General assembly construction data service encapsulation and construction method based on cloud native technology

Abstract

The invention relates to the technical field of ship manufacturing, and discloses a general assembly construction data service encapsulation and construction method based on a cloud native technology. The method comprises the following steps: dividing a ship final assembly construction data source into design, production, quality inspection, logistics and operation and maintenance data units, and determining a data type and an interface protocol; then, constructing an initial service model based on a micro-service architecture and a containerization technology; then, dynamic service packaging is carried out, and cross-model data routing and load balancing are achieved; dynamically allocating resources and scheduling tasks through a multi-target scheduling algorithm model; iteratively optimizing the service model by utilizing an incremental learning mechanism; and finally, realizing gray release and rolling update by continuously integrating and continuously delivering assembly line deployment to a cloud native platform. The ship general assembly construction data processing efficiency, the system performance and the service quality are improved, and the problems of disordered data management, low processing efficiency, unreasonable resource allocation and the like in a traditional mode are effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of shipbuilding technology, and specifically to a method for encapsulating and constructing assembly and construction data services based on cloud native technology. Background Art

[0002] In the field of shipbuilding and assembly, as the size and complexity of ships continue to increase, data management and processing face numerous challenges. Traditional data management methods are unable to meet the requirements of modern shipbuilding for data processing speed, accuracy, and collaboration.

[0003] From a data source perspective, shipbuilding involves multiple stages, including design, production, quality inspection, logistics, and operations and maintenance, each with a complex data type. The 3D design drawings, process parameters, and simulation data generated during the design phase come in diverse formats and large volumes. Production plans, equipment status, and process progress data during the production phase require real-time data. Quality inspection requires precise management of inspection standards, defect records, and compliance reports. Logistics involves complex data linkages between bills of materials, warehousing information, and transportation routes. Operations and maintenance require long-term storage and analysis of equipment operation logs, fault alarms, and maintenance records. This data is scattered across various systems and departments, lacking a unified and effective integration mechanism. This leads to severe data silos and makes data sharing and collaboration difficult.

[0004] In terms of data processing technology, traditional architectures struggle to cope with the real-time processing of large amounts of data and complex business logic. Shipbuilding requires real-time data for resource scheduling, task allocation, and quality control. However, traditional centralized architectures are prone to performance bottlenecks when processing highly concurrent data, making them unable to meet the demand for rapid response. For example, during peak production periods, large amounts of production and logistics data simultaneously flood the system. Traditional architectures can cause data processing delays, impacting production schedules.

[0005] Furthermore, traditional service construction and deployment methods lack flexibility and are difficult to adapt to the dynamic changes in the shipbuilding business. When shipbuilding requirements change, such as design modifications or production plan changes, traditional approaches require large-scale redevelopment and re-deployment of the entire system, which is costly and time-consuming. Furthermore, traditional approaches lack consideration for system scalability and compatibility, making it difficult to integrate new technologies and equipment, thus limiting innovation and development in shipbuilding technology.

[0006] Traditional resource management methods are unable to achieve dynamic and optimal resource allocation. During shipbuilding, the demand for computing and storage resources varies significantly at different stages, and traditional static resource allocation methods can easily lead to resource waste or shortages. High graphics processing capabilities are required during the design phase, while data storage and transmission capabilities are even more demanding during the production phase. Failure to dynamically adjust resources based on actual demand will degrade overall system performance and resource utilization. Summary of the Invention

[0007] The purpose of the present invention is to provide a cloud-native technology-based assembly construction data service encapsulation and construction method to solve the problems raised in the above background technology.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for packaging and constructing an assembly construction data service based on cloud native technology, the method comprising: Step S1: According to the ship assembly and construction process, the data source is divided into design data unit, production data unit, quality inspection data unit, logistics data unit and operation and maintenance data unit, and the data type and interface protocol of each unit are determined; Step S2: Based on the microservice architecture, construct the initial service model of each data unit, including the design data model, production execution model, quality inspection rule model, logistics scheduling model, and operation and maintenance monitoring model, and use containerization technology to achieve independent deployment and communication of each model; Step S3: Dynamically encapsulate the initial service model, use a lightweight data bus to integrate the input and output flows between models, and implement cross-model data routing and load balancing based on service mesh technology; Step S4: Based on real-time resource requirements, a multi-objective scheduling algorithm model is established, with container cluster resource occupancy, service response delay, and data throughput as optimization variables, to perform dynamic resource allocation and task scheduling under preset constraints; Step S5: Iteratively optimize the service model based on the incremental learning mechanism, combining historical operation data with real-time feedback results to update model parameters and scheduling strategies; Step S6: Deploy the optimized service model to the cloud-native platform through the continuous integration and continuous delivery pipeline, and implement grayscale release and rolling updates of the service model based on the version control mechanism.

[0009] Preferably, in step S1, the design data unit includes three-dimensional design drawings, process parameters and simulation data; the production data unit includes production plans, equipment status and process progress; the quality inspection data unit includes inspection standards, defect records and compliance reports; the logistics data unit includes bill of materials, warehousing information and transportation routes; the operation and maintenance data unit includes equipment operation logs, fault alarms and maintenance records; The interface protocols include RESTful API, message queue and streaming data transmission protocols.

