning model, and the gap deviation prediction model is used to predict the deviation between the basic clearance and the actual clearance, and the predicted gap deviation is used to correct the basic clearance, thereby further improving the calculation accuracy of the flange connection tightness.

[0018] The basic gap physical model and gap deviation prediction model constructed by the present invention are deployed in actual monitoring conditions to realize real-time and continuous calculation of flange connection tightness, ensuring the safety, reliability and service life of the equipment. It is suitable for flange connection status monitoring in wind power, aerospace, petrochemical and other fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only one embodiment of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 is a flow chart of a flange connection tightness calculation modeling method in an embodiment of the present invention; Figure 2 is a schematic diagram of a flange connection model in an embodiment of the present invention; Figure 3 It is a flow chart of a method for calculating the tightness of flange connection in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following is a clear and complete description of the technical solutions in the present invention in conjunction with the drawings in 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. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0023] Embodiment 1 Figure 1 The flow chart of the flange connection tightness calculation modeling method provided by the present invention is shown. Figure 1 As shown, the flange connection tightness calculation modeling method includes the following steps: Step A1: Obtain the historical clearance of flange joint surfaces under different temperatures and loads.

[0024] This embodiment takes the tower flange connection of MW-class wind turbines in wind farms as the research object, and collects 500 sets of historical gaps of flange joint surfaces under different temperatures and loads, where the gap of flange joint surface refers to the gap between the upper flange and the lower flange. The quartile method is used to detect anomalies of the 500 sets of historical gaps of flange joint surfaces under different temperatures and loads, and the outliers are removed; the historical gaps of flange joint surfaces after anomaly detection are smoothed using a sliding window to reduce the data noise level and improve the signal-to-noise ratio by about 35%. In this embodiment, the length of the sliding window is 5.

[0025] Step A2: Construct a basic gap physical model based on the historical gap of the flange joint surface under different temperatures and loads.

[0026] In a specific embodiment of the present invention, a basic gap physical model is constructed according to the historical gap of the flange joint surface under different temperatures and loads, specifically including: Step A2.1: Construct the flange finite element model.

[0027] The flange connection of the tower of a MW-class wind turbine in a wind farm is taken as the research object. The flange structural parameters are: outer diameter 2600mm, inner diameter 2400mm, thickness 80mm, number of bolts 128, and bolt specification M36. A flange three-dimensional model is constructed according to the flange structural parameters, and the flange three-dimensional model is meshed to obtain a flange finite element model. In this embodiment, the mesh unit type is selected as C3D8R, and the number of meshes is about 150,000.

[0028] The flange connection model includes an upper flange, a lower flange, and bolts connecting the upper flange and the lower flange. Figure 2 As shown in the figure, the flange finite element model is constructed based on the flange connection model.

[0029] Step A2.2: Based on the flange finite element model, simulation calculations are performed at different temperatures to obtain the simulated clearance of the flange joint surface at different temperatures; the simulated clearance of the flange joint surface at different temperatures is fitted to obtain a temperature-joint surface clearance fitting function.

[0030] According to the climatic conditions of the wind farm, the temperature range is set to -40℃ to 80℃; the material parameters adopt the nonlinear characteristics of Q345 steel, such as the thermal conductivity is 52W / (m·K) at 20℃, 45W / (m·K) at -40℃, and the specific heat capacity is 465J / (kg·K) at 20℃, which drops to 420J / (kg·K) at -40℃. Four temperature gradient conditions are set: 5℃ / h, 10℃ / h, 15℃ / h, and 20℃ / h to simulate different cooling rates.

[0031] Finite element simulation is performed on the flange finite element model under multiple temperature conditions to obtain the temperature field distribution cloud map. The simulated gap of the flange joint surface under different temperatures is extracted from the temperature field distribution cloud map, and then the temperature-joint surface gap fitting function is constructed, and its specific expression is: (1) in, It indicates the flange joint surface clearance caused by temperature, in mm; T indicates temperature, in °C; , , All represent fitting coefficients. In this embodiment, the fitting coefficients , , They are 0.00072, -0.0015 and 0.025 respectively.

