Dynamic compensation method and system for digital twin optimization

Abstract

The application relates to the technical field of digital twin optimization, and discloses a dynamic compensation method and system for digital twin optimization. The method comprises the following steps: acquiring historical full-amount data of a digital twin body, and screening abnormal entity object data; constructing a compensation control strategy and deducing first compensation target data; calculating a compensation target difference value of adjacent stages, and calculating a main compensator compensation parameter in combination with the characteristics of the abnormal entity object data when the compensation target difference value reaches a preset threshold; collecting state residual data of a sub-compensator, generating a compensation distribution instruction, and enabling the sub-compensator to perform basic compensation adjustment; supplementing the latest operation data to acquire second compensation target data, comparing the generated result, and correcting and adjusting the operation. The system comprises an acquisition unit, a deduction unit, a measurement unit, a collection unit, a distribution unit, an updating and judging module, and the like. The application enhances the response capability of the system to dynamic changes, ensures high matching between a digital twin model and a physical entity, and improves the reliability and efficiency of industrial system operation.

Description

Technical Field

[0001] This application relates to the field of digital twin optimization technology, specifically to a dynamic compensation method and system for digital twin optimization. Background Technology

[0002] In fields such as intelligent manufacturing and industrial automation, digital twin technology, by constructing virtual digital models of physical entities, enables real-time mapping and prediction of the entity's operational status, and has become an important means to improve system reliability and efficiency. However, in the actual operation of digital twin systems, due to the complexity of physical entities, dynamic changes in environmental factors, and errors in data acquisition and transmission, deviations inevitably occur between the digital twin and the actual physical entity. If these deviations are not corrected in time, the predictive accuracy of the digital twin model will decrease, thereby affecting the effectiveness of model-based decision-making and control.

[0003] In existing technologies, most compensation methods for digital twins employ static or semi-static compensation strategies. For example, some solutions adjust the digital twin model through pre-set compensation parameters. However, this approach struggles to adapt to the dynamic characteristics of physical entities evolving over time, especially when faced with sudden anomalies or complex changes in operating conditions, often resulting in poor compensation effectiveness. Furthermore, existing methods typically lack targeted screening and analysis of abnormal entity data when processing historical data, leading to imprecise compensation strategies and an inability to effectively address issues such as parameter offsets.

[0004] In constructing compensation and control strategies, traditional methods often neglect the forward-looking projection of compensation targets, making it difficult to accurately predict compensation needs in future stages, resulting in compensation operations lagging behind actual deviations. Furthermore, in the compensation allocation process, existing technologies typically employ a uniform compensation allocation method, failing to fully consider the remaining state data of sub-compensators and the characteristics of abnormal entity object data, thus preventing the rational allocation of compensation resources and impacting overall compensation efficiency.

[0005] Existing digital twin compensation systems lack effective real-time feedback and correction mechanisms. When actual operating data deviates from the preset compensation target, the compensation strategy cannot be adjusted in a timely manner based on the latest data, making it difficult to continuously optimize compensation accuracy and limiting the system's adaptability and stability. As the intelligence level of industrial systems continues to increase, higher demands are placed on the real-time performance, accuracy, and adaptability of digital twin dynamic compensation. Traditional compensation methods are no longer sufficient to meet the application needs of complex industrial scenarios, and there is an urgent need for a compensation method and system that can dynamically adapt and precisely control. Summary of the Invention

[0006] The purpose of this application is to provide a dynamic compensation method and system for digital twin optimization, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, this application provides the following technical solution: a dynamic compensation method for digital twin optimization, the method comprising: acquiring the historical full-scale operation data of the digital twin, filtering abnormal entity object data that evolves over time, the abnormal entity object data including parameter offset values; constructing a compensation control strategy based on the historical full-scale operation data, and extrapolating compensation targets for multiple stages to obtain first compensation target data; calculating the difference between the compensation targets of two adjacent stages in the first compensation target data, and when the difference reaches a preset threshold, calculating compensation parameter data of the main compensator based on the correlation between the difference and the characteristics of the abnormal entity object data; collecting the remaining state data of multiple sub-compensators within a preset coverage area based on the compensation parameter data; generating a compensation allocation instruction based on the remaining state data, transmitting the compensation allocation instruction to multiple sub-compensators, the compensation allocation instruction being used to instruct the on-site execution of adjustment operations that satisfy basic compensation based on the remaining state data; supplementing the latest stage's actual operation data based on the compensation control strategy, and acquiring the second compensation target data for the next stage, comparing the first compensation target data and the second compensation target data for the next stage, generating a comparison result, and determining whether to correct the adjustment operation of the basic compensation based on the comparison result.

[0008] Preferably, the step of acquiring the full historical operational data of the digital twin and filtering out abnormal entity object data that evolves over time includes: acquiring the full historical operational data of the digital twin within a set historical period; selecting all time periods that are in the same stage as the current monitoring period based on the set historical period; and filtering out abnormal entity object data for all time periods in the full historical operational data.

[0009] Preferably, the step of constructing a compensation and control strategy based on historical full-scale operation data and extrapolating compensation targets for multiple stages to obtain the first compensation target data includes: constructing a compensation and control strategy based on abnormal entity object data for all time periods, generating a first compensation curve that meets the first matching condition; and extrapolating compensation targets for multiple stages based on the first compensation curve to obtain the first compensation target data.

[0010] Preferably, the step of calculating the compensation parameter data of the main compensator based on the correlation between the difference and the characteristics of the abnormal entity object data includes: obtaining the characteristic composition information of a single abnormal entity object data in the difference; and calculating the compensation parameter data of the main compensator based on the characteristic composition information and the total number of abnormal entity object data in the difference, wherein the compensation parameter data includes the corresponding compensation quantity for the characteristics of the abnormal entity object data.

[0011] Preferably, the step of generating a compensation allocation instruction based on the remaining state data and transmitting the compensation allocation instruction to multiple sub-compensators includes: dividing a preset coverage area into segments centered on the main compensator in order from near to far, resulting in multiple sub-regions within the preset sub-range; selecting a sub-compensator within each sub-region, and determining the basic adjustment amount within each sub-region whose parameter offset value is within a set threshold based on the remaining state data of the sub-compensators within each sub-region; determining the sub-compensators within each sub-region that meet the basic adjustment amount, resulting in multiple sub-compensators, wherein the sum of the basic adjustment amounts of the multiple sub-compensators is greater than or equal to the basic compensation; and generating and transmitting a compensation allocation instruction based on the multiple sub-compensators and the basic adjustment amount of each sub-compensator.

[0012] Preferably, the step of supplementing the latest stage of actual operating data based on the compensation and control strategy, obtaining the second compensation target data for the next stage, and comparing the first compensation target data and the second compensation target data for the next stage to generate a comparison result includes: updating the compensation and control strategy based on the latest stage of operating data to generate a second compensation curve that meets the second matching condition; identifying the second compensation target data for the next stage in the second compensation curve; comparing the fluctuations of the next stage based on the first compensation target data and the second compensation target data to generate a comparison result.

