CN-121504419-B - Intelligent planning and aided design method for maintenance and guarantee tasks of ship equipment

CN121504419BCN 121504419 BCN121504419 BCN 121504419BCN-121504419-B

Abstract

In order to solve the defects of the prior art, the invention provides an intelligent planning and auxiliary design method for a maintenance and guarantee task of ship equipment. And collecting and marking technical documents for maintaining the ship equipment based on a maintenance support task element system for the ship equipment, and constructing a maintenance support task element diagram. Based on the maintenance support task element diagram, the nodes of the maintenance support task element diagram are used as learning samples of the graph convolution neural network for model training after data preprocessing. And carrying out intelligent planning and auxiliary design on the programming of the maintenance and guarantee tasks of the ship equipment based on the model obtained through training. The invention is based on a decision paradigm of knowledge driving and data driving fusion, effectively ensures time sequence compliance and resource suitability of the operation flow, and provides technical support for the full life cycle guarantee of ship equipment.

Inventors

DAI CHAO
Wu Shuochen
LIU SHULIN
FENG LINGLING
Deng Chenli
LIU YANG
ZUO JIAPENG
FANG XIAOTONG
Zhao Jingmeng
Li Lexiao

Assignees

中国船舶集团有限公司综合技术经济研究院

Dates

Publication Date: 20260505
Application Date: 20251020

Claims (7)