[0010] Preferably, in step S2: The design data model uses a parametric modeling approach to generate editable design services by combining geometric constraints and physical properties; The production execution model describes the conversion logic of the production process based on a finite state machine and triggers the process execution through an event-driven mechanism; The quality inspection rule model uses a rule engine to implement defect classification and priority determination; the logistics scheduling model generates optimal path planning based on a graph theory algorithm; The operation and maintenance monitoring model stores the device operation status through a time series database and triggers an early warning signal.

[0011] Preferably, in step S3, the dynamic service encapsulation adopts an adaptive data compression algorithm to encode the input and output streams, and dynamically adjusts the data transmission bandwidth based on the service level agreement; the service grid technology implements encryption and flow control of inter-service communication through the sidecar proxy, and supports a circuit breaker mechanism and retry strategy.

[0012] Preferably, in step S4, the multi-objective scheduling algorithm model adopts a non-dominated sorting genetic algorithm, with minimizing resource fragmentation rate, maximizing task completion rate and balancing node load as optimization goals, and generates a scheduling plan in combination with resource quota restrictions and task deadline constraints; The objective function of the multi-objective scheduling algorithm model is:

[0013] Among them, F represents the objective function, R f Indicates resource fragmentation rate, C t represents the task completion rate, L s represents the standard deviation of node load, α, β, γ are weight coefficients and satisfy .

[0014] Preferably, in step S5, the incremental learning mechanism adopts an online gradient descent algorithm to dynamically adjust the model weight according to the prediction error and resource consumption index of the service model, and retains the statistical characteristics of historical data through a sliding window mechanism to optimize training efficiency; The weight update formula of the online gradient descent algorithm is:

[0015] Among them, w t+1 represents the model weight at the t+1th iteration, w t represents the model weight of the t-th iteration, η is the dynamically adjusted learning rate, ∇J(w t ) is the gradient of the loss function with respect to the weight.

[0016] Preferably, in step S6, the continuous integration pipeline includes code static analysis, automated testing and image building modules; the continuous delivery pipeline implements zero-downtime updates of service models through a blue-green deployment strategy, and verifies the stability of the new version of the service based on a canary release mechanism.

[0017] Preferably, the design data model also integrates a topology optimization algorithm to automatically optimize the structural design according to material properties and load conditions; the production execution model supports virtual-reality mapping technology to synchronize the status data of physical equipment and virtual models in real time through the digital twin system.

[0018] Preferably, the adaptive data compression algorithm adopts a sparse representation method based on wavelet transform, and dynamically selects the compression level according to different data types; the service grid technology also supports distributed tracing functions, and records performance indicators and error logs across service call links through unique identifiers.

[0019] Preferably, the non-dominated sorting genetic algorithm introduces an elite retention strategy to retain the optimal solution set of each generation to accelerate convergence, and adopts a crowding comparison operator to maintain the diversity of the population and avoid local optimal solutions.

[0020] Compared with the prior art, the present invention has the following beneficial effects: In terms of data management and integration, data sources are precisely divided according to the ship assembly and construction process, and the data types and interface protocols of each data unit are clarified, effectively breaking down data silos. The clear definition of design data units, production data units, quality inspection data units, logistics data units, and operation and maintenance data units, combined with interface protocols such as RESTful API, message queues, and streaming data transmission protocols, enables efficient interaction and sharing of data in each link. The design department's three-dimensional design drawings can be transmitted to the production department in real time to guide production operations; the quality inspection department conducts quality control based on production data and testing standards, and promptly feeds back defect records to the production department for rectification; the logistics department arranges transportation and warehousing according to the production schedule and bill of materials. The smooth coordination of each link greatly improves the circulation efficiency and utilization value of data.

[0021] The initial service model was constructed based on a microservices architecture and containerization technology, enabling independent deployment and communication between models. The design data model utilizes a parametric modeling approach, combining geometric constraints with physical properties to generate editable design services. Designers can flexibly adjust design parameters and quickly generate multiple design solutions, improving design efficiency and quality. The production execution model uses a finite state machine to describe the production process transition logic. An event-driven mechanism triggers process execution, ensuring an orderly production process and reducing production delays and errors. The quality inspection rule model utilizes a rule engine to implement defect classification and priority determination, making quality inspection more targeted, prioritizing critical defects, and ensuring shipbuilding quality. The logistics scheduling model generates optimal path planning based on graph theory algorithms, reducing logistics costs and improving logistics efficiency. The operation and maintenance monitoring model uses a time-series database to store equipment operating status and trigger early warning signals, preventing equipment failures in advance, reducing equipment downtime, and lowering maintenance costs.

[0022] Dynamic service encapsulation technology utilizes an adaptive data compression algorithm and service mesh technology to improve data transmission and system performance. The adaptive data compression algorithm dynamically selects compression levels based on different data types, significantly reducing data transmission volume and alleviating network bandwidth pressure while ensuring data accuracy. Service mesh technology uses sidecar proxies to encrypt and control inter-service communication. It supports circuit breaking mechanisms and retry strategies to ensure system stability and reliability in complex network environments. Even if a service fails, the circuit breaker mechanism prevents the failure from spreading, and the retry strategy retryes requests after the failure is resolved, ensuring service continuity.