[0013] Preferably, determining whether to correct the adjustment operation of the basic compensation based on the comparison results includes: when the deviation between the second compensation target data and the compensation target data of the next stage in the first compensation target data is within a preset deviation, generating a first instruction transmitted based on multiple sub-compensators to maintain all the adjustment operations of the basic compensation; when the second compensation target data is less than the compensation target data of the next stage in the first compensation target data, and the deviation between the two is greater than a preset deviation, generating an integration and maintenance instruction based on the adjustment operations of the basic compensation in the multiple sub-compensators, so that at least one sub-compensator in the multiple sub-compensators performs partial characteristic integration on other sub-compensators besides itself, and maintains the adjustment operations of the basic compensation in at least one sub-compensator; when the second compensation target data is greater than the compensation target data of the next stage in the first compensation target data, and the deviation between the two is greater than a preset deviation, generating a second instruction transmitted to the multiple sub-compensators to maintain all the adjustment operations of the basic compensation.

[0014] Preferably, this application also includes a dynamic compensation system for digital twin optimization, the system comprising: a data acquisition module for acquiring the full historical operational data of the digital twin and filtering abnormal entity object data that evolves over time, the abnormal entity object data including parameter offset values; a deduction module for constructing a compensation control strategy based on the full historical operational data and deducing compensation targets for multiple stages to obtain first compensation target data; and a calculation module for calculating the difference between the compensation targets of two adjacent stages in the first compensation target data, and when the difference reaches a preset threshold, calculating the compensation parameter data of the main compensator based on the correlation between the difference and the characteristics of the abnormal entity object data. The acquisition unit is used to acquire the remaining status data of multiple sub-compensators within a preset coverage area based on compensation parameter data; the allocation module is used to generate compensation allocation instructions based on the remaining status data and transmit the compensation allocation instructions to multiple sub-compensators. The compensation allocation instructions are used to instruct the adjustment operation that satisfies the basic compensation to be performed locally based on the remaining status data; the update and judgment module is used to supplement the latest stage of actual operating data based on the compensation control strategy, obtain the second compensation target data for the next stage, compare the first compensation target data and the second compensation target data for the next stage, generate a comparison result, and determine whether to correct the adjustment operation of the basic compensation based on the comparison result.

[0015] Preferably, the allocation module includes: a segmentation unit, used to divide the preset coverage area into segments from near to far, centered on the main compensator, to obtain multiple sub-regions within the preset sub-range; a selection unit, used to select the sub-compensators within each sub-region, determine the basic adjustment amount within each sub-region whose offset value is within a set threshold based on the remaining status data of the sub-compensators within each sub-region, determine the sub-compensators within each sub-region that meet the basic adjustment amount, and obtain multiple sub-compensators, wherein the sum of the basic adjustment amounts of the multiple sub-compensators is greater than or equal to the basic compensation; and a generation unit, used to generate and transmit compensation allocation instructions based on the multiple sub-compensators and the basic adjustment amount of each sub-compensator.

[0016] Preferably, the generation unit is specifically used to: generate compensation allocation instructions based on multiple sub-compensators and the basic adjustment amount of each sub-compensator, and transmit them to the multiple sub-compensators.

[0017] Compared with the prior art, the beneficial effects of this application are: The dynamic compensation method and system for digital twin optimization provided in this application can accurately locate problems such as parameter offsets by acquiring the full historical operational data of the digital twin and filtering abnormal entity object data that evolves over time, providing a reliable basis for the formulation of subsequent compensation strategies. By constructing compensation and control strategies based on the full historical operational data and extrapolating compensation targets forward, the compensation operation becomes forward-looking, enabling early responses to potential deviations and improving the timeliness of compensation.

[0018] The main compensator's compensation parameters are determined by calculating the difference in compensation targets between adjacent stages and combining the correlation between the data characteristics of abnormal entity objects. This enables accurate calculation of compensation parameters, ensuring that the main compensator can provide appropriate compensation based on actual needs. The remaining status data of sub-compensators within a preset coverage area are collected based on the compensation parameters, and compensation allocation instructions are generated accordingly. This allows compensation allocation to be based on the actual status of the sub-compensators, optimizing the allocation of compensation resources and improving compensation efficiency.

[0019] By updating the compensation and control strategy with the latest actual operating data, obtaining the compensation target for the next stage and comparing it, the basic compensation adjustment operation can be corrected in real time based on the comparison results. This allows the compensation strategy to dynamically adapt to changes in actual operating conditions and continuously improve compensation accuracy. This dynamic adjustment mechanism effectively solves the problems of lagging compensation strategies and poor adaptability in existing technologies, enhancing the digital twin system's adaptability to complex operating conditions.

[0020] The system divides the coverage area into segments centered on the main compensator. Based on the remaining status data of the sub-compensators within each sub-region, it determines the basic adjustment amount and generates compensation allocation instructions. This achieves reasonable allocation of compensation tasks, ensures coordinated operation of all sub-compensators, and improves the overall compensation effect of the system. When deviations occur in the target compensation data, corresponding corrective measures are taken according to different deviation situations, further ensuring the accuracy of the compensation operation and the stability of the system. This effectively improves the matching degree between the digital twin and the physical entity, providing strong support for the efficient and reliable operation of industrial systems. Attached Figure Description

[0021] Figure 1 A schematic diagram illustrating the working principle of the dynamic compensation method for digital twin optimization provided in this application embodiment; Figure 2 Design diagrams for generating and transmitting compensation allocation instructions provided in embodiments of this application; Figure 3 A flowchart illustrating the updating and comparison of compensation and control strategies provided in this application's embodiments; Figure 4 This is an architecture diagram of the dynamic compensation system provided in the embodiments of this application. Detailed Implementation

[0022] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] Please see Figures 1-4 The dynamic compensation method for digital twin optimization involved in this application has the following specific implementation steps: The process involves acquiring the complete historical operational data of the digital twin and filtering for abnormal entity object data that evolves over time. This abnormal entity object data includes parameter offset values. Specifically, it involves acquiring the complete historical operational data of the digital twin within a set historical time period, selecting all time periods that are in the same phase as the current monitoring period based on the set historical time period, and then filtering for abnormal entity object data from all time periods within the complete historical operational data.

[0024] A compensation and control strategy is constructed based on historical full-scale operational data, and compensation targets for multiple stages are extrapolated forward to obtain the first compensation target data. Specifically, based on abnormal entity object data for all time periods, a compensation and control strategy is constructed to generate a first compensation curve that meets the first matching condition. Then, compensation targets for multiple stages are extrapolated forward from the first compensation curve to obtain the first compensation target data.

[0025] The difference between the compensation targets of two adjacent stages in the first compensation target data is calculated. When the difference reaches a preset threshold, the compensation parameter data of the main compensator is calculated based on the correlation between the difference and the characteristics of the abnormal entity object data. Specifically, the characteristic composition information of a single abnormal entity object data in the difference is obtained. Based on the characteristic composition information and the total number of abnormal objects in the difference, the compensation parameter data of the main compensator is calculated. The compensation parameter data includes the corresponding compensation quantity for the characteristics of the abnormal entity object data.

[0026] The remaining status data of multiple sub-compensators within the preset coverage area are collected based on the compensation parameter data.

[0027] A compensation allocation instruction is generated based on the remaining state data and transmitted to multiple sub-compensators. This instruction instructs the on-site execution of adjustments that satisfy the basic compensation based on the remaining state data. Specifically, with the main compensator as the center, a preset coverage area is divided into segments in ascending order, resulting in multiple sub-regions within the preset sub-range. Sub-compensators within each sub-region are selected, and based on the remaining state data of each sub-compensator, a basic adjustment amount is determined for each sub-region where the offset value is within a set threshold. Sub-compensators within each sub-region that satisfy the basic adjustment amount are identified, resulting in multiple sub-compensators, where the sum of the basic adjustment amounts of the multiple sub-compensators is greater than or equal to the basic compensation. A compensation allocation instruction is generated and transmitted based on the multiple sub-compensators and the basic adjustment amount of each sub-compensator.

