Thermocouple position layout optimization method based on feature point extraction

Abstract

The invention relates to a thermocouple position layout optimization method based on feature point extraction, and the method comprises the following steps: heating a patch and an adhesive through a hot patching instrument, obtaining the temperature field data of an adhesive interface based on a finite element simulation method, and extracting the temperature value and coordinates of each node; analyzing the time sequence change of the temperature field data, and screening the key frame with the maximum node temperature standard deviation; performing spatial interpolation on the node temperature data of the key frame to generate a two-dimensional temperature distribution matrix; feature points are extracted from the temperature distribution matrix through a combined algorithm of extreme value layering, OPTICS density clustering, gradient analysis and adaptive spatial distribution; and optimizing the thermocouple installation position according to the feature point distribution. Through the method, the heating conditions of the bonding interface and the patch are monitored and fed back by using the accurate number of the thermocouples and setting the accurate positions of the thermocouples, and meanwhile, the problem that uneven heating cannot be monitored due to the fact that the positions of the thermocouples are improperly set is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of aircraft composite material repair, and in particular to an optimization method for the layout of thermocouple positions based on feature point extraction. Background Art

[0002] With the wide application of composite materials in aircraft maintenance, the adhesive repair technology has developed rapidly. By arranging thermocouples in the repair area, the heating time and temperature can be monitored and controlled, so as to ensure the repair effect. The bonding process of the patch requires strict control of the heating temperature process parameters, and the heat patch instrument can provide a suitable thermal environment for the bonding between the patch and the skin. During the heating process, the uniformity and stability of the temperature at the bonding interface have a crucial impact on the bonding repair quality. Therefore, accurately measuring and controlling the temperature in the heating area is a key link in the bonding repair process.

[0003] During the aircraft bonding repair process, thermocouples are arranged in the heating area to monitor the temperature change in real time. When the temperature in the heating area is uneven or lower than the specified temperature, the thermocouple can feedback the measured temperature data to the control system of the heat patch instrument. Therefore, the layout position of the thermocouple is directly related to the accuracy and reliability of temperature measurement, and further affects the repair quality of the patch.

[0004] In the past aircraft bonding repair process, the layout position of the thermocouple was often determined according to experience or simple geometric layout. However, due to the complexity of the aircraft structure, the diversity of the patch shape, and the size difference of the heating blanket in the heat patch instrument, this simple geometric layout has limitations, and the thermocouple may not accurately reflect the temperature distribution in the heating area.

[0005] To improve the reliability of the heat repair process and the repair quality of the patch, it is necessary to optimize the layout position of the thermocouple. The current placement of the thermocouple position is only a simple geometric distribution. The purpose of this optimization is to enable the thermocouple to comprehensively and accurately monitor the temperature change in the heating area, while reducing the number of thermocouples to reduce the measurement cost and system complexity.

[0006] PYTHON is a powerful programming language. The temperature field distribution can be solved through numerical calculation, and the layout position of the thermocouple can be optimized by using a combination algorithm. ABAQUS is a powerful finite element analysis software that can simulate the heat conduction process during the bonding repair process and obtain the real-time temperature distribution data of the bonding interface. By the collaborative work of PYTHON and ABAQUS, making full use of the advantages of both, the efficient optimization of the layout position of the thermocouple can be achieved. Summary of the Invention

[0007] The technical problem to be solved by the present invention is: in order to solve the problem that the thermocouple layout in the above-mentioned background technology relies on experience or simple geometric distribution, cannot accurately reflect the temperature field changes in complex aircraft repair areas, and leads to low measurement reliability and high cost, a method for optimizing the thermocouple position layout based on feature point extraction is provided.

[0008] The technical solution adopted by the present invention to solve the technical problem is: a method for optimizing the position layout of thermocouples based on feature point extraction, characterized by comprising the following steps: S1. Heat the patch and adhesive using a thermal patching instrument, obtain the bonding interface temperature field data based on the finite element simulation method, and extract the temperature value and coordinates of each node; S2. Analyze the temporal changes of the temperature field data and select the key frame with the largest node temperature standard deviation; S3, performing spatial interpolation on the node temperature data of the key frame to generate a two-dimensional temperature distribution matrix; S4, extracting characteristic points from the temperature distribution matrix through a combination of extreme value stratification, OPTICS density clustering, gradient analysis and adaptive spatial distribution; S5. Optimize the thermocouple installation position according to the distribution of characteristic points.

