Method for monitoring deformation of a ship shafting installation based on three-dimensional scanning
Patent Information
- Application Number
- CN202611332342.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-31
- Publication Date
- 2026-11-03
AI Technical Summary
Existing monitoring methods cannot accurately trace the independent contribution of various process factors to deformation during the installation of ship shafting, resulting in a lack of targeted process intervention measures, low efficiency, and the risk of misjudgment.
A three-dimensional scanning-based method is adopted. By registering and Fourier series expansion of the three-dimensional scanning point cloud data during the shaft installation process, the deformation field is decomposed into multi-order harmonic components. The independent contribution components and their proportions of each process factor are obtained by non-negative least squares decomposition, and a deformation monitoring report is generated.
It enables accurate tracing of shaft deformation, identifies dominant deformation factors and their spatial distribution areas, reduces the risk of misjudgment relying on human experience, and improves the pertinence and efficiency of process intervention.
Smart Images
Figure CN123028527A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of shipbuilding technology, and more specifically, to a method for monitoring deformation of ship shafting installation based on three-dimensional scanning. Background Technology
[0002] The ship's shafting system undergoes several construction phases during installation, including bearing press-fitting, dock-mounted welding, and ship launching. Deformation at each stage is caused by the combined effects of various process factors: bearing press-fitting stress causes localized radial shrinkage deformation near the bearing housing; welding thermal deformation is distributed in a band along the weld seam and accompanied by angular deformation; and changes in dock-mounted load and launching buoyancy introduce overall bending deformation.
[0003] For the above installation process, existing monitoring methods use measurement to obtain the total axial offset and the total roundness change of the cross section at each stage, thereby assessing the overall deformation state of the shaft system.
[0004] However, existing monitoring methods can only obtain the total deformation at each stage, and cannot decompose the total deformation into independent contribution components of various process factors such as bearing press-fit stress, welding thermal deformation, load changes, and buoyancy conditions. When shaft deformation exceeds the limit, it is difficult to quickly locate the dominant deformation factor, process intervention measures lack specificity, and the formulation of repair plans relies on manual experience judgment, resulting in technical problems such as low efficiency and risk of misjudgment. Summary of the Invention
[0005] This invention provides a method for monitoring deformation during ship shafting installation based on three-dimensional scanning, solving the technical problem in related technologies that it is difficult to accurately trace the independent contribution of various process factors to deformation during shafting installation.
[0006] This invention discloses a method for monitoring the deformation of ship shafting installation based on three-dimensional scanning, including the following steps: performing registration processing on the three-dimensional scanning point cloud data of the shafting collected at each construction working condition node, and calculating the total deformation field data of the shafting relative to the initial reference state at each working condition node; Perform adjacent stage difference operations on the total deformation field data of the shaft system at each stage to obtain the incremental deformation field data between adjacent working condition nodes; Fourier series expansion is performed on each incremental deformation field data along the circumferential direction of the shaft system cross-section to obtain multi-order harmonic component data; The harmonic component feature vectors of each incremental deformation field are compared with the feature templates of each process factor in the pre-stored shaft process factor deformation mode feature library to obtain the matching degree score matrix. Based on the matching degree score matrix, non-negative least squares decomposition is performed on each incremental deformation field to obtain the independent contribution components and their proportion data of each process factor. Based on the independent contribution components, a cumulative superposition operation is performed to identify the dominant deformation factors and generate a shaft installation deformation monitoring report.
[0007] Furthermore, filtering, denoising, and multi-station stitching registration processing are performed on the 3D scanning point cloud data of the shaft system collected at each construction condition node. The center coordinate sequence and roundness parameter sequence of each cross section are extracted along the shaft system axis, and the axial offset distribution and cross-sectional roundness deformation increment distribution of each working condition node relative to the initial reference state are calculated. Among them, the initial reference state is the 3D scanning data of the shaft system components in the free state before bearing press-fitting; the multi-station stitching registration processing refers to the preset common target point on the shaft system surface, and the rigid body transformation registration of the local point cloud data collected by each measurement station based on the target point coordinates, so as to unify the multi-station point cloud data into the same coordinate system.
[0008] Furthermore, the incremental radial deformation of each cross section is expanded along the circumferential direction into a linear superposition of multiple harmonic components, and the amplitude and phase angle of each harmonic component are obtained. Among them, the zero-order component corresponds to the uniform radial expansion and contraction of the cross section, the first-order component corresponds to the overall eccentric displacement of the cross section, the second-order component corresponds to elliptic deformation, and the higher-order components correspond to local irregular deformation; the highest order of expansion is determined according to the number of sampling points in the circumferential direction of the cross section, and the highest order does not exceed half of the number of sampling points.
[0009] Furthermore, the amplitude values of each harmonic component are extracted from each cross section of each incremental deformation field, and arranged and combined along the axial direction according to the cross section order to form a feature vector; feature template vectors of each process factor are extracted from the feature library. The feature template vectors are composed of the standard harmonic amplitude values of the corresponding process factor distributed along the axial direction of each cross section, and the feature template vectors have the same dimension as the feature vectors; cosine similarity calculation is performed on the feature vectors and each feature template vector to obtain the matching degree score matrix between each incremental deformation field and each process factor feature template.