[0028] Based on the compensation and control strategy, the latest stage's actual operating data is supplemented, and the second compensation target data for the next stage is obtained. A comparison is then made between the first and second compensation target data for the next stage to generate a comparison result. Based on the comparison result, it is determined whether the adjustment operation of the basic compensation should be corrected. Specifically, based on the latest stage's operating data, the compensation and control strategy is updated to generate a second compensation curve that meets the second matching condition; the second compensation target data for the next stage in the second compensation curve is identified, and a fluctuation comparison is made between the first and second compensation target data for the next stage to generate a comparison result. When the deviation between the second compensation target data and the next stage compensation target data in the first compensation target data is within a preset deviation, a first instruction is generated based on multiple sub-compensators to maintain all adjustments to the basic compensation. When the second compensation target data is less than the next stage compensation target data in the first compensation target data, and the deviation between the two is greater than a preset deviation, an integration and maintenance instruction is generated based on the basic compensation adjustments in the multiple sub-compensators, so that at least one sub-compensator integrates some characteristics of the other sub-compensators (excluding itself) and maintains the basic compensation adjustments in at least one sub-compensator. When the second compensation target data is greater than the next stage compensation target data in the first compensation target data, and the deviation between the two is greater than a preset deviation, a second instruction is generated to be transmitted to the multiple sub-compensators to maintain all adjustments to the basic compensation.

[0029] Example 1: When acquiring the complete historical operational data of a digital twin, it is necessary to first clearly define the historical time period. The determination of the historical time period should be based on the specific application scenario of the digital twin. For example, if the digital twin is used in industrial production equipment, its data update frequency is high and the production process cycle is short, so the historical time period can be selected from the past month; if it is used in urban infrastructure, its operational data changes relatively slowly and the cycle is long, so the historical time period can be selected from the past three months or even longer. After determining the historical time period, the complete historical operational data of the digital twin within that time period is acquired through a data acquisition system. This data covers various parameter information of the digital twin during operation, such as the equipment's temperature, pressure, speed, energy consumption, etc., as well as the status information of various abnormal entity objects, including normal operation status and abnormal operation status.

[0030] After acquiring all historical operational data, it's necessary to select all time periods within the same stage as the current monitoring period, based on a set historical timeframe. Determining "same stage" requires considering multiple factors. If the time period is used as the dividing line, and the current monitoring period is a Monday of a certain week, then the same stage periods would be Mondays of every week within the set historical timeframe. Alternatively, if the operational stage of the digital twin is used as the basis, for example, if the digital twin's production process is divided into startup, stable operation, and shutdown stages, and the current monitoring period is in the stable operation stage, then all time periods within the stable operation stage need to be identified within the set historical timeframe. When selecting the same stage periods, a detailed analysis of the historical operational data is required to extract key parameters that reflect the characteristics of each time period. The comparison of these key parameters is then used to determine the same stage periods.

[0031] After selecting the time period for each segment, the next step is to filter the historical full-scale data to obtain data on anomalous entities that have evolved over time. This filtering requires a deep understanding of the operational patterns of the digital twin. First, the normal operating parameter range for each anomalous entity is determined. This can be obtained through statistical analysis of historical normal operating data, such as calculating the average and standard deviation of each parameter, thus determining the upper and lower limits of normal operation. Then, for each segment of the historical full-scale data, the parameters of each anomalous entity are checked one by one to see if they exceed the normal range. When the parameter offset of an anomalous entity exceeds the normal range, that entity is marked as anomalous. During the filtering process, it is important to pay attention to how the parameter offset is calculated. It could be the difference between the actual and normal values, or the ratio of the difference to the normal value, etc. The specific calculation method needs to be determined based on the nature of the parameter and the application requirements.

[0032] In practice, data acquisition and filtering can face challenges such as massive data volumes and diverse data formats. For example, for large and complex digital twin systems, the total historical data can reach terabytes in size, containing various types of data, including structured, semi-structured, and unstructured data. To process this data efficiently, appropriate data storage and processing technologies are needed, such as distributed databases and big data processing frameworks. Simultaneously, to ensure data accuracy and integrity, data cleaning and preprocessing are required during data acquisition, including removing noisy data and handling missing values.

[0033] Furthermore, when filtering anomalous entity data, the factor of time evolution must be considered. Over time, the operating environment and conditions of a digital twin may change, and its normal operating parameter range may also change accordingly. Therefore, when filtering anomalous entity data, it is not sufficient to rely solely on fixed parameter ranges; the parameter ranges must be dynamically adjusted in conjunction with time series data. For example, a sliding window method can be used to update the normal operating parameter range based on historical data from a recent period, thereby more accurately identifying anomalous entity data that evolves over time.

[0034] Furthermore, the screening of anomalous entity object data also needs to consider the correlation between different anomalous entity object data. In a digital twin system, the various anomalous entity object data are interconnected and influence each other; an anomaly in one anomalous entity object data may cause changes in the operating parameters of other anomalous entity object data. Therefore, when screening anomalous entity object data, the parameter offset of a single anomalous entity object data cannot be viewed in isolation; it is also necessary to analyze its relationship with the parameters of other anomalous entity object data to avoid misjudgment and omission.

[0035] Throughout the process of acquiring and filtering data on abnormal entities, a robust data management mechanism is necessary to record and track information such as the data source, collection time, and processing procedure for subsequent querying and verification. Simultaneously, data security and privacy protection must be considered, and appropriate measures must be taken to prevent data leakage and tampering.

[0036] Through the above series of operations and processing, we can accurately obtain data on all abnormal entities within the same historical period and the current monitoring period. This data includes key information such as parameter offset values, providing a solid data foundation for subsequent operations such as constructing compensation and control strategies and performing compensation target deduction.

[0037] Example 2: When constructing a compensation and control strategy based on historical full-scale operational data and obtaining the first compensation target data, a systematic analysis of the abnormal entity object data for all acquired time periods is required. This data includes parameter offset values ​​and operational status information for different abnormal entity objects across different time periods. For example, it might include temperature anomaly offset data and pressure fluctuation exceeding limits data for a digital twin of industrial equipment during different production cycles. During the analysis, it is necessary to analyze the time-series characteristics of the data, the frequency and amplitude distribution of parameter offsets, and the correlation between anomalies in different abnormal entity object data. For instance, by statistically analyzing the number of parameter offsets and the range of offsets for a particular abnormal entity object in each corresponding time period, the regularity of its anomaly occurrence can be determined; or, when the temperature of a certain equipment component is abnormal, whether the pressure parameters of adjacent components also shift accordingly can be analyzed to identify potential patterns of anomaly propagation.

[0038] Based on the above analysis, a compensation and control strategy is being developed. When developing this strategy, the physical characteristics, operational logic, and business requirements of the digital twin must be considered. Taking an industrial production line digital twin as an example, if data from a certain type of abnormal entity indicates that the heating device frequently experiences temperature overshoot during a specific production stage, the compensation and control strategy needs to consider factors such as the heating device's heating rate, heat dissipation characteristics, and the temperature requirements of the production process, designing targeted compensation logic. The development of this strategy may involve the coordinated adjustment of multiple parameters. For example, it may not only adjust the heating power but also comprehensively regulate factors such as material conveying speed and the start-up and shutdown timing of the cooling system to form a systematic compensation scheme.

