Ship product structure tree node matching method and device and electronic equipment
By integrating features such as substring inclusion relationship, LCS overlap rate, and relative edit distance, and combining them with a neural network model, the problem of node naming differences in the ship product structure tree was solved, achieving efficient and accurate node matching.
Patent Information
- Application Number
- CN202511709395.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-20
- Publication Date
- 2026-02-17
AI Technical Summary
Existing string similarity algorithms cannot effectively identify significant differences in the naming of nodes in the ship product structure tree, and lack the ability to adapt to knowledge in the ship domain, resulting in low matching accuracy.
By employing features such as substring inclusion relationship, longest common subsequence overlap rate, and relative edit distance, combined with a pre-trained neural network model, the system automatically learns ship-specific weights for node matching.
It significantly reduces the workload of manual comparison, improves matching efficiency and accuracy, reduces human and time costs, and adapts to the language habits of enterprise engineers.
Smart Images
Figure CN121542480A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of ship data processing, and more particularly to a ship product structure tree node matching method and device and electronic equipment. BACKGROUND
[0002] In a ship design and manufacturing enterprise, with the deepening of informatization construction, a plurality of information management systems such as product data management, manufacturing execution system, and maintenance management system have been developed around the core data model of the ship product structure tree (PST). However, due to different construction periods, different focuses, and implementation by different teams, there are great differences in the node naming of the underlying product structure tree. The differences are caused by: 1, different habits of engineers, such as “left engine” and “P-SIDE MAIN ENGINE” which are actually the same node; 2, different levels of description: such as “diesel engine” and “marine high-speed diesel engine”; 3, mixed English and Chinese: such as “radar mast” and “radar mast”.
[0003] At present, when performing system integration and data connection, the node mapping relationship is mainly established by manual comparison. However, the number of nodes of the ship product structure tree is usually thousands to tens of thousands, manual matching is extremely low in efficiency, high in cost, and prone to errors, which has become a bottleneck restricting enterprise data fusion and intelligent manufacturing. The existing general string similarity algorithm (such as edit distance, Jaccard similarity, etc.) cannot understand the characteristics of professional terms in the ship field, especially a single algorithm cannot cope with all types of naming differences, and lacks adaptive ability to the knowledge in the ship field, resulting in low matching accuracy. SUMMARY
[0004] In view of the defects of the prior art, the purpose of the present application is to provide a ship product structure tree node matching method, device and electronic equipment, which aims to solve the problems of great differences in the naming of the nodes of the ship product structure tree and low matching accuracy caused by the lack of adaptive ability to the knowledge in the ship field which cannot be well recognized by the existing string similarity algorithm.
[0005] To achieve the above purpose, in a first aspect, the present application provides a ship product structure tree node matching method, comprising: comparing the to-be-matched node string and each string in the preset node string set character by character respectively to obtain the substring inclusion relationship, the Longest Common Subsequence (LCS) overlap rate and the relative edit distance corresponding to each string in the preset node string set and the to-be-matched node string respectively; inputting the substring inclusion relationship, the LCS coincidence rate and the relative edit distance corresponding to each string in the set of preset node strings and the to-be-matched node string respectively into a pre-trained neural network model to obtain a comprehensive similarity score output by the neural network model corresponding to each string in the set of preset node strings and the to-be-matched node string respectively. obtaining a matching result of the to-be-matched node string based on the comprehensive similarity score corresponding to each string in the set of preset node strings and the to-be-matched node string respectively.
[0006] The application fuses three complementary features of substring inclusion relationship, LCS coincidence rate and relative edit distance, wherein the substring inclusion relationship is used to capture the relationship of abbreviations and abbreviations, the LCS coincidence rate is not sensitive to word sequence changes and can capture the consistency of core words, the relative edit distance is normalized at the set level and can better handle the difference in string length and the uneven distribution of the set and reflect the relative similarity degree of the string in the whole, and the powerful nonlinear fitting capability of the neural network model is used to automatically learn the specific weight in the ship field, so that the complex naming rules in the ship field can be accurately captured, and thousands of node matches can be processed at one time, which can significantly reduce the workload of manual comparison, improve the matching efficiency, reduce the labor cost and time cost, and improve the accuracy of node matching.
