CN-122021299-A - Heating system source side automatic simulation modeling platform based on intelligent agent

CN122021299ACN 122021299 ACN122021299 ACN 122021299ACN-122021299-A

Abstract

The invention discloses an intelligent-based automatic simulation modeling platform for a heating system source side, which comprises a simulation modeling intention recognition unit, a source side device modeling unit, a source side device operation simulation model unit and a source side device operation simulation model unit, wherein the simulation modeling intention recognition unit is used for recognizing intention of simulation modeling personnel by utilizing a large language model, the source side device modeling unit is internally provided with a device modeling intelligent body and is used for calling the modeling large model to automatically establish a source side device model consistent with source side devices according to a preset modeling thinking chain by adopting a modularized modeling method and training fine tuning, the source side device operation simulation model unit is internally provided with a device operation simulation intelligent body and is used for calling the simulation large model to automatically establish a source side device operation simulation model by utilizing a computer program according to the preset simulation thinking chain, and the source side device operation three-dimensional visual modeling unit comprises a three-dimensional visual model establishment module, wherein the three-dimensional visual modeling intelligent body is internally arranged in the source side device modeling unit and is used for calling the three-dimensional modeling large model to form a source side device three-dimensional visual model according to the preset three-dimensional visual thinking chain.

Inventors

SHI WEI
ZHU BEI

Assignees

浙江英集动力科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. An intelligent agent-based heating system source side automatic simulation modeling platform is characterized by comprising: The simulation modeling intention recognition unit is used for recognizing the intention of a simulation modeling person by using the trained large language model, and comprises a heating system source side equipment modeling intention, a source side equipment operation simulation intention and a source side equipment operation three-dimensional visual modeling intention, and the simulation modeling intention is transferred into a corresponding simulation modeling unit according to intention recognition information; the simulation modeling unit comprises a source side device modeling unit, a source side device operation simulation model unit and a source side device operation three-dimensional visual modeling unit; The source side equipment modeling unit is internally provided with an equipment modeling intelligent body and is used for calling a trained modeling large model to process static drawing information of source side equipment and an equipment operation process flow knowledge base according to a preset modeling thinking chain, identifying each equipment simulation target entity and entity parameter information, a connection relation among the entities and an entity operation process flow, automatically matching and calling an equipment element library, a hydraulic-thermal model library, an electric component library and an automatic control component library, automatically establishing a physical model consistent with the source side equipment by adopting a modularized modeling method, and also used for calling a database to establish the source side equipment model after training and fine-tuning the physical model by utilizing virtual data and actual data; The source side equipment operation simulation model unit is internally provided with an equipment operation simulation intelligent body and is used for calling a trained simulation big model to analyze operation simulation modeling problems and simulation modeling targets according to a preset simulation thinking chain, and then calling the source side equipment model to perform source side equipment operation simulation to obtain observation parameters and action parameters of the equipment operation simulation intelligent body, wherein the observation parameters and the action parameters are used as input and output parameters of the simulation model, and the source side equipment operation simulation model is built automatically by using a computer program; The source side equipment operation three-dimensional visual modeling unit comprises a data acquisition preprocessing module, a data processing module and a data processing module, wherein the data acquisition preprocessing module is used for acquiring structural design drawings and equipment position layout drawings of all equipment on the source side, inputting the structural design drawings and the equipment position layout drawings into a trained multi-mode large model for training, and analyzing the design structures and design parameters of all equipment, connection information among all equipment and position layout information; The system comprises a source side equipment three-dimensional visual model, a source side equipment operation simulation model, a three-dimensional visual model establishing module and a simulation optimizing operation module, wherein the three-dimensional visual model establishing module is internally provided with a three-dimensional visual modeling intelligent body and is used for calling a trained three-dimensional modeling big model to convert acquired equipment information into three-dimensional modeling data according to a preset three-dimensional visual thinking chain, performing depth extraction and fusion on the data by using a set modeling tool algorithm, a modeling rule and a three-dimensional model library, and further calling the source side equipment model and the source side equipment operation simulation model to acquire equipment operation monitoring data and simulation optimizing operation data, and performing data analysis and graphic display according to a data visual library to form the source side equipment three-dimensional visual model.