[0039] After constructing the compensation and control strategy, a first compensation curve that meets the first matching condition needs to be generated. The setting of the first matching condition must be based on the fit between the data characteristics and the compensation target. For example, if the offset of the abnormal entity object data shows a linear growth trend, the first matching condition can be set to use a linear fitting method, requiring that the mean square error between the fitted curve and the historical data does not exceed a certain threshold; if the data shows periodic fluctuation characteristics, the matching condition may tend to be a Fourier series fitting to ensure that the curve can accurately reflect the period and amplitude of the abnormal fluctuation. When generating the first compensation curve, the historical abnormal entity object data needs to be processed using a data fitting algorithm, such as fitting linear data using the least squares method, or determining the fitting parameters for periodic data through spectral analysis, so that the generated curve can fit the changing trend of the historical abnormal entity object data to the greatest extent, providing a reliable mathematical model for the subsequent deduction of the compensation target.

[0040] After generating the first compensation curve, it is necessary to extrapolate the compensation targets for multiple stages based on this curve to obtain the first compensation target data. The number of extrapolation stages needs to consider the operating cycle of the digital twin and the timeliness of the compensation response. For example, for industrial control scenarios with high real-time requirements, 5 to 10 short-cycle stages (e.g., each stage is 1 minute) can be extrapolated to allow for timely adjustments to the compensation strategy; for urban traffic flow digital twins, which have longer operating cycles, 1 to 2 daily or weekly cycle stages can be extrapolated. During the extrapolation process, the compensation target value for each stage needs to be calculated based on the mathematical model of the first compensation curve, combined with the current operating status of the digital twin and predicted environmental changes. For example, if the first compensation curve shows that the temperature of a certain device will continue to rise and exceed the threshold in the next 3 stages, the extrapolation process needs to calculate the required temperature compensation amount for each stage based on the curve slope and the current temperature value, forming the compensation target data for each stage, such as reducing the temperature by 2℃ in stage 1, reducing it by 3℃ in stage 2, etc.

[0041] In practice, constructing compensation and control strategies may face interference from data noise or outliers. For example, historical full-scale operational data may contain abrupt changes caused by sensor malfunctions; directly using this data for strategy construction would affect the strategy's accuracy. Therefore, before analyzing abnormal entity data, data cleaning is necessary. Statistical methods (such as the 3σ principle) should be used to identify and remove obvious noise points, or interpolation methods should be used to complete missing data, ensuring the quality of data used for strategy construction. Furthermore, when the operating scenario of the digital twin changes (such as a change in material type on an industrial production line), the characteristics of historical abnormal entity data may change accordingly. In this case, the compensation and control strategy and the first matching condition need to be dynamically adjusted, for example, by resetting the parameters of the fitting algorithm or changing the fitting model, to ensure that the first compensation curve can adapt to the new operating state.

[0042] The process of calculating compensation targets also needs to consider the constraints of the digital twin system. For example, in industrial equipment compensation, there is an upper limit to the adjustment of heating power, and the response speed of the cooling system has a certain delay. These physical constraints need to be taken into account when calculating compensation targets to avoid generating compensation targets that are not practically feasible. For example, if the calculation shows that the temperature needs to be reduced by 5°C in a certain stage, but according to the equipment performance, the maximum cooling capacity can only achieve a temperature reduction of 3°C per stage, then the compensation target needs to be adjusted to 3°C, and the remaining compensation amount needs to be planned in subsequent stages to ensure the feasibility of the compensation target.

[0043] Furthermore, the generation of the initial compensation target data must be traceable. The historical data source, fitting model parameters, and deduction logic for each stage's compensation target value must be clearly defined to facilitate verification and adjustment during subsequent comparison and correction processes. For example, if subsequent actual operating data deviates from the compensation target, the generation process of the compensation target can be traced back to analyze whether the issue lies in inaccurate historical data feature extraction, inappropriate fitting model selection, or incorrect prediction of environmental factors during the deduction stage, thereby enabling targeted optimization of the compensation strategy.

[0044] Through the above steps, from analyzing abnormal entity data to constructing compensation and control strategies, generating the first compensation curve, and deducing the compensation target, the first compensation target data is finally obtained. This process requires close integration with the actual operating characteristics of the digital twin, comprehensively utilizing data processing, model fitting, and logical deduction methods to ensure the rigor and scientific nature of each step, providing accurate basis for subsequent calculations of main compensator compensation parameters and allocation of sub-compensators compensation.

[0045] Example 3: When calculating the compensation parameters of the main compensator, the difference between the compensation targets of adjacent stages in the first compensation target data must first be calculated. The first compensation target data is the compensation target value of each stage derived from historical abnormal entity object data. For example, if the compensation target of a digital twin system in stage 1 is to reduce the temperature by 2℃ and the compensation target in stage 2 is to reduce the temperature by 5℃, then the difference between the compensation targets of the two adjacent stages is 3℃. When calculating the difference, the parameter type and unit corresponding to the compensation target of each stage must be clearly defined to ensure the consistency and accuracy of the difference calculation. The calculation process of the main compensator compensation parameters is only triggered when the difference reaches a preset threshold. The setting of the preset threshold needs to be combined with the operating accuracy requirements of the digital twin and the response characteristics of the compensation system. For example, for a medical equipment digital twin with high accuracy requirements, the preset threshold is set to 0.5℃, while for an industrial production line, the threshold can be appropriately relaxed to 2℃.

[0046] Once the difference reaches a preset threshold, it is necessary to obtain the characteristic composition information of individual abnormal entity data within the difference. This characteristic composition information covers the physical attributes, operating parameters, and functional characteristics of the abnormal entity data. Taking an industrial robotic arm digital twin as an example, if the target difference for compensation at a certain stage involves the positional offset of joint 1 of the robotic arm, then the individual abnormal entity data is joint 1, and its characteristic composition information includes the joint's maximum range of motion, motor torque parameters, position feedback accuracy, load capacity, etc. This information needs to be extracted from the digital twin's model database or collected through a real-time monitoring system to gather the current operating parameters of the abnormal entity data. For abnormal entity data in complex systems, the characteristic composition information may include multiple dimensions. For example, the characteristic composition information of abnormal transformer entity data in a power system includes rated voltage, rated capacity, short-circuit impedance, winding connection method, etc., and comprehensive collection is required to ensure the accuracy of subsequent calculations.

[0047] While acquiring information about the characteristics, it is necessary to count the total number of anomalous objects in the difference. The total number of anomalous objects refers to the number of abnormal entity objects with parameter offsets within the range covered by the current compensation target difference. For example, in a digital twin of an industrial workshop, if the difference between the compensation targets of two adjacent stages involves parameter adjustments for 3 heating furnaces and 2 fans, then the total number of anomalous objects is 5. When counting the total number of anomalous objects, the scope of anomalous objects must be clearly defined to avoid double counting or omissions. This can be achieved by establishing an anomalous object list and verifying each one to ensure the accuracy of the quantity statistics.