[0009] Through steps S1 to S5, fully automatic optimization from temperature field modeling to thermocouple layout is achieved, replacing the traditional manual experience layout method; the comprehensive timing analysis in step S2, the spatial interpolation in step S3, and the multi-algorithm fusion in step S4 improve the completeness of temperature field feature recognition by 40% compared with a single method; it is applicable to patches of arbitrary shapes and breaks through the limitations of traditional geometric layout.

[0010] According to one embodiment of the present invention, step S2 includes: Calculate the standard deviation of the node temperature in each frame and select the key frame with the largest deviation.

[0011] The standard deviation threshold is used to automatically capture moments of drastic temperature fluctuations, and key frames are intelligently filtered from massive simulation data to reduce the amount of calculation, ensure that thermocouples cover the worst working conditions, and reduce temperature monitoring blind spots.

[0012] According to one embodiment of the present invention, step S3 includes: The temperature data of the bonding surface nodes are projected onto a two-dimensional plane, and a continuous temperature field is generated through spatial interpolation.

[0013] Spatial interpolation realizes a continuous temperature field, and the local temperature reconstruction error is small. The generated two-dimensional matrix can directly output the industrial standard thermal map, providing smooth input for subsequent gradient analysis and reducing noise interference.

[0014] According to one embodiment of the present invention, step S4 includes: S41. Divide intervals based on the quantiles of temperature values and extract the extreme points of each interval; S42. Identify the cluster center points of the temperature anomaly regions through the OPTICS density clustering algorithm; S43. Calculate the temperature gradient field and screen the top k mutation points in terms of gradient magnitude ranking; S44. Through adaptive spatial distribution, merge and deduplicate the feature points to achieve a balanced spatial distribution of the feature points.

[0015] The extreme value stratification in step S41 ensures the coverage of high and low temperature regions and improves the detection rate of extreme temperatures; the OPTICS density clustering in step S42 is used to identify irregular hot spot regions to improve the positioning accuracy of the anomaly regions; the gradient analysis in step S43 captures the mutations at the material interfaces to improve the sensitivity of gradient changes; the adaptive spatial distribution in step S44 performs the merging and deduplication of multi-source feature points to achieve a balanced spatial distribution of the feature points.

[0016] According to an embodiment of the present invention, in the OPTICS density clustering algorithm in step S42, the temperature values and coordinate information are combined for clustering analysis, which specifically includes: Each node is represented as a feature vector [x, y, z], which includes its coordinates (x, y) and temperature value t; Normalize the feature vectors to ensure that features with different dimensions are on the same scale; In the three-dimensional feature space, the Euclidean distance metric method is used to calculate the similarity between nodes, and the calculation formula is: , where the node p = [x p , y p , t p , and the node q = [x q , y q , t q .

[0017] The three-dimensional feature vector [x, y, t] synchronously considers the spatial position and temperature value, and has a higher clustering accuracy than two-dimensional coordinate clustering. The normalization processing of temperature values solves the problem of dimension mismatch; it automatically identifies the number of clusters to adapt to patches of different sizes.

[0018] According to an embodiment of the present invention, in step S43, calculating the temperature gradient field specifically is: Spatial weighted gradient analysis, adding spatial density weights to the gradient points to reduce the importance of gradient points in sparse regions; Through data distribution, adjust the gradient density threshold to accurately extract the feature points and draw the places where temperature mutations occur in the temperature field.

[0019] Density weight suppresses edge noise and the effective feature point extraction rate; a gradient vector map can be generated to guide process adjustment.

[0020] According to an embodiment of the present invention, gradient points are selected in the high-temperature, medium-temperature, and low-temperature regions respectively.

[0021] According to an embodiment of the present invention, the step S44 further includes: Multi-source feature point fusion is performed through adaptive spatial distribution control to merge duplicate feature points; spatial de-duplication is performed on the extracted feature points, and points with a mutual distance greater than a set value are retained.

[0022] Advantages of the present invention: (1) Refined temperature field simulation. Through finite element analysis using ABAQUS, the temperature distribution in the heat repair area can be accurately calculated, thereby avoiding omissions or unreasonable layouts of thermocouples in these key parts and improving the accuracy of temperature monitoring; (2) Using a combined algorithm, multiple optimization objectives can be considered simultaneously, such as temperature measurement accuracy, the number of thermocouples, layout uniformity, etc.; during the optimization process, the importance of different objectives can be balanced by setting weight coefficients; (3) Through the combination of Python and ABAQUS, an automated process for optimizing the layout of thermocouples can be achieved; (4) This technology can adapt to the requirements of adhesive repair work with different structures and patch shapes through finite element analysis using ABAQUS; it can also be extended to other adhesive repair work that requires accurate temperature monitoring. By adjusting the finite element model and optimization algorithm, the layout optimization of thermocouples under different application scenarios can be achieved. Description of the Drawings

[0023] The present invention will be further described below in conjunction with the drawings and embodiments.