[0007] According to the ship product structure tree node matching method provided in the application, the neural network model is a multilayer perceptron composed of an input layer, a hidden layer and an output layer, the input layer includes three nodes for inputting the substring inclusion relationship, the LCS coincidence rate and the relative edit distance respectively, the hidden layer uses a ReLU activation function, and the output layer uses a Sigmoid activation function.
[0008] According to the ship product structure tree node matching method provided in the application, the training process of the neural network model includes: labeling the historical node matching data as a training set; training the neural network model based on the training set by using a back propagation algorithm.
[0009] The application trains the model based on the historical node matching data, so that the matching strategy can be adaptively fitted to the specific language habits of the enterprise engineering personnel, and the training is performed by using the back propagation algorithm, so that the training efficiency is high and the generalization ability is strong.
[0010] According to the ship product structure tree node matching method provided in the application, before the to-be-matched node string and each string in the set of preset node strings are compared character by character, the method further includes: Perform one or more of the following operations on each string in the string to be matched and the preset set of node strings: Convert all uppercase English characters to lowercase; Remove special characters; Standardized spaces.
[0011] This application preprocesses each string in the string to be matched and the preset string set to avoid interference from uppercase and lowercase letters, special characters and spaces in the matching results, thereby improving the accuracy of character matching.
[0012] Secondly, this application provides a device for matching nodes in a ship product structure tree, comprising: The feature extraction module is used to compare the string of the node to be matched with each string in the preset string set character by character to obtain the substring inclusion relationship, LCS overlap rate and relative edit distance of the string of the node to be matched and each string in the preset string set respectively; The scoring module is used to input the substring inclusion relationship, LCS overlap rate and relative edit distance corresponding to each string in the preset node string set into a pre-trained neural network model to obtain the comprehensive similarity score corresponding to each string in the preset node string set output by the neural network model. The matching module is used to obtain the matching result of the node string to be matched based on the comprehensive similarity score corresponding to each string in the preset node string set.
[0013] Thirdly, this application provides an electronic device, comprising: at least one memory for storing a program; and at least one processor for executing the program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to execute the ship product structure tree node matching method described in the first aspect or any possible implementation thereof.
[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to execute the ship product structure tree node matching method described in the first aspect or any possible implementation of the first aspect.
[0015] Fifthly, this application provides a computer program product that, when run on a processor, causes the processor to execute the ship product structure tree node matching method described in the first aspect or any possible implementation of the first aspect.
[0016] It is understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.
[0017] Overall, the technical solutions conceived in this application have the following beneficial effects compared with the prior art: This application integrates three complementary features: substring inclusion relationship, LCS overlap rate, and relative edit distance. The substring inclusion relationship is used to capture relationships such as abbreviations and shorthand. The LCS overlap rate is insensitive to word order changes and can capture the consistency of core words. The relative edit distance is normalized at the set level, which can better handle sets with large differences in string length and uneven distribution and reflect the relative similarity of strings in the whole. Furthermore, it utilizes the powerful nonlinear fitting ability of the neural network model to automatically learn the specific weights of the shipbuilding field, which can accurately capture the complex naming rules of the shipbuilding field and process tens of thousands of node matches at one time. On the one hand, it significantly reduces the workload of manual comparison, improves matching efficiency, and reduces labor and time costs. On the other hand, it can improve the accuracy of node matching. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0019] Figure 1 This is a flowchart illustrating the ship product structure tree node matching method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the algorithm flow provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the neural network model provided in the embodiments of this application; Figure 4 This is a schematic diagram of the model training process provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of the ship product structure tree node matching device provided in the embodiments of this application; Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation
[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0021] In this article, the term "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The symbol " / " in this article indicates that the related objects are in an "or" relationship; for example, A / B means A or B.
[0022] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.
[0023] In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more, for example, multiple processing units means two or more processing units, multiple elements means two or more elements, etc.
[0024] Next, combined Figures 1-4 The method for matching ship product structure tree nodes provided in the embodiments of this application is described.
