A Method and Apparatus for Optimizing Drug Column Measurement Points Based on Hybrid Clustering

Abstract

This application discloses a method for optimizing propellant grain measurement points based on hybrid clustering, belonging to the field of engine propellant grain monitoring technology. The method includes: acquiring point cloud data of the propellant grain surface; obtaining a propellant grain mechanical response value feature set based on the point cloud data, the feature set including propellant grain strain, propellant grain shell strain, and propellant grain interface stress; performing initial density clustering and representative sample screening on the propellant grain surface point cloud data to obtain a sample set for each response value feature in the propellant grain mechanical response value feature set; and inputting the sample set for each response value feature in the propellant grain mechanical response value feature set into a hybrid distance metric model to obtain the optimal combination of measurement points on the propellant grain surface, which includes the optimal measurement points for propellant grain strain, propellant grain shell strain, and propellant grain interface stress. This method improves the accuracy and reliability of engine propellant grain monitoring.

Description

Technical Field

[0001] This application belongs to the field of engine propellant monitoring technology, and particularly relates to a method and device for optimizing propellant measurement points based on hybrid clustering. Background Technology

[0002] The structural integrity of the propellant grain casing in a solid rocket motor directly determines the safety and reliability of missiles and spacecraft. Some solid rocket motor failures stem from the failure to promptly monitor damage to the casing structure, with nearly half of these accidents traceable to early warning failures caused by an inappropriate layout of monitoring points. Therefore, optimizing the arrangement of propellant grain monitoring points has significant engineering value in improving monitoring efficiency and reducing mission risks.

[0003] In existing technologies, the placement of measuring points for propellant charges mainly relies on methods such as uniform grid placement, manual experience, and full-point placement. The uniform grid placement method, a widely used early strategy, divides the shell surface into a regular grid at fixed intervals and deploys sensors at each node. While this method ensures uniform spatial coverage, it ignores the non-uniformity of the physical field distribution. Manual experience relies on the professional judgment of engineers, selecting "critical locations" such as geometrical abrupt changes and historical fault points, and adding redundant measuring points using a safety factor method. This method has significant subjective limitations, and overly conservative placement strategies often lead to cost overruns. Full-point placement, as the theoretically optimal solution, achieves full-area monitoring by deploying sensors at all candidate locations. However, its high cost and data processing bottlenecks severely restrict its engineering practicality.

[0004] The inherent limitations of traditional measurement point placement methods have spurred an urgent need for intelligent optimization technologies. However, existing intelligent optimization methods still have significant shortcomings: traditional clustering algorithms, such as the single K-Means algorithm, may select non-surface points inside the propellant grain and fail to consider the mechanical response weights, leading to insufficient monitoring of highly sensitive areas; while optimization methods such as genetic algorithms are computationally expensive and cannot meet the real-time requirements of engineering projects. Existing methods suffer from high computational complexity, low measurement point placement efficiency, and an inability to accurately capture the key mechanical response regions of the propellant grain in optimizing the placement of measurement points. Summary of the Invention

[0005] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a method and apparatus for optimizing propellant measurement points based on hybrid clustering, which improves the accuracy and reliability of engine propellant monitoring.

[0006] Firstly, this application provides a method for optimizing drug column measurement points based on hybrid clustering, the method comprising:

[0007] Acquire point cloud data of the propellant grain surface, and obtain a mechanical response value feature set of the propellant grain based on the point cloud data of the propellant grain surface. The mechanical response value feature set of the propellant grain includes propellant grain strain, propellant grain shell strain and propellant grain interface stress.

[0008] Initial density clustering and representative sample screening are performed on the point cloud data of the drug grain surface to obtain a sample set of each response value feature in the drug grain mechanical response value feature set;

[0009] The sample set of each response value feature in the mechanical response value feature set of the propellant grain is input into the hybrid distance metric model to obtain the optimal combination of measurement points on the propellant grain surface. The optimal combination of measurement points on the propellant grain surface includes the optimal measurement point of propellant grain strain, the optimal measurement point of propellant grain shell strain, and the optimal measurement point of propellant grain interface stress.

[0010] According to one embodiment of this application, the step of performing initial density clustering and representative sample screening based on the point cloud data of the propellant grain surface to obtain a sample set for each response value feature in the propellant grain mechanical response value feature set includes:

[0011] The Z-score normalization method was used to normalize the point cloud data of the propellant grain surface to obtain the feature set of mechanical response values ​​of the propellant grain;

[0012] For each response value feature in the pharmacopoeia mechanical response value feature set, the MiniBatchKMeans algorithm is used to perform initial density clustering and representative sample screening to obtain a sample set for each response value feature in the pharmacopoeia mechanical response value feature set.

[0013] According to one embodiment of this application, the step of performing initial density clustering and representative sample screening using the MiniBatchKMeans algorithm to obtain a sample set for each response value feature in the pharmacokinetics mechanical response value feature set includes:

[0014] The MiniBatchKMeans algorithm was used to perform initial density clustering dimensionality reduction on the point cloud data of the drug grain surface, dividing the point cloud data of the drug grain surface into multiple initial density clusters, each of which contains at least 15 data points.

[0015] The point with the most significant mechanical response is selected from each initial density cluster as a sample point to obtain a sample set of each response value feature in the mechanical response value feature set of the drug column.

[0016] According to one embodiment of this application, the step of inputting a sample set of each response value feature from the propellant mechanical response value feature set into a hybrid distance metric model to obtain the optimal combination of measurement points on the propellant surface includes:

[0017] The sample set of each response value feature in the pharmacopoeia mechanical response value feature set is input into the hybrid distance metric model to calculate the distance matrix between all sample points;

[0018] The k-medoids++ algorithm is used to initialize the cluster centers;

[0019] Based on the distance matrix between all sample points, the mixed distance from the sample point to the current center point is recalculated. The cluster center point and cluster affiliation relationship are updated iteratively. When the change of the cluster center is less than the threshold or the preset maximum number of iterations is reached, the iteration stops, and the optimal combination of measurement points on the drug column surface is obtained.

[0020] According to one embodiment of this application, the construction process of the hybrid distance metric model includes:

[0021] A preset weighting coefficient is set for each response value feature in the set of mechanical response value features of the drug column;

[0022] Spatial constraints are set based on the geometrical uniformity of the propellant grain surface;

[0023] Based on the clustering of high stress and high strain regions, response enhancement terms are set;

[0024] A dynamic weighting mechanism is set based on the balance between spatial coverage and response sensitivity;

[0025] A hybrid distance metric model is constructed based on the preset weight coefficients, spatial constraint terms, response enhancement terms, and dynamic weighting mechanism.