2. The heating system source-side automated simulation modeling platform of claim 1, wherein the simulation modeling intent recognition unit is in communication connection with the source-side device modeling unit, the source-side device operation simulation model unit, and the source-side device operation three-dimensional visualization modeling unit, respectively, and the source-side device modeling unit, the source-side device operation simulation model unit, and the source-side device operation three-dimensional visualization modeling unit are in communication connection.
3. The automated simulation modeling platform of a heating system source side of claim 1, wherein the identifying the intent of a simulation modeler using the trained large language model comprises: Presetting corresponding intention characteristic labels and a domain rule base aiming at modeling intention of source side equipment of a heating system, running simulation intention of the source side equipment and running three-dimensional visual modeling intention of the source side equipment, wherein the intention characteristic labels comprise keyword labels and action labels of various intentions of the source side equipment; the method comprises the steps of performing fine adjustment on a basic large language model by utilizing historical simulation modeling demand data, enabling the large language model to learn semantic knowledge in the field of simulation modeling of source side heating equipment, adding intention feature labels and corresponding prompt words of a field rule base to the historical simulation modeling demand data input by simulation modeling personnel, guiding the large language model to analyze the prompt words according to preset logic, combining the semantic knowledge of fine adjustment learning, and outputting intention classification results and key basis; When the demand data input by the simulation modeling personnel has fuzzy expression, the intention ambiguity is eliminated by calling the context information and triggering a question back mechanism.
4. The automated simulation modeling platform of a heating system source side of claim 1, wherein the building of a physical model consistent with source side equipment comprises: Performing field fine adjustment through historical modeling cases based on a general modeling large model to form a modeling large model which is adaptive to the modeling field of equipment at a source side of a heating system, meanwhile, disassembling artificial modeling logic of a simulation modeling person into a quantifiable modeling step sequence, and forming a prompt word and a reasoning operation process of injecting modeling thinking chains into the modeling large model; Analyzing a static drawing of the source side equipment by using the trained modeling large model, identifying an entity of the source side equipment by target detection, extracting equipment parameter labels, analyzing equipment connection relations by using a graph neural network, extracting the entity relations and entity operation process flows in an equipment operation process flow knowledge base according to a knowledge graph technology, and forming an equipment entity modeling data packet containing each equipment simulation target entity, entity parameter information, entity connection relations and entity operation process flows; The equipment entity modeling data packet is respectively and automatically matched with an equipment element library, a hydraulic-thermal model library, an electric component library and an automatic control component library, if the equipment entity modeling data packet is successfully matched, corresponding equipment elements, electric elements, automatic control components and a thermal hydraulic model are directly obtained by using a modularized modeling method, and a physical model consistent with source side equipment is established; if the matching is unsuccessful, the custom creation of the elements and the components is carried out or the script language is used for writing.
5. The automated simulation modeling platform of a heating system source side according to claim 1, wherein the creating the source side device model after the invoking database trains and trims the physical model using the virtual data and the actual data comprises: generating virtual data of the whole operating condition of the source side equipment by using a physical model operating mechanism and historical operating boundary adjustment of the source side equipment and a Monte Carlo simulation method; Driving and calibrating a physical model by utilizing part of actual data, and dynamically adjusting the weight of the virtual data according to the threshold value of the actual data quantity and the distribution deviation index of the virtual data and the actual data; screening key parameters to be optimized and fine-tuned from the calibrated physical model, and constructing a loss function comprising mechanism loss, smooth loss and data loss; And performing first training fine tuning on the physical model by utilizing virtual data and solving a loss function, learning general model characteristics and acquiring preliminary preferred key parameters, then establishing a source side equipment model after fine tuning, performing second training fine tuning on the source side equipment model after fine tuning by utilizing actual data and solving the loss function, enhancing the real operation generalization capability of the model and acquiring final preferred key parameters, and establishing a final source side equipment model.
6. The automated simulation modeling platform for a heating system source side according to claim 5, wherein during the second training and fine tuning, parameters of the first training are used as initial values, learning rate of a bottom layer of the model is reduced, only actual data features are learned by a top layer, learning rate of the bottom layer is gradually increased, and parameter smooth transition is performed.