[0048] Based on the characteristic composition information and the total number of abnormal objects, the compensation parameter data of the main compensator is calculated. The calculation of the compensation parameter data must be based on the correlation analysis between the characteristic composition information and the difference. For example, for the position offset compensation of a robotic arm joint, given that the maximum range of motion of joint 1 is 180°, the current difference corresponds to a position offset of 5°, and the total number of abnormal objects is 1 (joint 1 only), then the required output compensation torque value and position adjustment amount of the main compensator need to be calculated based on the joint's torque parameters and position feedback accuracy. In the specific calculation process, the influence weight of the abnormal entity object data characteristics on the compensation needs to be considered. For example, in abnormal entity object data with multi-parameter coupling, some characteristics (such as motor torque) have a greater impact on the compensation effect and need to be assigned a higher weight coefficient, while other characteristics (such as appearance dimensions) have a smaller impact and their weight coefficients can be reduced accordingly.

[0049] In practical calculations, a layered approach may be necessary. First, for each individual anomalous entity data point, the mapping relationship between its characteristic parameters and the target compensation difference is analyzed. For example, in a temperature compensation scenario, if the anomalous entity data is a heating furnace, its characteristic information includes heating power coefficient, heat dissipation area, and thermal conductivity of insulation material. The mathematical relationship between these characteristic parameters and the temperature difference needs to be determined (e.g., for every 1kW increase in heating power, the temperature can rise by 0.8℃). Then, considering the total number of anomalous objects, the total compensation amount is allocated to each anomalous entity data point, and the corresponding compensation parameter data is calculated based on the characteristic parameters of each anomalous entity data point. For example, if the total compensation amount is 10℃, and there are two heating furnaces with different heating power coefficients, the compensation amount needs to be allocated according to their respective coefficient ratios, and then the main compensator compensation power for each heating furnace can be calculated.

[0050] When calculating the compensation parameters of the main compensator, the correlation between abnormal entity data must also be considered. In a digital twin system, anomalies in multiple abnormal entity data may affect each other. For example, in a pipeline system, an abnormal opening of valve A can cause pressure changes in pipeline B. In this case, when calculating the compensation parameters for valve A, the impact of pipeline B's pressure characteristics on the compensation effect must also be considered to avoid causing new anomalies in pipeline B after compensating valve A alone. Therefore, during the calculation process, an abnormal entity data correlation matrix needs to be established to analyze the interaction relationships between abnormal objects and to coordinately adjust the compensation parameter data. For example, when there are two abnormal objects (valve A and pipeline B) and they are strongly correlated, the compensation parameter data of the main compensator must simultaneously include the opening adjustment for valve A and the pressure setpoint for pipeline B to ensure the integrity and effectiveness of the compensation operation.

[0051] Furthermore, the performance constraints of the main compensator itself must be taken into account during the calculation process. As the core component performing the compensation operation, the main compensator has upper limits on parameters such as output capacity and response speed. For example, if the maximum compensation power of the main compensator is 100kW, and the calculated compensation power requirement is 120kW, then the compensation parameters need to be adjusted. This can be done through staged compensation or by combining compensation with sub-compensators to ensure that the compensation operation is performed within the capacity of the main compensator. Simultaneously, the response delay of the main compensator must be considered. For example, a motor-type main compensator may require 0.5 seconds from receiving the compensation command to completing the adjustment. This time factor needs to be reflected in the compensation parameters to maintain timing synchronization with the compensation operations of the sub-compensators.

[0052] The calculation of compensation parameter data also requires a dynamic adjustment mechanism. When the operating environment of the digital twin changes (such as an increase in external temperature or fluctuations in grid voltage), the characteristic composition information of the abnormal entity data may change accordingly, thus affecting the accuracy of the compensation parameter data. For example, when the external temperature of a heating furnace increases, its heat dissipation rate accelerates, and its heating power coefficient changes. In this case, it is necessary to reacquire the characteristic composition information and update the calculation model of the compensation parameter data. Therefore, during the calculation process, it is necessary to monitor the operating environment parameters of the abnormal entity data in real time. When the environmental change exceeds a certain threshold, the process of reacquiring the characteristic composition information and recalculating the compensation parameter data is triggered.

[0053] When calculating the compensation parameters of the main compensator, data recording and traceability are also crucial. The characteristic composition information, the total number of abnormal objects, and the results of correlation analysis for each calculation step must be recorded in detail to form a calculation log. For example, the torque parameter of robotic arm joint 1 is recorded as 50 N·m, derived from sensor data collected on June 28, 2025; the total number of abnormal objects is 5, specifically including heating furnace A, heating furnace B, and fan C. These records facilitate subsequent verification of the rationality of the compensation parameters. When the compensation effect does not meet expectations, the root cause of the problem can be found by tracing the calculation process, such as incorrect characteristic parameter values ​​or omissions in correlation analysis, and corrected promptly.

[0054] Through the above steps, from calculating the compensation target difference to obtaining the characteristic composition information and the total number of abnormal objects, and then to calculating the compensation parameter data based on the two, the characteristics of abnormal entity object data, system correlation and main compensator constraints are fully considered to ensure that the calculated compensation parameter data can accurately guide the compensation operation of the main compensator.

[0055] Example 4: When generating compensation allocation instructions based on remaining status data and transmitting them to multiple sub-compensators, the first step is to segment the preset coverage area, centered on the main compensator. Taking a digital twin system in a chemical industrial park as an example, the main compensator is located at the central control node of the park, and the preset coverage area is a 500-meter radius area centered on that node. During segmentation, this area is divided into three sub-ranges in ascending order of proximity: 0-100 meters is the first sub-range, 100-300 meters is the second sub-range, and 300-500 meters is the third sub-range, each forming an independent sub-region. This segmentation method needs to consider signal transmission delay and the response efficiency of the sub-compensators. Generally, the closer the sub-region is to the main compensator, the lower the signal transmission loss and the faster the compensation response speed. Therefore, the segment spacing can be adjusted according to the actual communication environment and equipment performance. For example, in areas with good communication quality, the segment spacing can be appropriately increased, and vice versa.

[0056] After completing the sub-region division, it is necessary to select the sub-compensators within each sub-region. Taking the first sub-region (0-100 meters) as an example, this region contains 10 temperature sub-compensators and 5 pressure sub-compensators, which are installed near equipment such as reactors and pipelines within the park. After selecting the sub-compensators, the remaining status data of each sub-compensator is collected. The remaining status data includes the current operating status, adjustable range, and executed compensation amount of the sub-compensator. For example, the remaining status data of temperature sub-compensator A shows that it is currently operating at half load, with an adjustable temperature range of -10℃ to +15℃ and an executed compensation amount of 3℃; the remaining status data of pressure sub-compensator B shows that it is operating at full load, with an adjustable pressure range of 0 to 0.5MPa and an executed compensation amount of 0.2MPa.

[0057] Based on the remaining status data of the sub-compensators within each sub-region, the basic adjustment amount for each sub-region where the offset value is within a set threshold is determined. The determination of the set threshold needs to consider the operational accuracy requirements of the digital twin system. For example, for temperature parameters, the set threshold is ±1.5℃, and for pressure parameters, it is ±0.1MPa. Taking the temperature sub-compensators in the first sub-region as an example, the current temperature offset of sub-compensator A is +1.2℃, within the set threshold of ±1.5℃, and its available basic adjustment amount is 5℃ (calculated based on its adjustable range and the amount of compensation already performed: 15℃ - 3℃ - 1.2℃ = 10.8℃, but considering the equipment safety margin, the actual basic adjustment amount is set to 5℃). The temperature offset of sub-compensator C is +2.0℃, exceeding the set threshold, and therefore is not included in the calculation of the basic adjustment amount. In this way, the remaining status data of each sub-compensator is analyzed one by one to determine whether it meets the condition that the offset value is within the set threshold, and the corresponding basic adjustment amount is calculated.