[0025] Figure 1 This is a flowchart illustrating the ship product structure tree node matching method provided in this application embodiment, as shown below. Figure 1 As shown, the method includes the following steps: Step S100: Compare the string to be matched with each string in the preset string set character by character to obtain the substring inclusion relationship, LCS overlap rate and relative edit distance of the string to be matched and each string in the preset string set respectively. Optionally, the preset node string set can be any set of product structure tree node names in any ship information management system to be integrated, and the node string to be matched can be the product structure tree node name to be matched. For ease of description, the preset node string set is defined as the target set T, where the node string is t, and the node string to be matched is the string s in the source node set S.
[0026] Figure 2 This is a schematic diagram of the algorithm flow provided in the embodiments of this application, such as... Figure 2 As shown, each character in string s is compared one by one with each character in each string t in the target set T to obtain the substring inclusion relationship f_sub, LCS overlap rate f_lcs, and relative edit distance f_red.
[0027] The substring inclusion relation f_sub is a binary feature (0 or 1). If s is a substring of t, or t is a substring of s, then f_sub = 1; otherwise, it is 0. This feature is used to capture abbreviations, shorthand, and other related relationships.
[0028] The LCS overlap rate f_lcs is calculated by taking the length of the longest common subsequence of s and t and then normalizing it: f_lcs = 2×LCS(s,t) / (len(s)+len(t)). This feature is not sensitive to word order changes and can capture the consistency of core words.
[0029] The formula for calculating the relative edit distance f_red is as follows: f_red=1-[ED(s,t)-ED_min(s,T)] / [ED_max(s,T)-ED_min(s,T)+ε], where ε is a very small constant to prevent the denominator from being 0. This feature is normalized at the set level and can better reflect the relative similarity between s and t in the whole.
[0030] Step S110: Input the substring inclusion relationship, LCS overlap rate and relative edit distance corresponding to each string in the target node string and the preset node string set into the pre-trained neural network model to obtain the comprehensive similarity score corresponding to each string in the target node string and the preset node string set output by the neural network model.
[0031] like Figure 2 As shown, the feature vector consisting of three features [f_sub, f_lcs, f_red] is used as the input of the pre-trained neural network model to obtain a comprehensive similarity score Score(s,t) between 0 and 1 from the output of the neural network.
[0032] Optionally, the neural network model can be any neural network model, such as a multilayer perceptron, a convolutional neural network, etc., and this application does not limit it.
[0033] By training a neural network model, it learns the optimal and most complex nonlinear combination weight relationship of three features f_sub, f_lcs, and f_red in the specific task of matching nodes in the ship product structure tree. The trained neural network model is obtained. Then, for each node s in the source set S, s is paired with each node t in the target set T. For each pair (s, t), the three-dimensional features f_sub, f_lcs, and f_red are extracted to construct the feature vector [f_sub, f_lcs, f_red], which is then input into the trained neural network model. The forward propagation is used to calculate the comprehensive similarity score Score(s, t).
[0034] Step S120: Based on the comprehensive similarity score corresponding to each string in the target node string and the preset node string set, obtain the matching result of the target node string.
[0035] Optionally, after obtaining the list of similarity scores between node s and all nodes in the target set T, they can be sorted from high to low according to Score(s,t).
[0036] Optionally, based on actual needs, the highest-ranked node or the top N (e.g., Top 3 or Top 5) nodes can be selected as the recommended matching item for node s to complete the association.
[0037] The ship product structure tree node matching method provided in this application integrates three complementary features: substring inclusion relationship, LCS overlap rate, and relative edit distance. The substring inclusion relationship is used to capture relationships such as abbreviations and shorthand. The LCS overlap rate is insensitive to word order changes and can capture the consistency of core words. The relative edit distance is normalized at the set level, which can better handle sets with large differences in string length and uneven distribution and reflect the relative similarity of strings in the whole. Furthermore, it utilizes the powerful nonlinear fitting ability of neural network models to automatically learn specific weights in the ship domain, which can accurately capture the complex naming rules in the ship domain and can process tens of thousands of node matches at once. On the one hand, it significantly reduces the workload of manual comparison, improves matching efficiency, and reduces labor and time costs. On the other hand, it can improve the accuracy of node matching.