[0010] Furthermore, the characteristic harmonic orders and axial attenuation laws corresponding to various process factors recorded in the feature library include: bearing press-fit stress is dominated by the second-order component as the dominant harmonic order, and its amplitude decreases exponentially from the bearing seat position to both sides along the axial direction; welding thermal deformation is dominated by the second-order and third-order components as the dominant harmonic orders, and its amplitude is concentrated within a finite width range on both sides of the weld; the load carried in the dock is dominated by the first-order component as the dominant harmonic order, and its amplitude is continuously and gradually distributed along the axial direction; the change in buoyancy conditions during launching is dominated by the first-order component as the dominant harmonic order, and its amplitude is distributed in a low-frequency sinusoidal fluctuation along the axial direction.
[0011] Furthermore, based on the matching score matrix, process factors with matching scores greater than a preset threshold are selected as candidate factor sets. A template matrix is constructed using the feature template vectors corresponding to the candidate factor sets. Under the constraint that the contribution coefficients of each candidate process factor are non-negative, the contribution coefficient vector is solved to minimize the squared L2 norm between the product of the template matrix and the contribution coefficient vector and the feature vector of the incremental deformation field. Based on the contribution coefficient vector, the ratio of the product of the contribution coefficient of each process factor and the modulus of its corresponding feature template vector to the sum of the corresponding products of all candidate process factors is taken as the contribution proportion of each process factor.
[0012] Furthermore, the preset threshold is determined based on the cosine similarity between the feature templates of each process factor in the feature library, and is set to half of the maximum cosine similarity between any two different process factor feature template vectors in the feature library. Cross-validation is performed on process factors whose matching scores are within a preset range near the preset threshold: the process factor is included in and excluded from the candidate factor set respectively, and the residual norm is calculated for the non-negative least squares decomposition results in both cases. The decomposition scheme with the smaller residual norm is selected as the final result.
[0013] Furthermore, for each process factor, its contribution components at each operating condition node are accumulated sequentially to obtain the cumulative contribution curve of each process factor; multinomial regression fitting is performed on the cumulative contribution curve of each process factor to obtain the trend prediction value of the cumulative deformation of each factor; when the trend prediction value exceeds the graded early warning threshold of the corresponding process factor at the subsequent operating condition node to be executed, the early warning level and the operating condition node number that triggered the early warning are marked in the monitoring report; the process factor with the largest cumulative contribution ratio is identified as the dominant deformation factor, its spatial distribution area coordinates are extracted, and the dominant deformation factor identifier and its spatial distribution area coordinates are correlated with the pre-stored process parameter adjustment strategy library to output targeted process intervention suggestions for the dominant deformation factor; a shaft installation deformation monitoring report containing the shaft offset source analysis results at each stage, the contribution ratio of each process factor, and targeted intervention suggestions is generated.
[0014] Furthermore, it also includes: performing analysis on the residual vector of the non-negative least squares decomposition, where the residual vector is the difference between the incremental deformation field feature vector and the product of the template matrix and the contribution coefficient vector; when the ratio of the norm of the residual vector to the norm of the incremental deformation field feature vector exceeds a preset residual threshold, the incremental deformation field is marked as having unidentified deformation components, and a suggestion to supplement the feature library is marked in the monitoring report.
[0015] This invention discloses a ship shafting installation deformation monitoring system based on three-dimensional scanning, used to execute the above method, including: a point cloud processing module, used to perform registration processing on the shafting three-dimensional scanning point cloud data collected at each construction working condition node, and calculate the total shafting deformation field data of each working condition node relative to the initial reference state; The incremental deformation field calculation module is used to perform differential calculations between adjacent stages on the total deformation field data of the shaft system at each stage, and to obtain the incremental deformation field data between adjacent working condition nodes. The harmonic decomposition module is used to perform Fourier series expansion on the data of each incremental deformation field along the circumferential direction of the shaft system cross section to obtain multi-order harmonic component data; the pattern matching module is used to perform similarity calculation between the feature vector of the harmonic components of each incremental deformation field and the feature template of each process factor in the pre-stored shaft system process factor deformation pattern feature library to obtain the matching degree score matrix. The contribution decomposition module is used to perform non-negative least squares decomposition on each incremental deformation field based on the matching degree score matrix to obtain the independent contribution components and their proportion data of each process factor. The monitoring report generation module is used to perform cumulative superposition calculations based on independent contribution components, identify the dominant deformation factors, and generate a shaft installation deformation monitoring report.