[0032] The simulation results show that when the temperature drops from 20°C to -20°C, the maximum change in the simulated gap of the flange joint surface is 0.028mm; when the temperature gradient reaches 20°C / h, the local temperature difference of the flange can reach 15°C, resulting in uneven thermal stress, causing the simulated gap of the flange joint surface to increase to 0.035mm.

[0033] Step A2.3: Based on the flange finite element model, simulation calculations of different loads are performed to obtain the simulated clearance of the flange joint surface under different loads; the simulated clearance of the flange joint surface under different loads is fitted to obtain a load-joint surface clearance fitting function.

[0034] The load in this embodiment includes transverse load and axial load. Simulation calculation under transverse load: The range of the lateral load is 0-100kN, and 10 lateral load conditions are set at intervals of 10kN. Considering the elastic-plastic deformation characteristics of the flange material and the contact surface friction coefficient, the contact surface friction coefficient is set to 0.15. The deformation and gap distribution of the flange joint surface under each lateral load condition are calculated by the finite element method, and the simulated gap of the flange joint surface under different lateral loads is obtained, and then the lateral load-joint surface gap fitting function is constructed. The specific expression is: (2) in, Indicates the flange joint surface clearance caused by the lateral load, in mm; Indicates the lateral load, in kN; , , All represent fitting coefficients. In this embodiment, the fitting coefficients , , They are 0.0000023, 0.00085 and 0.002 respectively.

[0035] The simulation results show that when the lateral load reaches 50kN, the flange joint surface begins to separate locally; when the load reaches 90kN, the maximum simulation gap of the flange joint surface reaches 0.12mm.

[0036] Simulation calculation under axial load: The range of axial load is 0-200kN, and 10 axial load working points are set at intervals of 20kN; the stress conditions of the flange under different preloads (set at 60%, 80%, 100%, and 120% of the standard preload, that is, simulating different bolt connection states) are simulated; the deformation of the flange joint surface and the change of the leakage path under each axial load condition are calculated by the finite element method, and the simulated clearance of the flange joint surface under different axial loads is obtained, and then the axial load-joint surface clearance fitting function is constructed, and its specific expression is: (3) in, Indicates the flange joint surface clearance caused by axial load, unit: mm; Indicates axial load, unit kN; , , All represent fitting coefficients. In this embodiment, the fitting coefficients , , They are 0.000015, 0.00063 and 0.001 respectively.

[0037] The simulation results show that under standard preload, an obvious leakage path begins to appear when the axial load reaches 120kN; when the preload is reduced by 20%, this critical value drops to 90kN.

[0038] Step A2.4: Calculate the flange joint surface gap caused by temperature and the flange joint surface gap caused by load based on the historical gaps of the flange joint surface under different temperatures and loads, as well as the temperature-joint surface gap fitting function and the load-joint surface gap fitting function.

[0039] Substitute the temperature and load in the historical gap of the flange joint surface under different temperatures and loads obtained in step A1 into the temperature-joint surface gap fitting function and the load-joint surface gap fitting function (including the transverse load-joint surface gap fitting function and the axial load-joint surface gap fitting function), respectively, to obtain the flange joint surface gap caused by temperature and the flange joint surface gap caused by load. In this way, multiple groups of temperatures and the flange joint surface gaps caused by them, loads and the flange joint surface gaps caused by them can be obtained, which can be expressed as .

[0040] Step A2.5: Obtain a basic clearance physical model based on the flange joint surface clearance caused by temperature, the flange joint surface clearance caused by load, and the flange joint surface historical clearance.

[0041] The least square method is used to fit multiple groups of temperatures and the flange joint surface gaps caused by them, loads and the flange joint surface gaps caused by them, and a basic gap physical model is obtained. In this embodiment, the specific expression of the basic gap physical model is: (4) in, Indicates the foundation gap; , , Both represent weight coefficients, that is, the contribution weights of temperature, lateral load and axial load to the flange joint clearance. In this embodiment, the weight coefficient , , They are 0.35, 0.42, and 0.23 respectively. The root mean square error of the basic gap physical model tested separately is 0.0125mm.

[0042] Step A3: Calculate the gap deviation under different temperatures and loads based on the basic gap physical model and the historical gaps of the flange joint surfaces under different temperatures and loads.

