Model management and collaborative analysis method and system based on digital twinning

Abstract

The invention provides a model management and collaborative analysis method and system based on digital twinning. The method comprises the steps that S1, a digital twinning 3D model of a target object is constructed; s2, collecting point location data of the target object at a fixed frequency, and storing the point location data in an IHD real-time database; s3, constructing different business models, including a data model and a rule model; s4, selecting a business model according to a demand, and calculating a corresponding business index based on the point location data in the IHD real-time database; s5, pushing the calculated business indexes to the digital twinborn 3D model; s6, the digital twin 3D model restores the internal attributes of the target object based on the collected point location data of the target object and the received business indexes; the invention provides a unified digital twinborn model management and collaborative analysis framework for a complex industrial system, and aims to solve the problems of heterogeneous and complex tasks, dynamic environment change and real-time interaction existing in digital twinborn at present.

Description

Technical Field

[0001] The present invention relates to the field of intelligent manufacturing technology, and in particular to a model management and collaborative analysis method and system based on digital twins. Background Art

[0002] In modern industry, the normal operation and efficient maintenance of equipment are key factors for enterprises to improve production efficiency and reduce operating costs. However, traditional equipment maintenance methods are often based on a fixed schedule or after equipment failure, which may lead to excessive maintenance or production losses due to unexpected downtime. Digital twins simulate the entire production process, identify bottlenecks and optimize resource allocation, allowing enterprises to monitor equipment status in real time, analyze operating data, and achieve accurate predictive maintenance.

[0003] Although digital twin systems can monitor, analyze and optimize physical entities, they face problems in practical applications such as heterogeneous complexity of tasks, dynamic changes in the environment and real-time interaction that need to be solved urgently. Therefore, it is crucial to provide a unified model management and collaborative scheduling framework for complex industrial environments, which can integrate models and data from different fields, break information silos, and achieve cross-domain collaborative analysis.

[0004] This paper proposes a model management and collaborative analysis method based on digital twins. According to business needs, heterogeneous tasks are uniformly packaged and automatically issued, and collaborative analysis is performed using rule models and data models. This capability not only reduces the need for manual intervention, but also greatly improves the efficiency and accuracy of the system in processing heterogeneous tasks, and has the characteristics of automation, efficiency, and flexibility. Summary of the invention

[0005] In view of the defects in the prior art, the purpose of the present invention is to provide a model management and collaborative analysis method and system based on digital twins.

[0006] A model management and collaborative analysis method based on digital twins provided by the present invention includes: Step S1: construct a digital twin 3D model of the target object; Step S2: collecting target object point data at a fixed frequency and saving it in the IHD real-time database; Step S3: construct different business models, including: data model and rule model; Step S4: Select a business model according to the demand, and calculate the corresponding business indicators based on the point data in the IHD real-time database; Step S5: Push the calculated business indicators to the digital twin 3D model; Step S6: The digital twin 3D model restores the intrinsic properties of the target object based on the collected point data of the target object and the received business indicators; The data model is a mathematical or statistical model built based on point data (i.e., variable data collected during the operation of equipment or systems) for business objectives, which enables the organic organization, analysis, prediction, and anomaly detection of data. The construction of the data model is based on multidimensional data in industrial scenarios, and provides support for business decision-making and operation optimization through data cleaning, integration, modeling, and analysis; The rule model is based on a Boolean function, and obtains an alarm state based on the rule according to the point data of the target object; Preferably, the step S2 comprises: collecting the target object point data from the PLC or WinCC system at a fixed frequency and saving it in a data table of the IHD real-time database; based on the IHD real-time database, obtaining the real-time data and historical data of the point through the http interface; the structure of the data table comprises: sensor ID, timestamp, measurement value, status code; wherein the status code indicates the quality of data collection; The target object point data includes: the physical properties of the driven target object, including: speed, temperature and pressure.

