Standardized service construction system for connecting application service and model service

Abstract

The invention provides a standardized service construction system for connecting an application service and a model service. The standardized service construction system comprises a data layer configured to set target model parameters required by an application service side and a model service side; a model layer configured to store operational models in the iron and steel industry; the servitization core layer is configured to perform unified parameter mapping processing and parameter conversion processing on target model parameters, perform standardized packaging on a process model, an optimization model and a diagnosis model so as to convert an operation model into a micro-service form, and perform unified management of a whole life cycle on the packaged model; the uniform interface layer is configured to provide a uniform access interface of the servitization core layer, the application layer and the model layer, and an access interface for inputting data to be tested into the servitization core layer; and the application layer is configured to provide a visual management platform for a user. Through the system, the problem that a unified and effective system is lacked between existing application services and model services is solved.

Description

Technical Field

[0001] This invention relates to the field of industrial internet technology, and in particular to a standardized service building system that connects application services and model services. Background Technology

[0002] With the deepening of intelligent manufacturing and industrial internet strategies, the steel industry is accelerating its transformation towards digitalization, networking, and intelligence. In this process, utilizing advanced technologies such as artificial intelligence (AI), big data, and cloud computing to build various process models, optimization models, and diagnostic models has become a key means to improve production efficiency, ensure product quality, and reduce energy consumption. These models are widely applied throughout the entire process, including blast furnace ironmaking, converter steelmaking, continuous casting, hot rolling, and cold rolling, covering multiple core business scenarios such as process optimization, predictive equipment maintenance, surface quality inspection, and energy balance scheduling.

[0003] However, in current steel enterprise practices, the development, management, and application of model services generally suffer from the following prominent problems, which severely restrict the depth and breadth of intelligent applications: The high integration complexity creates "model silos": various models are developed by different teams at different times, using different technical frameworks (such as Python, TensorFlow / PyTorch, MATLAB, R, etc.). Their input / output formats, communication protocols (such as gRPC, RESTful, MQTT, etc.), and dependent environments differ. This means that every new application system (such as MES, EMS, equipment management platform) that needs to call a model must perform a large amount of repetitive and tedious adaptation and integration development work, essentially building an independent "chimney" for each model. This results in high coupling between systems and an exponential increase in integration complexity.

[0004] Lack of unified standards leads to poor reusability and collaboration: Due to the absence of unified standards for model service access, encapsulation, and management, models cannot be shared and circulated within the enterprise as reusable assets. A quality inspection model may be required by multiple systems (such as production line MES and plant-wide quality management systems), but it cannot be efficiently developed once and called multiple times. At the same time, cross-process model collaboration (such as using steelmaking model output to provide input for rolling model) is difficult to achieve due to inconsistent interfaces, hindering the realization of end-to-end optimization.

[0005] Operation and maintenance management is difficult, and lifecycle management is lacking: Model services are deployed in a decentralized manner, lacking centralized monitoring, governance, and operation and maintenance capabilities. Model version upgrades, iterations, and decommissioning may cause unpredictable system compatibility issues, posing high risks and involving cumbersome processes. At the same time, there is a lack of effective monitoring methods for model call performance, load status, and quality of service (QoS), making it difficult to guarantee the high reliability and stability required by the production system.

[0006] Low resource utilization and insufficient scalability: Model services are typically deployed tied to specific applications, making dynamic scheduling and elastic scaling based on global load impossible. In high-concurrency scenarios, this easily becomes a performance bottleneck; during idle periods, it wastes computing resources. Traditional deployment methods struggle to adapt to the fluctuating computing power demands caused by the volatility of steel production. Summary of the Invention

[0007] This application provides a standardized service building system that connects application services and model services to solve the problem of the lack of a unified and effective system between existing application services and model services.