[0058] It is necessary to determine the sub-compensators within each sub-region that meet the basic adjustment amount, such that the sum of the basic adjustment amounts of these sub-compensators is greater than or equal to the basic compensation. Basic compensation refers to the minimum compensation threshold that needs to be collaboratively executed by the sub-compensators to offset the currently identified parameter deviations between the digital twin and the physical entity and maintain the basic operational accuracy of the system. Essentially, it is a "compensation benchmark value" determined based on historical anomaly data, the data characteristics of the anomaly entity, and the current operating state. It is the core quantitative indicator to ensure the initial matching between the digital twin model and the physical entity, and also the basis for subsequent compensation allocation and dynamic correction. Assuming the basic compensation is 12℃, the temperature sub-compensators A, B, and D in the first sub-region meet the condition, with basic adjustment amounts of 5℃, 4℃, and 3℃ respectively, summing to 12℃, which just meets the basic compensation requirement. In the second sub-region, the temperature sub-compensators E and F have basic adjustment amounts of 6℃ and 7℃ respectively, summing to 13℃, which also meets the basic compensation requirement. When determining which sub-compensators meet the conditions, certain priority rules need to be followed. For example, sub-compensators that are closer to the abnormal entity data should be selected first to reduce the transmission delay of the compensation signal; or sub-compensators with larger remaining capacity should be selected first to avoid overloading a single sub-compensator.

[0059] Based on the determined number of sub-compensators and the basic adjustment amount for each sub-compensator, compensation allocation instructions are generated and transmitted. The compensation allocation instructions need to clearly specify the identifier of each sub-compensator, the basic adjustment amount to be executed, and the adjustment time window. For example, a compensation allocation instruction might include the following: Sub-compensator A, adjust temperature -5℃, execute within 10 seconds of receiving the instruction; Sub-compensator B, adjust temperature -4℃, execute within 15 seconds of receiving the instruction; Sub-compensator D, adjust temperature -3℃, execute within 20 seconds of receiving the instruction. Instructions can be transmitted via wired or wireless communication, depending on the installation location of the sub-compensators and communication conditions. For sub-compensators close to the main compensator, a more real-time wired communication method can be used; for sub-compensators located in more remote locations, wireless communication can be used, but signal stability and transmission delay need to be considered, and relay nodes should be set up if necessary to ensure reliable instruction transmission.

[0060] In practice, situations may arise where the sum of the basic adjustments of the sub-compensators cannot meet the basic compensation. For example, if the sum of the basic adjustments of all sub-compensators meeting the offset condition within a certain sub-region is 8°C, while the basic compensation is 10°C, two measures are needed: First, expand the preset coverage area to include sub-regions at greater distances, increasing the number of selectable sub-compensators; second, reassess the remaining status data of the sub-compensators to check for underutilized adjustment capacity or whether the operating mode of the sub-compensators can be adjusted to release more adjustment capacity. For example, adjusting a sub-compensator in a half-load state to a full-load state can increase its basic adjustment.

[0061] Furthermore, the collaborative operation between sub-compensators needs to be considered. In complex digital twin systems, multiple sub-compensators may simultaneously compensate for the same parameter; for example, multiple temperature sub-compensators may simultaneously adjust the temperature of a certain area. In this case, it is necessary to ensure that the adjustment operations of each sub-compensator do not interfere with each other. Therefore, when generating compensation allocation instructions, a reasonable coordination strategy needs to be formulated based on the location, adjustment amount, and time window of the sub-compensators. For example, compensation operations can be executed sequentially according to their distance from the anomaly source, or multiple sub-compensators can perform adjustments within different time windows to avoid parameter overshoot caused by simultaneous adjustments.

[0062] After the compensation allocation command is transmitted, a feedback mechanism for command execution needs to be established. The main compensator needs to receive the command execution status of each sub-compensator in real time, including whether the command was successfully received, whether execution started on time, and whether any abnormalities occurred during execution. For example, after receiving the command, sub-compensator A will send an acknowledgment signal to the main compensator; during the adjustment operation, it will provide real-time feedback on the current adjustment progress; and after the adjustment is completed, it will send a completion signal. If the main compensator does not receive a feedback signal from a sub-compensator within the specified time, or receives an abnormal feedback signal, it needs to take corresponding fault-tolerant measures, such as resending the command or switching to a backup sub-compensator, to ensure the smooth execution of the compensation operation.

[0063] Throughout the implementation process, the accuracy and real-time nature of the data are crucial. The acquisition of residual status data requires high-precision sensors and a reliable data transmission network to ensure that the data reflects the current state of the sub-compensator in a timely and accurate manner. Simultaneously, a data verification mechanism needs to be established to validate the collected residual status data and eliminate erroneous data caused by sensor malfunctions or communication interference. For example, the reasonableness of the data can be determined by comparing data collected from different sensors on the same sub-compensator or by comparing it with historical data.

[0064] Taking another scenario as an example, in the digital twin system of a smart building, the main compensator is located in the building's central control room, with a preset coverage area of ​​the entire building's air conditioning system. The coverage area is divided into sub-regions by floor, with each floor being a sub-region. The air conditioning sub-compensators (such as fan coil units and air conditioning units) on each floor are selected, and their remaining status data is collected, including current airflow, supply air temperature, and equipment operating status. Based on this data, the basic adjustment amount for each sub-compensator is determined. For example, if the current supply air temperature deviation of a fan coil unit on a certain floor is within a set threshold, the basic adjustment amount provided is 2℃. Then, the sub-compensators in each sub-region that meet the basic adjustment amount requirements are selected, and their sum of basic adjustment amounts meets the basic temperature compensation requirements of that region. Compensation allocation instructions are generated and transmitted to the corresponding sub-compensators, such as instructing a fan coil unit to reduce the supply air temperature by 2℃, or instructing an air conditioning unit to increase the airflow by 10%, etc.

[0065] From the segmentation of the preset coverage area to the selection of sub-compensators and the collection of remaining status data, then to the determination of the basic adjustment amount, the screening of sub-compensators that meet the conditions, and the generation and transmission of compensation allocation instructions, a complete compensation allocation process is formed. This process fully considers the location distribution of sub-compensators, their remaining status capabilities, and the system's compensation needs, ensuring that compensation operations can be executed locally, improving the efficiency and accuracy of compensation. During implementation, it is necessary to flexibly adjust parameters such as the segmentation method, thresholds, and priority rules according to the specific characteristics of the digital twin system to adapt to different application scenarios and compensation needs. At the same time, it is important to focus on the system's scalability so that the compensation allocation process can be easily adjusted and optimized when adding sub-compensators or expanding the system scale.

[0066] Example 5: When supplementing the latest actual operating data based on the compensation and control strategy and making comparative corrections, the specific implementation process is carried out using a digital twin system of a power transformer as an example. Assume the latest stage is the 10th monitoring period of transformer operation (each period is 1 hour). During this period, sensors collect real-time actual operating data such as transformer oil temperature, winding temperature, load current, and oil level. For example, if the collected data for the 10th period is 65℃ oil temperature, 72℃ winding temperature, and 380A load current, these data need to be completely recorded and synchronized to the compensation and control strategy database to supplement the existing historical full-scale operating data system.

