An intelligent simulation system for running state of key equipment in production line

Abstract

This invention belongs to the field of equipment simulation and discloses an intelligent simulation system for the operating status of key equipment in a production line. The system includes a preprocessing module, a slicing module, a calculation module, and a parameter update module. The preprocessing module preprocesses data obtained from sensors to obtain preprocessed data, and then fuses the preprocessed data with business context data to obtain a standard twin data package. The slicing module performs condition slicing on the standard twin data package obtained within a continuous time period to obtain a data judgment set divided according to the operating condition cycle. The calculation module obtains a parameter deviation indicator sequence based on the data judgment set. The parameter update module updates the simulation parameters according to the parameter deviation indicator sequence. The beneficial effects of this invention include: dynamic updating of simulation parameters, improving model simulation accuracy; realizing a knowledge loop from data to model; and supporting predictive maintenance and adaptive scheduling.

Description

Technical Field

[0001] This invention relates to the field of equipment simulation, and in particular to an intelligent simulation system for the operating status of key equipment in a production line. Background Technology

[0002] Digital twin technology, as a core means of realizing the integration of cyber-physical systems, has been widely applied to the planning, simulation, monitoring, and optimization of production lines. It achieves pre-production simulation, analysis, and decision support by constructing a dynamic model in virtual space that is highly faithfully mapped to the physical production line. Currently, typical implementation schemes in this field usually include: establishing a three-dimensional geometric model and layout of the production line; defining simulation logic and material flow based on product processes (e.g., by creating "process flow groups" with specific product and assembly attributes to distinguish different product types, as disclosed in CN118296836A); and integrating logistics equipment and control logic to form a workable virtual simulation model. Furthermore, to achieve monitoring functions, some advanced solutions collect real-time operating data of the physical production line through data interfaces and drive the virtual model to perform synchronous visualization operation (as described in CN112486131A), thereby achieving transparent perception of the production status.

[0003] However, when addressing the increasing demands for flexibility and high-mix production in modern aircraft maintenance parts production lines, existing digital twin technology has revealed significant limitations. The core issue lies in the lack of staticity and adaptability of the simulation model, specifically manifested in the following aspects: Simulation parameters are fixed and cannot reflect the true dynamic evolution of equipment performance: Existing simulation models rely heavily on preset fixed parameters or statistical values ​​based on historical data. For example, key parameters such as cycle time, mean time between failures (MTBF), and mean time to repair (MTTR) are set to constant values ​​or follow a fixed statistical distribution after model initialization. This prevents the model from sensing and responding to performance changes in physical equipment caused by wear, aging, sudden failures, or efficiency fluctuations during actual operation. The "equipment" in the model always operates in an ideal or initial state, and the deviation between the simulation results (such as capacity prediction and bottleneck analysis) and the actual performance of the physical production line gradually increases over time, ultimately leading to distorted decision-making.

[0004] The application of data is superficial, lacking a closed loop for autonomous transformation from data to model knowledge: Although existing technologies can collect real-time data to drive model operation or for visual monitoring, the data flow is usually unidirectional. The system only uses the data to "present" its state or makes simple comparisons with preset thresholds to trigger alarms, but fails to deeply mine the knowledge contained in the data, such as equipment performance degradation and process deviations, and fails to use this knowledge to autonomously and continuously correct and update the intrinsic parameters and logic of the simulation model. A "perception-analysis-evolution" intelligent closed loop has not been established between the model and the physical entity, and its fidelity maintenance relies entirely on manual intervention, resulting in a low level of intelligence.

[0005] In summary, existing digital twin technologies for production lines suffer from significant fidelity degradation due to the static nature of their models during long-term operation and in dynamic production environments. This severely limits the value of digital twin technology in advanced applications such as predictive maintenance, adaptive scheduling, and precise capacity assessment. Therefore, there is an urgent need for a technical solution that enables digital twin models to possess online self-learning and self-evolution capabilities, allowing them to dynamically adapt to changes in the physical system and maintain high fidelity at all times, thereby providing reliable and agile decision support for intelligent manufacturing. Summary of the Invention

[0006] The purpose of this invention is to disclose an intelligent simulation system for the operating status of key equipment in a production line, thereby solving the technical problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides an intelligent simulation system for the operating status of key equipment in a production line, including a preprocessing module, a slicing module, a calculation module, and a parameter update module; The preprocessing module is used to preprocess the data obtained by the sensor, obtain preprocessed data, and fuse the preprocessed data with the business context data to obtain a standard twin data package; The slicing module is used to perform operating condition slicing on standard twin data packets obtained in a continuous time period to obtain a data judgment set divided according to the operating condition cycle. The calculation module is used to obtain the parameter deviation indicator sequence based on the data judgment set; The parameter update module is used to update the simulation parameters according to the parameter deviation indication sequence to obtain the updated simulation parameters.