[0008] The system includes: The data layer is configured to set the target model parameters required by the application service side and model service side in the steel industry; the target model parameters are parameters after unified parameter mapping and parameter transformation processing. The model layer is configured to store operational models in the steel industry; the operational models include process models, optimization models, and diagnostic models; the process models, optimization models, and diagnostic models are all heterogeneous models. The service-oriented core layer is configured to perform unified parameter mapping and parameter transformation on the unprocessed target model parameters; to standardize and encapsulate the process model, the optimization model, and the diagnostic model to transform them into a unified microservice form, and to perform unified lifecycle management on the encapsulated models. A unified interface layer is configured to provide a unified access interface for the service-oriented core layer, the application layer, and the model layer, as well as an access interface for inputting test data into the service-oriented core layer. The application layer is configured to provide a visual management platform to users; the target model parameters include the parameters required for the process model, optimization model, and diagnostic model to run, as well as the parameters required for the application layer to run.

[0009] Preferably, the target model parameters include service-side parameters and model-side parameters. The service-side parameters include parameters of the steel enterprise data platform, L1 / L2 process control system, LIMS quality inspection system, and MES material tracking system. The model-side parameters include raw material composition parameters, process parameters, and quality data parameters.

[0010] Preferably, the data to be tested includes bulk raw material category parameters, product identification parameters, material tracking parameters, special event parameters, and quality indicator parameters; The bulk raw material category parameters are used to dynamically adjust the calculation parameters based on the operating model according to the currently used raw material varieties; The product identification parameters are used to select the corresponding calculation formula based on the steel type and specifications according to the operating model; The material tracking parameters are used for full-process quality traceability based on the operating model; The special event parameters are used to switch to the corresponding processing mode based on the special event code according to the operating model; The quality indicator parameters are used to determine the product grade based on the real-time quality indicators according to the operating model.

[0011] Preferably, the unified interface layer includes a layer management interface and a test data interface; The layer management interface is configured to provide a unified access interface for the service-oriented core layer, the application layer, and the model layer; The test data interface includes a bulk commodity category interface, a product identification interface, a material tracking interface, a special event interface, and a quality indicator interface; the test data interface is configured to input the bulk raw material category parameters, the product identification parameters, the material tracking parameters, the special event parameters, and / or the quality indicator parameters through the bulk commodity category interface, the product identification interface, the material tracking interface, the special event interface, and / or the quality indicator interface.

[0012] Preferably, the service-oriented core layer is further configured as follows: The parameters of the target model are validated and compliance checked. The parameters of the target model that have completed parameter validation and compliance checks are enhanced and context is constructed to obtain the first parameter set; The first parameter set is subjected to parameter mapping and model adaptation processing to obtain the second parameter set; the model adaptation processing is based on the process model, the optimization model, and / or the diagnostic model. The second parameter set is then distributed to the corresponding model.

[0013] Preferably, the service-oriented core layer is further configured as follows: The application layer service call request is obtained through the unified interface layer; The service call request is parsed to extract standard HTTP header information and business data, and general parameters are separated from steel industry-specific parameters.

[0014] Preferably, the parameter verification and the compliance check are performed independently of each other; The parameter verification includes verifying the structure and type of the target model parameters according to predefined steel industry data standards; The compliance check includes verifying the rationality and logic of the target model parameter values ​​based on business rules and context.

[0015] Preferably, the service-oriented core layer includes: A model service engine module is configured to standardize and encapsulate the process model, the optimization model, and the diagnostic model. The model service management module is configured to perform unified management of the encapsulated model throughout its entire lifecycle. A steel industry characteristic adaptation module is configured to perform unified parameter mapping and parameter transformation processing on the unprocessed target model parameters.

[0016] Preferably, the steel industry characteristic adaptation module includes: A raw material and fuel knowledge base, configured to store the physicochemical properties, process behavior models, and cost-benefit matrices of various raw materials and fuels; A special event handler is configured to dynamically switch the model's behavior mode and correct parameters when the model is in an abnormal production state. A quality rule engine, configured to logically centralize and standardize dispersed quality judgments; A parameter mapper configured to perform parameter mapping processing on the target model parameters using a configurable mapping rule template.