1. The intelligent planning and aided design method for the maintenance and guarantee tasks of the ship equipment is characterized by comprising the following steps of: Constructing a maintenance support task element system oriented to ship equipment, wherein the maintenance support task element system comprises two aspects of maintenance support operation sequences and maintenance support operation attributes; Based on a maintenance support task element system oriented to ship equipment, collecting and marking a ship equipment maintenance technical document, and constructing a maintenance support task element diagram, wherein: The method for collecting and marking the technical documents for maintaining the ship equipment based on the maintenance support task element system for the ship equipment comprises the following steps: Collecting technical documents of maintenance of ship equipment; Marking 6-dimensional information of a ship equipment maintenance technical document from 'maintenance operation', 'maintenance operation step', 'maintenance personnel specialty', 'maintenance personnel skill level', 'maintenance equipment/tool' and 'spare parts and consumed materials' based on a maintenance support task element system for the ship equipment; sequentially arranging the documents from top to bottom according to the sequence of the prior art documents; Constructing a directed graph consisting of nodes and edges based on a ship equipment maintenance technical document with completed labels, and connecting the directed graph based on an association relationship to obtain a maintenance and guarantee task element graph; nodes of the directed graph represent maintenance information; the edges of the directed graph represent the sequence relationship of two adjacent maintenance operations in one maintenance task; Based on the maintenance support task element diagram, the nodes of the maintenance support task element diagram are used as learning samples of the graph convolution neural network to input the learning samples into the graph convolution neural network for model training after data preprocessing; performing maintenance guarantee task element completion and unknown maintenance guarantee task element prediction on the compilation of the ship equipment maintenance guarantee task based on the model obtained through training; The maintenance support task element completion includes: inputting the task list as a seed node into a model obtained by training; The model obtained through training carries out comparison analysis on the related information in the task list from two aspects of the maintenance and guarantee operation sequence and the maintenance and guarantee operation attribute, carries out element completion on the missing part and predicts probability distribution on the supplemented element information; normalizing the supplemented element information, and selecting the element information with the highest probability as a prediction result to supplement the task list; the unknown maintenance support task element prediction includes: Inputting a model obtained by training by taking the currently known maintenance operation step as a seed node; the step of performing the next maintenance operation is obtained as a prediction step by the model obtained through training, and the probability that each prediction step is the next maintenance operation step is predicted; When the prediction probability is larger than the prediction step of the set threshold value, judging that maintenance sequence association exists, and outputting the prediction step; and (3) taking the obtained prediction step as a seed node to input a model obtained by training, repeating the process to generate a maintenance task sequence, and calculating the joint probability of the whole path until the link prediction probability is lower than a set threshold value and/or the number of inference steps reaches a preset upper limit, thereby completing the prediction of unknown maintenance guarantee task elements.
2. The intelligent planning and aided design method of ship equipment maintenance and assurance task according to claim 1, wherein the maintenance and assurance operation sequence is maintenance operation steps with irreversible sequential relationship contained in maintenance tasks, and the maintenance and assurance operation attribute is attribute information contained in the maintenance operation steps.
3. The intelligent planning and aided design method of a maintenance and guarantee task for ship equipment according to claim 2, wherein the maintenance and guarantee operation attribute comprises: the maintenance personnel professional attribute refers to the requirement of maintenance personnel professional skills in the maintenance operation step; The maintenance personnel skill level attribute refers to the requirement of professional skill title or level of the maintenance personnel in the maintenance operation step; The attribute of the guarantee equipment/tool refers to the required guarantee equipment or tool in the maintenance operation step; spare parts and consumable materials, which are spare parts and consumable materials required in the maintenance operation step.
4. The intelligent planning and aided design method for the maintenance and guarantee task of the ship equipment according to claim 1, wherein the method for preprocessing the data of the nodes of the maintenance and guarantee task element diagram comprises the following steps: embedding word information of each node in the maintenance support task element diagram into word vector characteristics, so that the text of each node in the maintenance support task element diagram is converted into word vectors; Constructing index mapping information of nodes for each node based on the converted word vector data, distributing digital identifiers, constructing index mapping information of relations for relations between the nodes, and creating mapping from a source node to a target node; creating a characteristic data set based on the nodes, the index mapping information of the nodes, the relation among the nodes and the index mapping information of the relation; The feature data set is divided into a training set, a verification set and a test set according to the ratio of 7:2:1, and the training set, the verification set and the test set are used as input information of a subsequent model.
5. The intelligent planning and aided design method of a ship equipment maintenance support task according to claim 1, wherein the graph roll-up neural network comprises: An input layer for receiving node characteristics and relationship information between nodes; The first graph convolution layer is used for performing first dimension conversion on the information of the input layer and accessing an activation function; the second graph convolution layer is used for further processing the data output by the first graph convolution layer and keeping the data dimension unchanged; the output layer includes a link prediction branch and an attribute prediction branch.
6. The intelligent planning and aided design method of ship equipment maintenance and guarantee task according to claim 5, wherein the link prediction branch is that the dimension characteristics of two nodes are spliced, and the dimension characteristics are mapped to 1-dimension output through a linear layer to represent the probability of the existence of the relationship between the adjacent nodes; The attribute prediction branch is used for directly mapping node data output by the second graph convolution layer to 4-dimensional output through a linear layer to represent the prediction probability of 4 maintenance attributes, wherein the 4 maintenance attributes are maintenance personnel professional attributes, maintenance personnel skill level attributes, guarantee equipment/tool attributes, spare parts and consumable material attributes.
7. The intelligent planning and aided design method of ship equipment maintenance and assurance task according to claim 6, characterized in that given maintenance and assurance task element graph g= (V, E, a, R), where V is node set, E is edge set, a is node type set, R is edge type set, for any node V E V its type is denoted as τ (V) E a, for any edge e= (s, t) E its type is denoted as Φ (E) E R, where the graph convolution neural network comprises: The input layer is used for constructing a type-specific linear transformation layer set aiming at the node types in the maintenance and guarantee task element diagram, and an independent linear mapping function is defined for each input node type tau by adopting a formula: one (I) In formula one: Embedding features for the initial text of the node; A weight matrix specific to the type; a bias vector specific to a type; The mapped unified dimension hidden characteristics; the graph convolution layer comprises two layers, and the calculation process can be formally expressed as: Two kinds of In the formula II, l represents the network layer number; Indicating that the type of the target node v is Is defined by a set of neighboring nodes of the network, Based on edge type The attention weight of the calculation is calculated, Is a ternary parameter matrix depending on the source node type, the target node type and the edge type; Activating a function for a ReLU; The attribute prediction branch of the output layer includes: For any attribute node Its characteristic vector The node text description is obtained by encoding by a bidirectional encoder module, and the processing mode of the output dimension is shown as a formula III: Three kinds of In three, embedding of nodes based on given maintenance steps Each attribute category is then predicted by a multi-class decoder as shown in equation four: Four kinds of Fourth middle A predictive probability distribution vector for a class 4 attribute for a given repair step node, For a given weight matrix of repair step nodes, For a given offset vector of the repair step node, Is a four-dimensional vector space; the link prediction branch of the output layer includes: First, the final embedded representation of two nodes u and v is given Where u is the source node type, v is the destination node type, Is 128-dimensional vector space; then based on the formula five, the two are spliced to form a joint representation: Five kinds of Five middle energizer 256-Dimensional vector space; finally, mapping the activated function layer into scalar scores, and decoding through an autoregressive decoder executing step six, so that the next task sequence text of the maintenance task can be predictably generated: Six-piece valve Wherein the method comprises the steps of For the predicted next task sequence text, In the form of a matrix of link weights, Is a link offset vector.