[0016] The beneficial effects of this invention are as follows: This invention decomposes the deformation of the shaft system cross section into multi-order harmonic components by performing Fourier series expansion on the incremental deformation field between each construction node along the circumferential direction of the cross section. This allows the deformation caused by different process factors to exhibit distinguishable characteristic distributions in the harmonic domain. Then, cosine similarity matching and non-negative least squares decomposition are used to obtain the independent contribution components of each process factor. This solves the technical problem that existing monitoring methods cannot decompose the total deformation into independent contribution components of each process factor. It achieves the technical effect of being able to directly identify the dominant deformation factor and its spatial distribution area from the decomposition results, enabling process intervention suggestions to be directed to specific dominant deformation factors, and reducing the risk of misjudgment caused by relying on human experience. Attached Figure Description
[0017] Figure 1 This is a flowchart of a method for monitoring ship shafting installation deformation based on three-dimensional scanning, provided in an embodiment of the present invention. Detailed Implementation
[0018] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.
[0019] At least one embodiment of the present invention discloses a method for monitoring ship shafting installation deformation based on three-dimensional scanning, see [link to relevant documentation]. Figure 1 This includes the following steps: Step S1: Obtain the total deformation field data of the shaft system at each construction condition node.
[0020] The 3D scanning point cloud data of the shaft system collected at each construction condition node is filtered, denoised, and multi-station stitching and registration are processed. The center coordinate sequence and roundness parameter sequence of each cross section are extracted along the shaft system axis. The axial offset distribution and cross section roundness deformation increment distribution of each construction condition node relative to the initial reference state are calculated to obtain the total deformation field data of the shaft system at each stage.
[0021] It should be noted that the above-mentioned initial reference state is the three-dimensional scanning data of the shaft system components in their free state before the bearing is press-fitted. The center coordinates and roundness parameters of each section of the shaft system in the initial reference state serve as the reference zero points for the calculation of deformation in subsequent stages.
[0022] It should be noted that the above-mentioned multi-station splicing and registration process refers to setting a common target point on the surface of the shaft system, performing rigid body transformation registration on the local point cloud data collected by each measurement station based on the target point coordinates, unifying the point cloud data of multiple stations into the same coordinate system, and obtaining complete point cloud data of the shaft system surface.
[0023] Step S2: Calculate the incremental deformation field data between adjacent working condition nodes.
[0024] Perform adjacent-stage difference operations on the total deformation field data of the shaft system at each stage to obtain the incremental deformation field data between adjacent working condition nodes. Among these... Incremental deformation field of working nodes For the first Total deformation field of the stage With the Total deformation field of the stage Cross-sectional and point-by-point differences: in, For the first Total deformation field data of the shaft system at the working node. For the first Total deformation field data of the shaft system at the working node. For the first The incremental deformation field at the working condition node. The incremental deformation field reflects the new shaft deformation introduced by specific process operations at this working condition stage.
[0025] Step S3: Perform a Fourier series expansion on the incremental deformation field to obtain multi-harmonic component data.
[0026] Fourier series expansions were performed on the incremental deformation field data along the circumferential direction of the shaft system cross-section, decomposing the radial deformation of the cross-section into multi-order harmonic components. For the shaft system... Incremental radial deformation of each cross section Its Fourier series expansion is: in, The angle is in the circumferential direction of the cross section. The zeroth-order component amplitude, Numbering the harmonic order, and The first The cosine and sine coefficients of the first harmonic component. This represents the highest order of expansion. The amplitudes of each component are... and phase angle They are respectively: in, For the first The amplitude of the first harmonic component, For the first Phase angle of the first harmonic component.
[0027] Obtain the amplitude and phase angle data of the multi-harmonic components of each order of incremental deformation field.
[0028] It should be noted that the zero-order component corresponds to uniform radial expansion and contraction of the cross section, reflecting the overall change in shaft diameter; the first-order component corresponds to the overall eccentric displacement of the cross section, reflecting the axial offset; the second-order component corresponds to elliptic deformation, reflecting uneven pressing or welding shrinkage; and the higher-order components correspond to local irregular deformation, reflecting local heat input or local load concentration.
[0029] It should be noted that the highest order of the above expansion The number of sampling points is determined based on the cross-sectional sampling points. When sampling is uniformly performed along the circumferential direction of the cross-section... At point 1, The value does not exceed This satisfies the requirements of the Nyquist sampling theorem.
[0030] Step S4: Perform variant pattern matching based on the feature library to obtain the matching degree score matrix.
[0031] The model matching is performed based on a pre-stored feature library of deformation patterns of shaft system process factors. The feature library of deformation patterns of shaft system process factors records the characteristic harmonic order, the attenuation law of characteristic amplitude along the axial direction, and the characteristic spatial distribution range corresponding to typical process factors such as bearing press-fit stress, welding thermal deformation, dock load, and changes in buoyancy conditions during launching.
[0032] Step S4 may specifically include: Step S401: For each cross section of each incremental deformation field, extract the amplitude of each harmonic component, and arrange and combine them along the axial direction according to the cross section order to form an eigenvector. eigenvectors By the The harmonic amplitudes of all sections in the incremental deformation field of the working node are arranged in axial order, and the dimension is... ,in The total number of cross sections, This is the highest order of harmonic expansion.