[0026] According to one embodiment of this application, the initialization of cluster centers using the k-medoids++ algorithm includes:

[0027] Calculate the strain response value for each sample point in the sample set and sort them according to the magnitude of the response value;

[0028] The sample point with the largest response value is used as the cluster center.

[0029] According to one embodiment of this application, the method further includes:

[0030] The optimal combination of measurement points on the surface of the drug grain was imported using the Matplotlib tool. Combined with the point cloud data of the drug grain surface, a three-dimensional measurement point distribution map and a cluster center distribution map with mechanical response feature labels were generated in a three-dimensional rectangular coordinate system.

[0031] A histogram of response values ​​was plotted based on the optimal combination of measuring points on the surface of the propellant grain.

[0032] Based on the three-dimensional measurement point distribution map, the spatial distribution characteristics of the optimal measurement point combination on the drug column surface are obtained; based on the cluster center distribution map, the mechanical response sensitivity of the optimal measurement point combination on the drug column surface is obtained; and based on the response value histogram, the response value coverage of the optimal measurement point combination on the drug column surface is obtained.

[0033] Based on the aforementioned spatial distribution characteristics, mechanical response sensitivity, and response value coverage, numerical analysis and simulation experiments were used to verify the response value distribution of the optimal combination of measuring points on the surface of the propellant grain.

[0034] Secondly, this application provides a device for optimizing drug column measurement points based on hybrid clustering, the device comprising:

[0035] The acquisition module is used to acquire point cloud data of the propellant grain surface and obtain a mechanical response value feature set of the propellant grain based on the point cloud data of the propellant grain surface. The mechanical response value feature set of the propellant grain includes propellant grain strain, propellant grain shell strain and propellant grain interface stress.

[0036] The sampling module is used to perform initial density clustering and representative sample screening on the point cloud data of the drug grain surface to obtain a sample set of each response value feature in the drug grain mechanical response value feature set;

[0037] The generation module is used to input the sample set of each response value feature in the mechanical response value feature set of the propellant into the hybrid distance metric model to obtain the optimal combination of measurement points on the propellant surface. The optimal combination of measurement points on the propellant surface includes the optimal measurement point of propellant strain, the optimal measurement point of propellant shell strain, and the optimal measurement point of propellant interface stress.

[0038] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the drug column measurement point optimization method based on hybrid clustering as described in the first aspect above.

[0039] Fourthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the drug column measurement point optimization method based on hybrid clustering as described in the first aspect above.

[0040] Fifthly, this application provides a chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the drug column measurement point optimization method based on hybrid clustering as described in the first aspect.

[0041] In a sixth aspect, this application provides a computer program product, including a computer program that, when executed by a processor, implements the drug column measurement point optimization method based on hybrid clustering as described in the first aspect above.

[0042] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application.

[0043] The present invention provides a method for optimizing drug column measurement points based on hybrid clustering, which has the following advantages over existing technologies:

[0044] (1) This invention obtains a mechanical response value feature set by acquiring point cloud data of the propellant surface and analyzes it, which can optimize the arrangement of propellant measuring points in multiple dimensions. By performing initial density clustering on the point cloud data of the propellant surface and selecting representative samples, redundant data is effectively reduced and computational efficiency is improved. By using a hybrid distance metric model to perform in-depth analysis on mechanical characteristics such as propellant strain, propellant shell strain and propellant interface stress, mechanically sensitive areas can be captured more accurately, and the optimal combination of measuring points for each response value feature of the propellant surface can be obtained more accurately. This improves the reliability of solid fuel propellant monitoring, increases the measuring point capture rate in the critical strain region, and the output measuring points are located on the solid surface of the propellant, reducing engineering errors caused by virtual measuring points.

[0045] (2) This invention effectively optimizes the selection of measuring points on the surface of the aerosol column by inputting a sample set of mechanical response characteristics of the aerosol column into a hybrid distance metric model and calculating the distance matrix between sample points. By using the k-medoids++ algorithm to initialize the cluster centers and iteratively updating the cluster centers and cluster affiliation, the error caused by the initial center point selection can be effectively reduced, and the accuracy of the clustering results can be improved. Through the synergistic optimization of initial density clustering and K-Medoids point selection, combined with a hybrid distance metric model specifically for aerosol columns, the efficiency of aerosol column measuring point layout and monitoring accuracy are improved, the computational complexity is reduced, and the coverage of sensitive areas is increased.

[0046] (3) This invention performs initial density clustering dimensionality reduction on the point cloud data of the propellant surface, dividing the point cloud data of the propellant surface into multiple initial density clusters. The point with the most significant mechanical response is selected from each cluster as the sample point, which effectively optimizes the sample selection of the propellant mechanical response value feature set. By adopting the MiniBatchKMeans algorithm, the efficiency of sample selection is improved, the consumption of computational resources is reduced, the error in the initial clustering results is reduced, the representativeness of the sample points and the computational efficiency of the propellant measurement points are improved, and the measurement points are located on the propellant surface and priority is given to capturing the mechanical response sensitive area, which is suitable for stability monitoring of solid rocket motor propellant. Attached Figure Description

[0047] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0048] Figure 1 This is one of the flowcharts illustrating the drug column measurement point optimization method based on hybrid clustering provided in the embodiments of this application;

[0049] Figure 2 This is a data cloud diagram of the mechanical response value characteristics of the propellant column provided in the embodiments of this application;

[0050] Figure 3 This is the second flowchart illustrating the drug column measurement point optimization method based on hybrid clustering provided in the embodiments of this application;

[0051] Figure 4 A three-dimensional visualization of the distribution of measurement points provided in the embodiments of this application;

[0052] Figure 5 A three-dimensional clustering distribution diagram provided for embodiments of this application;

[0053] Figure 6 The response value histogram provided in the embodiments of this application;

[0054] Figure 7 A comparison of the error index distribution obtained from simulation experiments of the hybrid clustering drug column measurement point optimization method and the uniform grid method provided in the embodiments of this application;

[0055] Figure 8 This is a schematic diagram of the structure of the drug column measurement point optimization device based on hybrid clustering provided in the embodiments of this application;

[0056] Figure 9 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0057] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0058] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0059] The following description, in conjunction with the accompanying drawings, details the drug column measurement point optimization method based on hybrid clustering, the drug column measurement point optimization device based on hybrid clustering, the electronic device, and the readable storage medium provided in this application, through specific embodiments and application scenarios.