[0043] The temperature and load in the historical gap of the flange joint surface under different temperatures and loads obtained in step A1 are respectively substituted into the temperature-joint surface gap fitting function and the load-joint surface gap fitting function to obtain the flange joint surface gap caused by temperature and the flange joint surface gap caused by load; then the flange joint surface gap caused by temperature and the flange joint surface gap caused by load are substituted into the basic gap physical model to obtain the basic gap; the difference between the historical gap of the flange joint surface and the basic gap is the gap deviation under this set of temperatures and loads.

[0044] Step A4: Construct a sample data set based on the gap deviation under different temperatures and loads.

[0045] In a specific embodiment of the present invention, a sample data set is constructed according to the gap deviation under different temperatures and loads, specifically including: Step A4.1: Construct a feature vector for each set of temperature and load.

[0046] This embodiment not only considers the influence of temperature and load, but also the interactive influence of temperature and load. Therefore, the eigenvector includes temperature, load and the interactive value of temperature and load, which can be expressed as the eigenvector .

[0047] Step A4.2: To eliminate the dimension effect, the feature vector X is normalized.

[0048] Step A4.3: The standardized feature vector and its gap deviation constitute a sample, thereby obtaining a sample data set.

[0049] The principal component analysis was used to evaluate the feature contribution of the feature vector, and the results showed that the cumulative contribution of the first three principal components reached 97.8%. Although the dimension can be reduced to 3 principal components, the original physical characteristics are maintained to ensure the interpretability of the model. The 6-dimensional feature vector after normalization is retained as the input of the deep learning model, and the corresponding gap deviation is used as the expected label of the deep learning model.

[0050] The number of samples in the sample data set constructed based on the historical gaps of flange joint surfaces under different temperatures and loads obtained in step A1 is 500. The sample data set containing 500 samples is divided into a training set (350), a validation set (100), and a test set (50) in a ratio of 7:2:1. In order to enhance the robustness of the deep learning model, random perturbations (i.e., gap deviations) that conform to engineering practice are added to each basic gap in the training set, thereby expanding the training set to 700 samples.

[0051] Step A5: Deploy a deep learning model and train the deep learning model using a sample data set to obtain a gap deviation prediction model.

[0052] In this embodiment, the deep learning model uses a neural network model, which includes an input layer, a hidden layer, a Dropout layer, and an output layer connected in sequence. The number of nodes in the input layer is set according to the dimension of the input feature vector. In this embodiment, the input layer contains 6 nodes; the hidden layer adopts a 2-layer structure, each layer contains 8 nodes, and uses a ReLU activation function; the output layer contains 1 node for outputting the gap deviation, that is, the correction amount for the basic gap. A Dropout layer is added between the hidden layer and the output layer to increase the generalization ability.

[0053] In a specific embodiment of the present invention, the neural network model is trained using a training set, specifically including: Step A5.1: Input the feature vectors in the training set into the neural network model; Step A5.2: The neural network model extracts and predicts the feature vector to obtain the prediction gap deviation; Step A5.3: Use the improved LASSO regression loss function to calculate the loss error between the gap deviation corresponding to the feature vector and the predicted gap deviation, adjust the parameters of the neural network model according to the loss error, and implement the training of the neural network model.

[0054] In this embodiment, the neural network model is trained using the Adam optimizer (learning rate 0.001, batch size 64). The early stopping strategy is used, and the training is stopped when 15 consecutive rounds of training cannot improve the performance of the neural network model. The improved LASSO regression loss function is used for loss calculation. The improved LASSO regression loss function introduces L2 regularization on the basis of the original LASSO regression loss function, and L2 regularization is used to better capture the interaction between influencing factors.

[0055] In a specific embodiment of the present invention, there are multiple deep learning models (for example, 5), the structures of the multiple deep learning models are the same and the initial weights are different, and the multiple deep learning models are trained, verified and tested respectively using sample data sets to obtain multiple gap deviation prediction models. In this embodiment, a random seed is used to initialize the weights of each deep learning model.

[0056] When there are multiple gap deviation prediction models, the final predicted gap deviation is equal to the weighted average of the output results of the multiple gap deviation prediction models. The weight of the output result of each gap deviation prediction model in the weighted average calculation is determined according to the performance of the gap deviation prediction model on the validation set. The better the performance on the validation set, the greater the weight, and the sum of all weights is equal to 1.