[0017] As described above, this application provides a standardized service construction system connecting application services and model services. The system includes a data layer configured to set target model parameters required by the application service side and model service side in the steel industry; these target model parameters are parameters after unified parameter mapping and transformation processing; a model layer configured to store operating models in the steel industry; these operating models include process models, optimization models, and diagnostic models; the process models, optimization models, and diagnostic models are all heterogeneous models; and a service-oriented core layer configured to perform unified parameter mapping processing on the unprocessed target model parameters. The system includes parameter conversion processing; standardized encapsulation of the process model, optimization model, and diagnostic model to transform them into a unified microservice format, and unified lifecycle management of the encapsulated models; a unified interface layer configured to provide unified access interfaces for the service-oriented core layer, application layer, and model layer, as well as an access interface for inputting test data into the service-oriented core layer; and an application layer configured to provide a visual management platform for users. The target model parameters include the parameters required for the process model, optimization model, and diagnostic model to run, as well as the parameters required for the application layer to run. This application solves the problem of the lack of a unified and effective system between existing application services and model services through the above system. Attached Figure Description

[0018] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A schematic diagram of a standardized service construction system for connecting application services and model services according to this application; Figure 2 A schematic diagram of the unified interface layer in a standardized service construction system that connects application services and model services, as described in this application; Figure 3 This is a schematic diagram of the service-oriented core layer in a standardized service building system that connects application services and model services, as described in this application. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] It should be noted that the brief descriptions of terms in this application are only for the convenience of understanding the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise stated, these terms should be understood in their ordinary and common meaning.

[0022] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0023] Figure 1 This is a schematic diagram of a standardized service building system for connecting application services and model services, as described in this application.

[0024] See Figure 1 As can be seen, this embodiment provides a standardized service building system for connecting application services and model services, the system comprising: The data layer is configured to set the target model parameters required by the application service side and model service side in the steel industry; the target model parameters are parameters after unified parameter mapping and parameter transformation processing.

[0025] Specifically, in this embodiment, the data layer is the foundation of the technical architecture, used to set and store the parameters required by the application service side and the model service side, wherein the parameters are parameters after unified parameter mapping and parameter transformation processing.

[0026] The target model parameters include service-side parameters and model-side parameters. The service-side parameters include parameters of the steel enterprise data platform, L1 / L2 process control system, LIMS quality inspection system, and MES material tracking system. The model-side parameters include raw material composition parameters, process parameters, and quality data parameters.

[0027] The system also includes: The model layer is configured to store operational models in the steel industry; the operational models include process models, optimization models, and diagnostic models; the process models, optimization models, and diagnostic models are all heterogeneous models.

[0028] Specifically, in this embodiment, the model layer is used to store all the models required in the system. Corresponding to the steel industry, the models include process models, optimization models, and diagnostic models, and the process models, optimization models, and diagnostic models are all heterogeneous models.

[0029] The system also includes: The service-oriented core layer is configured to perform unified parameter mapping and parameter transformation on the unprocessed target model parameters; to standardize and encapsulate the process model, the optimization model, and the diagnostic model to transform them into a unified microservice form; and to perform unified lifecycle management on the encapsulated models.

[0030] Specifically, in this embodiment, the service-oriented core layer is the core layer of this embodiment. The service-oriented core layer is used to uniformly manage the model and the parameters required by the model. Since different models and different parameter specifications are different, it is necessary to uniformly standardize and encapsulate all models through the service-oriented core layer, thereby transforming the model into a unified microservice form and performing unified lifecycle management of the encapsulated model; and to uniformly map and transform the target model parameters required by the model, thereby achieving unified management of all models and parameters with the same standard.

[0031] Furthermore, in some embodiments, the service-oriented core layer is also configured as follows: The parameters of the target model are validated and compliance checked.

[0032] Specifically, in this embodiment, the parameter verification and the compliance check are performed independently of each other. The parameter verification includes verifying the structure and type of the target model parameters according to predefined steel industry data standards. The compliance check includes verifying the rationality and logic of the target model parameter values ​​based on business rules and context.

[0033] For example, the parameter verification can be understood as verifying the format of the parameters, such as: Verify whether the materialLotId (material batch number) conforms to the company's coding rules (such as the fixed-length format of "P" + year, month, day + serial number).