[0007] Preferably, step S3 comprises: Step S3.1: Construct a data model management system, including: data model classification management, data model path configuration, data model dynamic loading and loading status management; The data model classification management is to classify the data models using a dictionary structure, and establish a mapping relationship between configuration parameters and detection functions for each type of data model; different data models construct different parameter configurations; The data model path configuration is to obtain the weight file paths of different types of data models based on the configuration parameters of each type of data model, and store the obtained paths in the corresponding dictionary structure; The data model is dynamically loaded by checking the model path defined in the configuration file, loading the formulated data model detection function, and loading the trained weight file based on the detection function; The loading state management is to record the loading and unloading state of the data model through the state dictionary; Step S3.2: Building a rule model management system, including: rule definition, trigger configuration, rule parsing and logical operations; The trigger configuration includes: a time-driven mechanism, an event-driven mechanism, and a driving mechanism combining events and time; The rule analysis and the logic operation are to analyze the rules and trigger configurations in the rule model, and execute corresponding tasks based on the rule judgment function.

[0008] Preferably, the step S4 comprises: Step S4.1: Build a system status management system, including: initialization management when the server starts, including: current status check, log recording, exception handling and global status update; Step S4.2: Construct a multi-task scheduling management system, including: task definition, task status management, task parameter persistence and task batch management; Step S4.3: Heterogeneous tasks are issued. According to different task requirements, the corresponding data model, rule model, or data model and rule model are called to calculate the corresponding business indicators based on the point data in the IHD real-time database.

[0009] Preferably, step S5 includes: storing the calculated business indicators in a db2 database, and then using a kafka message queue to push the latest business indicators to the digital twin 3D model.

[0010] A model management and collaborative analysis system based on digital twins provided by the present invention includes: Module M1: Build a digital twin 3D model of the target object; Module M2: collects target object point data at a fixed frequency and saves it in the IHD real-time database; Module M3: Build different business models, including data model and rule model; Module M4: Select the business model according to the demand and calculate the corresponding business indicators based on the point data in the IHD real-time database; Module M5: Push the calculated business indicators to the digital twin 3D model; Module M6: The digital twin 3D model restores the intrinsic properties of the target object based on the collected point data of the target object and the received business indicators; The data model is a mathematical or statistical model built based on point data (i.e., variable data collected during the operation of equipment or systems) for business objectives, which enables the organic organization, analysis, prediction, and anomaly detection of data. The construction of the data model is based on multidimensional data in industrial scenarios, and provides support for business decision-making and operation optimization through data cleaning, integration, modeling, and analysis; The rule model is based on a Boolean function, and obtains an alarm state based on the rule according to the point data of the target object; Preferably, the module M2 includes: collecting target object point data from the PLC or WinCC system at a fixed frequency and saving it in a data table of the IHD real-time database; based on the IHD real-time database, obtaining real-time data and historical data of the point through the http interface; the structure of the data table includes: sensor ID, timestamp, measurement value, status code; wherein the status code indicates the quality of data collection; The target object point data includes: the physical properties of the driven target object, including: speed, temperature and pressure.

[0011] Preferably, the module M3 comprises: Module M3.1: Build a data model management system, including: data model classification management, data model path configuration, data model dynamic loading and loading status management; The data model classification management is to classify the data models using a dictionary structure, and establish a mapping relationship between configuration parameters and detection functions for each type of data model; different data models construct different parameter configurations; The data model path configuration is to obtain the weight file paths of different types of data models based on the configuration parameters of each type of data model, and store the obtained paths in the corresponding dictionary structure; The data model is dynamically loaded by checking the model path defined in the configuration file, loading the formulated data model detection function, and loading the trained weight file based on the detection function; The loading state management is to record the loading and unloading state of the data model through the state dictionary; Module M3.2: Build a rule model management system, including: rule definition, trigger configuration, rule parsing and logical operations; The trigger configuration includes: a time-driven mechanism, an event-driven mechanism, and a driving mechanism combining events and time; The rule analysis and the logic operation are to analyze the rules and trigger configurations in the rule model, and execute corresponding tasks based on the rule judgment function.