[0008] Preferably, it also includes a model update module; The model update module is used to update the digital twin based on the updated simulation parameters to obtain the digital twin with updated parameters.

[0009] Preferably, it also includes a simulation optimization module; The simulation optimization module is used to input the planned work order into the updated digital twin to obtain high-fidelity simulation results.

[0010] Preferably, the data obtained by the sensor includes vibration frequency, current, rotational speed, and voltage.

[0011] Preferably, the data acquired by the sensor is preprocessed to obtain preprocessed data, including: The data obtained from the sensor are sequentially processed by clock synchronization, outlier removal, and missing value imputation to obtain preprocessed data.

[0012] Preferably, the business context data includes work order information, process recipe, and status word.

[0013] Preferably, the standard twin data packets obtained within a continuous time period are subjected to operating condition slicing processing to obtain a data judgment set divided according to the operating condition cycle, including: Get the time tstarti when the status word changes from idle to running; Get the status word `tendi`, which indicates the time it takes for the status word to change from running to idle. Store the standard twin data packets obtained within the time interval tstarti to tendi into a temporary set; If the process recipes of the standard twin data packets in the temporary set have the same operation code and the status word is "running", then the temporary set will be used as the data judgment set.

[0014] Preferably, obtaining the parameter deviation indicator sequence based on the data judgment set includes: Obtain the preprocessed data from each standard twin data packet in the data judgment set; Extract the first feature value from the preprocessed data; Using a digital twin with unupdated parameters, simulate and run a process with the same process code as the process recipe in the data judgment set, and obtain the second feature value after the simulation is completed; Calculate the relative deviation between the first eigenvalue and the second eigenvalue to obtain the parameter deviation indication; The parameter deviation indication values ​​for the same operating cycle are stored in the parameter deviation indication value sequence.

[0015] Preferably, the simulation parameters are updated according to the parameter deviation indicator sequence to obtain the updated simulation parameters, including: Obtain the simulation parameters of the digital twin whose parameters have not been updated; If a trend drift is detected based on the parameter deviation indicator sequence, the simulation parameters are updated based on at least two parameter update strategies to obtain the updated simulation parameters.

[0016] Preferably, the simulation parameters are updated based on at least two parameter update strategies to obtain the updated simulation parameters, including: Calculate the adaptive weights of the calculation results for each parameter update strategy; The calculation results obtained by various parameter update strategies based on adaptive weights are fused to obtain the updated simulation parameters.

[0017] Beneficial effects: This invention addresses the highly mixed and flexible production characteristics of aircraft component production lines by proposing a digital twin simulation system with online self-learning and adaptive capabilities. Its core beneficial effects include: Dynamically updated simulation parameters improve the accuracy of model simulation. By collecting real-time equipment operating data (such as vibration, current, and rotational speed), the system can automatically detect changes in the physical equipment's performance (such as wear, aging, and efficiency fluctuations) and dynamically update simulation parameters (such as cycle time and failure rate). This allows the digital twin model to continuously closely approximate the actual equipment state, avoiding the accumulation of simulation deviations caused by the fixed parameters in traditional static models.

[0018] Achieving a closed knowledge loop from data to models: The system not only monitors data but also transforms real-time data into a basis for adjusting simulation parameters through a process of "condition slicing → deviation calculation → parameter update → model reconstruction," forming an intelligent closed loop of "perception-analysis-evolution." In the multi-variety, small-batch production of aircraft parts, the system can quickly adapt to process changes, improving the reliability of simulation results and their decision support value.

[0019] Supports predictive maintenance and adaptive scheduling: Because the model evolves in sync with the physical system, it can more accurately predict production capacity, analyze bottlenecks, and test scheduling schemes. For example, in the production of aircraft maintenance parts, it can provide early warnings of equipment performance degradation, optimize production scheduling, reduce downtime, and improve the overall production line response speed and reliability. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of an intelligent simulation system for the operating status of key equipment in a production line according to the present invention.

[0022] Figure 2This is a schematic diagram of the method for updating simulation parameters according to the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0024] like Figure 1 As shown in one embodiment, the present invention provides an intelligent simulation system for the operating status of key equipment in a production line, including a preprocessing module, a slicing module, a calculation module, and a parameter update module.

[0025] The simulation system of this invention first models the physical entities in the aircraft parts production line before simulation, generating a digital twin. The physical entities include the assembly line, equipment in each production stage, etc.