[0034] Verify that the steelGrade exists in the standard grade enumeration list.

[0035] Verify that eventStartTime conforms to ISO 8601 or the date and time format specified by the company.

[0036] For example, the compliance check can be understood as checking the value of a parameter, such as: Verify whether the combination of rawMaterialType and materialGrade is valid (e.g., whether the category "Grade 1 Anthracite" exists).

[0037] Verify that the heatNumber is valid and traceable in the current process context.

[0038] For the qualityIndexValue, check whether its value is within a reasonable physical range (e.g., the tensile strength value cannot be negative or exceed the known limits of the steel).

[0039] The service-oriented core layer is also configured as follows: The parameters of the target model that have completed parameter validation and compliance checks are enhanced and context is constructed to obtain the first parameter set.

[0040] Specifically, in this embodiment, the system uses verified parameters as "keys" to automatically retrieve related data from systems such as the data platform, MES, and LIMS to construct a complete, model-oriented computational context. For example, when a request includes a materialLotId, the system will automatically query all process parameters and inspection results of the batch of material and inject them into the request as implicit enhancement parameters, without requiring explicit transmission from the application layer.

[0041] The service-oriented core layer is also configured as follows: The first parameter set is subjected to parameter mapping and model adaptation processing to obtain the second parameter set; the model adaptation processing is based on the process model, the optimization model and / or the diagnostic model.

[0042] Specifically, in this embodiment, during this step, the verified and enhanced standardized industry parameters are converted by the parameter mapper in the industry characteristic adaptation module into the input format expected by the downstream specific model. This ensures that heterogeneous models (which may be developed by different teams on different technology stacks) can be seamlessly integrated into this system.

[0043] For example, the standard productSpec "12×1500×C" is parsed and mapped to the thickness=12, width=1500, coilFlag=true required for model A, and {'dim:'12x1500', type:'Coil'} required for model B.

[0044] The service-oriented core layer is also configured as follows: The second parameter set is then distributed to the corresponding model.

[0045] Specifically, in this embodiment, the adapted parameters are distributed to the corresponding model service instances. The model service engine is responsible for executing the calls and managing resource allocation and timeout control during the computation process. The raw computation results returned by the model are repackaged by the service engine into a unified data structure that conforms to the previously defined response specifications. At the same time, information such as the requestId associated with the request, processing status, and data source identifier used for enhancement are also returned, providing a clear audit and diagnostic link for the application layer.

[0046] The system also includes: A unified interface layer is configured to provide a unified access interface for the service-oriented core layer, the application layer, and the model layer, as well as an access interface for inputting test data into the service-oriented core layer.

[0047] Specifically, in this embodiment, the unified interface layer uses an API gateway as its core component, providing a unified access point including industry-enhanced interfaces.

[0048] The data to be tested includes bulk raw material category parameters, product identification parameters, material tracking parameters, special event parameters, and quality indicator parameters. The bulk raw material category parameters are used to dynamically adjust calculation parameters based on the currently used raw material varieties according to the operating model. The product identification parameters are used to select the corresponding calculation formula based on the steel grade and specifications according to the operating model. The material tracking parameters are used for full-process quality traceability based on the operating model. The special event parameters are used to switch to the corresponding processing mode based on the special event code according to the operating model. The quality indicator parameters are used to determine the product grade based on the real-time quality indicators according to the operating model.

[0049] Example parameters for bulk raw materials and fuels: Parameter names: rawMaterialType (raw material type), materialGrade (material grade); Example: {"rawMaterialType":"COKE","materialGrade":"FIRST_GRADE"}; Function: Allows the model to dynamically adjust calculation parameters based on the type of raw materials and fuels currently used, such as adjusting the coal ratio calculation based on the coke grade in the blast furnace model.

[0050] Product identification parameters: Parameter names: steelGrade (steel grade), productSpec (product specification); Example: {"steelGrade":"Q235B","productSpec":"12×1500×C"}; Function: The rolling force model and cooling model can automatically select the appropriate calculation formula according to the steel grade and specifications.