[0012] Preferably, the module M4 comprises: Module M4.1: Build a system status management system, including: initialization management when the server starts, including: current status check, logging, exception handling and global status update; Module M4.2: Build a multi-task scheduling management system, including: task definition, task status management, task parameter persistence and task batch management; Module M4.3: Heterogeneous tasks are issued. According to different task requirements, the corresponding data model, rule model, or data model and rule model are called to calculate the corresponding business indicators based on the point data in the IHD real-time database.

[0013] Preferably, the module M5 includes: storing the calculated business indicators in a db2 database, and then using a kafka message queue to push the latest business indicators to the digital twin 3D model.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention builds a unified digital twin model management and collaborative analysis framework to realize data processing and collaborative analysis in complex industrial environments; 2. Combine the task scheduling module with the digital twin model to dynamically adjust the task execution order and resource allocation, so that the industrial system can respond to changes in the external environment in real time; 3. Integrate multiple business models and link them with digital twin models to support cross-domain and multi-dimensional collaborative analysis to form a comprehensive assessment of the product's operating status. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings: Figure 1 Flowchart of the model management and collaborative analysis method based on digital twins in the present invention Figure 2 It is a data model parameter configuration diagram in the present invention; Figure 3 It is a parameter configuration diagram of the rule model in the present invention; Figure 4 It is the multi-task scheduling management screen in the present invention; Figure 5 This is a 3D simulation sample picture of the digital twin in the present invention. DETAILED DESCRIPTION

[0016] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0017] Example 1 The present invention provides a model management and collaborative analysis method and system based on digital twins, which combines digital twin technology with task scheduling and business models to realize data processing and collaborative analysis in complex industrial environments.

[0018] like Figure 1 As shown, a model management and collaborative analysis method based on digital twins includes the following steps: Step 1: Create a digital twin 3D model of the product. The product can be a factory, production line, or equipment. Step 2: Collect product point data from the PLC or WinCC system at a fixed frequency and save it in the IHD real-time database. Provide an http interface to obtain real-time and historical data of the point. The data table structure includes sensor ID, timestamp, measurement value, and status code, where the status code indicates the quality of data collection.

[0019] Step 3: Configure business model parameters and data points. Business models include data-driven models and rule-driven models.

[0020] Step 3.1: Build a data model management system, including model classification management, model path configuration, model dynamic loading, and loading status management; The dictionary structure is used to classify models, and a mapping relationship between configuration parameters and detection functions is established for each type of model. Each type of model has corresponding detection functions and state management logic.

[0021] The paths of different types of model weights are read from the given configuration parameters, and the paths are stored in the corresponding dictionaries. Different model types have different parameter configuration logics. In this embodiment, the data model is divided into a prediction model and an anomaly detection model. The prediction model includes single signal prediction, multi-signal prediction, and online incremental learning model. The prediction model configuration parameters include model name, algorithm type, input length, output length, detection frequency, and data point position. The anomaly detection model configuration parameters include: model name, algorithm type, and data point position.

[0022] like Figure 2 As shown in the figure, after the model configuration parameters are set, a dynamic loading mechanism is used to load the specified model detection function by checking the model path defined in the configuration file instead of directly loading all models. This makes the system lighter at startup, and only loads specific models when needed.

[0023] Initialize a state dictionary for each model to record the loading and unloading status of the model. A state of 0 means that the model is not loaded, and a state of 1 or n means that the model has been loaded 1 or n times. Each unloading state is reduced by 1. The loading process of the model is controlled by the state variable to avoid repeated loading of the model and save computing resources.

[0024] Step 3.2: Build a rule model management system, including rule definition, trigger configuration, rule parsing and logical operations.