[0060] Among them, the drug column measurement point optimization method based on hybrid clustering can be applied to the terminal, specifically executed by the hardware or software in the terminal.

[0061] The terminal includes, but is not limited to, portable communication devices such as mobile phones or tablets with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads). It should also be understood that, in some embodiments, the terminal may not be a portable communication device, but rather a desktop computer with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads).

[0062] The following embodiments describe a terminal including a display and a touch-sensitive surface. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, mouse, and joystick.

[0063] The pharmacopoeia optimization method based on hybrid clustering provided in this application embodiment can be executed by an electronic device or a functional module or entity in an electronic device that can implement the pharmacopoeia optimization method based on hybrid clustering. The electronic devices mentioned in this application embodiment include, but are not limited to, mobile phones, tablets, computers, cameras and wearable devices. The pharmacopoeia optimization method based on hybrid clustering provided in this application embodiment will be described below using an electronic device as the execution subject as an example.

[0064] Figure 1 This is one of the flowcharts illustrating the drug column measurement point optimization method based on hybrid clustering provided in this application embodiment, such as... Figure 1 As shown, the drug column measurement point optimization method based on hybrid clustering includes steps 110, 120 and 130.

[0065] Step 110: Obtain point cloud data of the propellant grain surface, and obtain a mechanical response value feature set of the propellant grain based on the point cloud data of the propellant grain surface. The mechanical response value feature set of the propellant grain includes propellant grain strain, propellant grain shell strain and propellant grain interface stress.

[0066] It's easy to understand that acquiring point cloud data of the propellant grain surface involves taking a solid rocket motor propellant grain-shell assembly as an example. The propellant grain has dimensions of 400mm in diameter and 1000mm in length, and is made of HTPB (Hydroxyl-terminated polybutadiene) composite propellant. The shell material is 30CrMnSiA alloy steel, and the interface material is an epoxy resin layer with a thickness of 0.5~1.2mm. The mechanical response data of the propellant grain comes from the propellant grain strain, shell strain, and interface stress. The propellant grain strain focuses on the distribution of principal stresses on the surface, the shell strain focuses on the circumferential strain concentration area, and the interface stress focuses on the peak shear stress. Figure 2 This is a data cloud diagram of the mechanical response value characteristics of the propellant column provided in the embodiments of this application. Figure 2 Figure (a) in the figure is a cloud map of the strain data of the propellant column. Figure 2 Figure (b) in the figure is a cloud map of the shell strain data. Figure 2 Figure (c) shows the interface stress data cloud map. A cylindrical coordinate system aligned with the symmetry axis of the propellant is established. The rectangular coordinate system (X,Y,Z) of the point cloud on the surface of the propellant is converted into a cylindrical coordinate system (R,θ,X). The X-axis coincides with the symmetry axis of the propellant. The design variable for the optimized arrangement of the propellant measuring points is the high-density point cloud data of the three.

[0067] Step 120: Perform initial density clustering and representative sample screening on the point cloud data of the drug grain surface to obtain a sample set of each response value feature in the drug grain mechanical response value feature set;

[0068] In some embodiments, the initial density clustering and representative sample screening based on the point cloud data of the propellant grain surface to obtain a sample set for each response value feature in the propellant grain mechanical response value feature set includes:

[0069] The Z-score normalization method was used to normalize the point cloud data of the propellant grain surface to obtain the feature set of mechanical response values ​​of the propellant grain;

[0070] For each response value feature in the pharmacopoeia mechanical response value feature set, the MiniBatchKMeans algorithm is used to perform initial density clustering and representative sample screening to obtain a sample set for each response value feature in the pharmacopoeia mechanical response value feature set.

[0071] It should be noted that the initial collected point cloud data of the drug column surface is high-density point cloud data. Each point contains three-dimensional spatial coordinates (X, Y, Z) and the corresponding response value (stress or strain value). To unify the dimensions, the Z-score standardization method is used to normalize the response value, unifying the dimensions of the spatial coordinates and the mechanical response value. A preset weighting coefficient is applied to the mechanical response value features, for example, the weighting coefficient is set to 2.5. The standardized mechanical response value is multiplied by the weighting factor to guide the subsequent clustering results to prioritize the selection of mechanically sensitive regions.

[0072] In some embodiments, the initial density clustering and representative sample screening using the MiniBatchKMeans algorithm to obtain a sample set for each response value feature in the pharmacokinetics mechanical response value feature set includes:

[0073] The MiniBatchKMeans algorithm was used to perform initial density clustering dimensionality reduction on the point cloud data of the drug grain surface, dividing the point cloud data of the drug grain surface into multiple initial density clusters, each of which contains at least 15 data points.

[0074] The point with the most significant mechanical response is selected from each initial density cluster as a sample point to obtain a sample set of each response value feature in the mechanical response value feature set of the drug column.

[0075] For example, the point cloud data is divided into 400-600 initial density clusters using the MiniBatchKMeans algorithm (e.g., dividing the point cloud into 512 initial density clusters). Each initial density cluster contains at least 15 data points. After the initial density clustering is completed, the points with the most significant mechanical response are extracted from each cluster as representative samples. The selection criteria are the top 10% of points in terms of comprehensive response index. If there are multiple candidate points, the points closer to the geometric center of the cluster are selected first. This yields a sample set of each response value feature in the mechanical response value feature set of the drug column.

[0076] In this embodiment, by performing initial density clustering dimensionality reduction on the point cloud data of the propellant surface, the point cloud data of the propellant surface is divided into multiple initial density clusters. Points with the most significant mechanical response are selected from each cluster as sample points, effectively optimizing the sample selection of the propellant mechanical response value feature set. By employing the MiniBatchKMeans algorithm, the efficiency of sample selection is improved, computational resource consumption is reduced, errors in the initial clustering results are reduced, the representativeness of sample points and the computational efficiency of propellant measurement points are improved, and the measurement points are located on the propellant surface, prioritizing the capture of mechanically sensitive regions, making it suitable for stability monitoring of solid rocket motor propellant grains.

[0077] In this embodiment, by normalizing the point cloud data on the surface of the propellant column and performing initial density clustering on the mechanical response value features of each propellant column, the sample selection of the mechanical response value feature set of the propellant column is effectively optimized, the efficiency of measurement point optimization and computational performance are improved, the computation time is reduced, and the retention rate of mechanical response feature information is increased.