[0033] Step S402: Extract feature template vectors for each process factor from the feature library. ,in Number the process factors. Each feature template vector. The dimension is also It is composed of the standard harmonic amplitude values corresponding to the process factors distributed along the axial direction of each cross section.
[0034] Step S403: Convert the feature vector With each feature template vector Perform cosine similarity calculation to obtain the matching score. ,in For the first Incremental deformation field of working node and the first Matching score of process factor feature template.
[0035] Obtain the matching degree score matrix between each incremental deformation field and the feature template of each shaft system process factor. .
[0036] It should be noted that the specific definitions of the characteristic harmonic order and axial attenuation law corresponding to each process factor in the above feature library are as follows: the bearing press-fit stress is dominated by the second-order component as the dominant harmonic order, and its amplitude decreases exponentially from the bearing seat position to both sides along the axial direction; the welding thermal deformation is dominated by the second-order and third-order components as the dominant harmonic order, and its amplitude is concentrated within a finite width range on both sides of the weld; the load mounted in the dock is dominated by the first-order component as the dominant harmonic order, and its amplitude is continuously and slowly distributed along the axial direction; the change in buoyancy conditions during launching is dominated by the first-order component as the dominant harmonic order, and its amplitude is distributed in a low-frequency sinusoidal fluctuation along the axial direction.
[0037] Step S5: Perform non-negative least squares decomposition based on the matching degree score matrix to obtain the contribution components of each process factor.
[0038] Based on the matching degree score matrix Non-negative least squares decomposition is performed on each incremental deformation field. The measured incremental deformation field is approximated by a linear combination of characteristic templates of each shaft system process factor, and the contribution coefficients of each process factor are solved.
[0039] Step S5 may specifically include: Step S501: For the first The incremental deformation field of the working node is selected based on the matching degree score matrix. Process factors exceeding a preset threshold are used as a set of candidate factors.
[0040] Furthermore, the aforementioned preset threshold is determined based on the minimum cosine similarity between the feature templates of each process factor in the feature library. The value is half of the maximum cosine similarity between any two different process factor feature template vectors in the feature library, so that the screening threshold can distinguish between actual matching and inherent correlation between templates.
[0041] Step S502: Construct a template matrix using the feature template vectors corresponding to the candidate factor set. Solve the nonnegative least squares optimization problem: in, This represents the contribution coefficient vector for each candidate process factor. It is a matrix composed of candidate process factor feature template vectors arranged column by column. For the first The eigenvectors of the incremental deformation field at the working node. The square of the L2 norm of the vector. Nonnegativity constraint. Ensure that the contribution of each process factor is a physically reasonable non-negative value.
[0042] Furthermore, the above non-negative least squares optimization problem is solved using the activity set method: All contribution coefficients are initialized to zero. Iteratively, the variable that maximizes the negative gradient component of the objective function is moved from the zero-constraint set to the free variable set. Unconstrained least squares solutions are performed on the free variable set. When a negative value appears in the solution, the corresponding variable is moved back to the zero-constraint set and the solution is repeated. The iteration terminates when all solutions for all free variables are non-negative and the gradient components corresponding to all variables in the zero-constraint set are non-negative, and the final contribution coefficient vector is output. .
[0043] Step S503: Based on the contribution coefficient vector Calculate the independent contribution components and their proportions for each process factor: in, For the first Contribution coefficient of process factors For the first The modulus of the feature template vector of process factors The total number of candidate process factors. For the summation index of candidate process factors, For the first Contribution coefficient of process factors For the first The modulus of the feature template vector of process factors For the first The contribution percentage of process factors.
[0044] Obtain the independent contribution components and their proportions of each factor in the incremental deformation of bearing press stress, welding thermal deformation, load and buoyancy conditions during each construction stage.
[0045] In this embodiment of the application, in order to avoid process factors whose matching scores are close to the threshold boundary being incorrectly excluded or included, based on step S501, cross-validation is also performed on process factors whose matching scores are within a preset range near the threshold: the boundary factor is included in and excluded from the candidate set respectively, the residual norm is calculated for the non-negative least squares decomposition results in both cases, and the decomposition scheme with the smaller residual norm is selected as the final result.
[0046] Step S6: Generate a shaft installation deformation monitoring report.
[0047] The independent contribution components of each process factor are cumulatively superimposed along the construction condition node sequence to obtain the cumulative deformation contribution curve of each process factor during the entire shaft system installation process.
[0048] Step S6 may specifically include: Step S601: For each process factor, sum up its contribution components at each operating node in chronological order to obtain the cumulative contribution curve of each process factor. ,in For the first Process factors in the first The contribution coefficient of the working node, For the summation index of the working condition nodes, This is the node number for the current operating condition.
[0049] Step S602: Identify the process factor with the largest cumulative contribution as the dominant deformation factor and extract its spatial distribution area coordinates.