[0025] Each rule can be viewed as a Boolean function, the input is the product point data, and the output is the alarm status (true / false) based on the rule. In the embodiment, the configured rule types are as follows:

[0026] Different trigger configurations have different rule execution modes. In the embodiment, the trigger configuration types include time-driven mechanism, event-driven mechanism, and event-time combined driven mechanism, and the corresponding status codes are 0, 1, and 2, respectively, as follows:

[0027] like Figure 3 As shown, the incoming rules and trigger configurations are parsed and the corresponding tasks are executed. For some rule types, there may be nested sub-rule configurations. The function parses these nested structures by recursively calling itself. The start event and end event in the trigger configuration are essentially rules and can be recursively parsed. Among them, the trigger configuration parameters, including monitoring start time, monitoring end time, delayed detection time, detection frequency, detection point, start event, end event, detection duration, etc., all need to be converted into a unified time unit (such as seconds) for processing.

[0028] Rules can be combined into complex rules through logical operations such as "AND", "OR", and "NOT". For complex rules, it is necessary to traverse the rule configuration and execute the corresponding rule judgment function. Judge the exception according to the rule type, and summarize the judgment results according to the relationship between the rules (AND / OR).

[0029] For AND relationship: All rules must trigger an exception to be considered as an overall exception. Formula:

[0030] For OR relationship: As long as one rule triggers an exception, it is an overall exception. Formula:

[0031] Step 4: Build a multi-task scheduling management system to implement heterogeneous task distribution and collaborative analysis of business models.

[0032] Step 4.1: Build a system status management system: The initialization process when the server starts involves checking the current system status, logging, exception handling, and global status updates.

[0033] Define the system status check result set: All_Status = {S1, S2, S3, S4, S5}, where S1 is the real-time database connection status, S2 is the rule task scheduling connection status, S3 is the algorithm library connection status, S4 is the anomaly detection model task scheduling connection status, and S5 is the prediction model task scheduling connection status. When all states are connected normally, the system initialization process is completed, otherwise it exits abnormally and records the log.

[0034] Step 4.2 builds a multi-task scheduling management system, including task definition, task status management, task parameter persistence and task batch management.

[0035] The scheduling task can be expressed as a function S(w, r, t, p), where w represents the warning ID, r represents the business model configuration, t represents the trigger configuration, and p represents the priority setting of the task. High-priority tasks preempt resources in scheduling, and their execution order is guaranteed by the priority queue. The priority p of the task will be dynamically adjusted according to the scheduling system, and critical tasks will be inserted into the high-priority queue in real time to ensure their rapid response and execution. The system monitors the load status of the Worker (such as CPU usage, memory usage, task queue length, etc.) in real time through status monitoring. When the node load is too high, non-critical tasks in task S(w, r, t, p) are transferred to other idle nodes first to avoid task backlogs caused by resource bottlenecks.

[0036] If the task scheduling is successful, the task ID is returned; otherwise, a failure status is returned. In the embodiment, the scheduling tasks are divided into three categories according to the business model: monitoring tasks based on expert rules, abnormal monitoring tasks based on data models, and monitoring tasks based on trend prediction and rule coordination.

[0037] During the task scheduling process, the system decides whether to terminate the task according to the current state of the task and updates the task status information in the database. The corresponding task status codes are 0 to 4. In the embodiment, 5 task states are set, and the corresponding task status codes are as follows:

[0038] When processing business models, scheduling tasks involve conditions such as abnormal duration and abnormal number, and these status information needs to be persisted through Redis. In an embodiment, for the management of abnormal duration and abnormal number, Redis key-value pairs are used for persistence: k is the key name, usually a combination of the warning ID and the task type, and v is the last status value, such as the last abnormal occurrence time and abnormal number.

[0039] like Figure 4As shown in the figure, after the tasks are created, they can be managed in batches. Suppose the set of all tasks to be managed in the scheduling engine is T={t1,t2,…,tn}, where tn represents the nth task. Select the operation o∈O to be performed, such as o=start, o=stop, or o=delete. Apply operation o to each task tj∈S in the selected subset S to form a new task set T′=A(S,o); when operation o=start, the task status in S is updated to "start"; when operation o=stop, the task status in S is updated to "stop"; when operation o=delete, the tasks in S are removed from T.

[0040] Step 4.3: Heterogeneous task distribution and business model collaborative analysis.