[0078] Step 130: Input the sample set of each response value feature in the mechanical response value feature set of the propellant into the hybrid distance metric model to obtain the optimal combination of measurement points on the propellant surface. The optimal combination of measurement points on the propellant surface includes the optimal measurement point of propellant strain, the optimal measurement point of propellant shell strain, and the optimal measurement point of propellant interface stress.

[0079] In some embodiments, the step of inputting a sample set of each response value feature from the propellant grain mechanical response value feature set into a hybrid distance metric model to obtain the optimal combination of measurement points on the propellant grain surface includes:

[0080] The sample set of each response value feature in the pharmacopoeia mechanical response value feature set is input into the hybrid distance metric model to calculate the distance matrix between all sample points;

[0081] The k-medoids++ algorithm is used to initialize the cluster centers;

[0082] Based on the distance matrix between all sample points, the mixed distance from the sample point to the current center point is recalculated. The cluster center point and cluster affiliation relationship are updated iteratively. When the change of the cluster center is less than the threshold or the preset maximum number of iterations is reached, the iteration stops, and the optimal combination of measurement points on the drug column surface is obtained.

[0083] Finally, a hybrid distance metric model for the measurement points of the drug column is constructed by weighted combination of spatial Euclidean distance and response value difference. The sample set of each response value feature in the drug column mechanical response value feature set is input into the hybrid distance metric model, the distance matrix between all sample points is calculated, and after the distance matrix is ​​constructed, the k-medoids++ algorithm is used to cluster the sample set.

[0084] In some embodiments, initializing cluster centers using the k-medoids++ algorithm includes:

[0085] Calculate the strain response value for each sample point in the sample set and sort them according to the magnitude of the response value;

[0086] The sample point with the largest response value is used as the cluster center.

[0087] For example, Figure 3This is the second flowchart of the drug column measurement point optimization method based on hybrid clustering provided in this application embodiment. In the initialization stage, the k-medoids++ strategy is used to select 12 initial center points from 512 sample points. The selection probability of the initial point is positively correlated with its strain response value. In the iterative optimization stage, the batch size is set to 1000 and the maximum number of iterations is 50. The center point update and cluster allocation operations are performed alternately. In each iteration, the hybrid distance from the sample point to the current center point is recalculated and the cluster affiliation is updated until the objective function, that is, the rate of change of the total hybrid distance within the cluster is less than 0.1% or the maximum number of iterations is reached (50). The number of optimal measurement point combinations is 10-20, preferably 12. The 12 cluster center points finally output are the optimal measurement point combinations on the drug column surface. Their spatial distribution meets the following engineering constraints: (1) The minimum Euclidean distance between any two measurement points is ≥45mm; (2) The surface curvature radius of all measurement points is ≥5mm.

[0088] In some embodiments, a hybrid distance metric model and a hybrid clustering method are used to optimize the arrangement of measuring points on the propellant grains of a certain type of solid rocket motor. A finite element model is established using Abaqus software, and automated mesh generation, boundary and load condition settings, and solution are performed. Finite element verification shows that the optimized measuring point layout can accurately capture 98.7% of the critical strain region (ε≥2.5%) on the propellant grain surface, with a coverage rate of 93.5% for the high strain region (ε≥3.0%). The spatial distribution of measuring points is uniform, with a maximum spacing deviation not exceeding 15 mm and a minimum radius of curvature greater than 8 mm, meeting engineering installation requirements.

[0089] In this embodiment, by calculating the strain response value of each sample point in the sample set and sorting them according to the size of the response value, the sample point with the largest response value is selected as the cluster center point, which can effectively improve the initial accuracy of clustering. By reasonably selecting the cluster center point, the clustering process is optimized, the error caused by the initial cluster center point selection is reduced, and the stability and efficiency of the subsequent clustering process are improved. This method is suitable for the automated selection of key parameter monitoring points of solid fuel propellant grains, and improves the accuracy and reliability of propellant grain structural health monitoring.

[0090] In this embodiment, by inputting a sample set of the mechanical response characteristics of the propellant grain into a hybrid distance metric model and calculating the distance matrix between sample points, the selection of measuring points on the propellant grain surface can be effectively optimized. By using the k-medoids++ algorithm to initialize cluster centers and iteratively updating the cluster centers and cluster affiliation relationships, the error caused by the initial center point selection can be effectively reduced, improving the accuracy of the clustering results. Through the synergistic optimization of initial density clustering and K-Medoids point selection, combined with a hybrid distance metric model specifically for propellant grains, both the efficiency of propellant grain measuring point placement and monitoring accuracy are improved, reducing computational complexity and increasing the coverage of sensitive areas.

[0091] According to the hybrid clustering-based propellant grain measurement point optimization method provided in this application, the mechanical response value feature set obtained by acquiring point cloud data of the propellant grain surface is analyzed, which can optimize the arrangement of propellant grain measurement points in multiple dimensions. By performing initial density clustering on the point cloud data of the propellant grain surface and selecting representative samples, redundant data is effectively reduced and computational efficiency is improved. By using a hybrid distance metric model to perform in-depth analysis on mechanical characteristics such as propellant grain strain, propellant grain shell strain, and propellant grain interface stress, mechanically sensitive areas can be captured more accurately, and more precise optimal measurement point combinations for each response value feature of the propellant grain surface can be obtained. This improves the reliability of solid fuel propellant grain monitoring, increases the measurement point capture rate in the critical strain region, and the output measurement points are located on the solid surface of the propellant grain, reducing engineering errors caused by virtual measurement points.

[0092] In some embodiments, the construction process of the hybrid distance metric model includes:

[0093] A preset weighting coefficient is set for each response value feature in the set of mechanical response value features of the drug column;

[0094] Spatial constraints are set based on the geometrical uniformity of the propellant grain surface;

[0095] Based on the clustering of high stress and high strain regions, response enhancement terms are set;

[0096] A dynamic weighting mechanism is set based on the balance between spatial coverage and response sensitivity;

[0097] A hybrid distance metric model is constructed based on the preset weight coefficients, spatial constraint terms, response enhancement terms, and dynamic weighting mechanism.

[0098] It is easy to understand that the optimization of measuring points on the surface of the propellant grain needs to meet two core requirements at the same time: uniformity of spatial coverage and sensitivity to mechanical response. Traditional Euclidean distance measurement only considers geometric proximity, which is difficult to effectively capture high strain areas. On the other hand, relying solely on the difference in response values ​​will lead to spatial clustering of measuring points. This application constructs a hybrid distance measurement model by coupling spatial distance and mechanical response characteristics, which prioritizes mechanically sensitive areas while ensuring the rationality of the spatial distribution of measuring points.