[0057] The basic gap physical model (i.e., formula (1) to formula (4)) and the gap deviation prediction model are packaged in a programmatic manner to form a software tool for real-time flange connection tightness calculation and solution.

[0058] Embodiment 2 Figure 3 The flowchart of the flange connection tightness calculation method provided by the present invention is shown. Figure 3 As shown, the flange connection tightness calculation method includes the following steps: Step B1: Call the basic gap physical model and the gap deviation prediction model.

[0059] Among them, the basic gap physical model (i.e., formula (1) to formula (4)) and the gap deviation prediction model are constructed using the flange connection tightness calculation modeling method of an embodiment of the present application.

[0060] Step B2: Obtain the real-time temperature and real-time load of the flange.

[0061] Taking the tower flange of a wind turbine as an example, sensors are installed at the flange connection to collect the real-time temperature and load of the flange. The sensors include temperature sensors, lateral load sensors, and axial load sensors. Specifically, 8 PT100 platinum resistance temperature sensors (accuracy ±0.1°C) are evenly installed along the circumference of the flange to monitor the temperature field distribution of the flange; 4 strain gauge sensors (range 0-120kN, accuracy ±0.5%FS) are evenly installed along the circumference of the flange to monitor the lateral load; 4 piezoelectric sensors (range 0-250kN, accuracy ±0.3%FS) are evenly installed along the circumference of the flange to monitor the axial load. The sampling frequency of all sensors is not less than 200Hz to meet the real-time monitoring requirements.

[0062] The real-time temperature is acquired through 8 PT100 platinum resistance temperature sensors, and the final real-time temperature is equal to the average value of each real-time temperature; the real-time lateral load is acquired through 4 strain gauge sensors, and the final real-time lateral load is equal to the average value of each real-time lateral load; the real-time axial load is acquired through 4 piezoelectric sensors, and the final real-time axial load is equal to the average value of each real-time axial load.

[0063] Step B3: Input the real-time temperature and real-time load into the foundation gap physical model to obtain the real-time foundation gap.

[0064] Substitute the final real-time temperature into formula (1) to obtain the flange joint surface clearance caused by temperature; substitute the final real-time lateral load into formula (2) to obtain the flange joint surface clearance caused by lateral load; substitute the final real-time axial load into formula (3) to obtain the flange joint surface clearance caused by axial load; finally, substitute the flange joint surface clearance caused by temperature, the flange joint surface clearance caused by lateral load and the flange joint surface clearance caused by axial load into formula (4) to obtain the real-time basic clearance.

[0065] Step B4: Preprocess the real-time temperature and real-time load to obtain a real-time feature vector.

[0066] According to the processing method of step A4.1 and step A4.2 in the first embodiment of the present application, the real-time temperature and the real-time load are preprocessed to obtain a real-time feature vector.

[0067] Step B5: Input the real-time feature vector into the gap deviation prediction model to obtain the real-time gap deviation.

[0068] When there are multiple gap deviation prediction models, the real-time feature vector is input into each gap deviation prediction model, and then the output results of each gap deviation prediction model are weighted averaged to obtain the final real-time gap deviation.

[0069] Step B6: Calculate the real-time gap of the flange joint surface according to the real-time basic gap and the real-time gap deviation to realize the calculation of the tightness of the flange connection.

[0070] The sum of the real-time basic clearance and the final real-time clearance deviation is the real-time clearance of the flange joint surface.

[0071] Step B7: Issue an early warning based on the calculated real-time clearance of the flange joint surface.

[0072] In a specific embodiment of the present invention, the early warning based on the calculated real-time gap between flange joint surfaces specifically includes: When the real-time gap between the flange joint surfaces is less than the first gap threshold, the tightness of the flange connection is normal; When the first gap threshold ≤ the real-time gap of the flange joint surface < the second gap threshold, attention should be paid to the tightness of the flange connection, and a warning prompt should be issued; When the second gap threshold ≤ the real-time gap of the flange joint surface < the third gap threshold, a flange connection tightness warning prompt is issued; When the real-time gap between the flange joint surfaces is ≥ the third gap threshold, a flange connection tightness danger warning is issued.