7. The heating system source-side automated simulation modeling platform of claim 1, wherein the building of the source-side equipment operation simulation model comprises: Performing field fine adjustment through historical source side equipment simulation cases based on a general simulation big model to form a simulation big model which is adaptive to the simulation field of the source side equipment of the heating system, meanwhile, disassembling artificial simulation logic of simulation modeling personnel into a quantifiable simulation step sequence, and forming a prompt word and a reasoning operation process of injecting a simulation thinking chain into the simulation big model; Calling a simulation big model to receive an input operation simulation demand, analyzing an operation simulation modeling problem and a simulation modeling target by combining a preset simulation thinking chain, identifying a simulation object, a simulation working condition, a target index, a constraint condition and generating a simulation model scheme, and converting the simulation object, the simulation working condition, the target index and the constraint condition into a simulation structural instruction; the simulation large model calls a source side equipment model to perform simulation according to a simulation structural instruction, and equipment runs a simulation intelligent body to obtain a simulation object observation parameter and an action parameter, wherein the observation parameter is a simulation object running state parameter, a performance parameter and a boundary parameter which are used as output parameters of the simulation model; The simulation big model drives a computer program and selects a simulation modeling tool according to the simulation type of a simulation model scheme in a simulation structural instruction, takes input and output parameters and simulation working conditions of the simulation model as simulation model variables, embeds a related calculation module of simulation target indexes and constraint conditions at the same time, simulates according to a simulation time step in the simulation model scheme until the whole simulation period is completed, automatically generates a visual chart, and constructs a source side device operation simulation model capable of directly executing a program file.
8. The automated simulation modeling platform of a heating system source side according to claim 1, wherein the obtaining the structural design drawing and the equipment position layout drawing of each equipment of the source side, inputting the structural design drawing and the equipment position layout drawing into the trained multi-mode large model for training, analyzing the design structure and the design parameters of each equipment, the connection information and the position layout information between each equipment, and comprises: Selecting a multi-modal large model supporting image-text cross-modal understanding, performing domain training fine adjustment on the multi-modal large model by utilizing structural design drawings and equipment position layout drawings of all equipment at a source side, and meanwhile, disassembling logic of a manual analysis drawing into an executable step sequence to form a prompt word and an reasoning operation process of injecting a analysis thinking chain into the multi-modal large model; Calling a multi-mode large model, identifying equipment entities, equipment design structures and extracting equipment design parameters through visual feature matching and text labeling, analyzing line features in a drawing, combining connection point labeling, obtaining a connection mode and connection attribute between equipment, converting pixel coordinates of the equipment into physical coordinates based on a drawing coordinate system and a scale, and generating structural description of position layout information.
9. The automatic simulation modeling platform of the heating system source side according to claim 1, wherein the calling the trained three-dimensional modeling big model converts the acquired equipment information into three-dimensional modeling data according to a preset three-dimensional visual thinking chain, and then performs depth extraction and fusion on the data by using a set modeling tool algorithm, a modeling rule and a three-dimensional model library, so as to automatically establish the three-dimensional model of the source side equipment, and the automatic simulation modeling platform comprises the following steps: Selecting a basic three-dimensional modeling large model, performing training fine adjustment by utilizing three-dimensional modeling historical data of source side equipment, and meanwhile, disassembling an artificial three-dimensional modeling flow into a quantifiable step sequence to form a prompt word and a three-dimensional visual thinking chain, and injecting the prompt word and the three-dimensional visual thinking chain into an reasoning operation process of the three-dimensional modeling large model; calling a three-dimensional modeling large model to perform structural integration on the equipment information acquired by the data acquisition preprocessing module to form an equipment three-dimensional modeling data packet comprising source side equipment geometric characteristic data, attribute association data and space layout data; and selecting an adaptive modeling tool algorithm according to the three-dimensional modeling requirement, automatically calling an equipment foundation module matched with an equipment three-dimensional modeling data packet from a preset three-dimensional model library according to a general modeling rule and an equipment exclusive modeling rule to perform equipment main body assembly, connecting part assembly and accessory part assembly, and automatically establishing a source side equipment three-dimensional model.
10. The automated simulation modeling platform of a heating system source side according to claim 1, wherein the calling the source side device model and the source side device operation simulation model to obtain each device operation monitoring data and simulation optimization operation data, and performing data analysis and graphic display according to a data visualization library to form a three-dimensional visualization model of the source side device comprises: The three-dimensional modeling large model calls the source side equipment model and the source side equipment operation simulation model to obtain equipment operation monitoring data and simulation optimization operation data, wherein the equipment operation monitoring data comprise equipment real-time state parameters, historical operation state parameters, alarm information and energy consumption data; Binding the acquired data with a three-dimensional model of the source side equipment, presetting association rules of the data and visual features of the three-dimensional model, calling a visual library to convert the data into a perceivable dynamic visual effect, and carrying out data analysis and graphic display to form the three-dimensional visual model of the source side equipment.

