Ship design knowledge pushing method and system based on artificial intelligence
By building deep learning models and multiple iterative updates, personalized and precise push of ship design knowledge is achieved, solving the problems of inefficient and insufficient personalized support for design engineers to acquire knowledge, and improving design quality and efficiency.
Patent Information
- Application Number
- CN202411217832.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-02
- Publication Date
- 2024-12-03
AI Technical Summary
In existing ship design, design engineers find it difficult to obtain personalized and accurate knowledge push services, and traditional information acquisition methods are inefficient and cannot fully cover technological development and cutting-edge knowledge.
Using an artificial intelligence-based method, a deep learning model is constructed, and multiple iterative updates are carried out based on ship design history search data and user feedback data to form an optimal knowledge push model to achieve personalized and accurate knowledge push.
It realizes personalized and precise push of ship design knowledge, improves design quality and efficiency, integrates artificial intelligence technology, and solves the problems of inefficient information acquisition and insufficient personalized support in traditional methods.
Smart Images

Figure CN119066977A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of ship design technology, and in particular, to a method and system for pushing ship design knowledge based on artificial intelligence. Background Art
[0002] With the booming global marine economy and the continuous advancement of technological innovation, the marine engineering and shipbuilding industries are undergoing unprecedented changes. At present, in this industry, not only are ship designs required to have higher safety, economy and environmental protection, but also higher requirements are put forward for their automation and intelligence. As the key force driving the development of this industry, the improvement and updating of the knowledge system of ship design engineers is crucial to improving the quality, efficiency and digitalization of ship design.
[0003] However, in current ship design practice, design engineers face many challenges in obtaining and applying key information such as design standards, process knowledge, quality cases, etc. Traditional ways of obtaining information, such as consulting paper materials, attending industry conferences, or relying on personal experience accumulation, are not only inefficient, but also difficult to fully cover the ever-changing technological developments and cutting-edge knowledge. In addition, existing knowledge consulting systems often lack personalized and intelligent support, and cannot provide accurate knowledge push services based on the specific needs and personal preferences of design engineers. Summary of the invention
[0004] The purpose of this application is to provide a method and system for pushing ship design knowledge based on artificial intelligence to solve the current problem that ship knowledge cannot be pushed to different objects in a personalized and accurate manner.
[0005] In order to achieve the above objectives, this application adopts the following technical solutions:
[0006] A ship design knowledge push method based on artificial intelligence comprises: constructing a deep learning model according to ship design historical search data; setting and defining a ship knowledge push service based on the deep learning model, and obtaining user feedback data according to the ship knowledge push service; dividing a time axis into a number of update cycles, performing multiple rounds of iterative updates on the deep learning model based on the user feedback data, and obtaining an optimal ship knowledge push model; and using the optimal ship knowledge push model to complete the ship knowledge push.
[0007] In some embodiments, constructing a deep learning model based on the ship design history search data includes: collecting the ship design history search data; determining the knowledge tower in the deep learning model based on the attribute characteristics of the ship design history search data; the deep learning model includes a plurality of knowledge towers, and the knowledge tower in the deep learning model is divided into a plurality of knowledge tower sub-modules according to the attribute characteristics of the ship design history search data; processing the ship design history search data and the knowledge tower sub-modules into a data set according to the corresponding relationship, and constructing the deep learning model based on the knowledge tower and the corresponding data set.
[0008] In some embodiments, the ship design history search data and the knowledge tower submodule are processed into a data set according to the correspondence, and the deep learning model is constructed based on the knowledge tower and the corresponding data set, specifically including: processing the attribute features of the ship design history search data into a state vector through natural language processing technology, and arranging and combining the state vectors according to the correspondence to form a data set corresponding to the knowledge tower; inputting each of the data sets into a corresponding scoring model, and fitting all the scoring models to form a deep learning quality scoring model, i.e., the deep learning model.
[0009] In some embodiments, the timeline is divided into several update cycles, and the deep learning model is iteratively updated for multiple rounds based on the user feedback data to obtain an optimal ship knowledge push model, including: if the update cycle is a micro-update cycle, the deep learning model is micro-updated and iterated according to the first user feedback data, and the first user feedback data is single-cycle user feedback data within the micro-update cycle.