[0073] In this embodiment, the first gap threshold is 0.05mm, the second gap threshold is 0.08mm, and the third gap threshold is 0.10mm. The present invention establishes a multi-level warning mechanism to achieve monitoring and early warning. When the real-time gap of the flange joint surface is greater than the third gap threshold, it is considered to be insufficient tightness; the real-time gap of the flange joint surface is calculated every 0.1s to achieve continuous monitoring and calculation of the tightness of the flange connection.

[0074] The calculation method of the present invention was applied to 10 wind turbines in a wind farm and operated for 12 months, achieving remarkable results: 7 abnormal flange connection conditions were successfully warned, avoiding unexpected equipment shutdowns; the calculation accuracy reached 91.5%, which is about 55% higher than the traditional method; the equipment availability increased by 2.3%, and the annual maintenance cost decreased by 18.7%.

[0075] Compared with traditional methods, the monitoring accuracy of the present invention is improved by more than 50%, and the response time is shortened to milliseconds; the calculation result of the joint surface clearance is more than 90% consistent with the actual working conditions. The calculation method of the present invention has wide applicability and significant application value, and can be appropriately adjusted according to the characteristics of different industries and application scenarios, effectively improving the safety, reliability and service life of flange connections of various types of equipment.

[0076] Embodiment 3 The flange connection tightness calculation system provided by the present invention comprises a data access layer, a calculation layer and a result output layer.

[0077] The data access layer is used to obtain the real-time temperature and real-time load of the flange, and transmit the real-time temperature and real-time load to the calculation layer.

[0078] The computing layer is deployed with a basic gap physical model and a gap deviation prediction model, and is used to input the real-time temperature and real-time load into the basic gap physical model to obtain the real-time basic gap; pre-process the real-time temperature and real-time load to obtain a real-time feature vector; input the real-time feature vector into the gap deviation prediction model to obtain the real-time gap deviation; calculate the real-time gap of the flange joint surface according to the real-time basic gap and the real-time gap deviation. Among them, the basic gap physical model and the gap deviation prediction model are constructed using the flange connection tightness calculation modeling method in Example 1 of the present application.

[0079] The result output layer is used to display the real-time temperature and real-time load of the flange and the calculated real-time gap of the flange joint surface.

[0080] The flange connection tightness calculation system of the present invention is built on a cloud architecture and adopts a three-layer architecture design, namely, a data access layer, a calculation layer, and a result output layer. The data access layer is used for reading and preprocessing sensor data, the calculation layer is used for calculating the flange joint surface gap based on sensor data, and the result output layer is used for visualizing the calculation results. The flange connection tightness calculation system uses C++ to develop the core algorithm and Python to develop the interface layer, supporting multi-threaded parallel computing.

[0081] At the data access layer, an edge computing unit is deployed. The edge computing unit is located at the bottom of the tower and performs preliminary filtering on the sensor data, removes obvious outliers, and then encrypts and transmits the data to the computing layer through the 4G network. The average data transmission delay is 32ms, which meets the real-time requirements; at the same time, the sensor data is stored in the MySQL database. In this embodiment, the Kalman filter is used for preliminary filtering to effectively reduce signal noise; outliers exceeding 3 times the standard deviation are automatically identified and removed to ensure data reliability.

[0082] At the computing layer, the basic gap physical model (i.e., formula (1) to formula (4)) and the gap deviation prediction model are deployed. The Intel MKL matrix operation library is used to optimize the calculation process, and the speed of large matrix operations is increased by about 2.8 times. Tests show that the single calculation delay is controlled within 125ms, which meets the real-time monitoring requirements. In order to ensure the reliability of calculation, a redundancy mechanism is implemented at the computing layer. When the main computing unit fails, the backup computing unit automatically takes over the calculation task. A complete exception handling process is designed to automatically identify abnormal input data and take compensation measures. The monitoring platform performs self-calibration every 24 hours to ensure long-term stable operation.

[0083] In the result output layer, a web interface is designed to display the temperature, load and calculated flange joint surface gap in real time, supporting PC and mobile terminal access.