[0038] In some embodiments, the neural network model is a multilayer perceptron, consisting of an input layer, a hidden layer, and an output layer. The input layer includes three nodes, which are used to input substring inclusion relationship, LCS overlap rate, and relative edit distance, respectively. The hidden layer uses the ReLU activation function, and the output layer uses the Sigmoid activation function.
[0039] Figure 3 This is a schematic diagram of the structure of the neural network model provided in the embodiments of this application, such as... Figure 3 As shown, in one embodiment of this application, a multilayer perceptron model is used as the neural network model. Its input layer contains 3 nodes, which are used to input f_sub, f_lcs and f_red respectively. The hidden layer contains 64 nodes and uses the ReLU activation function. The output layer contains 1 node and uses the Sigmoid activation function to output a comprehensive similarity score Score(s,t) between 0 and 1.
[0040] In some embodiments, the training process of a neural network model includes: Use the labeled historical node matching data as the training set; The neural network model is trained using the backpropagation algorithm based on the training set.
[0041] Figure 4 This is a schematic diagram of the model training process provided in the embodiments of this application, such as... Figure 4 As shown, historical node matching data is first collected as a training set, which contains a large number of (s, t) pairs. Experts label whether the matching is correct (1 for positive samples and 0 for negative samples). Then, the training set is used to train the constructed neural network model using a backpropagation algorithm (such as the Adam optimizer) with mean squared error or cross-entropy as the loss function. The model performance is then evaluated on the test set. If the performance is not up to standard, the model parameters are adjusted and the model is retrained until the performance meets the standard. The trained model weight parameters are saved as the trained neural network model.
[0042] In some embodiments, prior to step S100, the method further includes: For each string in the matching node string and the preset node string set, perform one or more of the following: Convert all uppercase English characters to lowercase; Remove special characters; Standardized spaces.
[0043] This application preprocesses each string in the string to be matched and the preset string set to avoid interference from uppercase and lowercase English letters, special characters and spaces in the matching results, thereby improving the accuracy of character matching.
[0044] In one embodiment of this application, taking the matching node strings "Design System PST" and "Production System PST" as an example, the matching steps are as follows: 1a. Data Preparation: Obtain the node set S = {"Port Main Engine", "Stern Tube", "Radar Mast"...} from the design system. Obtain the node set T = {"P-SIDE M / E", "Stern Tube Assembly", "PORT MAIN ENGINE", "Radar Mast", "WZGCGZ"...} from the production system; 2a. Model training: Train the neural network model using historical associated data or expert-annotated data until the model converges; 3a. Matching Reasoning: For the source node "Port Main Engine" (s) and the target node "P-SIDE M / E" (t), extract the following features: f_sub=0 (non-substring relationship); f_lcs=2×0 / (4+9)=0 (no common subsequence); calculate the edit distance between "Port Main Engine" and all nodes in T. Assuming ED_min=10, ED_max=50, and ED=12 with "P-SIDE M / E", then f_red=1-(12-10) / (50-10+0.001) ≈ 0.95. Input the feature vector [0, 0, 0.95] into the trained neural network model. The model outputs a comprehensive score of Score = 0.88. 4a. Sorting Output: Similarly, calculate the score of "Port Main Engine" and other nodes in T. Assuming the score with "PORT MAINENGINE" is 0.92, then after sorting, "PORT MAIN ENGINE" is the best match.
[0045] Table 1 is a schematic diagram of the matching and sorting results. The final sorting results are shown in Table 1.
[0046] Table 1. Schematic diagram of matching and sorting results
[0047] The following describes the ship product structure tree node matching device provided in this application. The ship product structure tree node matching device described below can be referred to in correspondence with the ship product structure tree node matching method described above.
[0048] Figure 5 This is a schematic diagram of a ship product structure tree node matching device provided in an embodiment of this application, as shown below. Figure 5 As shown, the device 500 includes: The feature extraction module 510 is used to compare the string of the node to be matched with each string in the preset string set character by character to obtain the substring inclusion relationship, LCS overlap rate and relative edit distance of the string of the node to be matched and each string in the preset string set respectively; The scoring module 520 is used to input the substring inclusion relationship, LCS overlap rate and relative edit distance of each string in the pre-trained neural network model into the string to be matched and each string in the preset string set, respectively, to obtain the comprehensive similarity score of the string to be matched and each string in the preset string set output by the neural network model. The matching module 530 is used to obtain the matching result of the node string to be matched based on the comprehensive similarity score corresponding to each string in the preset node string set.