[0024] Figure 1 is a flowchart of a specific implementation manner of the present invention.

[0025] Figure 2 is a line graph of the standard deviation of the temperature values of all frame nodes in a specific implementation manner of the present invention.

[0026] Figure 3 is a flowchart of extracting the temperature values of the feature frames and generating a temperature heat map in a specific implementation manner of the present invention.

[0027] Figure 4 is the temperature heat map of the feature frames in a specific implementation manner of the present invention.

[0028] Figure 5 is the algorithm flowchart of extracting feature points in a specific implementation manner of the present invention.

[0029] Figure 6 It is the position coordinate diagram of feature points in a specific embodiment of the present invention. Specific embodiment

[0030] Now, the present invention will be further described in detail with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0031] Such as Figure 1 shown, the optimization method for the layout position of thermocouples based on feature point extraction includes the following steps: S1. Heat the patch and the adhesive with a heat compensator, and obtain the temperature field data of the bonding interface based on the finite element simulation method, and extract the temperature values and coordinates of each node; S2. Analyze the temporal variation of the temperature field data, and screen the key frames with the largest standard deviation of the node temperatures; S3. Perform spatial interpolation on the node temperature data of the key frames to generate a two-dimensional temperature distribution matrix; S4. Extract feature points from the temperature distribution matrix through a combined algorithm of extreme value stratification, OPTICS density clustering, gradient analysis, and adaptive spatial distribution; S5. Optimize the installation position of the thermocouple according to the distribution of the feature points.

[0032] For the extraction of the node temperature values of all frames during the heat compensation process: First, extract the node temperature values of the specified instance from the ODB file of the ABAQUS software. Specifically: use the ABAQUS software to draw the temperature contour diagram during the heat compensation process of the patch, and then save the file to the OBD file; use the openOdb function of PYTHON to open the ODB file at the specified path, and then obtain the node information through the instance name BUPIAN-1 and the temperature field variable NT11, and use the PYTHON code frame.fieldOutputs['NT11'].getSubset(position=NODAL) to extract the temperature values of all node positions in each frame.

[0033] Then, through the code temp_data={value.nodeLabel: value.data for value in temperature_field.values}, create a dictionary temp_data to store the temperature data with the node label as the key and the temperature value as the value.

[0034] Traverse all nodes again to obtain the labels and coordinates; use csv_filename = f'surface_temperatures_frame_{frame_index}.csv' to generate the CSV file name based on the current frame index, and save the node data to the CSV file.

[0035] Finally, first convert all node temperature values into a numerical array and use the nanstd function of numpy to calculate the temperature range of each frame. The specific implementation in the PYTHON language is as follows: temp_values=df.iloc[:, temp_column] max_temp=temp_values.max() min_temp=temp_values.min() temp_range=max_temp-min_temp frame_range_pairs.append( (frame_index, temp_range)).

[0036] Use matplotlib to plot the relationship between the number of frames and the temperature standard deviation. The result is as Figure 2 shown. Select the frame number with the largest standard deviation as the moment that can best reflect the most uneven heating temperature during the heat repair process. Select this frame number to obtain the characteristic point positions of the thermocouple, which can prominently reflect the temperature change problem in the heating area.

[0037] Regarding the plotting of the heat map of node temperature values: As Figure 3 shown, first, use the open0db() function to open the DAT and OBD files, obtain the node data of the specified instance, extract the node IDs and the corresponding temperature values; use csv.writer() to create a CSV file, write the header and each row of data, including the node label, coordinates, and temperature; extract the data points from the CSV file , where k = 1, 2,..., N.

[0038] Then, use np.unique() to obtain the unique X and Y values, that is, use np.meshgrid() to create a grid and generate an XY two-dimensional grid: , ; X,Y=np.meshgrid( , ); Use a triple loop to fill the temperature grid according to the values of X and Y, interpolate the discrete temperature data onto the grid, and generate a temperature distribution matrix ; For each grid point , find the corresponding temperature value ; If it exists satisfy and , then .

[0039] Finally, a scatter plot is drawn, using the NumPy and Matplotlib libraries to process and plot the data, realizing the visualization of the data, as Figure 4 shown.

[0040] For extracting feature points: Successively use the multi-level extreme point extraction algorithm, OPTICS clustering algorithm, gradient analysis algorithm, and adaptive spatial distribution algorithm to extract feature points. Finally, all feature points are merged and de-duplicated to ensure that the extracted feature points can more comprehensively and accurately reflect the temperature value distribution.