[0010] In some embodiments, the micro-update iteration process includes: adding random perturbations to the model parameters of the deep learning model based on the neural network of the deep learning model to obtain new model parameters and a corresponding exploration network; generating ship knowledge recommendation lists by the neural network of the deep learning model and the exploration network respectively, cross-merging the ship knowledge recommendation lists to form a new ship knowledge recommendation list and recommending it to users; obtaining first user feedback data, which is the user's feedback on the new ship knowledge recommendation list. If the performance of the exploration network is better than that of the neural network of the deep learning model, the deep learning model corresponding to the exploration network is used as the model for subsequent update iterations; if the performance of the neural network of the deep learning model is better than that of the exploration network, the deep learning model is continued to be used as the model for subsequent update iterations.
[0011] In some implementations, the method further includes: if the update cycle is a major update cycle, performing a major update iteration on the model obtained after the micro-update iteration according to second user feedback data, wherein the second user feedback data is all user feedback data before the major update cycle.
[0012] In some embodiments, the main update iteration process includes: obtaining all user feedback data before the main update cycle, and analyzing the user feedback data; screening valid data based on the analysis results, adding the valid data to partially valid ship design historical search data, training the model obtained after the micro-update iteration, and using the model obtained after the training as the basis for subsequent update iterations.
[0013] The second aspect of the present application provides an artificial intelligence-based ship knowledge push system, including a model initialization module, a data accumulation module, a model upgrade module and a knowledge push module, wherein the model initialization module is used to construct a deep learning model based on the ship design history search data; the data accumulation module is used to set up and define a ship knowledge push service based on the deep learning model, and obtain user feedback data according to the ship knowledge push service; the model upgrade module is used to divide the time axis into several update cycles, and perform multiple rounds of iterative updates on the deep learning model based on the user feedback data to obtain an optimal ship knowledge push model; the knowledge push module is used to complete the ship knowledge push using the optimal ship knowledge push model.
[0014] A third aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program runs on a processor, the processor executes the method described in the first aspect.
[0015] The fourth aspect of the present application provides an electronic device, comprising: at least one memory for storing programs; and at least one processor for executing the programs stored in the memory. When the programs stored in the memory are executed, the processor is used to execute the method described in the first aspect.
[0016] Compared with the prior art, the present invention has the following advantages:
[0017] The present application provides a method and system for pushing ship knowledge based on artificial intelligence, which uses historical data to build a neural network deep learning model, and after accumulating certain user feedback, it realizes online updating of the model according to the reinforcement learning process in the algorithm, and finally forms a self-optimized intelligent push of ship design knowledge. The technical solution of the present application is based on artificial intelligence technology, which makes it completely unmanned for different objects to obtain ship knowledge and solve design problems, and can predict their preferences or research reversely according to the search scenarios of different objects to achieve personalized and precise push, breaking away from the past practice of manually collecting data, manually consulting questions and matching fixed knowledge models, and opening up a new path for the integration of the ship industry and artificial intelligence technology. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0019] Figure 1 A schematic diagram showing a process of a method for pushing ship knowledge based on artificial intelligence according to an embodiment of the present application is shown;
[0020] Figure 2 A schematic diagram showing a flow chart of a method for pushing ship knowledge based on artificial intelligence according to another embodiment of the present application is shown;
[0021] Figure 3 A schematic diagram showing the structure of a ship knowledge push system based on artificial intelligence according to an embodiment of the present application is shown;
[0022] Figure 4 A structural diagram of an artificial intelligence-based ship knowledge push system according to another embodiment of the present application is shown. DETAILED DESCRIPTION
[0023] The technical solutions of the specific implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings. These implementation methods are only used to illustrate the present application, but not to limit the present application.
[0024] The term "and / or" in this application is a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. In the present invention, the symbol " / " indicates that the associated objects are in an "or" relationship, for example, "A / B" means "A or B".
[0025] The terms "first" and "second" in the specification and claims of this application are used to distinguish different objects rather than to describe a specific order of the objects. For example, "first response message" and "second response message" are used to distinguish different response messages rather than to describe a specific order of the response messages.
[0026] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be considered as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.
[0027] In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more than two. For example, multiple processing units refer to two or more processing units, etc., and multiple elements refer to two or more elements, etc.
[0028] The following is a detailed description of an artificial intelligence-based ship knowledge push method and system of the present application in conjunction with the accompanying drawings.