[0026] The preprocessing module is used to preprocess the data obtained by the sensor, obtain preprocessed data, and fuse the preprocessed data with the business context data to obtain a standard twin data package.

[0027] Preferably, the data obtained by the sensor includes vibration frequency, current, rotational speed, and voltage.

[0028] The data acquired by the sensor includes raw time-series signals directly collected from various sensors of the physical entity, denoted as Sraw(tp), where tp is the physical world time.

[0029] Preferably, the data acquired by the sensor is preprocessed to obtain preprocessed data, including: The data obtained from the sensor are sequentially processed by clock synchronization, outlier removal, and missing value imputation to obtain preprocessed data.

[0030] Further clock synchronization and processing includes: At the data acquisition gateway, a high-precision physical timestamp tp is added to each Sraw(tp) data point. A mapping relationship ts=f(tp) between tp and the digital twin simulation time ts is maintained (usually a linear relationship). This process unifies the asynchronously acquired physical data onto the simulation time axis ts of the digital twin.

[0031] This invention maps asynchronous data from different sensors onto the simulation timeline of a digital twin, ensuring data temporal consistency. In aircraft component production lines, with numerous devices and heterogeneous sensors, temporal alignment is fundamental for subsequent condition slicing and feature extraction, preventing analysis errors caused by time deviations.

[0032] Furthermore, outlier removal includes: For the sensor data stream aligned to ts, online filtering is performed using the "3σ principle based on dynamic threshold". For data within a short-term data window from a specific sensor, its mean m and standard deviation std are calculated. If the data value does not belong to [m-3std, m+3std], the data is considered an outlier and is removed.

[0033] This invention employs a dynamic threshold 3σ principle to filter abnormal data caused by sensor noise, transient interference, etc. For example, when collecting vibration frequencies, sudden electromagnetic interference may cause data spikes. Removing these spikes can prevent them from misleading the calculation of characteristic values ​​(such as the effective value of vibration) and improve data quality.

[0034] Further, missing value imputation processing includes: For data loss caused by outlier removal or communication interruption, a "forward valid value filling" strategy is adopted, which fills the data with the previous valid value that was not removed in the data stream to ensure data continuity.

[0035] For data loss caused by communication interruptions or abnormal data removal, this invention employs a forward valid value imputation strategy to ensure data continuity. During continuous production, data interruptions may lead to incomplete condition slices; imputation maintains the integrity of the data stream and supports subsequent continuous analysis.

[0036] Preferably, the business context data includes work order information, process recipe, and status word.

[0037] Work order information: Cworkorder, including product model PN, planned quantity, etc.

[0038] Process recipe: Crecipe, including the operation code (OPCode) currently being performed by the equipment.

[0039] Status word: Cstateplc, obtained from the device PLC (e.g., running 1, idle 0, fault 2).

[0040] Furthermore, the preprocessed data and business context data are fused to obtain a standard twin data package, including: Let D(ts) represent the original data packet: D(ts)={ts,DeviceID,DataType,Value,Csnapshot} Csnapshot={PN,OPCode,Cstateplc}.

[0041] DeviceID is the device ID, DataType is the data type, and Value is the data value.

[0042] By fusing sensor preprocessing data with business context (work orders, process recipes, equipment status words) into structured, semantically enhanced data packets, each data entry not only contains physical quantity values ​​but also is associated with its production background (such as the product model being processed, process code, and equipment status). This allows for: It supports refined data analysis by process, by equipment, and by product; Provide clear state transition basis for subsequent operating condition slices (such as "idle → running"); Enhance the interpretability of simulation results and facilitate the tracing of data sources and production context.

[0043] The slicing module is used to perform operating condition slicing on standard twin data packets obtained in a continuous time period to obtain a data judgment set divided according to the operating condition cycle.

[0044] Preferably, the standard twin data packets obtained within a continuous time period are subjected to operating condition slicing processing to obtain a data judgment set divided according to the operating condition cycle, including: Get the time tstarti when the status word changes from idle to running; Get the status word `tendi`, which indicates the time it takes for the status word to change from running to idle. Store the standard twin data packets obtained within the time interval tstarti to tendi into a temporary set; If the process recipes of the standard twin data packets in the temporary set have the same operation code and the status word is "running", then the temporary set will be used as the data judgment set.

[0045] Since standard twin data packets are continuously generated, the aforementioned data judgment set is also continuously generated. The data is first sliced ​​based on the status word, and then the temporary set is used as the data judgment set based on whether the process codes are identical.