Description

Intelligent planning and aided design method for maintenance and guarantee tasks of ship equipment Technical Field The invention belongs to the technical field of ship maintenance, and particularly relates to an intelligent planning and aided design method for a maintenance and guarantee task of ship equipment. Background The current ship equipment system presents a high integration characteristic, and various ship systems are composed of multi-level heterogeneous components to form a composite framework. Such complex equipment systems expose significant process coordination challenges in maintenance operations, namely, single equipment overhaul often requires disassembly and reassembly operations across multiple subsystems. Because of strict logic dependency relationship in ship maintenance procedures, the conventional manual arrangement mode is extremely easy to generate procedure logic fracture or operation node dislocation, and the arrangement efficiency and the execution reliability of maintenance regulations are seriously affected. From the maintenance engineering implementation dimension analysis, the ship equipment guarantee activity has a multidimensional coupling characteristic that each operation unit needs to follow a strict topological execution sequence in a time sequence constraint layer. In the resource coordination level, the dynamic matching of elements such as multi-seed technical team, special detection instrument, heterogeneous guarantee supplies and the like is related. The problem of combined optimization under space-time-resource double constraint exceeds the capability boundary of traditional manual experience decision, and systematic risks such as misallocation of resources, frequent process conflicts and the like commonly exist, so that the improvement of comprehensive guarantee efficiency of offshore equipment is severely restricted. Disclosure of Invention Aiming at the problems existing in the prior art, the invention provides an intelligent planning and auxiliary design method for a maintenance and guarantee task of ship equipment, which comprises the following steps: and constructing a maintenance and guarantee task element system for ship equipment, wherein the maintenance and guarantee task element system comprises two aspects of maintenance and guarantee operation sequences and maintenance and guarantee operation attributes. And collecting and marking technical documents for maintaining the ship equipment based on a maintenance support task element system for the ship equipment, and constructing a maintenance support task element diagram. Based on the maintenance support task element diagram, the nodes of the maintenance support task element diagram are used as learning samples of the graph convolution neural network for model training after data preprocessing. And carrying out intelligent planning and auxiliary design on the programming of the maintenance and guarantee tasks of the ship equipment based on the model obtained through training. Further, the maintenance and guarantee operation sequence is maintenance and guarantee operation steps with irreversible sequential relation and included in maintenance tasks, and the maintenance and guarantee operation attribute is attribute information in the maintenance and guarantee operation steps. Further, the maintenance and guarantee operation attribute includes: the maintenance staff professional attribute refers to the requirement of maintenance staff professional skills in the maintenance and guarantee operation step. The maintenance personnel skill level attribute refers to the requirement of the maintenance personnel professional skill title or level in the maintenance guarantee operation step. The guarantee equipment/tool attribute refers to the guarantee equipment or tool required in the maintenance and guarantee operation step. Spare parts and consumable materials, which are spare parts and consumable materials required in the maintenance and guarantee operation step. Further, the method for collecting and labeling the technical documents for maintaining the ship equipment based on the maintenance support task element system for the ship equipment comprises the following steps: and collecting technical documents of maintenance of ship equipment. And marking the information of 6 dimensions of the ship equipment from 'maintenance and guarantee operation', 'maintenance and guarantee operation step', 'maintenance personnel specialty', 'maintenance personnel skill level', 'guarantee equipment/tool' and 'spare parts and consumed materials' based on a maintenance and guarantee task element system for the ship equipment. Sequentially arranged from top to bottom according to the sequence of the prior art documents. Further, the method for constructing the maintenance and guarantee task element diagram comprises the steps of constructing a directed diagram consisting of nodes and edges based on a ship equipment maintenance techn