[0041] After heterogeneous tasks are issued, the data model and rule model can be executed independently. When the data model and rules are analyzed collaboratively, the applicability of both needs to be verified. Check whether the label in the rule configuration exists in the model configuration parameter. The matching process can be expressed as: M=m(C,T), where C represents the model parameter dictionary, T represents the rule configuration dictionary, and M represents the matching result (True or False). When the matching result is True, the task is issued for execution; when the matching result is False, the task is not issued and the matching failure status information is returned.

[0042] Step 5: The business model execution results are saved in the DB2 database, and the Kafka message queue is used to push the latest result data to the digital twin 3D model.

[0043] Step 6: Figure 5 As shown in the figure, the digital twin constructed in this embodiment receives and verifies the dynamic data transmitted by the background, and restores the intrinsic properties of the product. The dynamic data includes two parts: one part is the point data collected in real time, which is used to drive the physical properties of the product, such as speed, temperature, pressure, etc.; the other part is the execution result of the business model, which is used to predict the potential state of the product, such as whether there are potential anomalies, future change trends, fault diagnosis results, etc.

[0044] The present invention also provides a model management and collaborative analysis system based on digital twins. The model management and collaborative analysis system based on digital twins can be implemented by executing the process steps of the model management and collaborative analysis method based on digital twins, that is, those skilled in the art can understand the model management and collaborative analysis method based on digital twins as a preferred implementation mode of the model management and collaborative analysis system based on digital twins.

[0045] Those skilled in the art know that, in addition to realizing the system and its various devices, modules, and units provided by the present invention in a purely computer-readable program code, the system and its various devices, modules, and units provided by the present invention can be completely realized in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps to realize the same functions. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered as a hardware component, and the devices, modules, and units included therein for realizing various functions can also be regarded as structures within the hardware component; the devices, modules, and units for realizing various functions can also be regarded as both software modules for realizing the method and structures within the hardware component.

[0046] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A model management and collaborative analysis method based on digital twins, characterized in that: include: Step S1: construct a digital twin 3D model of the target object; Step S2: collecting target object point data at a fixed frequency and saving it in the IHD real-time database; Step S3: construct different business models, including: data model and rule model; Step S4: Select a business model according to the demand, and calculate the corresponding business indicators based on the point data in the IHD real-time database; Step S5: Push the calculated business indicators to the digital twin 3D model; Step S6: The digital twin 3D model restores the intrinsic properties of the target object based on the collected point data of the target object and the received business indicators; The data model is a mathematical or statistical model built based on point data for business objectives, which realizes the organic organization, analysis, prediction and anomaly detection of data; The rule model is based on a Boolean function, and obtains an alarm state based on the rule according to the point data of the target object.

2. The model management and collaborative analysis method based on digital twins according to claim 1 is characterized in that: The step S2 includes: collecting target object point data from the PLC or WinCC system at a fixed frequency and saving it in a data table of the IHD real-time database; based on the IHD real-time database, obtaining real-time data and historical data of the point through the http interface; the structure of the data table includes: sensor ID, timestamp, measurement value, status code; wherein the status code indicates the quality of data collection; The target object point data includes: the physical properties of the driven target object, including: speed, temperature and pressure.

3. The model management and collaborative analysis method based on digital twins according to claim 1 is characterized in that: The step S3 comprises: Step S3.1: Construct a data model management system, including: data model classification management, data model path configuration, data model dynamic loading and loading status management; The data model classification management is to classify the data models using a dictionary structure, and establish a mapping relationship between configuration parameters and detection functions for each type of data model; different data models construct different parameter configurations; The data model path configuration is to obtain the weight file paths of different types of data models based on the configuration parameters of each type of data model, and store the obtained paths in the corresponding dictionary structure; The data model is dynamically loaded by checking the model path defined in the configuration file, loading the formulated data model detection function, and loading the trained weight file based on the detection function; The loading state management is to record the loading and unloading state of the data model through the state dictionary; Step S3.2: Building a rule model management system, including: rule definition, trigger configuration, rule parsing and logical operations; The trigger configuration includes: a time-driven mechanism, an event-driven mechanism, and a driving mechanism combining events and time; The rule analysis and the logic operation are to analyze the rules and trigger configurations in the rule model, and execute corresponding tasks based on the rule judgment function.

