A digital twin model intelligent correction method and system based on multi-source data fusion

Abstract

The application discloses a kind of based on multi-source data fusion's digital twinborn model intelligent correction method and system, belong to digital twinborn simulation calibration technical field.The present digital twinborn model is prone to produce virtual-real deviation for long-term operation, artificial calibration lags behind, correction dimension is single, and the technical problem of poor dynamic adaptability, the application is standardized by fusing physical entity sensing, working condition, environment and historical multi-source data and completing preprocessing, simultaneously realizes the time sequence comparison of model simulation data and field measured data, and the hierarchical classification determination of parameter drift, structural deviation, time sequence synchronization deviation is completed based on deviation feature;Different deviation types are adopted Differentiated intelligent correction strategy, respectively complete model parameter setting, structure weight optimization, time sequence delay compensation, and the correction result is solidified through closed-loop iterative verification mechanism.The application can realize the automatic, high-precision and dynamic intelligent correction of digital twinborn model throughout its life cycle, effectively improve the consistency of virtual-real state mapping and simulation accuracy, reduce the cost of artificial operation and maintenance, adapt to various industrial digital twinborn application scenarios such as intelligent manufacturing, energy equipment and engineering monitoring, and have good engineering application and popularization value.

Description

Technical Field

[0001] This invention relates to the fields of digital twin modeling, industrial simulation, and synchronous calibration of virtual and real states, specifically to an intelligent correction method and system for digital twin models based on multi-source data fusion. Background Technology

[0002] Digital twin technology constructs digital mirror images of physical equipment, production conditions, and engineering structures to achieve state mapping, simulation, and operational monitoring between physical and digital spaces. It is now widely used in industrial scenarios such as intelligent manufacturing, energy equipment, engineering construction, and urban operation and maintenance. The mapping accuracy and simulation accuracy of the digital twin model are the core foundation for ensuring the reliable operation of state monitoring, fault prediction, and operational optimization functions. Once established, existing digital twin models typically have fixed parameters, simulation structures, and operational logic. However, during the long-term operation of the physical entity, its operational characteristics gradually change due to factors such as equipment wear, component aging, environmental changes, operating condition switching, and load fluctuations. A fixed-configuration digital twin model cannot adjust synchronously with these changes, causing a gradual discrepancy between simulation data and actual field measurements. This discrepancy accumulates over time, leading to decreased simulation accuracy and a mismatch between the virtual and real-world mapping. Currently, the industry's primary method for correcting model deviations is manual, periodic calibration. Technicians periodically compare on-site measured data with model simulation data and manually adjust model parameters based on engineering experience. This method suffers from significant lag, high labor costs, and reliance on personnel experience, resulting in poor consistency and low accuracy. It cannot meet the correction requirements of continuous, dynamic equipment operation. Furthermore, existing correction methods mostly adjust only single parameter deviations, failing to differentiate between different types of deviations such as parameter drift, structural adaptation, and timing synchronization. This lack of specificity hinders comprehensive, high-precision dynamic calibration of the model and fails to meet the requirements of long-term stable operation in engineering applications. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of existing technologies, such as lagging model correction, low calibration accuracy, high dependence on manual intervention, single deviation identification, and inability to dynamically adapt to changes in working conditions. It provides an intelligent correction method and system for digital twin models based on multi-source data fusion, which realizes automatic identification, classification correction, closed-loop verification, and dynamic iteration of deviations in digital twin models, thereby improving the consistency of virtual-real mapping and the accuracy of model simulation.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides an intelligent correction method for digital twin models based on multi-source data fusion. First, the multi-source operational data from the field is standardized and preprocessed to ensure the integrity of the data sequence and the uniformity of dimensions. Then, the model deviation characteristics are obtained through virtual-real synchronous simulation comparison, and the deviation type and level are classified and determined according to the deviation change law. Corresponding correction strategies are adopted for different deviation types to complete model parameter tuning, structural weight adaptation, and time delay compensation, respectively. Finally, the correction effect is verified through a closed-loop verification mechanism. If the accuracy requirements are not met, automatic iterative optimization is performed, ultimately realizing adaptive correction of the digital twin model throughout its entire life cycle. Furthermore, the present invention is equipped with a dedicated correction system, which, through the coordinated operation of multiple modules, realizes the fully automated operation of data processing, deviation identification, intelligent correction, iterative verification, data storage and anomaly warning, without the need for continuous manual intervention, thus greatly improving the efficiency of model calibration and long-term operating accuracy.

[0005] Compared with the prior art, the present invention has the following advantages: 1. This invention achieves unified and standardized processing of multi-source operational data, effectively eliminating identification errors caused by data loss, noise interference, and timing misalignment, and providing a reliable data foundation for accurate determination of model deviation. 2. This invention adopts a deviation classification and identification method, which can accurately distinguish three common model errors: parameter drift, structural deviation, and timing synchronization deviation. This solves the problem that traditional correction methods cannot specifically handle multiple types of deviations, and the correction is more targeted. 3. This invention adopts a classification and adaptation correction strategy, which simultaneously completes model correction from three dimensions: parameters, structure, and time series. This solves the defects of traditional single-dimensional correction, such as incompleteness and poor adaptability, and effectively improves the accuracy of virtual-real mapping. 4. The present invention sets up a closed-loop iterative verification mechanism, which can automatically verify the correction effect and iteratively optimize, and can adapt to complex scenarios such as long-term aging of physical entities and dynamic fluctuations in working conditions, ensuring the stability and accuracy of the model in long-term operation. 5. This invention features fully automated execution without the need for manual calibration, reducing the operation and maintenance costs of the digital twin system and making it suitable for large-scale applications of digital twin scenarios for various industrial equipment and engineering structures. Attached Figure Description

[0006] Figure 1 This is a flowchart illustrating the overall process of this invention. Figure 2 This is a block diagram of the intelligent correction system architecture of the present invention.