[0029] Figure 1 FIG. 1 shows a method for pushing ship knowledge based on artificial intelligence according to an embodiment of the present application. Figure 1 As shown, the ship knowledge push method based on artificial intelligence in the embodiment of the present application includes the following steps:
[0030] A deep learning model is constructed based on the historical search data of ship design. Specifically, historical search data related to various specialties of ship design are collected. The historical search data of ship design is the question data raised by ship designers when searching the knowledge base. According to the attribute characteristics of the historical search data of ship design, the knowledge tower in the deep learning model is determined; according to the attribute characteristics of the historical search data of ship design, the knowledge tower in the deep learning model is determined; the deep learning model includes several knowledge towers, and the knowledge tower in the deep learning model is divided into several knowledge tower sub-modules according to the attribute characteristics of the historical search data of ship design. The historical search data of ship design and the knowledge tower sub-modules are processed into a data set according to the corresponding relationship, and the deep learning model is constructed based on the knowledge tower and the corresponding data set.
[0031] In one embodiment, the collected historical information related to various professions of ship design is divided into a user tower and an item tower according to the dual-tower structure of DQN (Deep Q-Network). Among them, the user tower includes user features and professional features, and the item tower includes action features and knowledge features. Each feature includes several sub-modules. For example, user features can be divided into technical directors, chief engineers, deputy chief engineers, assistant engineers and designers according to the staff rank. Professional features can be divided into structure, interior, electrical, piping and iron outfit according to different ship tooling. Action features include interactive features of user search knowledge, and knowledge features include historical databases and open source databases.
[0032] The attribute features of the ship design history search data are processed into state vectors through natural language processing technology, and the state vectors are arranged and combined according to the corresponding relationship to form a data set corresponding to the knowledge tower; each data set is input into the corresponding scoring model, and all the scoring models are fitted to form a deep learning quality scoring model, namely, a deep learning model.
[0033] Specifically, the aforementioned features can be processed into state vectors through natural language processing technology, and the corresponding relationships of each submodule are arranged and combined to form a data set. Among them, the user features and professional features in the data set are input into the user network as combined features, and the evaluation score is output to obtain the value score model; the user features, professional features, action features and knowledge features are combined and input into the item network, and the advantage score is output to obtain the advantage score model. The value score model and the advantage score model are fitted to form the final reinforcement learning quality score model, namely the DQN deep learning model, referred to as the deep learning model.
[0034] Based on the deep learning model DQN, a ship knowledge push service is set and defined, and user feedback data is obtained according to user feedback. In one embodiment, a ship knowledge push service for a time period can be set and defined. In a specific embodiment, Figure 2 As shown, this time period can be Figure 2 The stage before T1 in the time axis. During this time period, the ship knowledge push service will push relevant knowledge or information to users based on the recommendations or suggestions generated by the deep learning model. The pushed content may include design suggestions, reference cases or technical documents, etc. The specific content depends on the application scenario. After the user interacts with the pushed knowledge, user feedback data is collected. The user feedback data may include multiple indicators such as user satisfaction with the recommended content, click-through rate, dwell time, conversion rate, etc.
[0035] User feedback data is an important basis for evaluating the performance of deep learning models and is also the basis for subsequent model updates and upgrades. When this time period expires, it will trigger the evaluation of deep learning model performance and analysis of user feedback data to prepare for subsequent model updates. The length of this time period can be adjusted according to actual conditions to ensure that enough data can be collected during this time period to support model updates.
[0036] The time axis is divided into several update cycles, and the current deep learning model is updated multiple times iteratively based on user feedback data to obtain the optimal ship knowledge push model to implement ship knowledge push. In a micro-update cycle, in one embodiment, the micro-update cycle can be Figure 2 From the starting point of the middle timeline to stage T2, based on the deep reinforcement network (DRN) competitive gradient descent algorithm, the current deep learning model is micro-updated according to user usage data.
[0037] The micro-update process includes micro-updating the current deep learning model according to the first user feedback data within the micro-update cycle, and the first user feedback data is the single-cycle user feedback data within the micro-update cycle. Specifically, based on the agent network Q of the deep learning model obtained in step S1, a random perturbation is added to the parameter W of the deep learning model to obtain a new model parameter W', and the corresponding exploration network Q is obtained. The ship knowledge recommendation lists generated by the current agent network Q and the exploration network Q' are cross-fused to form a new ship knowledge recommendation list and recommended to the user. The first user feedback data is obtained, and the first user feedback data is the user's feedback on the new ship knowledge recommendation list. If the performance of the exploration network Q is better than that of the current agent network Q, the deep learning model corresponding to the exploration network is used as the model for subsequent update iterations. If the performance of the current agent network Q is better than that of the exploration network Q', the deep learning model is used as the model for subsequent update iterations.