[0084] The calculation system also has a rationality check mechanism. If the change rate of the calculation result exceeds three times the historical data, a secondary verification process is triggered. Verification tests show that the mechanism successfully identified 98.5% of calculation anomalies, effectively improving system reliability.

[0085] For software architecture, the system of the present invention adopts modular design, supports multi-threaded parallel computing, and improves processing efficiency; implements model hot update mechanism, supports online optimization and adjustment; and standardizes the design interface to facilitate integration with other systems. For algorithm optimization and acceleration, the present invention applies matrix operation library to optimize the calculation process; introduces GPU acceleration characteristics to improve the calculation speed of neural network model; implements incremental calculation scheme, only processes changing data; and controls the calculation delay within 200ms to meet real-time requirements. For reliability, the present invention adopts redundancy mechanism to prevent single point failure; designs abnormal handling process to ensure stable operation of the system; establishes regular self-check and self-calibration functions; supports offline calculation mode to deal with network interruption.

[0086] Embodiment 4 An embodiment of the present invention also provides an electronic device, which includes: a memory, a processor, and a computer program / instructions stored in the memory, and the processor executes the computer program / instructions to implement the flange connection tightness calculation modeling method or the flange connection tightness calculation method in the embodiment of the present application.

[0087] Although not shown, the electronic device includes a processor that can perform various appropriate operations and processes according to the programs and / or data stored in the read-only memory (ROM) or the programs and / or data loaded from the storage part into the random access memory (RAM). The processor can be a multi-core processor or can include multiple processors. In some embodiments, the processor can include a general-purpose main processor and one or more special coprocessors, such as a central processing unit, a graphics processing unit (GPU), a neural network processor (NPU), a digital signal processor (DSP), etc. In RAM, various programs and data required for device operation are also stored. The processor, ROM, and RAM are connected to each other via a bus. The input / output (I / O) interface is also connected to the bus.

[0088] The processor and memory are used together to execute the program / instructions stored in the memory. When the program / instructions are executed by the computer, the methods, steps or functions described in the above embodiments can be implemented.

[0089] Although not shown, an embodiment of the present invention further provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the flange connection tightness calculation modeling method or flange connection tightness calculation method in the embodiment of the present application.

[0090] Readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0091] Although not shown, an embodiment of the present invention further provides a computer program product, including: a computer program / instruction, which, when executed by a processor, implements the flange connection tightness calculation modeling method or flange connection tightness calculation method in the embodiment of the present application.

[0092] What is disclosed above is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or modifications within the technical scope disclosed in the present invention, which should be covered within the protection scope of the present invention.

Claims

1. A flange connection tightness calculation modeling method, characterized in that: The modeling method comprises: Obtain the historical clearance of flange joint surfaces under different temperatures and loads; The basic gap physical model is constructed based on the historical gap of the flange joint surface under different temperatures and loads; Calculate the gap deviation under different temperatures and loads according to the basic gap physical model and the historical gaps of the flange joint surfaces under different temperatures and loads; Construct a sample data set based on gap deviations at different temperatures and loads; Deploy a deep learning model, and use the sample data set to train the deep learning model to obtain a gap deviation prediction model.

2. The flange connection tightness calculation modeling method according to claim 1 is characterized in that: The basic gap physical model is constructed according to the historical gap of the flange joint surface under different temperatures and loads, including: Constructing a flange finite element model; wherein the flange finite element model includes an upper flange, a lower flange, and bolts connecting the upper flange and the lower flange, and the gap between the upper flange and the lower flange is a flange joint surface gap; Based on the flange finite element model, simulation calculations are performed at different temperatures to obtain the simulated clearances of the flange joint surfaces at different temperatures; the simulated clearances of the flange joint surfaces at different temperatures are fitted to obtain a temperature-joint surface clearance fitting function; Based on the flange finite element model, simulation calculations of different loads are performed to obtain the simulated clearance of the flange joint surface under different loads; the simulated clearance of the flange joint surface under different loads is fitted to obtain a load-joint surface clearance fitting function; The flange joint surface gap caused by temperature and the flange joint surface gap caused by load are calculated according to the historical gap of the flange joint surface under different temperatures and loads, as well as the temperature-joint surface gap fitting function and the load-joint surface gap fitting function; According to the flange joint surface gap caused by temperature, the flange joint surface gap caused by load and the flange joint surface historical gap, the basic gap physical model is obtained.