[0051] Material tracking parameters: Parameter names: materialLotId (material batch number), heatNumber (furnace number); Example: {"materialLotId":"P20231027001","heatNumber":"23102708"}; Function: Enables end-to-end quality traceability; the model service can obtain complete data on the material at each process stage through this parameter.

[0052] Special event parameters: Parameter names: specialEventCode (special event code), eventStartTime (event start time); Example: {"specialEventCode":"BF_REPAIR","eventStartTime":"2023-10-2708:00:00"}; Function: The model can switch to the corresponding processing mode according to special event codes, such as when the blast furnace is shut down, the model enters the heat preservation calculation mode.

[0053] Quality indicator parameters: Parameter names: qualityIndexName (quality index name), qualityIndexValue (quality index value); Example: {"qualityIndexName":"tensileStrength","qualityIndexValue":"450"}; Function: The quality assessment model can determine the product grade based on real-time quality indicators.

[0054] The system also includes: The application layer is configured to provide a visual management platform to users; the target model parameters include the parameters required for the process model, optimization model, and diagnostic model to run, as well as the parameters required for the application layer to run.

[0055] Specifically, in this embodiment, the application layer is the direct platform of the system for users. The application layer may include various production management systems of steel enterprises, thereby providing users with a visual management platform and improving the convenience of the system.

[0056] Figure 2 This is a schematic diagram of the unified interface layer in a standardized service construction system that connects application services and model services, as described in this application.

[0057] See Figure 2 It can be seen that, further, in some embodiments, the unified interface layer includes a layer management interface and a test data interface; The layer management interface is configured to provide a unified access interface for the service-oriented core layer, the application layer, and the model layer; The test data interface includes a bulk commodity category interface, a product identification interface, a material tracking interface, a special event interface, and a quality indicator interface; the test data interface is configured to input the bulk raw material category parameters, the product identification parameters, the material tracking parameters, the special event parameters, and / or the quality indicator parameters through the bulk commodity category interface, the product identification interface, the material tracking interface, the special event interface, and / or the quality indicator interface.

[0058] Specifically, in this embodiment, the layer management interface corresponds to the unified access interface of the service-oriented core layer, the application layer, and the model layer, while the test data interface is the input interface corresponding to the bulk raw material category parameters, the product identification parameters, the material tracking parameters, the special event parameters, and / or the quality indicator parameters.

[0059] The test data interface includes a bulk product category interface, a product identification interface, a material tracking interface, a special event interface, and a quality indicator interface, meaning that corresponding parameters can be input through independent interfaces.

[0060] Figure 3 This is a schematic diagram of the service-oriented core layer in a standardized service building system that connects application services and model services, as described in this application.

[0061] See Figure 3 Furthermore, in some embodiments, the service-oriented core layer includes: A model service engine module is configured to standardize and encapsulate the process model, the optimization model, and the diagnostic model. The model service management module is configured to perform unified management of the encapsulated model throughout its entire lifecycle.

[0062] Specifically, in this embodiment, the model service engine module and the model service management module work together to manage the model, and the specific management method is not limited here.

[0063] The service-oriented core layer also includes: A steel industry characteristic adaptation module is configured to perform unified parameter mapping and parameter transformation processing on the unprocessed target model parameters.

[0064] Specifically, in this embodiment, this module is the core component for achieving deep integration of industry know-how and IT technology in this technical solution. It is not a simple parameter forwarder, but an intelligent conversion hub embedded with steelmaking process knowledge. Its core mission is to "translate" standardized industry interface parameters into specific instructions that can be "understood" by various heterogeneous downstream models.

[0065] Furthermore, in some embodiments, the steel industry characteristic adaptation module includes: A raw material and fuel knowledge base is configured to store the physicochemical properties, process behavior models, and cost-benefit matrices of various raw materials and fuels.

[0066] Specifically, in this embodiment, the raw material knowledge base is a dynamic and evolving process parameter database that goes beyond simple material classification and stores the physicochemical properties, process behavior models, and cost-benefit matrices of various raw materials.