4. The model management and collaborative analysis method based on digital twins according to claim 1, characterized in that: The step S4 comprises: Step S4.1: Build a system status management system, including: initialization management when the server starts, including: current status check, log recording, exception handling and global status update; Step S4.2: Construct a multi-task scheduling management system, including: task definition, task status management, task parameter persistence and task batch management; Step S4.3: Heterogeneous tasks are issued. According to different task requirements, the corresponding data model, rule model, or data model and rule model are called to calculate the corresponding business indicators based on the point data in the IHD real-time database.

5. The model management and collaborative analysis method based on digital twins according to claim 1 is characterized in that: The step S5 includes: storing the calculated business indicators in the db2 database, and then using the kafka message queue to push the latest business indicators to the digital twin 3D model.

6. A model management and collaborative analysis system based on digital twins, characterized in that: include: Module M1: Build a digital twin 3D model of the target object; Module M2: collects target object point data at a fixed frequency and saves it in the IHD real-time database; Module M3: Build different business models, including data model and rule model; Module M4: Select the business model according to the demand and calculate the corresponding business indicators based on the point data in the IHD real-time database; Module M5: Push the calculated business indicators to the digital twin 3D model; Module M6: The digital twin 3D model restores the intrinsic properties of the target object based on the collected point data of the target object and the received business indicators; The data model is a mathematical or statistical model built based on point data for business objectives, which realizes the organic organization, analysis, prediction and anomaly detection of data; The rule model is based on a Boolean function, and obtains an alarm state based on the rule according to the point data of the target object.

7. The digital twin-based model management and collaborative analysis system according to claim 6, characterized in that: The module M2 includes: collecting target object point data from the PLC or WinCC system at a fixed frequency and saving it in the data table of the IHD real-time database; based on the IHD real-time database, obtaining the real-time data and historical data of the point through the http interface; the structure of the data table includes: sensor ID, timestamp, measurement value, status code; wherein the status code indicates the quality of data collection; The target object point data includes: the physical properties of the driven target object, including: speed, temperature and pressure.

8. The digital twin-based model management and collaborative analysis system according to claim 6, characterized in that: The module M3 comprises: Module M3.1: Build a data model management system, including: data model classification management, data model path configuration, data model dynamic loading and loading status management; The data model classification management is to classify the data models using a dictionary structure, and establish a mapping relationship between configuration parameters and detection functions for each type of data model; different data models construct different parameter configurations; The data model path configuration is to obtain the weight file paths of different types of data models based on the configuration parameters of each type of data model, and store the obtained paths in the corresponding dictionary structure; The data model is dynamically loaded by checking the model path defined in the configuration file, loading the formulated data model detection function, and loading the trained weight file based on the detection function; The loading state management is to record the loading and unloading state of the data model through the state dictionary; Module M3.2: Build a rule model management system, including: rule definition, trigger configuration, rule parsing and logical operations; The trigger configuration includes: a time-driven mechanism, an event-driven mechanism, and a driving mechanism combining events and time; The rule analysis and the logic operation are to analyze the rules and trigger configurations in the rule model, and execute corresponding tasks based on the rule judgment function.

9. The digital twin-based model management and collaborative analysis system according to claim 6, characterized in that: The module M4 comprises: Module M4.1: Build a system status management system, including: initialization management when the server starts, including: current status check, logging, exception handling and global status update; Module M4.2: Build a multi-task scheduling management system, including: task definition, task status management, task parameter persistence and task batch management; Module M4.3: Heterogeneous tasks are issued. According to different task requirements, the corresponding data model, rule model, or data model and rule model are called to calculate the corresponding business indicators based on the point data in the IHD real-time database.

10. The digital twin-based model management and collaborative analysis system according to claim 6, characterized in that: The module M5 includes: storing the calculated business indicators in the db2 database, and then using the kafka message queue to push the latest business indicators to the digital twin 3D model.

Images