3. The flange connection tightness calculation modeling method according to claim 1 is characterized in that: A sample data set was constructed based on gap deviations at different temperatures and loads, including: Constructing a characteristic vector according to each set of temperature and load; wherein the characteristic vector includes temperature, load and interaction value between temperature and load; Performing standardization on the feature vector; The standardized feature vector and its gap deviation constitute a sample, and then a sample data set is obtained.

4. The flange connection tightness calculation modeling method according to claim 1 is characterized in that: There are multiple deep learning models, each of which has the same structure and different initial weights, and each of the multiple deep learning models is trained using the sample data set; When each gap deviation prediction model is applied to perform real-time calculation, the final real-time gap deviation is equal to the weighted average of the real-time gap deviations output by each gap deviation prediction model.

5. The flange connection tightness calculation modeling method according to any one of claims 1 to 4, characterized in that: Using the sample data set to train the deep learning model specifically includes: Inputting the feature vectors in the sample data set into the deep learning model; The deep learning model extracts and predicts the feature vector to obtain a predicted gap deviation; The improved LASSO regression loss function is used to calculate the loss error between the gap deviation corresponding to the feature vector and the predicted gap deviation, and the parameters of the deep learning model are adjusted according to the loss error to realize the training of the deep learning model. Among them, the improved LASSO regression loss function introduces L2 regularization on the basis of the original LASSO regression loss function.

6. A method for calculating the tightness of flange connection, characterized in that: The calculation method includes: Calling a basic gap physical model and a gap deviation prediction model; wherein the basic gap physical model and the gap deviation prediction model are constructed using the flange connection tightness calculation modeling method according to any one of claims 1 to 5; Get the real-time temperature and load of the flange; Inputting the real-time temperature and the real-time load into a foundation gap physical model to obtain a real-time foundation gap; Preprocessing the real-time temperature and the real-time load to obtain a real-time feature vector; Inputting the real-time feature vector into a gap deviation prediction model to obtain a real-time gap deviation; The real-time gap of the flange joint surface is calculated according to the real-time basic gap and the real-time gap deviation.

7. The flange connection tightness calculation method according to claim 6, characterized in that: The calculation method also includes providing an early warning based on the calculated real-time clearance of the flange joint surface, specifically including: When the real-time gap between the flange joint surfaces is less than the first gap threshold, the tightness of the flange connection is normal; When the first gap threshold ≤ the real-time gap of the flange joint surface < the second gap threshold, a flange connection tightness attention prompt is issued; When the second gap threshold ≤ the real-time gap of the flange joint surface < the third gap threshold, a flange connection tightness warning prompt is issued; When the real-time gap between the flange joint surfaces is ≥ the third gap threshold, a flange connection tightness danger warning is issued.

8. A flange connection tightness calculation system, characterized in that: The computing system includes a data access layer, a computing layer and a result output layer; The data access layer is used to obtain the real-time temperature and real-time load of the flange, and transmit the real-time temperature and real-time load to the calculation layer; The computing layer is deployed with a basic gap physical model and a gap deviation prediction model, and is used to input the real-time temperature and real-time load into the basic gap physical model to obtain a real-time basic gap; pre-process the real-time temperature and real-time load to obtain a real-time feature vector; input the real-time feature vector into the gap deviation prediction model to obtain a real-time gap deviation; calculate the real-time gap of the flange joint surface according to the real-time basic gap and the real-time gap deviation; wherein the basic gap physical model and the gap deviation prediction model are constructed using the flange connection tightness calculation modeling method described in any one of claims 1 to 5; The result output layer is used to display the real-time temperature and real-time load of the flange and the calculated real-time gap of the flange joint surface.

9. An electronic device comprising a memory, a processor, and a computer program / instruction stored in the memory, characterized in that: The processor executes the computer program / instructions to implement the flange connection tightness calculation modeling method as described in any one of claims 1 to 5, or the flange connection tightness calculation method as described in claim 6 or 7.

10. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the flange connection tightness calculation modeling method as described in any one of claims 1 to 5, or the flange connection tightness calculation method as described in claim 6 or 7 is implemented.

Images