[0067] The raw materials and fuels knowledge base not only defines the categories of "primary coke" and "secondary coke", but also links the standard values ​​and fluctuation ranges of key indicators such as fixed carbon content, sulfur content, reactivity (CRI) and post-reaction strength (CSR).

[0068] When the parameter {"rawMaterialType":"COKE","materialGrade":"FIRST_GRADE"} is received, the knowledge base will automatically provide a set of benchmark calculation coefficients and parameter adjustment ranges for the blast furnace fuel ratio optimization model, so that the model can perform accurate calculations under the current raw material conditions.

[0069] The raw material knowledge base enables quantitative management of the impact of raw material quality fluctuations on key production indicators, enabling model calculations to move from "empirical estimation" to "data-driven".

[0070] The steel industry characteristic adaptation module also includes: A special event handler is configured to dynamically switch the model's behavior mode and correct parameters when the model is in an abnormal production state.

[0071] Specifically, in this embodiment, the special event processor is a rule engine with embedded metallurgical working condition logic, which is responsible for dynamically switching the behavior mode and correcting parameters of the model under abnormal production conditions.

[0072] The special event handler predefines standardized handling plans for special events such as BF_REPAIR (blast furnace shutdown), CON_CAST_BREAK (continuous casting interruption), and ROLL_CHANGE (roll replacement).

[0073] When the processor recognizes "specialEventCode":"BF_REPAIR", it sends an instruction to the blast furnace thermal model to switch it from the normal full air volume calculation mode to the heat preservation and pressure preservation calculation mode, and automatically adjusts the weights and boundary conditions of a series of related parameters such as air temperature and coal ratio.

[0074] The special event handler ensures the applicability and robustness of the model under complex working conditions, and avoids outputting incorrect guidance due to model mismatch in the event of abnormal events.

[0075] The steel industry characteristic adaptation module also includes: A quality rule engine is configured to logically centralize and standardize dispersed quality judgments.

[0076] Specifically, in this embodiment, the quality rule engine is a configurable decision-making hub based on product quality standards and expert experience, which centralizes and standardizes the dispersed quality judgment logic.

[0077] The quality rule engine has pre-built quality decision trees and mathematical criteria for different steel grades and product specifications.

[0078] When the engine receives data {"qualityIndexName":"tensileStrength","qualityIndexValue":"450"} from the inspection line, it will automatically call the rule set for "Q235B" steel grade, compare the measured value with the standard range (such as ≥375MPa), and combine it with other indicators (such as yield strength and elongation) to finally output a comprehensive judgment conclusion and basis for "qualified", "downgraded" or "scraped".

[0079] The quality rule engine automates, standardizes, and unmannes product quality judgment, significantly improving quality consistency and reducing the risk of human error.

[0080] The steel industry characteristic adaptation module also includes: A parameter mapper configured to perform parameter mapping processing on the target model parameters using a configurable mapping rule template.

[0081] Specifically, in this embodiment, the parameter mapper is a technical "protocol converter" that solves the "language incompatibility" problem between standardized interfaces and heterogeneous model inputs through configurable mapping rule templates.

[0082] The parameter mapper maintains a model parameter mapping table that clearly defines how to decompose, transform, and assign the unified parameters of the API layer to specific model inputs.

[0083] Mapping example: When the input is {"productSpec":"12×1500×C"}, the mapper will perform the following operations in parallel: The rolling force model is analyzed to have a thickness of 12 (mm), a width of 1500 (mm), and a type of coil.

[0084] The target strip thickness was determined to be 12 for the cooling control model, and the recommended water flow curve number was retrieved from the knowledge base accordingly.

[0085] Generate standard product size description fields for the MES system.

[0086] The parameter mapper decouples the standardization of business interfaces from the diversification of model implementations, greatly improving the overall flexibility and scalability of the system and supporting the rapid access of new models.