[0099] Specifically, this hybrid distance model achieves dual optimization objectives through the following techniques:

[0100] Spatial constraints This ensures the uniform geometric distribution of measuring points on the surface of the propellant column, avoiding monitoring blind spots;

[0101] Response reinforcement This guides the measuring points to converge on high-stress, high-strain areas, thereby improving monitoring sensitivity;

[0102] Dynamic weighting mechanism By adjusting the coefficient Achieving a balance between spatial coverage and response sensitivity.

[0103] A hybrid distance metric model specifically for propellant column measuring points is constructed, and the calculation formula is shown below:

[0104]

[0105] in, and These are the weighting coefficients, and ; and Let be the spatial coordinate vectors of the i-th sample point and the j-th sample point, respectively. and Let be the mechanical response values ​​of the i-th sample point and the j-th sample point, respectively, and D be the mixing distance between the i-th sample point and the j-th sample point.

[0106] In this embodiment, by setting a preset weighting coefficient for each response value feature in the propellant's mechanical response feature set, and combining spatial constraint terms, response enhancement terms, and a dynamic weighting mechanism, the selection accuracy of measurement points on the propellant surface can be effectively improved. By spatially constraining the geometrical uniformity of the propellant surface and enhancing the response in high-stress, high-strain regions, the response sensitivity and spatial coverage of the measurement points are improved. A weighted hybrid distance model of spatial Euclidean distance and mechanical response difference is constructed. By dynamically adjusting the weighting coefficients, the measurement points can be guided to preferentially cluster in stress concentration regions, thus optimizing the measurement point layout on the propellant surface.

[0107] In some embodiments, the method further includes:

[0108] The optimal combination of measurement points on the surface of the drug grain was imported using the Matplotlib tool. Combined with the point cloud data of the drug grain surface, a three-dimensional measurement point distribution map and a cluster center distribution map with mechanical response feature labels were generated in a three-dimensional rectangular coordinate system.

[0109] A histogram of response values ​​was plotted based on the optimal combination of measuring points on the surface of the propellant grain.

[0110] Based on the three-dimensional measurement point distribution map, the spatial distribution characteristics of the optimal measurement point combination on the drug column surface are obtained; based on the cluster center distribution map, the mechanical response sensitivity of the optimal measurement point combination on the drug column surface is obtained; and based on the response value histogram, the response value coverage of the optimal measurement point combination on the drug column surface is obtained.

[0111] Based on the aforementioned spatial distribution characteristics, mechanical response sensitivity, and response value coverage, numerical analysis and simulation experiments were used to verify the response value distribution of the optimal combination of measuring points on the surface of the propellant grain.

[0112] It is easy to understand that a three-dimensional measurement point distribution map can reflect the spatial distribution characteristics of the measurement points, a cluster center distribution map can reflect the mechanical response sensitivity of the measurement points, and a response value distribution histogram can quantify the response value coverage of the measurement points.

[0113] The optimal combination of measurement points on the surface of the drug column was imported using the Matplotlib tool. Combined with the point cloud data of the drug column surface, a three-dimensional measurement point distribution map and a cluster center distribution map with mechanical response feature markers were generated in a three-dimensional rectangular coordinate system. The distribution of response values ​​was verified by numerical analysis and the distribution of error index was verified by simulation experiments.

[0114] The analysis process for the 3D measurement point distribution map and cluster center distribution map specifically includes:

[0115] Figure 4 This application provides a three-dimensional visualization of the measurement point distribution map in an embodiment of the application, wherein... Figure 4 Figure (a) shows the distribution of measurement points for the three-dimensional visualization of the strain of the propellant column. Figure 4 Figure (b) shows the distribution of three-dimensional visualization measurement points for shell strain. Figure 4 Figure (c) shows the distribution of three-dimensional visualization measurement points for interface stress.

[0116] Figure 5 The three-dimensional clustering distribution map provided in the embodiments of this application, wherein Figure 5 Figure (a) shows the three-dimensional cluster distribution of strain in the propellant column. Figure 5 Figure (b) shows the three-dimensional cluster distribution of shell strain. Figure 5 Figure (c) shows the three-dimensional cluster distribution of interfacial stress.

[0117] Taking the strain data of the propellant charge and the obtained optimized measuring points as an example, Table 1 is a table of coordinates of the 12 optimal measuring points for propellant charge strain provided in the embodiments of this application, wherein, Let θ be the radial distance in cylindrical coordinates, θ be the azimuth angle, and z be the height. The three-dimensional measurement point distribution map and the cluster center distribution map are shown below. Figure 4 Figure (a) in the middle and Figure 5As shown in Figure (a), the optimized measuring point layout scheme exhibits significant spatial distribution characteristics and response patterns in the strain monitoring of the propellant charge, fully demonstrating the algorithm's ability to accurately capture the stress field of complex structures. From a three-dimensional spatial distribution perspective, the measuring points mainly form three distinct monitoring clusters: a high-stress front-end region, a transition monitoring zone, and a foundation stability region. These distributions form a good mapping relationship with the actual stress state of the shell. In the front-end region of the shell, the measuring points exhibit typical stress concentration characteristics. The triangular monitoring network formed by the measuring points in this region provides crucial data support for assessing the reliability of the connectors. The transition monitoring zone exhibits gradient-changing response characteristics. The measuring point distribution in the foundation stability region has significant spatial stratification characteristics. A low-response monitoring surface is formed near the base, while a jump in response value is observed in the far-end region. This axial gradient change accurately captures the attenuation law of the shell's bending deformation.

[0118] Table 1. Coordinates of the 12 measurement points for optimal strain of the propellant column.