[0038] In a major update cycle, in one embodiment, the major update cycle may be as follows Figure 2From the starting point of the middle timeline to the T3 stage, the micro-updated model is mainly updated according to all user feedback data. The main update process is included in the main update cycle, and the model obtained after the micro-update iteration is mainly updated according to the second user feedback data. The second user feedback data is all user feedback data before the main update cycle. Specifically, the user feedback data accumulated in all stages before the main update cycle is obtained, and the user activity analysis is performed. Valid data is filtered based on the analysis results. Valid data is the feedback of the user on the recommendation results after receiving the design knowledge recommendation list. The feedback includes clicks or ignores. The valid data is added to part of the valid ship design historical search data, and the model obtained after the micro-update iteration is trained. The model obtained after the training is completed is used as the basis for subsequent update iterations and replaces the current model. A new cycle starts until the optimal ship knowledge push model is obtained to achieve accurate ship knowledge push.
[0039] like Figure 3 As shown, the second aspect of the present application provides a ship knowledge push system based on artificial intelligence, including a model initialization module, a data accumulation module, a model upgrade module and a knowledge push module. Among them, the model initialization module is used to search data based on the history of ship design and build a deep learning model. Collect historical information related to various majors in ship design, and divide the collected historical information related to various majors in ship design into a user tower and an item tower according to the dual-tower structure of DQN (Deep Q-Network). Among them, the user tower includes user features and professional features, and the item tower includes action features and knowledge features, and each feature includes several sub-modules.
[0040] The above features are processed into state vectors through natural language processing technology, and the corresponding relationships of each submodule are arranged and combined to form a data set. Among them, the user features and professional features in the data set are input into the user network as combined features, and the evaluation score is output. The user features, professional features, action features and knowledge features are combined and input into the item network, and the advantage score is output. The output evaluation score and advantage score are fitted to form the final reinforcement learning quality score model, namely the DQN deep learning model, referred to as the deep learning model.
[0041] The data accumulation module is used to set and define ship knowledge promotion services based on deep learning models to obtain user feedback data. Figure 2 In the offline part, you can set and define a time period of ship knowledge push service, and train the deep learning model based on user feedback data, that is, user usage logs. The user feedback data may include multiple indicators such as user satisfaction with the recommended content, click-through rate, dwell time, conversion rate, etc.
[0042] The model upgrade module is used to divide the timeline into several update cycles, and perform multiple rounds of iterative updates on the current deep learning model based on user feedback data to obtain the optimal ship knowledge push model. The knowledge push module is used to complete the ship knowledge push with the optimal ship knowledge push model. Figure 2 In the online part shown, taking a cycle as an example, the time axis can be divided into T1, T2, T3, T4 and T5 according to the time stage, and each stage corresponds to a candidate set, which is a data set obtained by accumulating user feedback data. Based on the DRN algorithm, a strategy is formulated and the ship knowledge content pushed to the user is decided, and updated according to the user feedback data. In one embodiment, the micro-update cycle can be Figure 2 From the start point of the timeline to stage T2, the current deep learning model is micro-updated based on the user feedback data of this stage. The main update cycle can be Figure 2 From the starting point of the middle timeline to the T3 stage, the model after the micro-update is mainly updated based on all user feedback data before T3.
[0043] like Figure 4 As shown in the figure, user feedback data will be fed back to the current model as positive or negative incentives. After receiving the user feedback data, the current model will change and update the current state according to the user feedback data, and perform model training to update the model. Then, the push mechanism will continue to repeat the steps of "sort-push-feedback" until the overall click-through rate of the pushed content or user retention is achieved.
[0044] The third aspect of the present application provides a computer-readable storage medium, on which a program of a ship knowledge push method based on artificial intelligence is stored, so that the ship knowledge push method based on artificial intelligence provided in the first aspect of the present application is executed. As for the aforementioned computer-readable storage medium, those skilled in the art can understand that the implementation example of the functions of the aforementioned system and each unit can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, it executes an implementation example including the functions of the aforementioned system and each unit. The aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM or optical disk.
[0045] The fourth aspect of the present application provides an electronic device, comprising: at least one memory for storing programs; at least one processor for executing the programs stored in the memory. The electronic device may be a server. The electronic device comprises a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device comprises a non-volatile and / or volatile storage medium, an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, the functions or steps of the ship knowledge push method based on artificial intelligence of the present application are realized.
[0046] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present application. These improvements and substitutions should also be regarded as the scope of protection of the present application.