[0087] This embodiment has the following advantages: 1. In terms of technology integration and efficiency: This invention significantly reduces system integration complexity and development costs: By providing a unified RESTful API interface and protocol conversion capabilities, it transforms the original point-to-point integration model, which required extensive custom development, into a standard, declarative service call model. Application system developers can quickly integrate AI capabilities without needing to delve into the technical details of the underlying model, resulting in a model integration efficiency improvement of over 60% and a substantial reduction in development and maintenance workload.

[0088] This invention achieves "plug-and-play" and efficient reuse of model capabilities: The standardized middleware layer created by this invention enables various heterogeneous models to become service assets that can be uniformly discovered, managed, and invoked. Once a model (such as a blast furnace condition prediction model) is encapsulated and registered, it can be reused simultaneously by multiple different application systems (such as expert systems, operator platforms, and scheduling centers), completely breaking down "model silos," avoiding redundant development, and significantly improving the return on investment of model assets.

[0089] 2. In terms of system performance and reliability: It ensures high availability and high reliability for production-level applications: Through built-in load balancing, automatic failover, circuit breaker degradation and elastic scaling mechanisms, this invention can effectively cope with single point of failure and sudden flow surges, ensuring the continuous and stable operation of model services (such as converter end-point control) on which key production processes depend, and meeting the stringent requirements of 24 / 7 uninterrupted operation of steel production.

[0090] Improved overall resource utilization and system elasticity: Based on containerization and Kubernetes orchestration technologies, the model service can dynamically schedule computing resources according to actual load, achieving second-level scaling up and down. It ensures service performance during peak production periods and automatically releases resources during idle periods, resulting in an average increase in infrastructure resource utilization of 30%-50%.

[0091] 3. At the level of operation and maintenance management and governance: A unified and visualized governance system for the entire lifecycle of model services has been implemented: The model service governance center provides centralized monitoring, unified configuration, and granular management capabilities for all model services across the entire plant. Operations personnel can easily complete version upgrades, canary releases, performance monitoring, and troubleshooting, achieving standardization and automation of operations and maintenance work, improving operational efficiency by over 50%, and significantly reducing management chaos and the risks of human error.

[0092] A comprehensive security and audit control system has been established: through the authentication, authorization, traffic control and security audit functions of the unified gateway, fine-grained access control and full-process traceability of model service call behavior have been achieved, effectively protecting the enterprise's core process data and algorithm model assets, and meeting the high standards of the steel industry for production data security.

[0093] 4. At the level of business empowerment and strategy: This invention accelerates the agile iteration and innovation of AI-enabled businesses: It significantly shortens the cycle from model development to production deployment, allowing data scientists to focus on algorithm innovation without worrying about deployment and integration issues. Enterprises can thus test and implement new AI applications more quickly, rapidly respond to market changes and business needs, and build core competitiveness in the intelligent era.

[0094] By connecting data flow and business flow, the foundation for end-to-end optimization is laid: standardized interfaces enable easy collaboration of model services across processes and systems (such as using the output of the steelmaking model as the input of the rolling model), providing a crucial technical foundation for the ultimate realization of integrated intelligent optimization and decision-making throughout the entire process from molten iron to finished product, which is difficult to achieve with traditional integration methods.

[0095] For ease of explanation, the above description has been provided in conjunction with specific embodiments. However, the discussion in some embodiments is not intended to be exhaustive or to limit the embodiments to the specific forms disclosed above. Various modifications and variations can be obtained based on the above teachings. The selection and description of the above embodiments are for the purpose of better explaining the contents of this disclosure, thereby enabling those skilled in the art to better utilize the embodiments.

Claims

1. A standardized service building system for connecting application services and model services, characterized in that, The system includes: The data layer is configured to set the target model parameters required by the application service side and model service side in the steel industry; the target model parameters are parameters after unified parameter mapping and parameter transformation processing. The model layer is configured to store operational models in the steel industry; the operational models include process models, optimization models, and diagnostic models; the process models, optimization models, and diagnostic models are all heterogeneous models. The service-oriented core layer is configured to perform unified parameter mapping and parameter transformation on the unprocessed target model parameters; to standardize and encapsulate the process model, the optimization model, and the diagnostic model to transform them into a unified microservice form, and to perform unified lifecycle management on the encapsulated models. A unified interface layer is configured to provide a unified access interface for the service-oriented core layer, the application layer, and the model layer, as well as an access interface for inputting test data into the service-oriented core layer. The application layer is configured to provide a visual management platform to users; the target model parameters include the parameters required for the process model, optimization model, and diagnostic model to run, as well as the parameters required for the application layer to run.