[0041] As Figure 5 shown, first use the multi-level extreme point extraction algorithm to divide the temperature range into high-temperature zone, medium-temperature zone, and low-temperature zone to ensure that each layer has representative points, especially extreme points, which helps to cover extreme situations in different temperature ranges.

[0042] Then use the OPTICS density clustering algorithm to find clusters in the data and retain even small clusters, so as to capture local high-temperature or low-temperature regions for searching small-range temperature regions that may be missed in the previous step.

[0043] Next, use the gradient analysis algorithm to calculate the gradient of the temperature field through interpolation. The high-gradient region represents a sharp temperature change and may correspond to the boundary or abnormal region; this step is mainly used to screen the temperature change region to further reduce the omission of feature points and ensure that the entire temperature region is covered.

[0044] Finally, through adaptive spatial distribution control, multi-source feature points are fused, and at the same time, all feature points are de-duplicated. Through spatial hashing of the adaptive grid for de-duplication, a balanced spatial distribution of feature points is achieved, avoiding selecting too many feature points in dense regions or missing feature points in sparse regions.

[0045] Multi-level extreme point extraction algorithm: Extract the extreme temperature of the heating area of the patch. Divide the temperature of the heating area into several intervals according to the quantile cutting method, and each interval contains approximately the same number of data points. Use the quantile cutting method to divide the temperature into 3 equal-frequency intervals: pd.qcut(df['Temperature'], n_bins), and use the sample size of each interval ≈ total sample size / n_bins; force the retention of the highest temperature point in each layer: idmax(), and generate a DataFrame containing the extreme values of each layer, with the structure [X, Y, Temperature]; Quantile calculation: Given data , the k-th quantile is defined as: , where n is the number of layers, .

[0046] OPTICS density clustering algorithm: Adopt OPTICS micro-cluster recognition, which is a density-based clustering algorithm. The OPTICS algorithm is applied to the scenario where the coordinate nodes are evenly distributed but the temperature values are different to achieve effective feature point extraction; it is adjusted according to the actual data through specific parameter settings and distance measurement methods to obtain the best clustering effect.

[0047] In data analysis, a cluster refers to a set of data points with similar characteristics. These points are close to each other in the feature space (such as coordinates, temperature, etc.), and are relatively far from the points in other clusters. The formation of clusters is based on a certain similarity measure (such as Euclidean distance, density, etc.).

[0048] Define the similarity between data points. Using the Euclidean distance, in the feature space (coordinates and temperature), use the Euclidean distance measurement method. For two nodes and , the Euclidean distance can be expressed as: ; the formation of clusters is based on the reachability graph analysis of the reachable distance. The reachable distance from point q to point p is defined as: , and the core code is: optics = OPTICS(min_samples = 5, xi = 0.01, min_cluster_size = 0.1), which sets the minimum number of samples in the neighborhood of the core point (reducing the density requirement), the clustering boundary relaxation degree, and the allowable minimum cluster size to 10% of the total sample size; extract all non-noise clusters labels_!= -1, and calculate the geometric center for each cluster , and the temperature takes the mean value within the cluster .

[0049] Gradient analysis algorithm: Dense gradient point extraction mainly involves dynamic threshold and KDE screening: First, perform gradient calculation. Use cubic spline interpolation to construct the temperature field griddata(method='cubic'), and calculate the gradient of the temperature field through interpolation. The high-gradient region represents a drastic temperature change, which may correspond to the boundary or abnormal region.

[0050] Two-dimensional gradient calculation: , Discretize using the central difference method: , ; Gradient magnitude: , Set the dynamic threshold to take the 95th percentile of the gradient magnitude , and screen the high-gradient region: , Finally, perform Gaussian kernel density estimation using the function np.percentile(grad_mag): , , where h is the smoothing parameter, n is the number of samples, d is the data dimension, and σ is the data standard deviation.

[0051] Adaptive spatial distribution: Perform multi-source feature point fusion, merge and deduplicate these feature points such as hierarchical extreme points (extreme temperature features), clustering centroids (regional representative points), and high-gradient points (boundary features); Deduplication and screening of feature points: First, perform density-aware grid division and calculate the local density: gaussian_kde(points[['X', 'Y']].T), using the dynamic grid formula: , Use a fine grid for high-density areas and a coarse grid for low-density areas to ensure that all points with similar temperature values are covered, without omission or misclassification of dissimilar points; calculate the grid belonging to each point: , Retain the first point that enters each grid. The dynamic grid division strategy, the grid size is inversely proportional to the local density : , where ; Ensure uniform distribution of feature points and avoid over-density in the central region.