Description

Heating system source side automatic simulation modeling platform based on intelligent agent Technical Field The invention belongs to the technical field of heating system simulation modeling, and particularly relates to an intelligent agent-based heating system source side automatic simulation modeling platform. Background The heating system source side equipment comprises a boiler, a steam turbine, a thermoelectric unit, a coal-fired unit, a heat pump, an energy storage device and the like, the equipment is different in type and complex in equipment operation structure, simulation modeling staff is required to have rich experience knowledge and proficiency in field knowledge when traditional simulation modeling is carried out, professional modeling tools are used skillfully, data input is complex, modeling efficiency is low, each link is required to be connected manually, iteration cost is high, a simulation process is disjointed with an actual system, and dynamic adaptability is weak. With the application of new artificial intelligence technologies such as intelligent agents, large models and the like, the traditional simulation modeling paradigm is being remodeled, and the core advantages are represented in breakthrough of complexity, efficiency and intelligence. How to apply the intelligent body and the large model to the automatic simulation modeling of the source side equipment of the heating system improves the simulation modeling efficiency, improves the intellectualization and the precision of the simulation modeling, has high fidelity, reduces the time cost and the like, reduces the use difficulty of non-professional staff, and is the problem to be solved urgently at present. Based on the technical problems, a new intelligent agent-based heating system source side automatic simulation modeling platform needs to be designed. Disclosure of Invention The invention aims to solve the technical problems of overcoming the defects of the prior art, and provides an intelligent heating system source side automatic simulation modeling platform which can improve the heating system source side simulation modeling efficiency by utilizing intelligent and large model technology and simultaneously improve the accuracy, the intellectualization and the high fidelity of the simulation modeling. In order to solve the technical problems, the technical scheme of the invention is as follows: the invention provides an intelligent agent-based heating system source side automatic simulation modeling platform, which comprises: The simulation modeling intention recognition unit is used for recognizing the intention of a simulation modeling person by using the trained large language model, and comprises a heating system source side equipment modeling intention, a source side equipment operation simulation intention and a source side equipment operation three-dimensional visual modeling intention, and the simulation modeling intention is transferred into a corresponding simulation modeling unit according to intention recognition information; the simulation modeling unit comprises a source side device modeling unit, a source side device operation simulation model unit and a source side device operation three-dimensional visual modeling unit; The source side equipment modeling unit is internally provided with an equipment modeling intelligent body and is used for calling a trained modeling large model to process static drawing information of source side equipment and an equipment operation process flow knowledge base according to a preset modeling thinking chain, identifying each equipment simulation target entity and entity parameter information, a connection relation among the entities and an entity operation process flow, automatically matching and calling an equipment element library, a hydraulic-thermal model library, an electric component library and an automatic control component library, automatically establishing a physical model consistent with the source side equipment by adopting a modularized modeling method, and also used for calling a database to establish the source side equipment model after training and fine-tuning the physical model by utilizing virtual data and actual data; The source side equipment operation simulation model unit is internally provided with an equipment operation simulation intelligent body and is used for calling a trained simulation big model to analyze operation simulation modeling problems and simulation modeling targets according to a preset simulation thinking chain, and then calling the source side equipment model to perform source side equipment operation simulation to obtain observation parameters and action parameters of the equipment operation simulation intelligent body, wherein the observation parameters and the action parameters are used as input and output parameters of the simulation model, and the source side equipment operation simulation model is built automatically by usi