[0119]

[0120] Taking the shell strain data and the obtained optimized measurement points as an example, Table 2 is a table of coordinates of the 12 optimal measurement points for shell strain provided in the embodiments of this application. The three-dimensional measurement point distribution map and the cluster center distribution map are shown below. Figure 4 Figure (b) in the middle and Figure 5 As shown in Figure (b), the optimized measuring point layout scheme exhibits a clear spatial distribution pattern in shell strain monitoring, with the layout of each measuring point highly consistent with the structural characteristics and stress properties of the shell. From a three-dimensional spatial perspective, the measuring points are mainly distributed in the key functional areas of the shell, with a clear overall layout logic, effectively capturing the strain state of different areas and identifying potential high-stress risk zones. In the X-axis direction, the measuring point distribution shows a significant asymmetry, mainly concentrated in the middle of the shell, indicating that this area is the main load-bearing area of ​​the shell. Among them, there is some stress fluctuation in the front connection area of ​​the shell. In contrast, the overall stress in the middle load-bearing area is more uniform. In the Y-axis direction, the measuring point distribution shows targeted monitoring of the shell edge areas. The stress state of the positive edge is stable. The stress of the negative edge is slightly higher than that of the positive area, possibly related to external constraints or loads on that side. In the Z-axis direction, the measuring point distribution exhibits a layered monitoring characteristic, mainly divided into shallow and deep monitoring zones. The shallow area is close to the shell surface and has a lower response value, mainly used for monitoring the surface strain state. The relatively higher response values ​​in deeper regions indicate a more complex stress distribution in the core load-bearing area inside the shell. This layered layout strategy effectively achieves comprehensive monitoring from the surface to the interior, ensuring the visualization of stress transmission paths.

[0121] Table 2. Coordinates of the 12 Optimal Measurement Points for Shell Strain

[0122]

[0123] Taking the interface stress data and the obtained optimized measuring points as an example, Table 3 is a table of coordinates of the 12 optimal measuring points for interface stress provided in the embodiments of this application. The three-dimensional measuring point distribution map and the cluster center distribution map are shown below. Figure 4 Figure (c) in the middle and Figure 5 As shown in Figure (c), the optimized measuring point layout scheme exhibits a clear spatial distribution pattern in interface stress monitoring, and the measuring point layout corresponds well with the structural characteristics and stress distribution of the shell. From a three-dimensional spatial perspective, the measuring points are mainly concentrated in three key stress regions of the shell: the high stress concentration region, the transition transfer region, and the stable bearing region. The measuring point distribution density and response value characteristics of each region intuitively reflect the stress state of the shell. In the X-axis direction, the measuring points exhibit a clear asymmetric distribution. The high stress concentration region accurately captures the stress concentration phenomenon in the front-end connection area of ​​the shell. The transition transfer region is affected by the internal reinforcing structure, resulting in significant stress fluctuations. The stable bearing region shows a lower response value, indicating that the stress in this region is relatively uniform. In the Y-axis direction, the measuring point distribution exhibits significant edge effect characteristics. Stress concentration exists at the positive edges. The stress state at the negative edges is more complex. The measuring point response values ​​in the central region are generally low, verifying the stable stress state of the shell's central region. In the Z-axis direction, the measuring point distribution exhibits typical layered monitoring characteristics. The measuring points in the shallow region are mainly used to monitor surface stress fluctuations, and the response values ​​remain at a low level. The measuring points in the middle layer accurately reflect the complex stress state of the core load-bearing area of ​​the shell. The measuring points in the deeper layer generally have lower response values, providing reliable benchmark reference data for the system.

[0124] Table 3. Coordinates of the 12 measuring points with optimal interface stress

[0125]

[0126] The Matplotlib tool was used to import the measured point response value data optimized by hybrid clustering, generate a response value distribution histogram, and perform numerical analysis to quantify the response value coverage and evaluate the information retention efficiency. The specific process of numerical analysis of the response value distribution histogram includes:

[0127] Figure 6 The response value histogram provided in the embodiments of this application, wherein Figure 6 Figure (a) in the figure is a histogram of the strain response values ​​of the propellant column. Figure 6 Figure (b) is a histogram of the shell strain response values. Figure 6 Figure (c) is a histogram of the interface stress response values.

[0128] Taking the response data of the obtained strain optimization measurement points of the propellant column as an example, such as Figure 6As shown in Figure (a), the response value distribution exhibits a typical bimodal characteristic: the low-value cluster (<0.01) contains 7 measurement points, constituting the system's baseline signal; the high-value cluster (>0.01) has 5 measurement points that clearly identify key monitoring locations. The highest response value is 14 times the lowest value; this magnitude difference not only verifies the algorithm's sensitivity in identifying abnormal areas but also provides clear direction for subsequent structural optimization. The new measurement point layout scheme, through a non-uniform sampling strategy, achieves more comprehensive monitoring coverage than the conventional uniform layout with the same number of measurement points.

[0129] Taking the response data of the obtained shell strain optimization measurement points as an example, such as Figure 6 As shown in Figure (b), the response value distribution characteristics reveal that the response values ​​of the vast majority of measuring points are concentrated in the range of 1E-06 to 1E-05, forming the stability baseline of the system. Two main peaks indicate a potentially high risk of stress concentration in these areas. In contrast, the measuring points with extremely low response values ​​provide a benchmark for judging the system's stability. Although the overall response values ​​are low, the relative differences between measuring points still clearly reflect the stress characteristics of the shell, verifying the algorithm's positioning accuracy and the effectiveness of the optimized layout.

[0130] Taking the response data of the obtained interface stress optimization measurement points as an example, such as Figure 6 As shown in (c), the system exhibits a clear gradient characteristic in terms of the overall response value distribution. Measurement points in the low-value region (<0.0002) form the stable baseline of the system, while those in the medium-value region (0.0002~0.0004) reflect the stress distribution under conventional loads. Measurement points in the high-value region (>0.0004) accurately pinpoint high-stress risk areas. The three major stress peak points highly coincide with the high-stress areas predicted by finite element analysis, fully verifying the positioning accuracy of the optimization algorithm. Compared to the traditional uniform measurement point layout, this non-uniform sampling strategy achieves more accurate stress monitoring coverage with the same number of measurement points, providing a more reliable basis for the structural optimization and safety assessment of the shell.

[0131] Furthermore, the specific process of verification through simulation experiments includes:

[0132] In one embodiment of this application, a simulation proxy model of the drug column is reconstructed using Python. The measurement points obtained by the drug column measurement point optimization method based on hybrid clustering are used to calculate the error. The uniform grid method commonly used in measurement point layout is compared with the error index distribution comparison chart drawn using Matplotlib.

[0133] Furthermore, the specific process of analyzing the error index distribution comparison chart includes:

[0134] The evaluation used three commonly used metrics: Mean Absolute Error (MAE), Mean Square Error (MSE), and Root Mean Square Error (RMSE). Relevant results are shown in Table 4. Figure 7 As shown in Table 4, the error performance of the two methods combined with simulation is illustrated. Figure 7 This is a comparison chart of the error index distribution obtained from simulation experiments of the hybrid clustering drug column measurement point optimization method and the uniform grid method provided in the embodiments of this application. Figure 7 It can be seen that the measurement point layout scheme based on cluster optimization shows significant advantages in all error indicators.