2. The standardized service construction system for connecting application services and model services according to claim 1, characterized in that, The target model parameters include service-side parameters and model-side parameters. The service-side parameters include parameters of the steel enterprise data platform, L1 / L2 process control system, LIMS quality inspection system, and MES material tracking system. The model-side parameters include raw material composition parameters, process parameters, and quality data parameters.

3. The standardized service construction system for connecting application services and model services according to claim 1, characterized in that, The data to be tested includes parameters of bulk raw materials and fuels, product identification parameters, material tracking parameters, special event parameters, and quality indicator parameters. The bulk raw material category parameters are used to dynamically adjust the calculation parameters based on the operating model according to the currently used raw material varieties; The product identification parameters are used to select the corresponding calculation formula based on the steel type and specifications according to the operating model; The material tracking parameters are used for full-process quality traceability based on the operating model; The special event parameters are used to switch to the corresponding processing mode based on the special event code according to the operating model; The quality indicator parameters are used to determine the product grade based on the real-time quality indicators according to the operating model.

4. A standardized service construction system for connecting application services and model services according to claim 3, characterized in that, The unified interface layer includes a layer management interface and a test data interface; The layer management interface is configured to provide a unified access interface for the service-oriented core layer, the application layer, and the model layer; The test data interface includes a bulk commodity category interface, a product identification interface, a material tracking interface, a special event interface, and a quality indicator interface; the test data interface is configured to input the bulk raw material category parameters, the product identification parameters, the material tracking parameters, the special event parameters, and / or the quality indicator parameters through the bulk commodity category interface, the product identification interface, the material tracking interface, the special event interface, and / or the quality indicator interface.

5. A standardized service construction system for connecting application services and model services according to claim 1, characterized in that, The service-oriented core layer is also configured as follows: The parameters of the target model are validated and compliance checked. The parameters of the target model that have completed parameter validation and compliance checks are enhanced and context is constructed to obtain the first parameter set; The first parameter set is subjected to parameter mapping and model adaptation processing to obtain the second parameter set; the model adaptation processing is based on the process model, the optimization model, and / or the diagnostic model. The second parameter set is then distributed to the corresponding model.

6. A standardized service construction system for connecting application services and model services according to claim 5, characterized in that, The service-oriented core layer is also configured as follows: The application layer service call request is obtained through the unified interface layer; The service call request is parsed to extract standard HTTP header information and business data, and general parameters are separated from steel industry-specific parameters.

7. A standardized service construction system for connecting application services and model services according to claim 5, characterized in that, The parameter verification and the compliance check are performed independently of each other; The parameter verification includes verifying the structure and type of the target model parameters according to predefined steel industry data standards; The compliance check includes verifying the rationality and logic of the target model parameter values ​​based on business rules and context.

8. A standardized service construction system for connecting application services and model services according to claim 1, characterized in that, The service-oriented core layer includes: A model service engine module, configured to standardize and encapsulate the process model, the optimization model, and the diagnostic model; The model service management module is configured to perform unified management of the encapsulated model throughout its entire lifecycle. A steel industry characteristic adaptation module is configured to perform unified parameter mapping and parameter transformation processing on the unprocessed target model parameters.

9. A standardized service construction system for connecting application services and model services according to claim 8, characterized in that, The steel industry characteristic adaptation module includes: A raw material knowledge base, configured to store the physicochemical properties, process behavior models, and cost-benefit matrices of various raw materials; A special event handler is configured to dynamically switch the model's behavior mode and correct parameters when the model is in an abnormal production state. A quality rule engine, configured to logically centralize and standardize dispersed quality judgments; A parameter mapper configured to perform parameter mapping processing on the target model parameters using a configurable mapping rule template.

Images