Claims
1. A ship design knowledge push method based on artificial intelligence, characterized in that: include: Build a deep learning model based on historical ship design search data; Setting and defining a ship knowledge push service based on the deep learning model, and obtaining user feedback data according to the ship knowledge push service; The time axis is divided into several update cycles, and the deep learning model is iteratively updated for multiple rounds based on the user feedback data to obtain an optimal ship knowledge push model; The ship knowledge push is completed by adopting the optimal ship knowledge push model.
2. The method for pushing ship design knowledge based on artificial intelligence according to claim 1, characterized in that: The deep learning model is constructed based on the historical search data of ship design, including: Collecting the ship design history search data; Determine a knowledge tower in the deep learning model according to the attribute characteristics of the ship design history search data; the deep learning model includes a plurality of knowledge towers, and the knowledge tower in the deep learning model is divided into a plurality of knowledge tower sub-modules according to the attribute characteristics of the ship design history search data; The ship design historical search data and the knowledge tower submodule are processed into a data set according to the corresponding relationship, and the deep learning model is constructed based on the knowledge tower and the corresponding data set.
3. The method for pushing ship design knowledge based on artificial intelligence as claimed in claim 2, characterized in that: Processing the ship design history search data and the knowledge tower submodule into a data set according to the corresponding relationship, and constructing the deep learning model based on the knowledge tower and the corresponding data set, specifically including: Processing the attribute features of the ship design history search data into state vectors through natural language processing technology, and arranging and combining the state vectors according to the corresponding relationship to form a data set corresponding to the knowledge tower; Each of the data sets is input into a corresponding scoring model, and all the scoring models are fitted to form a deep learning quality scoring model, namely the deep learning model.
4. The method for pushing ship design knowledge based on artificial intelligence according to claim 1, characterized in that: The time axis is divided into several update cycles, and the deep learning model is iteratively updated for multiple rounds based on the user feedback data to obtain an optimal ship knowledge push model, including: If the update cycle is a micro-update cycle, the deep learning model is micro-updated and iterated according to the first user feedback data, and the first user feedback data is single-cycle user feedback data within the micro-update cycle.
5. The method for pushing ship design knowledge based on artificial intelligence according to claim 4, characterized in that: The micro-update iteration process includes: Based on the neural network of the deep learning model, random perturbations are added to the model parameters of the deep learning model to obtain new model parameters and a corresponding exploration network; The neural network of the deep learning model and the exploration network respectively generate a ship knowledge recommendation list, cross-integrate the ship knowledge recommendation lists, form a new ship knowledge recommendation list and recommend it to the user; Acquire first user feedback data, where the first user feedback data is user feedback on the new ship knowledge recommendation list, and if the performance of the exploration network is better than the neural network of the deep learning model, use the deep learning model corresponding to the exploration network as a model for subsequent update iterations; If the performance of the neural network of the deep learning model is better than that of the exploration network, the deep learning model is used as the model for subsequent update iterations.
6. The method for pushing ship design knowledge based on artificial intelligence according to claim 4, characterized in that: Also includes: If the update cycle is a major update cycle, a major update iteration is performed on the model obtained after the micro-update iteration according to the second user feedback data, where the second user feedback data is all user feedback data before the major update cycle.
7. The method for pushing ship design knowledge based on artificial intelligence according to claim 6, characterized in that: The main update iteration process includes: Obtaining all user feedback data before the main update cycle and performing user feedback data analysis; Valid data is screened based on the analysis results, the valid data is added to partially valid historical search data of ship design, and the model obtained after the micro-update iteration is trained. The model obtained after the training is used as the basis for subsequent update iterations.
8. A ship design knowledge push system based on artificial intelligence, characterized in that: It includes model initialization module, data accumulation module, model upgrade module and knowledge push module, among which, The model initialization module is used to construct a deep learning model based on the ship design history search data; The data accumulation module is used to set and define a ship knowledge push service based on the deep learning model, and obtain user feedback data according to the ship knowledge push service; The model upgrade module is used to divide the time axis into several update cycles, perform multiple rounds of iterative updates on the deep learning model based on the user feedback data, and obtain an optimal ship knowledge push model; The knowledge push module is used to complete the ship knowledge push by adopting the optimal ship knowledge push model.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed on a processor, the processor is caused to execute the method according to any one of claims 1 to 7.
10. An electronic device, characterized in that: include: at least one memory for storing a program; At least one processor is used to execute the program stored in the memory, and when the program stored in the memory is executed, the processor is used to execute the method according to any one of claims 1 to 7.
Citation Information
Cited By
Intelligent naval vessel structure construction method and system based on AI large model
CN119849317A