Claims

1. A method for intelligent correction of digital twin models based on multi-source data fusion, characterized in that, Includes the following steps: Step 1: Data Acquisition and Standardization Processing. Real-time acquisition of sensor data, operating condition data, environmental parameters, and historical operating data during the operation of the physical entity. Outlier removal, noise filtering, missing data completion, and dimension normalization are performed on the raw acquired data to obtain time-aligned standardized measured data. Step 2: Simultaneous comparison of virtual and real data. Simultaneously perform simulation calculations using the current digital twin model to generate simulation data that corresponds to the time series and dimensions of the measured data. Establish the correspondence between the physical measured data and the model simulation data, and calculate the overall model deviation, time series deviation, and feature deviation parameters. Step 3: Determine the type of deviation. Based on the preset deviation feature rule library, match and identify the current virtual and real deviation features, distinguish three types of deviation: parameter drift deviation, model structure deviation, and timing synchronization deviation, and classify the deviation level according to the deviation amplitude to determine the priority of correction execution. Step 4: Classification and Adaptive Correction. Corresponding correction strategies are adopted for different types of deviations. The inherent operating parameters and coupling coefficients of the model are optimized and tuned. The weights of the model sub-modules and simulation logic are adapted and adjusted. The timing delay of virtual and real communication is dynamically compensated to complete the overall state correction of the digital twin model. Step 5: Closed-loop verification and iteration. The corrected digital twin model is put into online simulation. The measured data and simulation data are collected again to verify the deviation. When the deviation meets the preset accuracy requirements, the model parameters are fixed and the correction is completed. When the deviation does not meet the accuracy requirements, the comparison, judgment and correction steps are repeated to achieve closed-loop iterative optimization.

2. The intelligent correction method for digital twin models based on multi-source data fusion according to claim 1, characterized in that: In step one, outlier removal uses statistical discrimination to screen out abrupt interference data, noise filtering uses wavelet smoothing to remove high-frequency noise, and interpolation is used to fill in time-series gaps for monitoring data with different sampling frequencies, unifying data units and value ranges to ensure that the time-series axes of virtual and real data are completely consistent.

3. The intelligent correction method for digital twin models based on multi-source data fusion according to claim 1, characterized in that: In step three, parameter drift deviation manifests as a steady-state, persistent, small deviation; model structure deviation manifests as a systematic offset deviation under operating condition switching; and timing synchronization deviation manifests as data timing misalignment and instantaneous fluctuation deviation. The classification and determination are completed by the duration of the deviation, the fluctuation pattern, and the operating condition correlation characteristics.

4. The intelligent correction method for digital twin models based on multi-source data fusion according to claim 1, characterized in that: In step four, to address parameter drift deviations, a global optimization approach is used to optimize model physical parameters, operational coupling parameters, and boundary condition parameters; to address model structural deviations, the weight coefficients of each simulation submodule are adjusted based on historical operational deviation patterns to correct multi-field coupled simulation logic; and to address timing synchronization deviations, real-time transmission delay and simulation calculation delay are statistically analyzed to dynamically adjust the model simulation scheduling cycle and achieve virtual real-time synchronization.

5. The intelligent correction method for digital twin models based on multi-source data fusion according to claim 1, characterized in that: In step five, a fixed precision threshold and a maximum number of iterations are set. During the iteration process, the deviation change data and model correction parameters are recorded in real time. When the iteration exceeds the limit, an abnormal prompt is output and the original data is retained for subsequent model optimization iterations.

6. A digital twin model intelligent correction system based on multi-source data fusion, characterized in that, The method for performing the correction method according to any one of claims 1 to 5 includes a data preprocessing module, a virtual-real comparison module, a deviation identification module, an adaptive correction module, a closed-loop iterative verification module, and a data storage early warning module. The data preprocessing module is used to complete the collection, cleaning, completion and standardization of multi-source operating data; The virtual-real comparison module is used to synchronously call the digital twin model to perform simulation calculations, construct the correspondence between virtual and real data, and calculate the deviation index. The deviation identification module is used to classify deviation types and determine deviation levels, and output correction strategy instructions; The adaptive correction module is used to complete parameter tuning, structural adaptation and timing compensation correction according to the type of deviation. The closed-loop iterative verification module is used to complete the verification of the correction effect and iterative optimization control; The data storage and early warning module is used to store model iteration versions, correction logs, and deviation data, and to provide early warnings for abnormal operating conditions that exceed limits.

7. The intelligent correction system for digital twin models based on multi-source data fusion according to claim 6, characterized in that: The adaptive correction module includes a parameter correction unit, a structure optimization unit, and a timing compensation unit. Each unit operates independently and works in coordination to complete the correction of three types of deviations: parameter drift, structure adaptation, and timing synchronization.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent correction method for digital twin models based on multi-source data fusion as described in any one of claims 1 to 5.

Images