[0049] It should be understood that the above-described device is used to execute the methods in the above embodiments. The implementation principle and technical effect of the corresponding program modules in the device are similar to those described in the above methods. The working process of the device can be referred to the corresponding process in the above methods, and will not be repeated here.
[0050] Based on the methods in the above embodiments, Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown in the illustration, this application provides an electronic device that may include a processor 610, a communications interface 620, a memory 630, and a communication bus 640. The processor 610, communications interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions stored in the memory 630 to execute the ship product structure tree node matching method described in the above embodiment.
[0051] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the ship product structure tree node matching method described in the various embodiments of this application.
[0052] Based on the methods in the above embodiments, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to execute the ship product structure tree node matching method in the above embodiments.
[0053] Based on the methods in the above embodiments, this application provides a computer program product that, when run on a processor, causes the processor to execute the ship product structure tree node matching method in the above embodiments.
[0054] It is understood that the processor in the embodiments of this application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor can be a microprocessor or any conventional processor.
[0055] The method steps in this application embodiment can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.
[0056] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).
[0057] It is understood that the various numerical designations used in the embodiments of this application are merely for the convenience of description and are not intended to limit the scope of the embodiments of this application.
[0058] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.
Claims
1. A method for matching nodes in a ship product structure tree, characterized in that, include: The string to be matched is compared character by character with each string in the preset string set to obtain the substring inclusion relationship, LCS overlap rate and relative edit distance of the string to be matched and each string in the preset string set respectively; Input the substring inclusion relationship, LCS overlap rate and relative edit distance corresponding to each string in the preset node string set into the pre-trained neural network model to obtain the comprehensive similarity score corresponding to each string in the preset node string set output by the neural network model. Based on the comprehensive similarity score corresponding to each string in the preset node string set and the node string to be matched, the matching result of the node string to be matched is obtained.
2. The method for matching nodes in a ship product structure tree according to claim 1, characterized in that, The neural network model is a multilayer perceptron, consisting of an input layer, a hidden layer, and an output layer. The input layer includes three nodes, which are used to input the substring inclusion relationship, LCS overlap rate, and relative edit distance, respectively. The hidden layer uses the ReLU activation function, and the output layer uses the Sigmoid activation function.
3. The method for matching nodes in a ship product structure tree according to claim 1, characterized in that, The training process of the neural network model includes: Use the labeled historical node matching data as the training set; Based on the training set, the neural network model is trained using the backpropagation algorithm.
4. The method for matching nodes in a ship product structure tree according to claim 1, characterized in that, Before comparing the string to be matched with each string in the preset set of node strings character by character, the method further includes: Perform one or more of the following operations on each string in the string to be matched and the preset set of node strings: Convert all uppercase English characters to lowercase; Remove special characters; Standardized spaces.
5. A device for matching nodes in a ship product structure tree, characterized in that, include: The feature extraction module is used to compare the string of the node to be matched with each string in the preset string set character by character to obtain the substring inclusion relationship, LCS overlap rate and relative edit distance of the string of the node to be matched and each string in the preset string set respectively; The scoring module is used to input the substring inclusion relationship, LCS overlap rate and relative edit distance corresponding to each string in the preset node string set into a pre-trained neural network model to obtain the comprehensive similarity score corresponding to each string in the preset node string set output by the neural network model. The matching module is used to obtain the matching result of the node string to be matched based on the comprehensive similarity score corresponding to each string in the preset node string set.
6. An electronic device, characterized in that, include: At least one memory for storing computer programs; At least one processor is configured to execute a program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to perform the ship product structure tree node matching method as described in any one of claims 1-4.
7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is run on a processor, the processor performs the ship product structure tree node matching method as described in any one of claims 1-4.
8. A computer program product, characterized in that, When the computer program product is run on the processor, the processor performs the ship product structure tree node matching method as described in any one of claims 1-4.