[0023] The multi-objective scheduling algorithm model uses container cluster resource utilization, service response latency, and data throughput as optimization variables to achieve dynamic resource allocation and task scheduling. A non-dominated sorting genetic algorithm is used to generate a scheduling solution, combining resource quota limits with task deadline constraints. This effectively improves resource utilization, balances node loads, and maximizes task completion rates. During peak shipbuilding periods, computing and storage resources are rationally allocated to ensure on-time completion of production tasks while avoiding performance bottlenecks caused by excessive concentration of resources on certain nodes.

[0024] The service model is iteratively optimized based on an incremental learning mechanism. Model parameters and scheduling strategies are updated based on historical operational data and real-time feedback, enabling the system to continuously adapt to business changes. An online gradient descent algorithm dynamically adjusts model weights based on prediction error and resource consumption metrics. A sliding window mechanism retains the statistical characteristics of historical data to optimize training efficiency, continuously improving the service model's predictive accuracy and adaptability. As data accumulates throughout the shipbuilding process, the system automatically optimizes scheduling strategies and quality control models, enhancing overall service levels.

[0025] Continuous integration and delivery pipelines, along with version control mechanisms, enable efficient deployment and updates of service models. The continuous integration pipeline's code static analysis, automated testing, and image building modules ensure code quality and deployability. The continuous delivery pipeline implements a blue-green deployment strategy for zero-downtime updates. The canary release mechanism verifies the stability of new service versions, minimizing the impact of service updates on operations and ensuring the continued stable operation of data services during the shipbuilding process. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a working principle diagram of the method for encapsulating and constructing the final assembly construction data service based on cloud native technology described in the present invention; Figure 2 This is a diagram for implementing dynamic service encapsulation and service grid technology functions; Figure 3 Build a model for multi-objective scheduling algorithms and a strategy selection diagram; Figure 4 Flowchart for weight update and optimization of service model based on incremental learning. DETAILED DESCRIPTION

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0028] See also Figure 1-4 The present invention provides a technical solution: a method for packaging and constructing assembly construction data services based on cloud native technology, with the following specific steps: Step S1: Based on the ship assembly and construction process, the data sources are divided into five categories: design data unit, production data unit, quality inspection data unit, logistics data unit, and operation and maintenance data unit. At the same time, the data type of each unit is clarified. For example, the design data unit may contain the ship's three-dimensional design drawings, process parameters, and simulation data; the production data unit covers information such as production plans, equipment status, and process progress; the quality inspection data unit involves inspection standards, defect records, and compliance reports; the logistics data unit includes bills of materials, warehousing information, and transportation routes; and the operation and maintenance data unit includes equipment operation logs, fault alarms, and maintenance records. In addition, the interface protocols used for data interaction between the units are determined, such as RESTful API, message queues, and streaming data transmission protocols.

[0029] Step S2: Based on the concept of a microservices architecture, an initial service model is constructed for each data unit. The design data model, production execution model, quality inspection rule model, logistics scheduling model, and operation and maintenance monitoring model are constructed separately. During the construction process, containerization technology is utilized to enable each model to be independently deployed in different container environments. Information exchange between them is achieved through specific communication mechanisms, thereby improving the flexibility and scalability of the system.

[0030] Step S3: Dynamically encapsulate the constructed initial service model. During this process, a lightweight data bus is used to integrate the input and output flows between models, enabling efficient data transmission. Furthermore, service mesh technology is used to implement cross-model data routing and load balancing, ensuring stable system operation under high concurrency conditions.

[0031] Step S4: To more rationally utilize resources, a multi-objective scheduling algorithm model is established based on real-time resource requirements. Using container cluster resource utilization, service response latency, and data throughput as optimization variables, the algorithm dynamically allocates system resources and schedules tasks appropriately within pre-set constraints to improve overall system performance.

[0032] Step S5: Iteratively optimize the service model based on an incremental learning mechanism. Combining the data generated by the service model during its historical operation and the real-time feedback, the model parameters and scheduling strategies are updated, enabling the service model to continuously adapt to actual business changes and improve service quality.

[0033] Step S6: Deploy the optimized service model to the cloud-native platform through the continuous integration and continuous delivery pipeline. During the deployment process, grayscale releases and rolling updates of the service model are implemented based on the version control mechanism to ensure that the new service version can be smoothly launched and minimize the impact on the business.

[0034] The present invention will be further described below in conjunction with Examples 1 to 6: Example 1: In step S1 of the above-mentioned overall implementation scheme, the data type of each data unit is described in more detail. The three-dimensional design drawings in the design data unit are the basis of ship design. They accurately depict the shape, structure and positional relationship of each component of the ship, providing an intuitive reference for subsequent production and manufacturing. The process parameters include various technical parameters required in the ship construction process, such as the current and voltage parameters of the welding process, the speed parameters of the cutting process, etc. These parameters directly affect the construction quality of the ship. Simulation data is data obtained by simulating the performance of the ship under different working conditions through computer simulation technology, such as the resistance and stability of the ship during navigation, which helps to discover potential problems in the design in advance and optimize them.

[0035] The production data unit's production plan details the start and end times and task allocation for each phase of shipbuilding, ensuring smooth and orderly construction. Equipment status provides real-time information on the operational status of various equipment during the production process, such as whether the equipment is functioning properly and whether there are potential faults. This facilitates timely maintenance and ensures continuous production. The process progress records the completion progress of each production process, allowing managers to better understand the overall production situation.