[0050] Step S603: Perform a correlation query between the identifier of the dominant deformation factor and its spatial distribution area coordinates and the pre-stored shaft system process parameter adjustment strategy library, and output targeted process intervention suggestions for the dominant deformation factor.
[0051] Step S604: Generate a shaft installation deformation monitoring report that includes the source analysis results of shaft offset at each stage, the contribution ratio of each process factor, and directional intervention suggestions.
[0052] It should be noted that the directional intervention strategies recorded in the above-mentioned process parameter adjustment strategy library include: when the dominant deformation factor is bearing press-fit stress, the corresponding intervention strategy is to adjust the press-fit interference or add an annealing stress relief process; when the dominant deformation factor is welding hot deformation, the corresponding intervention strategy is to adjust the welding sequence or add a reverse deformation preset amount; when the dominant deformation factor is the load, the corresponding intervention strategy is to adjust the support point position or add temporary support; when the dominant deformation factor is the change in buoyancy conditions, the corresponding intervention strategy is to adjust the ballast water distribution.
[0053] In this embodiment of the application, in order to identify the changing trend of the cumulative contribution curve of each process factor, based on step S601, a multinomial regression fitting is performed on the cumulative contribution curve of each process factor to obtain the trend prediction value of the cumulative deformation of each factor. When the trend prediction value exceeds the graded early warning threshold of the corresponding process factor at the subsequent working condition node to be executed, the early warning level and the working condition node number that triggered the early warning are marked in the monitoring report.
[0054] Furthermore, the aforementioned graded early warning thresholds are determined based on the maximum allowable cumulative deformation of each process factor in the shaft system installation technical specifications, and are divided into three levels: the first-level early warning threshold is 60% of the maximum allowable cumulative deformation, the second-level early warning threshold is 80% of the maximum allowable cumulative deformation, and the third-level early warning threshold is the maximum allowable cumulative deformation itself.
[0055] In this embodiment of the application, in order to reduce the interference of the residual term in the single decomposition result on the monitoring conclusion, based on step S5, the residual vector of the non-negative least squares decomposition is further processed. Perform analysis. When the residual norm With the norm of the eigenvector of the incremental deformation field When the ratio exceeds the preset residual threshold, the incremental deformation field is marked as having unidentified deformation components, and the monitoring report suggests supplementing the feature library or conducting manual verification.
[0056] Furthermore, the aforementioned preset residual threshold is set to 0.2. That is, when the residual energy accounts for more than 20% of the total energy of the incremental deformation field, it is considered that the existing process factor templates in the feature library cannot fully explain the deformation at this stage, and there are deformation components not included in the feature library.
[0057] On the other hand, the present invention also proposes a ship shafting installation deformation monitoring system based on three-dimensional scanning, comprising: The point cloud processing module is used to perform filtering, noise reduction, and multi-station stitching and registration processing on the 3D scanning point cloud data of the shaft system collected from each construction condition node. It extracts the center coordinate sequence and roundness parameter sequence of each cross section along the shaft system axis, calculates the axial offset distribution and cross section roundness deformation increment distribution of each working condition node relative to the initial reference state, and obtains the total deformation field data of the shaft system at each stage.
[0058] The incremental deformation field calculation module is used to perform differential calculations between adjacent stages on the total deformation field data of the shaft system at each stage, and to obtain the incremental deformation field data between adjacent working condition nodes.
[0059] The harmonic decomposition module is used to perform Fourier series expansion on each incremental deformation field data along the circumferential direction of the axis cross section to obtain the amplitude and phase angle data of the multi-order harmonic components of each incremental deformation field.
[0060] The pattern matching module is used to perform cosine similarity calculation between the harmonic component feature vectors of each incremental deformation field and the feature templates of each process factor in the pre-stored shaft process factor deformation pattern feature library, and obtain the matching degree score matrix.
[0061] The contribution decomposition module is used to perform non-negative least squares decomposition on each incremental deformation field based on the matching degree score matrix to obtain the independent contribution components and their proportion data of each process factor.
[0062] The monitoring report generation module is used to perform cumulative superposition calculations on the independent contribution components of each process factor, identify the dominant deformation factors, and correlate them with targeted process intervention suggestions to generate a shaft installation deformation monitoring report.
[0063] By way of example, the system proposed in this invention may also include various features and combinations thereof in the method embodiments, which will not be elaborated here.
[0064] This invention decomposes the shaft system cross-sectional deformation into multi-order harmonic components with clear physical correspondences by performing a Fourier series expansion of the incremental deformation field along the circumferential direction of the cross-section at each construction stage node. This allows the deformation caused by different process factors, such as bearing press stress, welding thermal deformation, load variations, and buoyancy changes, to exhibit distinguishable characteristic distributions in the harmonic domain. Based on this, cosine similarity matching and non-negative least squares decomposition are used to quantitatively decompose the incremental deformation field at each stage into independent contribution components of each process factor. Therefore, when the shaft system deformation exceeds the limit, the dominant deformation factor with the largest cumulative contribution and its spatial distribution area can be directly identified from the decomposition results. This overcomes the limitation of existing monitoring methods that can only obtain the total deformation but cannot distinguish the independent contributions of each process factor, enabling subsequent process intervention suggestions to target specific dominant deformation factors and reducing the risk of misjudgment caused by relying on manual experience.