[0135] Table 4 Simulation Comparison Experiment Verification Table

[0136]

[0137] In strain monitoring of propellant charges, the MAE and RMSE of the clustering optimization scheme were 0.03779 and 0.03925, respectively, both lower than those of the uniform arrangement (0.04017 and 0.04054), corresponding to an error reduction of 5.9% and 3.2%, respectively. This indicates that the clustering algorithm can more effectively capture the spatial distribution of the strain field, especially in high gradient regions, and significantly improves the ability to identify local strain abrupt changes by adaptively adjusting the density of measuring points.

[0138] In shell strain monitoring, both methods showed relatively low overall error values, but the clustering method was still slightly better than the uniform arrangement. The MAE of the clustering method was 0.000484, and the MSE was 2.35 × 10⁻⁶. -7 Both are slightly smaller than the uniform method (MAE is 0.000490, MSE is 2.41×10). -7 This demonstrates that clustering methods exhibit more stable performance in fine strain monitoring tasks, primarily due to their intelligent identification of shell geometric boundaries and key stress areas.

[0139] In interface stress monitoring, the clustering method showed a smaller difference in MAE (0.000203 vs. 0.000196) compared to uniform arrangement, but performed better in RMSE (0.000231 vs. 0.000235), indicating that the clustering method has a certain advantage in long-term stability. Although the MAE was slightly higher, the lower RMSE indicated a more balanced overall error distribution, which helps to improve monitoring reliability.

[0140] Figure 7 The bottom right subplot shows the percentage improvement of the K-Medoids clustering method over the uniform grid method in three error metrics under three operating conditions. It can be seen that the clustering method exhibits comprehensive advantages in both propellant grain and shell operating conditions. While its MAE is slightly higher than the uniform grid method in interfacial stress, it still maintains optimization in RMSE and MSE, demonstrating an overall improvement in stability.

[0141] This application significantly improves the efficiency and computational performance of measurement point optimization through a hybrid clustering architecture, reducing computation time while ensuring the retention rate of mechanical response feature information, effectively solving the computational bottleneck problem in large-scale point cloud data processing. By constructing a weighted hybrid distance model of spatial Euclidean distance and mechanical response difference, and dynamically adjusting the weight coefficients, it accurately captures mechanically sensitive areas, improves monitoring reliability, increases the measurement point capture rate in critical strain areas, and ensures that the output measurement points are strictly located on the surface of the propellant grain, avoiding engineering errors caused by virtual measurement points. By combining 3D spatial clustering distribution visualization with finite element simulation verification, a multi-dimensional measurement point optimization effect evaluation system is established, comprehensively verifying the technical superiority and industrial applicability of this optimization layout method from three aspects: spatial coverage completeness, mechanical response sensitivity, and engineering feasibility.

[0142] In this embodiment, by analyzing the three-dimensional measurement point distribution map, cluster center distribution map, and response value histogram, the spatial distribution characteristics, mechanical response sensitivity, and response value coverage of the optimal measurement point combination on the propellant grain surface can be comprehensively evaluated, thereby verifying the rationality of the measurement point layout on the propellant grain surface. Through numerical analysis and simulation experiments, the response sensitivity and spatial coverage of the measurement point layout are improved. By combining three-dimensional spatial clustering distribution visualization with finite element simulation verification, a multi-dimensional measurement point optimization effect evaluation system is established. This system comprehensively verifies the technical performance and industrial applicability of the optimized layout method from three aspects: spatial coverage completeness, mechanical response sensitivity, and engineering feasibility.

[0143] The pharmacopoeia-based drug column measurement point optimization method provided in this application can be executed by a pharmacopoeia-based drug column measurement point optimization device. This application uses the pharmacopoeia-based drug column measurement point optimization device executing the method as an example to illustrate the pharmacopoeia-based drug column measurement point optimization device provided in this application.

[0144] This application also provides a drug column measurement point optimization device based on hybrid clustering, such as... Figure 8 As shown, the drug column measurement point optimization device based on hybrid clustering includes: acquisition module 810, sampling module 820 and generation module 830.

[0145] The acquisition module 810 is used to acquire point cloud data of the propellant grain surface and obtain a mechanical response value feature set of the propellant grain based on the point cloud data of the propellant grain surface. The mechanical response value feature set of the propellant grain includes propellant grain strain, propellant grain shell strain and propellant grain interface stress.

[0146] The sampling module 820 is used to perform initial density clustering and representative sample screening on the point cloud data of the drug column surface to obtain a sample set of each response value feature in the drug column mechanical response value feature set;

[0147] The generation module 830 is used to input the sample set of each response value feature in the mechanical response value feature set of the propellant into the hybrid distance metric model to obtain the optimal combination of measurement points on the propellant surface. The optimal combination of measurement points on the propellant surface includes the optimal measurement point of propellant strain, the optimal measurement point of propellant shell strain, and the optimal measurement point of propellant interface stress.

[0148] According to the hybrid clustering-based propellant grain measurement point optimization method provided in this application, the mechanical response value feature set obtained by acquiring point cloud data of the propellant grain surface is analyzed, which can optimize the arrangement of propellant grain measurement points in multiple dimensions. By performing initial density clustering on the point cloud data of the propellant grain surface and selecting representative samples, redundant data is effectively reduced and computational efficiency is improved. By using a hybrid distance metric model to perform in-depth analysis on mechanical characteristics such as propellant grain strain, propellant grain shell strain, and propellant grain interface stress, mechanically sensitive areas can be captured more accurately, and more precise optimal measurement point combinations for each response value feature of the propellant grain surface can be obtained. This improves the reliability of solid fuel propellant grain monitoring, increases the measurement point capture rate in the critical strain region, and the output measurement points are located on the solid surface of the propellant grain, reducing engineering errors caused by virtual measurement points.

[0149] The pharmacopoeia-based drug column measurement point optimization device provided in this application embodiment can achieve... Figures 1 to 7 The various processes implemented in the embodiment of the drug column measurement point optimization method based on hybrid clustering will not be described in detail here to avoid repetition.