[0036] The inspection standards for the quality inspection data unit define the quality inspection requirements for each link and component in the shipbuilding process and serve as the basis for determining ship compliance. The defect record details various issues discovered during the quality inspection process, including location, type, and severity, providing accurate information for subsequent rectification. The compliance report provides a comprehensive evaluation of whether the shipbuilding process complies with relevant standards and regulations.

[0037] The bill of materials in the logistics data unit details the names, specifications, and quantities of all the raw materials and components required for shipbuilding, serving as a crucial basis for procurement and distribution. Warehousing information records material storage locations and inventory quantities, facilitating material management and allocation. Transportation routes map the transportation routes from suppliers to shipyards and within shipyards to ensure timely and accurate delivery of materials.

[0038] The equipment operation log in the operation and maintenance data unit records various parameters and events during equipment operation, such as start-up time, stop time, operating time, temperature, pressure, and other parameters, as well as equipment failures and maintenance operation records. Fault alarms promptly issue alerts when equipment anomalies occur, notifying relevant personnel to address them, reducing the risk of equipment damage and production interruptions. Maintenance records detail information such as maintenance time, maintenance content, and maintenance personnel, helping to track equipment maintenance status and develop a reasonable maintenance plan.

[0039] In terms of interface protocols, RESTful API is simple and easy to expand. It is suitable for scenarios that do not have particularly high requirements for real-time data transmission, but have high requirements for the flexibility and versatility of data interaction. For example, the interaction between the design data unit and other units to obtain static data such as design drawings. The message queue is suitable for asynchronous communication scenarios and can effectively decouple different modules in the system. For example, when the production data unit generates new process progress information, it can notify the quality inspection data unit to perform quality inspection through the message queue. The streaming data transmission protocol is suitable for scenarios that process large amounts of real-time data, such as the real-time monitoring data of the equipment operating status in the operation and maintenance data unit, which can be efficiently transmitted to the cloud platform for analysis and processing through the streaming data transmission protocol.

[0040] Example 2: When constructing the initial service model in step S2, the design data model utilizes a parametric modeling approach. First, based on the ship design requirements, a series of parameters are determined, such as geometric parameters such as the ship's length, width, height, and draft, as well as physical parameters such as the mechanical properties and physical characteristics of the materials. These parameters are then combined with geometric constraints, such as those regarding the ship's structural connectivity and dimensional ratio requirements, and physical constraints such as the material's strength and stiffness. This approach generates an editable design service. In the design data model, the integration of parameters with geometric and physical constraints is reflected in the fact that parameters serve as variable carriers of constraint conditions, while geometric and physical constraints define the parameter value ranges and association rules. Specifically, geometric constraints, such as the ship's structural connectivity, are implemented by defining matching rules between parameters (e.g., the interface dimensions of adjacent segments must be equal to the corresponding parameters). Dimensional ratio requirements are implemented by setting proportional coefficients between parameters (e.g., the range of the ratio of deck length to ship width). Physical constraints, such as the material's strength and stiffness, are implemented by limiting the parameter values ​​(e.g., the minimum steel plate thickness must satisfy the strength calculation formula). During the generation of the editable design service, In the system, parameters are associated and stored with the aforementioned constraints. When a user adjusts a parameter, the system automatically verifies whether the parameter satisfies the associated geometric and physical constraints. If not, the system prompts the adjustment range. If so, the associated parameters are synchronously updated to maintain the constraint relationship. For example, in ship compartment design, when the user adjusts the compartment height parameter, the system automatically verifies whether the height matches the interface dimensions of the adjacent compartment (geometric constraint) and whether the corresponding compartment steel plate thickness parameter meets the structural strength requirements at that height (physical constraint). After the height is adjusted, the connection parameters of the adjacent compartments are automatically updated to ensure that the overall structure's geometric matching and physical performance meet the standards. Designers can adjust these parameters according to actual needs, quickly generate ship models for different design schemes, and can view the model's various performance indicators in real time, greatly improving design efficiency and quality.

[0041] The production execution model describes the conversion logic of the production process based on the finite state machine. In the finite state machine used in the production execution model, the states of the production process are divided into initial states (e.g., process waiting to start), intermediate states (e.g., process in progress), and terminal states (e.g., process completed). Each state is linked by preset transition conditions, and the transition logic is triggered by an event-driven mechanism. Specifically, the definition of a state is based on key process nodes (e.g., material preparation, equipment commissioning, operation execution, and result confirmation). Transition conditions include events such as completion signals from preceding processes, material arrival information, and equipment status feedback. When an event meets the preset conditions, the system automatically transitions from the current state to the next state, simultaneously recording the transition time and the triggering event. For example, in the hull block assembly process, the initial state is "waiting for materials." When the logistics data unit transmits an event signal indicating the arrival of a block component (and the verified quantity and specifications meet the production plan), the system triggers a state transition to "assembly in progress." When the assembly operation sensor returns a completion signal (and key dimensional inspection data meets standards), the state transitions to "waiting for quality inspection." If the quality inspection data unit returns a qualified result, the state finally transitions to "process completed." Throughout this process, the state transition logic strictly adheres to the process connection requirements in the production process specification to ensure the orderly execution of each process. The shipbuilding process is divided into multiple discrete process states, such as raw material cutting, component welding, and segmented assembly. Each process state has clear input and output conditions, and process execution is triggered by an event-driven mechanism. When the input conditions of a process are met, the corresponding event is triggered, and the system automatically executes the process and transitions to the next process state based on the execution results. When the raw materials are ready and the cutting equipment is operating normally, the raw material cutting process is triggered. After cutting is completed, the system transitions to the next process state according to preset logic, such as the preparation state for the component welding process.