[0065] The following application examples are proposed: The intermediate shaft section of a bulk carrier (Type A) at a shipyard is approximately 8 meters long. Four monitoring cross-sections (sections 1 to 4) are evenly distributed along the shaft system's axial direction, with a spacing of approximately 2 meters between the sections. The shaft system sequentially undergoes three construction phases: bearing press-fitting (phase 1), in-dock welding (phase 2), and ship launching (phase 3). A 3D laser scanner performs a full-circumference scan of the shaft system after each phase, sampling 36 points evenly along the circumference of each cross-section. The bearing housing is located near section 2, and the weld is located near section 3. The implementation process of each core step in this scenario is described below.
[0066] Before bearing press-fitting, a datum scan is performed on the intermediate shaft section in a free state to obtain initial datum state data. The center coordinates and roundness parameters of each cross-section are used as reference zero points for subsequent deformation calculations. After the three working condition nodes are completed, a three-dimensional scan of the shaft system is performed. The local point clouds collected by each measuring station are registered to the same coordinate system based on preset target points. The center coordinates and roundness parameters of four cross-sections are extracted along the axial direction, and the axial offset and cross-sectional roundness deformation increment of each working condition node relative to the initial datum state are calculated.
[0067] Table 1. Total Deformation Field Data of Shaft System at Working Condition Node 1 Table 2. Total Deformation Field Data of Shaft System at Working Condition Node 2 and Working Condition Node 3 A cross-sectional difference operation is performed on the total deformation field data of adjacent working node. The incremental deformation field of working node 1 is the difference between the total deformation field of working node 1 and the initial reference state (zero value), that is, the total deformation field of working node 1 itself. The incremental deformation field of working node 2 is the total deformation field of working node 2 minus the total deformation field of working node 1, reflecting the new deformation introduced during the welding stage. The incremental deformation field of working node 3 is the total deformation field of working node 3 minus the total deformation field of working node 2, reflecting the new deformation introduced during the launching stage.
[0068] Table 3 Incremental deformation field data between nodes under various working conditions For each cross-section in each incremental deformation field, the radial deformation data are expanded along the circumferential direction using a Fourier series expansion. 36 points are sampled along the circumferential direction for each cross-section, and the highest order expansion is performed. Take the 18th order. Extract the amplitude of each harmonic component. and phase angle Taking section 3 in the incremental deformation field of node 2 as an example, this section is located near the weld. After unfolding, the amplitudes of the second and third order components are significant, while the amplitudes of the other higher order components are relatively small.
[0069] Table 4. Harmonic component data of section 3 in the incremental deformation field at node 2 under working condition. For each of the four cross sections of the incremental deformation field, the amplitudes of each harmonic are arranged and combined in axial order to form eigenvectors. The dimension is 72 (4 sections multiplied by 18). Feature template vectors for four types of process factors—bearing press-fit stress, welding thermal deformation, load, and buoyancy condition changes—are extracted from the feature library. Each feature vector is compared with the cosine similarity of the feature vectors to obtain the matching score matrix.
[0070] Table 5 Matching score matrix The maximum cosine similarity between any two feature template vectors of different process factors in the feature library is 0.30, and the screening threshold is half of this, i.e., 0.15. For the incremental deformation field of working condition node 2, the candidate factors with a matching score greater than 0.15 are bearing press-fit stress (0.18) and welding thermal deformation (0.91). Among them, the bearing press-fit stress score of 0.18 is close to the threshold, so cross-validation is performed on it: non-negative least squares decomposition is performed to include and exclude this factor respectively, the residual norm is compared, and finally the bearing press-fit stress is included in the candidate set.
[0071] The template matrix is constructed using the feature template vectors corresponding to the candidate factors. For eigenvectors Solve the nonnegative least squares optimization problem to obtain the contribution coefficients of each process factor. Calculate the contribution percentage of each process factor based on the contribution coefficients and the feature template vector magnitude. .
[0072] Table 6. Decomposition results of the incremental deformation field at node 2 under working condition For the incremental deformation field at node 2 of the working condition, the ratio of the residual norm to the eigenvector norm of the non-negative least squares decomposition is 0.11, which is lower than the preset residual threshold of 0.20, indicating that the templates in the feature library can fully explain the deformation at this stage.
[0073] The contributions of each process factor at each operating node are accumulated over time to obtain a cumulative contribution curve. The process factor with the largest cumulative contribution is identified as the dominant deformation factor, and its spatial distribution region is extracted. Polynomial regression is performed on the cumulative contribution curve to predict the cumulative deformation trend at subsequent operating nodes.