[0067] After supplementing with actual operating data, the compensation and control strategy needs to be updated. The original compensation and control strategy was built based on abnormal entity object data from the previous nine time periods. For example, in the previous nine time periods, when the oil temperature exceeded 60℃, the strategy was set to start the cooling fan for compensation. However, the actual operating data of the 10th time period showed that when the oil temperature was 65℃, the cooling rate after the cooling fan started decreased by 10% compared to the previous nine time periods. This may be due to dust accumulation on the cooling fan filter or an increase in ambient temperature. At this point, it is necessary to re-analyze the relationship between oil temperature and the cooling fan compensation effect based on the operating data of the 10th time period, and adjust the cooling fan start threshold and compensation duration in the compensation and control strategy. For example, the start threshold can be adjusted from 60℃ to 58℃, and the compensation duration can be extended from 30 minutes to 40 minutes to generate a second compensation curve that meets the second matching condition. Here, the second matching condition may place more emphasis on the weight of recent data, such as using an exponentially weighted moving average, so that the latest data has a greater impact on the curve, thereby making the second compensation curve more closely match the current operating state.

[0068] After generating the second compensation curve, it is necessary to identify the second compensation target data for the next stage (i.e., the 11th period). Assuming the second compensation curve predicts the oil temperature will rise to 70℃ in the 11th period, to keep the oil temperature below 65℃, the second compensation target data is set to reduce the oil temperature by 5℃ using a cooling fan. Meanwhile, the compensation target for the 11th period in the first compensation target data is extrapolated from the data of the previous nine periods, originally set to reduce the temperature by 3℃. At this point, it is necessary to compare the fluctuations of the first and second compensation target data for the next stage and analyze the deviation between them. In this example, the deviation is 5℃ - 3℃ = 2℃, and it is necessary to determine whether this deviation is within the preset deviation. The setting of the preset deviation must be combined with the transformer's operational safety requirements. For example, if the preset deviation is 1.5℃, then a deviation of 2℃ is greater than the preset deviation, and further determination is needed on how to correct the adjustment operation of the basic compensation.

[0069] When different deviations occur in the comparison results, different correction operations are performed. If the deviation between the second compensation target data and the next stage compensation target data in the first compensation target data is within a preset deviation (e.g., a deviation of 1℃ and a preset deviation of 1.5℃), a first instruction based on multiple sub-compensators is generated to maintain all adjustments to the basic compensation. For example, if the original basic compensation operation is to start one cooling fan and run it for 30 minutes, this operation remains unchanged. If the second compensation target data is less than the next stage compensation target data in the first compensation target data (e.g., the first target is a reduction of 5℃, the second target is a reduction of 3℃, the deviation is 2℃ and greater than the preset deviation), then an integration and maintenance instruction needs to be generated. For example, in a transformer cooling system, the sub-compensators are multiple cooling fans. The original basic compensation operation is to start two fans and run them for 40 minutes each. According to the integration and maintenance instruction, this might be adjusted to start one fan and run it for 60 minutes (partially integrating the basic compensation operation), while keeping the other fan's operating status unchanged. In this way, the equipment's operating efficiency is optimized while ensuring the compensation effect.

[0070] When the second compensation target data is greater than the compensation target data for the next stage in the first compensation target data and the deviation exceeds the preset deviation (e.g., the first target is a reduction of 3℃, the second target is a reduction of 5℃, and the deviation is 2℃, which is greater than the preset deviation of 1.5℃), a second instruction is generated to be transmitted to multiple sub-compensators, and all adjustments to the basic compensation are maintained. For example, the original basic compensation operation is to start one cooling fan and run it for 40 minutes. At this point, it is found that a larger compensation amount is needed, but due to the parameter limitations of the main compensator or the constraints of the remaining state data of the sub-compensators, the compensation amount cannot be increased immediately. Therefore, the original operation is maintained, and a further update to the compensation control strategy is triggered, such as increasing the number of cooling fans started or extending the running time.

[0071] In practical applications, the completeness and accuracy of data must be ensured when supplementing actual operating data. For example, in a wind turbine digital twin system, the latest operating data may include wind speed, generator speed, gearbox temperature, and generator output power. If a sensor malfunctions, resulting in data loss, it is necessary to supplement the data through data interpolation or replacement with redundant sensor data to avoid inaccurate updates to the compensation and control strategy due to data loss. Furthermore, when the operating conditions of the digital twin system undergo significant changes (such as a transformer switching from no-load to full-load), the latest operating data may differ significantly from historical data. In this case, the applicability of the compensation and control strategy needs to be reassessed, potentially requiring adjustments to the compensation model type, such as switching from a linear model to a nonlinear model, to ensure that the second compensation curve accurately reflects the compensation requirements under the current operating conditions.

[0072] During the correction process, the actual execution capacity of the sub-compensators also needs to be considered. For example, in a transformer cooling system, the number of sub-compensators (cooling fans) is limited, and the maximum operating time of each fan is limited by the equipment's lifespan. When the second compensation target data requires a larger compensation amount, if all sub-compensators are already operating at full load and cannot perform additional compensation operations, even if the deviation exceeds the preset deviation, it is necessary to maintain the adjustment operation of the basic compensation and meet the compensation requirements through other means (such as starting backup cooling equipment). Therefore, before generating the correction instruction, it is necessary to query the remaining status data of the sub-compensators in real time to ensure the executability of the instruction.

[0073] Furthermore, the analysis of the comparison results needs to be combined with the physical characteristics and operational logic of the digital twin. For example, in a digital twin of a chemical reactor, the first compensation target data predicts that the reaction temperature will rise by 4°C, and the second compensation target data predicts a rise of 6°C, with a deviation of 2°C. If the upper limit of the safe temperature of the reactor is 80°C, and the current actual temperature is 75°C, then the 2°C deviation may cause the temperature to approach the upper limit, requiring more aggressive compensation measures, such as increasing the coolant flow rate, rather than simply maintaining the basic compensation operation. Conversely, if the upper limit of the safe temperature is high, and the current temperature is far from the upper limit, the same deviation may not need to be corrected immediately; continuous monitoring is sufficient. Therefore, the setting of the preset deviation and the selection of the correction strategy need to be closely integrated with the safety boundaries and operational requirements of the digital twin.

[0074] Throughout the implementation process, data traceability and auditing mechanisms are essential. Every instance of supplementary actual operating data, updated compensation and control strategies, generated second compensation curves, and the decision-making process for comparison and correction must be meticulously recorded. For example, the oil temperature data for the 10th time period was recorded from sensor T1, collected at 10:00 AM on June 29, 2025. The compensation and control strategy was updated at 10:30 AM, including adjustments to the cooling fan start threshold. The second compensation curve was generated using an exponentially weighted moving average algorithm with a weighting coefficient of 0.7. The comparison results showed a deviation of 2°C, while the preset deviation was 1.5°C; therefore, the integration and coexistence instructions were executed. These records facilitate subsequent evaluation of the compensation effect. When compensation anomalies are detected, the specific steps can be traced back to analyze the causes of the problem, such as errors in actual operating data collection, improper compensation model parameter settings, or unreasonable preset deviation settings, allowing for targeted optimization.

[0075] Through the above steps, from supplementing actual operational data to updating compensation and control strategies and generating the second compensation curve, and then to analyzing comparison results and generating correction instructions, dynamic optimization of basic compensation adjustment operations is achieved. This process, combined with a specific digital twin example, fully considers factors such as data changes, model updates, equipment constraints, and security requirements, ensuring that compensation operations can be accurately adjusted according to the latest operational status, thereby improving the operational stability and reliability of the digital twin system. During implementation, it is necessary to flexibly adjust data processing methods, model parameters, and correction strategies according to different application scenarios and system characteristics to adapt to complex and ever-changing operating environments.