[0042] The quality inspection rule model uses a rules engine to classify and prioritize defects. The rules engine pre-defines a series of quality inspection rules, such as those for classification based on defect size, location, and type. Larger defects in critical areas are classified as high-priority defects, while smaller defects in non-critical areas are classified as low-priority defects. When the quality inspection data unit receives a defect record, the rules engine classifies and prioritizes the defect based on pre-set rules. This allows quality management personnel to prioritize high-priority defects, improving the efficiency and relevance of quality inspection work.

[0043] The logistics scheduling model generates optimal path planning based on graph theory algorithms. The shipyard's internal logistics network is abstracted into a graph, where nodes represent logistics sites, such as warehouses, production workshops, and processing areas, and edges represent logistics paths. Each edge is assigned a weight, which can represent factors such as transportation distance, transportation time, and transportation cost. Using graph theory shortest path algorithms, such as Dijkstra's algorithm or A*, the optimal transportation path from the starting point to the destination is calculated, taking into account multiple factors. This optimizes the allocation of logistics resources, reduces logistics costs, and improves logistics efficiency. When the logistics scheduling model calculates the optimal path, parameterization is reflected in assigning a unique identifier and coordinate attributes to each physical node in the logistics network (such as raw material warehouses, temporary storage yards, production workshops, and inspection areas), and assigning weight parameters to the transportation paths (edges) between nodes. The weight parameters include transportation distance (unit: meter), transportation time (unit: minute), transportation cost (unit: yuan), and path congestion coefficient (unitless, range 0-1). The congestion coefficient is dynamically updated based on historical transportation data and real-time equipment occupancy. For example, taking the transportation of steel from the raw material warehouse to the welding workshop in shipbuilding as an example, the nodes involved include raw material warehouse A, transfer station B, and transport station C. , welding workshop C, the weight parameters of the paths between nodes are: the distance from A to B is 200 meters, the time is 10 minutes, the cost is 30 yuan, and the congestion coefficient is 0.2; the distance from B to C is 150 meters, the time is 8 minutes, the cost is 20 yuan, and the congestion coefficient is 0.1; the direct distance from A to C is 300 meters, the time is 25 minutes, the cost is 50 yuan, and the congestion coefficient is 0.8. The system calculates the comprehensive weight of each path through the graph theory algorithm (the weight parameters are superimposed according to the preset ratio), and it is concluded that the comprehensive weight of the ABC path is the lowest, so it is determined as the optimal transportation path. Compared with the direct path, this path shortens the transportation time by 7 minutes and reduces the cost by 20 yuan. In addition, the risk of transportation delays can be reduced due to the low congestion coefficient.

[0044] The operation and maintenance monitoring model uses a time-series database to store equipment operating status and trigger early warning signals. Time-series databases are specifically designed to store data with time-series characteristics and can efficiently process the large amounts of real-time data generated during equipment operation. The system collects equipment operating parameters, such as temperature, pressure, and speed, in real time and stores this data in a chronological order in the time-series database. Furthermore, corresponding thresholds are set. When equipment operating parameters exceed these thresholds, the system automatically triggers early warning signals, notifying operation and maintenance personnel to promptly address the situation, preventing equipment failures and ensuring the smooth progress of the shipbuilding process.

[0045] Example 3: During the dynamic service encapsulation process in step S3, an adaptive data compression algorithm is used to encode the input and output streams. This algorithm, based on the sparse representation method of wavelet transform, dynamically selects a compression level based on the data type. For data types such as images and videos, which have large data volumes and relatively low detail requirements, a higher compression level is selected to significantly reduce the amount of data transmitted. For critical text data such as process parameters and testing standards, where data accuracy is extremely high, a lower compression level is selected to ensure that key information is not lost during the compression and decompression process. The parameterization of the adaptive data compression algorithm is reflected in the compression level classification from 1 to 6, where level 1 is the lowest compression, only removing redundant whitespace characters and repeated identifiers from the data, and level 6 is the highest compression, preserving the data's outline features through wavelet transform while discarding detailed information. The correspondence between compression level and data type is preset in the system. Image data (such as 3D ship model renderings and workshop surveillance video frames) is associated with levels 5-6 by default, while critical text data (such as welding process parameter tables and non-destructive testing standard clauses) is associated with levels 1-2 by default. For example, for a 3D design rendering of a ship section (image data), the default level is 1-2. ), the system uses level 5 compression and extracts the edge contours and main structural features of the image through wavelet transform. The data volume is greatly reduced after compression. Although the detailed texture is slightly lost, it can still meet the visual reference needs of the production site; for text data such as "T-joint welding process parameters: current 280-320A, voltage 22-25V, welding speed 300-400mm / min", the system uses level 1 compression, only removing unnecessary spaces in the sentence. The data volume is slightly reduced after compression, and the numerical range and expression logic of the original parameters are completely retained, ensuring that the parameter calls during subsequent production execution are accurate.