[0150] In some embodiments, such as Figure 9 As shown, this application embodiment also provides an electronic device 900, including a processor 901, a memory 902, and a computer program stored in the memory 902 and executable on the processor 901. When the program is executed by the processor 901, it implements the various processes of the above-described embodiment of the drug column measurement point optimization method based on hybrid clustering and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0151] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0152] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described embodiment of the drug column measurement point optimization method based on hybrid clustering and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0153] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0154] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for optimizing drug column measurement points based on hybrid clustering.

[0155] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0156] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above-described embodiment of the drug column measurement point optimization method based on hybrid clustering, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0157] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a device-level chip, device chip, chip device, or on-chip device chip, etc.

[0158] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0159] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the drug column measurement point optimization method based on hybrid clustering of the various embodiments of this application.

[0160] In the description of this application, "first feature" and "second feature" may include one or more of the features.

[0161] In the description of this application, "multiple" means two or more.

[0162] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0163] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0164] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. A hybrid clustering based method for optimizing the measurement points of a propellant grain, characterized in that, The method comprises: obtaining a propellant grain surface point cloud data, and obtaining a propellant grain mechanical response value feature set based on the propellant grain surface point cloud data, wherein the propellant grain mechanical response value feature set comprises propellant grain strain, propellant grain shell strain and propellant grain interface stress; performing initial density clustering and representative sample screening on the propellant grain surface point cloud data to obtain a sample set of each response value feature in the propellant grain mechanical response value feature set; respectively inputting the sample set of each response value feature in the propellant grain mechanical response value feature set into a hybrid distance measurement model to obtain a propellant grain surface optimal measurement point combination, wherein the propellant grain surface optimal measurement point combination comprises a propellant grain strain optimal measurement point, a propellant grain shell strain optimal measurement point and a propellant grain interface stress optimal measurement point; the step of respectively inputting the sample set of each response value feature in the propellant grain mechanical response value feature set into the hybrid distance measurement model to obtain the propellant grain surface optimal measurement point combination comprises: inputting the sample set of each response value feature in the propellant grain mechanical response value feature set into the hybrid distance measurement model to calculate a distance matrix between all sample points; initializing a clustering center point by using a k-medoids++ algorithm; recomputing a hybrid distance of a sample point to a current center point based on the distance matrix between all sample points, and updating the clustering center point and cluster attribution by iteration, and stopping iteration when a clustering center change amplitude is less than a threshold value or a preset maximum iteration number is reached, to obtain the propellant grain surface optimal measurement point combination; the construction process of the hybrid distance measurement model comprises: setting a preset weight coefficient for each response value feature in the propellant grain mechanical response value feature set; setting a space constraint term based on geometric distribution uniformity of a propellant grain surface; setting a response intensification term based on the aggregation of high stress and high strain areas; setting a dynamic weight mechanism based on a balance between space coverage and response sensitivity; constructing the hybrid distance measurement model based on the preset weight coefficient, the space constraint term, the response intensification term and the dynamic weight mechanism.

2. The hybrid clustering based propellant grain measurement point optimization method of claim 1, wherein, the step of performing initial density clustering and representative sample screening on the propellant grain surface point cloud data to obtain the sample set of each response value feature in the propellant grain mechanical response value feature set comprises: performing normalization processing on the propellant grain surface point cloud data by using a Z-score standardization method to obtain the propellant grain mechanical response value feature set; performing initial density clustering and representative sample screening on each response value feature in the propellant grain mechanical response value feature set by using a MiniBatchKMeans algorithm to obtain the sample set of each response value feature in the propellant grain mechanical response value feature set.

3. The hybrid clustering based propellant grain measurement point optimization method of claim 2, wherein, the step of performing initial density clustering and representative sample screening on each response value feature in the propellant grain mechanical response value feature set by using the MiniBatchKMeans algorithm comprises: performing initial density clustering and dimensionality reduction on the propellant grain surface point cloud data by using the MiniBatchKMeans algorithm, and dividing the propellant grain surface point cloud data into a plurality of initial density clustering clusters, each initial density clustering cluster containing at least 15 data points; respectively selecting the most significant points of mechanical response from each initial density clustering cluster as sample points to obtain the sample set of each response value feature in the propellant grain mechanical response value feature set.

4. The hybrid clustering based propellant grain measurement point optimization method of claim 1, wherein, The k-medoids++ algorithm is used to initialize the cluster center point, comprising: calculating the strain response value of each sample point in the sample set, and sorting according to the response value size; the sample point with the largest response value is taken as the cluster center point.

5. The hybrid clustering based propellant grain measurement point optimization method of claim 1, wherein, The method further comprises: using the Matplotlib tool to import the optimal measuring point combination on the grain column surface, combining the grain column surface point cloud data, and generating a three-dimensional measuring point distribution map and a cluster center distribution map with mechanical response characteristic markers in a three-dimensional rectangular coordinate system; based on the optimal measuring point combination on the grain column surface, a response value histogram is drawn; based on the three-dimensional measuring point distribution map, the spatial distribution characteristics of the optimal measuring point combination on the grain column surface are obtained, based on the cluster center distribution map, the mechanical response sensitivity of the optimal measuring point combination on the grain column surface is obtained, and based on the response value histogram, the response value coverage rate of the optimal measuring point combination on the grain column surface is obtained; based on the spatial distribution characteristics, the mechanical response sensitivity and the response value coverage rate, the response value distribution of the optimal measuring point combination on the grain column surface is verified by numerical analysis and simulation experiment.

6. A hybrid clustering based propellant grain measurement point optimization device, which is implemented by using the hybrid clustering based propellant grain measurement point optimization method according to any one of claims 1 to 5, characterized in that, The device comprises: an acquisition module for acquiring grain column surface point cloud data, obtaining a grain column mechanical response value feature set based on the grain column surface point cloud data, the grain column mechanical response value feature set comprising grain column strain, grain column shell strain and grain column interface stress; a sampling module for performing initial density clustering and representative sample screening on the grain column surface point cloud data to obtain a sample set of each response value feature in the grain column mechanical response value feature set; a generation module for inputting the sample set of each response value feature in the grain column mechanical response value feature set into a hybrid distance measurement model respectively to obtain an optimal measuring point combination on the grain column surface, the optimal measuring point combination on the grain column surface comprising a grain column strain optimal measuring point, a grain column shell strain optimal measuring point and a grain column interface stress optimal measuring point.

7. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to realize the hybrid clustering-based grain column measuring point optimization method according to any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the hybrid clustering-based grain column measuring point optimization method according to any one of claims 1 to 5.

Images