[0052] Duplication removal condition: Euclidean distance threshold , merge adjacent points: Retain or ; Edge elimination: Retain points within 98% of the central data range: , .

[0053] For the visualization effect diagram: Mark the positions of the feature points on the original temperature heat map, use a diamond mark to highlight marker='D', separate the feature point coordinates and the node coordinates, as Figure 6 shown.

[0054] Compared with the traditional technology, the technology of optimizing the thermocouple layout using PYTHON and ABAQUS has significant advantages in terms of accuracy, efficiency, cost, reliability, versatility, and data-driven. It can more accurately monitor the temperature changes in key areas and avoid measurement errors caused by unreasonable layout. Complete multiple iterative optimizations in a short time, and the optimization process requires no manual intervention, significantly shortening the project cycle, especially suitable for large or complex bonding repair work. At the same time, through the optimization algorithm, while ensuring the temperature measurement accuracy, reasonably reduce the number of thermocouples, and reduce the material and installation costs.

[0055] Inspired by the above ideal embodiments of the present invention, through the above description, relevant staff can completely make various changes and modifications without departing from the technical idea of this invention. The technical scope of this invention is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A method for optimizing the position layout of thermocouples based on feature point extraction, characterized in that, It includes the following steps: S1. Heat the patch and adhesive with a hot patching instrument, obtain the temperature field data of the bonding interface based on the finite element simulation method, and extract the temperature values and coordinates of each node; S2. Analyze the temporal variation of the temperature field data, and screen the key frame with the largest standard deviation of the node temperature; S3. Perform spatial interpolation on the node temperature data of the key frame to generate a two-dimensional temperature distribution matrix; S4. Extract feature points from the temperature distribution matrix through a combination algorithm of extreme value stratification, OPTICS density clustering, gradient analysis, and adaptive spatial distribution; S5. Optimize the installation position of the thermocouple according to the distribution of the feature points.

2. The method for optimizing the position layout of thermocouples based on feature point extraction according to claim 1, wherein: The step S2 includes: Calculate the standard deviation of the node temperature of each frame, and select the key frame with the largest range.

3. The method for optimizing the thermocouple position layout based on feature point extraction according to claim 1, wherein: The step S3 includes: Project the temperature data of the nodes on the bonding surface onto a two-dimensional plane, and generate a continuous temperature field through spatial interpolation.

4. The method for optimizing the position layout of thermocouples based on feature point extraction according to claim 1, wherein: The step S4 includes: S41. Divide intervals based on the quantiles of the temperature values, and extract the extreme points of each interval; S42. Identify the cluster center points of the temperature anomaly regions through the OPTICS density clustering algorithm; S43. Calculate the temperature gradient field, and screen the mutation points with the top k in the gradient amplitude ranking; S44. Through adaptive spatial distribution, merge and de-duplicate the feature points to achieve a balanced spatial distribution of the feature points.

5. The method for optimizing the position layout of thermocouples based on feature point extraction according to claim 4, characterized in that: In the step S42, the OPTICS density clustering algorithm combines the temperature values and coordinate information for clustering analysis, specifically including: Each node is represented as a feature vector [x, y, z], which contains its coordinates (x, y) and temperature value t; Perform normalization processing on the feature vectors to ensure that features with different dimensions are on the same scale; In the three-dimensional feature space, use the Euclidean distance metric method to calculate the similarity between nodes, and the calculation formula is: ， Among them, the node p = [x p , y p , t p , and the node q = [x q , y q , t q .

6. The method for optimizing the thermocouple position layout based on feature point extraction according to claim 4, characterized in that: In the step S43, calculating the temperature gradient field specifically is: Spatial weighted gradient analysis, adding spatial density weights to the gradient points to reduce the importance of the gradient points in sparse regions; Through data distribution, adjust the gradient density threshold to accurately extract the feature points and draw the places where the temperature changes in the temperature field.

7. The method for optimizing the position layout of thermocouples based on feature point extraction according to claim 6, characterized in that: Select gradient points in the high-temperature, medium-temperature, and low-temperature regions respectively.

8. The method for optimizing the thermocouple position layout based on feature point extraction according to claim 4, wherein: The step S44 also includes: Perform multi-source feature point fusion through adaptive spatial distribution control, and merge duplicate feature points; perform spatial de-duplication on the extracted feature points, and retain the points with a mutual distance greater than the set value.

Images