[0046] Service mesh technology uses sidecar proxies to implement encryption and flow control for inter-service communication, and supports circuit breaker mechanisms and retry strategies. The service mesh technology uses a sidecar proxy to implement inter-service communication. When the production data unit requests material delivery instructions from the logistics scheduling model, the sidecar proxy first encrypts the transmitted data (including material model, required quantity, and delivery time limit) using the TLS 1.3 protocol, generates an encrypted data packet, and adds a unique checksum. Simultaneously, the sidecar proxy limits the number of requests per second from the production data unit to no more than 200 requests per second, based on the preset service level agreement. When the request volume reaches 250, the excess 50 requests are queued and buffered, processed sequentially. If the logistics scheduling model experiences three consecutive request timeouts (with a timeout threshold set to 2 seconds) due to excessive load, the sidecar proxy triggers a circuit breaker mechanism, suspending new requests to it for the next 10 seconds and returning a "service temporarily unavailable" response to the production data unit. After 10 seconds, the sidecar proxy initiates a retry strategy, initially making a exploratory request at 10% of the request volume (i.e., 20 requests per second). If five consecutive requests are successfully responded to, the proxy gradually returns to the normal request volume, ensuring the security, stability, and continuity of service communication. The sidecar proxy is deployed next to each service container and shares the same network namespace. When services communicate, the sidecar proxy first encrypts the communication data to prevent data theft or tampering during transmission, ensuring data security. Furthermore, the sidecar proxy limits and manages traffic between services based on pre-set traffic control policies, preventing system performance degradation caused by excessive traffic to a single service. When a service fails or times out, a circuit breaker automatically activates, temporarily cutting off requests to the faulty service to prevent the fault from spreading and impacting the normal operation of the entire system. Furthermore, when the fault is resolved, a retry strategy automatically attempts to resend the request to ensure service availability.

[0047] Service mesh technology also supports distributed tracing, which uses unique identifiers to record performance metrics and error logs across service call chains. During a service call, the system assigns each request a unique identifier, which is passed between services. The sidecar proxy records the request's performance metrics on each service node, such as processing time and response time, as well as any errors encountered during processing. By collecting and analyzing this data, we can clearly understand the operational status of each service in the system, quickly locate fault points, and facilitate system optimization and maintenance.

[0048] Embodiment 4: When establishing the multi-objective scheduling algorithm model in step S4, a non-dominated sorting genetic algorithm is used. The instantiation application of the non-dominated sorting genetic algorithm in multi-objective scheduling is that for the container cluster resource scheduling scenario in ship assembly and construction, the initial population is set to 50 scheduling schemes, each of which contains the resource allocation ratio (CPU ratio, memory ratio) and task priority ranking of the production execution model and the logistics scheduling model; the algorithm first calculates the resource fragmentation rate (the ratio of idle resources to total resources), task completion rate (the ratio of the number of tasks completed on time to the total number of tasks) and node load standard deviation (reflecting the difference in the load of each container node) of each scheme, and divides the scheme into different levels through non-dominated sorting, among which the Pareto optimal level is the lowest. The algorithm includes solutions with low resource fragmentation, high task completion rates, and small node load standard deviations. It also introduces an elite retention strategy, retaining the five best solutions in this layer directly to the next generation. It also uses a congestion comparison operator to screen non-optimal layer solutions, retaining evenly distributed solutions within the population to maintain diversity. After 20 iterations, the resulting scheduling solution allocates 30% of CPU resources to the production execution model and 25% to the logistics scheduling model, maintaining a low resource fragmentation rate, a high task completion rate, and a small node load standard deviation, achieving coordinated optimization of resource utilization and task execution efficiency. The algorithm aims to minimize resource fragmentation, maximize task completion rates, and balance node loads, combining resource quota limits with task deadline constraints to generate scheduling solutions.

[0049] The objective function is: Among them, F represents the objective function, which comprehensively reflects the degree of optimization of the algorithm for multiple optimization objectives. f Indicates the resource fragmentation rate. The resource fragmentation rate refers to the proportion of resources that cannot be effectively utilized due to unreasonable resource allocation in the resource allocation process. Reducing the resource fragmentation rate can improve resource utilization. t Represents the task completion rate, which is the ratio of the number of tasks actually completed to the number of tasks planned to be completed. Maximizing the task completion rate can ensure that the system can complete various tasks efficiently. s Represents the node load standard deviation, which is used to measure the degree of load balance of each node in the system. A smaller node load standard deviation means that the load of each node is relatively balanced, avoiding performance bottlenecks caused by excessive load on a node. α, β, and γ are weight coefficients. Their values ​​determine the relative importance of each optimization goal in the objective function and satisfy .

[0050] The non-dominated sorting genetic algorithm incorporates an elite retention strategy, retaining the optimal solution set during each evolutionary generation. This prevents the loss of excellent individuals during the evolutionary process, accelerates the algorithm's convergence, and more quickly finds a scheduling solution that meets multi-objective optimization requirements. Furthermore, a crowding comparison operator is used to maintain population diversity and prevent the algorithm from becoming trapped in a local optimum. This operator calculates the degree of crowding of individuals within their non-dominated solution set and prioritizes individuals with lower crowding for genetic manipulation. This allows the algorithm to explore more extensively in the search space, increasing the probability of finding the global optimal solution.