[0074] Table 7 Cumulative Contribution and Early Warning Status of Each Process Factor Welding thermal deformation contributed the largest cumulative percentage and was identified as the dominant deformation factor, with its spatial distribution concentrated in the weld area near section 3. When the predicted trend reached 60% of the maximum allowable cumulative deformation, a Level 1 warning was triggered. After consulting the process parameter adjustment strategy library, targeted intervention suggestions were output, including adjusting the welding sequence or increasing the anti-deformation preset amount. The final monitoring report included source analysis of axial offset at each stage, the distribution of the contribution percentage of each process factor, and targeted intervention suggestions for welding thermal deformation.
[0075] In the above implementation process, the data starts from the point cloud of the initial baseline state, and the total deformation field is obtained through 3D scanning of each working condition node. The incremental deformation field is extracted by differential extraction between adjacent stages, and then transformed into a harmonic domain feature vector through Fourier series expansion. After matching with the feature library template, the quantitative contribution of each process factor is obtained through non-negative least squares decomposition. Finally, the data is accumulated along the time series to form a trend curve and trigger an early warning. The output data of each step is directly used as the input data of the next step. The entire data flow gradually transforms from the original point cloud in the spatial domain into frequency domain feature expression, and then into the quantitative contribution ratio of each process factor, maintaining logical coherence and causal relationship.
[0076] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.
Claims
1. A method for monitoring ship shafting installation deformation based on three-dimensional scanning, characterized in that, Includes the following steps: Registration processing is performed on the 3D scanning point cloud data of the shaft system collected at each construction condition node, and the total deformation field data of the shaft system relative to the initial reference state is calculated for each condition node. Perform adjacent stage difference operations on the total deformation field data of the shaft system at each stage to obtain the incremental deformation field data between adjacent working condition nodes; Fourier series expansion is performed on each incremental deformation field data along the circumferential direction of the shaft system cross-section to obtain multi-order harmonic component data; The harmonic component feature vectors of each incremental deformation field are compared with the feature templates of each process factor in the pre-stored shaft process factor deformation mode feature library to obtain the matching degree score matrix. Based on the matching degree score matrix, non-negative least squares decomposition is performed on each incremental deformation field to obtain the independent contribution components and their proportion data of each process factor. Based on the independent contribution components, a cumulative superposition operation is performed to identify the dominant deformation factors and generate a shaft installation deformation monitoring report.
2. The method for monitoring ship shafting installation deformation based on three-dimensional scanning according to claim 1, characterized in that, The registration process is performed on the 3D scanning point cloud data of the shaft system collected at each construction condition node, and the total deformation field data of the shaft system relative to the initial reference state at each condition node is calculated, including: The 3D scanning point cloud data of the shaft system collected at each construction condition node is filtered, denoised and multi-station stitched and registered. The center coordinate sequence and roundness parameter sequence of each cross section are extracted along the shaft system axis. The axial offset distribution and cross section roundness deformation increment distribution of each construction condition node relative to the initial reference state are calculated. The initial reference state refers to the three-dimensional scanning data of the shaft system components in their free state before bearing press-fitting; the multi-station splicing registration process refers to setting a common target point on the surface of the shaft system, performing rigid body transformation registration on the local point cloud data collected by each measurement station based on the target point coordinates, and unifying the multi-station point cloud data into the same coordinate system.
3. The method for monitoring ship shafting installation deformation based on three-dimensional scanning according to claim 1, characterized in that, The process of performing a Fourier series expansion on each incremental deformation field data along the circumferential direction of the shaft system cross-section to obtain multi-order harmonic component data includes: The incremental radial deformation of each cross section is expanded along the circumferential direction into a linear superposition of multiple harmonic components, and the amplitude and phase angle of each harmonic component are obtained. Among them, the zero-order component corresponds to the uniform radial expansion and contraction of the cross section, the first-order component corresponds to the overall eccentric displacement of the cross section, the second-order component corresponds to the elliptic deformation, and the higher-order components correspond to the local irregular deformation; the highest order of expansion is determined according to the number of sampling points in the circumferential direction of the cross section, and the highest order does not exceed half of the number of sampling points.
4. The method for monitoring ship shafting installation deformation based on three-dimensional scanning according to claim 1, characterized in that, The step of performing similarity calculations between the harmonic component feature vectors of each incremental deformation field and the feature templates of each process factor in the pre-stored shaft process factor deformation mode feature library to obtain a matching score matrix includes: The amplitudes of each harmonic component are extracted from each cross section of each incremental deformation field, and arranged and combined along the axial direction according to the cross section order to form an eigenvector; The feature template vectors of each process factor are extracted from the feature library. The feature template vectors are composed of the standard harmonic amplitude values of the corresponding process factor distributed along the axial direction of each cross section. The feature template vectors have the same dimension as the feature vectors. The cosine similarity is calculated between the feature vector and each feature template vector to obtain the matching degree score matrix between each incremental deformation field and each process factor feature template.