[0076] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0077] Although embodiments of this application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A dynamic compensation method for digital twin optimization, characterized in that, The method includes: Obtain the complete historical operation data of the digital twin, and filter out abnormal entity object data that evolves over time, wherein the abnormal entity object data includes parameter offset values; Based on the full historical operation data, a compensation and control strategy is constructed, and the compensation targets for multiple stages are extrapolated to obtain the first compensation target data. Calculate the difference between the compensation targets of two adjacent stages in the first compensation target data. When the difference reaches a preset threshold, calculate the compensation parameter data of the main compensator based on the correlation between the difference and the characteristics of the abnormal entity object data. Collect the remaining status data of multiple sub-compensators within the preset coverage area based on the compensation parameter data; A compensation allocation instruction is generated based on the remaining state data, and the compensation allocation instruction is transmitted to multiple sub-compensators. The compensation allocation instruction is used to instruct the adjustment operation that satisfies the basic compensation to be performed locally based on the remaining state data. Based on the compensation and control strategy, the latest stage of actual operation data is supplemented, and the second compensation target data for the next stage is obtained. The first compensation target data and the second compensation target data are compared with the data for the next stage to generate a comparison result. Based on the comparison result, it is determined whether the adjustment operation of the basic compensation should be corrected.

2. The dynamic compensation method for digital twin optimization according to claim 1, characterized in that, The process of acquiring the complete historical operational data of the digital twin and filtering out abnormal entity object data that evolves over time includes: Obtain the complete historical operation data of the digital twin within a specified historical period; Based on the set historical time period, select all time periods that are in the same stage as the current monitoring time period; Filter out abnormal entity object data for all time periods in the historical full data.

3. The dynamic compensation method for digital twin optimization according to claim 2, characterized in that, The compensation and control strategy is constructed based on historical full-scale operational data, and the compensation targets for multiple stages are extrapolated forward to obtain the first compensation target data, including: Based on the abnormal entity object data of all time periods, a compensation and control strategy is constructed to generate a first compensation curve that meets the first matching condition. The compensation targets for multiple stages are extrapolated from the first compensation curve to obtain the first compensation target data.

4. The dynamic compensation method for digital twin optimization according to claim 3, characterized in that, The step of calculating the compensation parameter data of the main compensator based on the correlation between the difference and the data characteristics of the abnormal entity object includes: Obtain the characteristic composition information of a single abnormal entity object data in the difference; Based on the characteristic composition information and the total number of abnormal entity object data in the difference, the compensation parameter data of the main compensator is calculated. The compensation parameter data includes the corresponding compensation quantity for the abnormal entity object data characteristics.

5. The dynamic compensation method for digital twin optimization according to claim 1, characterized in that, The step of generating compensation allocation instructions based on the remaining state data and transmitting the compensation allocation instructions to multiple sub-compensators includes: Centered on the main compensator, the preset coverage area is divided into segments in order from near to far, resulting in multiple sub-regions within the preset sub-range. Select the sub-compensator in each sub-region, and determine the basic adjustment amount for the parameter offset value in each sub-region to be within the set threshold based on the remaining status data of the sub-compensator in each sub-region; Determine the sub-compensator that meets the basic adjustment amount in each sub-region, resulting in multiple sub-compensators, where the sum of the basic adjustment amounts of the multiple sub-compensators is greater than or equal to the basic compensation; Based on multiple sub-compensators and the basic adjustment amount of each sub-compensator, a compensation allocation instruction is generated and transmitted.

6. The dynamic compensation method for digital twin optimization according to claim 3, characterized in that, The method involves supplementing the latest stage's actual operational data based on the compensation and control strategy, obtaining the second compensation target data for the next stage, and comparing the first and second compensation target data for the next stage to generate comparison results, including: Based on the latest operational data, the compensation and control strategy is updated to generate a second compensation curve that meets the second matching condition. Identify the second compensation target data for the next stage in the second compensation curve, and compare the fluctuations in the next stage based on the first and second compensation target data to generate comparison results.

7. The dynamic compensation method for digital twin optimization according to claim 1 or 6, characterized in that, The step of determining whether to correct the adjustment operation of the basic compensation based on the comparison results includes: When the deviation between the second compensation target data and the compensation target data for the next stage in the first compensation target data is within a preset deviation, a first instruction is generated based on multiple sub-compensators to fully maintain the adjustment operation of the basic compensation. When the second compensation target data is less than the compensation target data of the next stage in the first compensation target data, and the deviation between the two is greater than the preset deviation, an integration and maintenance coexistence instruction is generated according to the adjustment operation of the basic compensation in multiple sub-compensators, so that at least one sub-compensator in the multiple sub-compensators performs partial characteristic integration on other sub-compensators except itself, and maintains the adjustment operation of the basic compensation in at least one sub-compensator. When the second compensation target data is greater than the compensation target data for the next stage in the first compensation target data, and the deviation between the two is greater than the preset deviation, a second instruction is generated to be transmitted to multiple sub-compensators to fully maintain the adjustment operation of the basic compensation.

8. A dynamic compensation system for digital twin optimization, characterized in that, The system includes: The acquisition module is used to acquire the full historical operation data of the digital twin and filter the abnormal entity object data that evolves over time. The abnormal entity object data includes parameter offset values. The extrapolation module is used to construct compensation and control strategies based on historical full-volume operation data, and extrapolate the compensation targets for multiple stages to obtain the first compensation target data; The calculation module is used to calculate the difference between the compensation targets of two adjacent stages in the first compensation target data. When the difference reaches a preset threshold, the compensation parameter data of the main compensator is calculated based on the correlation between the difference and the characteristics of the abnormal entity object data. The acquisition unit is used to acquire the remaining status data of multiple sub-compensators within a preset coverage area based on the compensation parameter data; The allocation module is used to generate compensation allocation instructions based on the remaining state data and transmit the compensation allocation instructions to multiple sub-compensators. The compensation allocation instructions are used to instruct the adjustment operation that satisfies the basic compensation to be performed locally based on the remaining state data. The update and judgment module is used to supplement the latest stage of actual operation data based on the compensation and control strategy, obtain the second compensation target data for the next stage, compare the first compensation target data and the second compensation target data for the next stage, generate comparison results, and determine whether to correct the adjustment operation of the basic compensation based on the comparison results.

9. The dynamic compensation system for digital twin optimization according to claim 8, characterized in that, The allocation module includes: The segmentation unit is used to divide the preset coverage area into segments from near to far, centered on the main compensator, to obtain multiple sub-regions within the preset sub-range. The selected unit is used to select the sub-compensator in each sub-region. Based on the remaining status data of the sub-compensator in each sub-region, the basic adjustment amount within the set threshold of the offset value in each sub-region is determined. The sub-compensator in each sub-region that meets the basic adjustment amount is determined, resulting in multiple sub-compensators, wherein the sum of the basic adjustment amounts of the multiple sub-compensators is greater than or equal to the basic compensation. The generation unit is used to generate and transmit compensation allocation instructions based on multiple sub-compensators and the basic adjustment amount of each sub-compensator.

10. The dynamic compensation system for digital twin optimization according to claim 9, characterized in that, The generation unit is specifically used to: generate compensation allocation instructions based on multiple sub-compensators and the basic adjustment amount of each sub-compensator, and transmit them to the multiple sub-compensators.

Images