[0051] Example 5: In step S5, the service model is iteratively optimized using an incremental learning mechanism using an online gradient descent algorithm. This algorithm dynamically adjusts model weights based on the service model's prediction error and resource consumption metrics. Prediction error refers to the difference between the model's predicted results and the actual results, while resource consumption metrics reflect the model's resource usage during operation, such as memory usage and CPU utilization.

[0052] The weight update formula is: Among them, w t+1 represents the model weight of the t+1th iteration, which is calculated based on the model weight of the tth iteration and the gradient update of this iteration. t represents the model weight of the tth iteration, which is the model weight obtained in the previous iteration. η is the dynamically adjusted learning rate. The learning rate determines the step size of each weight update and is dynamically adjusted according to the model training situation. If the model training error decreases rapidly, the learning rate is appropriately increased to speed up the training; if the model oscillates or the training error increases, the learning rate is reduced to ensure the stability of the training. ∇J(w t ) is the gradient of the loss function with respect to the weight, which reflects the rate at which the loss function changes with the weight. The direction of weight update is determined by calculating the gradient, so that the loss function is optimized in the direction of reduction.

[0053] A sliding window mechanism preserves the statistical characteristics of historical data to optimize training efficiency. During training, the sliding window mechanism retains only the most recent historical data. As time passes, the window slides forward, allowing new data to enter and older data to exit. This prevents excessive training time caused by excessive use of historical data while preserving the statistical characteristics of the data, enabling the model to better adapt to changing data trends and improving its predictive accuracy and adaptability.

[0054] Example 6: When deploying the service model through the continuous integration and continuous delivery pipeline in step S6: The continuous integration pipeline includes code static analysis, automated testing, and image building modules. The code static analysis module performs syntax checking, code standardization, and potential code defect detection on the written code, ensuring code quality and compliance, and reducing system failures caused by code quality issues. The automated testing module writes test cases for different service models, comprehensively testing the model's functionality, performance, and compatibility. When code changes occur, test cases are automatically run to promptly identify problems caused by code changes and ensure system stability. The image building module packages the tested code into container images for easy subsequent deployment on cloud-native platforms.

[0055] The continuous delivery pipeline achieves zero-downtime updates to service models through a blue-green deployment strategy and verifies the stability of new service versions based on a canary release mechanism. The blue-green deployment strategy involves maintaining two identical production environments simultaneously during the deployment process: one called the "blue environment," which runs the current version of the service; the other, the "green environment," where the new version of the service is deployed. During the switchover process, traffic is gradually switched from the blue environment to the green environment via a load balancer, enabling seamless service switching and ensuring that users experience no service interruptions during the update process. The canary release mechanism involves introducing a small amount of user traffic into the new service version after deployment for testing. This allows feedback from these users to be collected and evaluated for stability, performance, and other aspects of the new service version. If problems are discovered during testing, they can be promptly fixed to prevent the problem from spreading to a large number of users and ensure service quality.

[0056] The design data model also integrates a topology optimization algorithm to automatically optimize the structural design according to material properties and load conditions. The topology optimization algorithm is based on the principle of mathematical optimization. Under given material properties and load conditions, it optimizes the topological form of the structure, that is, the distribution of materials in the structure, so that the structure can achieve optimization goals such as lightest weight or maximum stiffness while meeting certain performance requirements. The specific application process of the topology optimization algorithm in ship structure design is as follows: taking the hull transverse support frame as an example, first input the material properties (such as the density of high-strength steel 7850kg / m³, yield strength 355MPa, elastic modulus 206GPa) and load conditions (such as the vertical load of 500kN and horizontal impact force of 100kN that the frame needs to withstand when the ship is fully loaded), and set performance constraints (such as maximum deformation ≤10mm, stress value ≤300MPa); the algorithm discretizes the frame into several units by establishing a finite element model, and uses the material density of each unit as the design variable (range 0- 1, 0 indicates a deleted element, 1 indicates a retained element), with minimizing the total mass of the structure as the optimization goal. During the iterative process, the algorithm first calculates the stress distribution of the initial structure, identifies low-stress elements with low stress values ​​(non-critical load-bearing areas), and gradually reduces their density to 0 to delete them. At the same time, elements in high-stress areas are retained and their distribution is optimized. After 100 iterations, the resulting frame structure meets the performance requirements of maximum deformation ≤8mm and stress ≤280MPa, while reducing material usage compared to the initial design, achieving the optimization goal of achieving the lightest weight while meeting performance standards. In ship design, using topology optimization algorithms to optimize the ship's structure based on the ship's operating requirements and the properties of the selected materials can not only improve the ship's performance, but also reduce material costs and improve the ship's economy.

[0057] The production execution model supports virtual-reality mapping technology, synchronizing the status data of physical equipment and virtual models in real time through a digital twin system. The digital twin system creates a virtual model corresponding to the physical equipment, using sensors to collect real-time operating status data of the physical equipment, such as its location, speed, and temperature, and synchronizes this data with the virtual model. The virtual model can also perform simulation analysis based on the received data, predicting potential equipment failures or performance changes and taking appropriate measures in advance. This enables remote monitoring and intelligent management of physical equipment, improving the visualization and controllability of the production process.