5. The method for monitoring ship shafting installation deformation based on three-dimensional scanning according to claim 4, characterized in that, The characteristic harmonic order and axial attenuation law corresponding to each process factor recorded in the feature library include: The bearing press-fit stress is dominated by the second-order harmonic order, and its amplitude decreases exponentially from the bearing housing to both sides along the axial direction; the welding thermal deformation is dominated by the second and third-order harmonic orders, and its amplitude is concentrated within a finite width range on both sides of the weld; the load carried in the dock is dominated by the first-order harmonic order, and its amplitude is continuously and gradually distributed along the axial direction; the change in buoyancy conditions during launching is dominated by the first-order harmonic order, and its amplitude is distributed in a low-frequency sinusoidal oscillation along the axial direction.
6. The method for monitoring ship shafting installation deformation based on three-dimensional scanning according to claim 1, characterized in that, The non-negative least squares decomposition of each incremental deformation field based on the matching degree score matrix is performed to obtain the independent contribution components and their proportions of each process factor, including: Based on the matching score matrix, process factors with matching scores greater than a preset threshold are selected as a set of candidate factors; A template matrix is constructed using the feature template vectors corresponding to the candidate factor set. Under the constraint that the contribution coefficients of each candidate process factor are non-negative, the contribution coefficient vector is solved to minimize the squared L2 norm between the product of the template matrix and the contribution coefficient vector and the feature vector of the incremental deformation field. Based on the contribution coefficient vector, the ratio of the product of the contribution coefficient of each process factor and the magnitude of its corresponding feature template vector to the sum of the products of all candidate process factors is taken as the contribution percentage of each process factor.
7. The method for monitoring ship shafting installation deformation based on three-dimensional scanning according to claim 6, characterized in that, The preset threshold is determined based on the cosine similarity between the feature templates of each process factor in the feature library, and is set to half of the maximum cosine similarity between any two different process factor feature template vectors in the feature library. Cross-validation is performed on process factors whose matching scores are within a preset range near the preset threshold: the process factor is included in and excluded from the candidate factor set respectively, the residual norm is calculated for the non-negative least squares decomposition results in both cases, and the decomposition scheme with the smaller residual norm is selected as the final result.
8. The method for monitoring ship shafting installation deformation based on three-dimensional scanning according to claim 1, characterized in that, The step of performing cumulative superposition calculations based on the independent contribution components, identifying the dominant deformation factors, and generating a shaft installation deformation monitoring report includes: For each process factor, its contribution at each operating point is accumulated over time to obtain the cumulative contribution curve of each process factor. Perform polynomial regression fitting on the cumulative contribution curves of each process factor to obtain the trend prediction value of the cumulative deformation of each factor; when the trend prediction value exceeds the graded early warning threshold of the corresponding process factor at the subsequent working condition node to be executed, mark the early warning level and the working condition node number that triggered the early warning in the monitoring report. The process factor with the largest cumulative contribution is identified as the dominant deformation factor. Its spatial distribution area coordinates are extracted. The dominant deformation factor identifier and its spatial distribution area coordinates are correlated with the pre-stored process parameter adjustment strategy library and output targeted process intervention suggestions for the dominant deformation factor. Generate a shaft installation deformation monitoring report that includes the source analysis results of shaft offset at each stage, the contribution ratio of each process factor, and targeted intervention suggestions.
9. The method for monitoring ship shafting installation deformation based on three-dimensional scanning according to claim 6, characterized in that, Also includes: Analysis is performed on the residual vector of the non-negative least squares decomposition, where the residual vector is the difference between the incremental deformation field feature vector and the product of the template matrix and the contribution coefficient vector; When the ratio of the norm of the residual vector to the norm of the feature vector of the incremental deformation field exceeds a preset residual threshold, the incremental deformation field is marked as having unidentified deformation components, and a suggestion to supplement the feature library is marked in the monitoring report.
10. A ship shafting installation deformation monitoring system based on three-dimensional scanning, used to execute the ship shafting installation deformation monitoring method based on three-dimensional scanning as described in any one of claims 1 to 9, characterized in that, include: The point cloud processing module is used to perform registration processing on the 3D scanning point cloud data of the shaft system collected from each construction condition node, and to calculate the total deformation field data of the shaft system relative to the initial reference state for each condition node. The incremental deformation field calculation module is used to perform differential calculations between adjacent stages on the total deformation field data of the shaft system at each stage, and to obtain the incremental deformation field data between adjacent working condition nodes. The harmonic decomposition module is used to perform Fourier series expansion on each incremental deformation field data along the circumferential direction of the shaft system cross section to obtain multi-order harmonic component data. The pattern matching module is used to perform similarity calculation between the harmonic component feature vectors of each incremental deformation field and the feature templates of each process factor in the pre-stored shaft process factor deformation pattern feature library, and obtain the matching degree score matrix. The contribution decomposition module is used to perform non-negative least squares decomposition on each incremental deformation field based on the matching degree score matrix to obtain the independent contribution components and their proportion data of each process factor. The monitoring report generation module is used to perform cumulative superposition calculations based on the independent contribution components, identify the dominant deformation factors, and generate a shaft installation deformation monitoring report.