[0058] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0059] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for packaging and constructing assembly construction data services based on cloud native technology, characterized in that: The following steps are involved: Step S1: According to the ship assembly and construction process, the data source is divided into design data unit, production data unit, quality inspection data unit, logistics data unit and operation and maintenance data unit, and the data type and interface protocol of each unit are determined; Step S2: Based on the microservice architecture, construct the initial service model of each data unit, including the design data model, production execution model, quality inspection rule model, logistics scheduling model, and operation and maintenance monitoring model, and realize the independent deployment and communication of each model through containerization technology; Step S3: Dynamically encapsulate the initial service model, use a lightweight data bus to integrate the input and output flows between models, and implement cross-model data routing and load balancing based on service mesh technology; Step S4: Based on real-time resource requirements, a multi-objective scheduling algorithm model is established, with container cluster resource occupancy, service response delay, and data throughput as optimization variables, to perform dynamic resource allocation and task scheduling under preset constraints; Step S5: Iteratively optimize the service model based on the incremental learning mechanism, combining historical operation data with real-time feedback results to update model parameters and scheduling strategies; Step S6: Deploy the optimized service model to the cloud-native platform through the continuous integration and continuous delivery pipeline, and implement grayscale release and rolling updates of the service model based on the version control mechanism.

2. The method for packaging and constructing assembly and construction data services based on cloud native technology according to claim 1, characterized in that: In step S1, the design data unit includes three-dimensional design drawings, process parameters and simulation data; the production data unit includes production plans, equipment status and process progress; the quality inspection data unit includes inspection standards, defect records and compliance reports; the logistics data unit includes bill of materials, warehousing information and transportation routes; the operation and maintenance data unit includes equipment operation logs, fault alarms and maintenance records; The interface protocols include RESTful API, message queue and streaming data transmission protocols.

3. The method for packaging and constructing assembly construction data services based on cloud native technology according to claim 1 is characterized in that: In the step S2: The design data model uses a parametric modeling approach to generate editable design services by combining geometric constraints and physical properties; The production execution model describes the conversion logic of the production process based on a finite state machine and triggers the process execution through an event-driven mechanism; The quality inspection rule model uses a rule engine to implement defect classification and priority determination; the logistics scheduling model generates optimal path planning based on a graph theory algorithm; The operation and maintenance monitoring model stores the device operation status through a time series database and triggers an early warning signal.

4. The method for packaging and constructing assembly and construction data services based on cloud native technology according to claim 1, characterized in that: In step S3, the dynamic service encapsulation uses an adaptive data compression algorithm to encode the input and output streams, and dynamically adjusts the data transmission bandwidth based on the service level agreement; the service mesh technology implements encryption and flow control of inter-service communication through the sidecar proxy, and supports a circuit breaker mechanism and retry strategy.

5. The method for packaging and constructing assembly and construction data services based on cloud native technology according to claim 1 is characterized by: In step S4, the multi-objective scheduling algorithm model adopts a non-dominated sorting genetic algorithm, with minimizing resource fragmentation rate, maximizing task completion rate and balancing node load as optimization goals, and generates a scheduling plan in combination with resource quota restrictions and task deadline constraints; The objective function of the multi-objective scheduling algorithm model is: ； Among them, F represents the objective function, R f Indicates resource fragmentation rate, C t represents the task completion rate, L s represents the standard deviation of node load, α, β, γ are weight coefficients and satisfy .

6. The method for packaging and constructing assembly and construction data services based on cloud native technology according to claim 1, characterized in that: In step S5, the incremental learning mechanism adopts an online gradient descent algorithm to dynamically adjust the model weight according to the prediction error and resource consumption index of the service model, and retains the statistical characteristics of historical data through a sliding window mechanism to optimize training efficiency; The weight update formula of the online gradient descent algorithm is: ； Among them, w t+1 represents the model weight at the t+1th iteration, w t represents the model weight of the t-th iteration, η is the dynamically adjusted learning rate, ∇J(w t ) is the gradient of the loss function with respect to the weight.

7. The method for packaging and constructing assembly and construction data services based on cloud native technology according to claim 1, characterized in that: In step S6, the continuous integration pipeline includes code static analysis, automated testing and image building modules; The continuous delivery pipeline implements zero-downtime updates of service models through a blue-green deployment strategy, and verifies the stability of new versions of services based on a canary release mechanism.

8. The method for packaging and constructing assembly and construction data services based on cloud native technology according to claim 3, characterized in that: The design data model also integrates a topology optimization algorithm to automatically optimize structural design based on material properties and load conditions; the production execution model supports virtual-reality mapping technology, synchronizing the status data of physical equipment and virtual models in real time through a digital twin system.

9. The method for packaging and constructing assembly and construction data services based on cloud native technology according to claim 4, characterized in that: The adaptive data compression algorithm adopts a sparse representation method based on wavelet transform and dynamically selects the compression level according to different data types; The service mesh technology also supports distributed tracing, which records performance indicators and error logs across service call links through unique identifiers.

10. The method for packaging and constructing assembly and construction data services based on cloud native technology according to claim 5, characterized in that: The non-dominated sorting genetic algorithm introduces an elite retention strategy to retain the optimal solution set of each generation to accelerate convergence, and adopts a crowding comparison operator to maintain the diversity of the population and avoid local optimal solutions.

Images