[0046] A working cycle is defined as the complete time interval from the start of a specific OPCode execution by a device (Cstateplc changes from "idle" to "running") to the completion of that operation (Cstateplc changes from "running" to "idle" or "waiting"). Based on the state transitions in D(ts), continuous working cycle data segments can be divided, and the data judgment set is represented by Segi: Segi = {D(ts) | ts ∈ [tstarti, tendi]} and OPCode and DeviceID are constant within the interval [tstarti, tendi].

[0047] The above process divides the continuous data stream into independent data segments according to the complete process execution cycle. Each segment corresponds to a specific production activity (such as processing a part), achieving the following effect: Focus on effective production phases: Eliminate non-production periods such as equipment idleness and debugging to ensure that the analyzed data reflects the true processing status; Supports process-level parameter comparison: facilitates comparison of actual characteristic values ​​with simulation values ​​for the same process, and calculation of parameter deviations; Adaptable to mixed production lines with multiple product types: In aircraft parts production, the same equipment may perform different processes. After slicing, it can be analyzed separately according to process codes, improving the targeting of parameter updates.

[0048] The calculation module is used to obtain the parameter deviation indicator sequence based on the data judgment set.

[0049] The calculation module mainly judges the degree of parameter deviation based on preprocessed data.

[0050] Preferably, obtaining the parameter deviation indicator sequence based on the data judgment set includes: The first step is to obtain the preprocessed data from each standard twin data packet in the data judgment set.

[0051] Extracting the first feature value Fi from the preprocessed data includes: For example: Beat characteristics: Ficycle = tendi - tstarti (actual time elapsed for this segment).

[0052] Vibration characteristics: Fivibration = RMS(Svibration(ts)), where ts is in the interval [tstarti, tendi].

[0053] The second step involves using the digital twin with unupdated parameters (i.e., the digital twin after the previous parameter update) to simulate a process with the same operation code as the process recipe in the data judgment set, and obtaining the second feature value Fisim after the simulation is completed, including: The digital twin simulates a virtual operating cycle with the same OPCode and context as Segi based on the current simulation parameters θcurr, and outputs the simulation characteristics Fisim for this virtual cycle. For example, the simulation cycle time is taken from the distribution N(μjcurr, σjcurr). 2 (a single sampling)

[0054] μjcurr represents the average cycle time of device j under the current simulation parameters, i.e., the average time to process one part, σjcurr 2 This represents the variance of the j-cycle of the equipment, reflecting the degree of fluctuation in processing time.

[0055] In the simulation, each time a virtual process is run, the system randomly selects a cycle value from the normal distribution to simulate the natural fluctuations that exist in actual production (such as material differences, operation fine-tuning, etc.).

[0056] The first and second eigenvalues ​​have the same data type.

[0057] Calculate the relative deviation between the first eigenvalue and the second eigenvalue to obtain the parameter deviation indication, including: δi=(Fi-Fisim) / Fisim For different characteristics (such as beat, vibration), the corresponding δicycle and δivibration can be calculated.

[0058] The parameter deviation indication values ​​for the same operating cycle are stored in the parameter deviation indication value sequence.

[0059] The parameter deviation indicator sequence is represented as {δi}, where each δi is associated with a specific device j, parameter type (such as θcyclej), and operating condition i.

[0060] This invention achieves the following effects by continuously collecting parameter deviation indicators δi from multiple operating cycles to form a time series {δi}: While a single bias may be influenced by random factors, a sequence can reveal a systematic shift; it provides a data foundation for subsequent "trend drift" detection; and it allows for parameter adjustment based on the sequence, avoiding over-updating due to a single anomaly.

[0061] Furthermore, the current simulation parameters include at least one of the following parameters: θcyclej: The probability distribution parameter of the time (cycle time) required for equipment / station j to process a single workpiece (e.g., following a normal distribution N(μj,σj)). 2 The mean μj and standard deviation σj at that time.

[0062] θfailurej: Fault model parameters of equipment / workstation j (such as the failure rate λj when the fault interval time follows an exponential distribution Exp(λj).

[0063] θtransportk: Transmission time parameter of transmission device k.

[0064] θqualitym: The threshold parameter for detecting anomalies at quality inspection point m.

[0065] The parameter update module is used to update the simulation parameters according to the parameter deviation indication sequence to obtain the updated simulation parameters.

[0066] Preferably, the simulation parameters are updated according to the parameter deviation indicator sequence to obtain the updated simulation parameters, including: Obtain the simulation parameters of the digital twin whose parameters have not been updated; If a trend drift is detected based on the parameter deviation indicator sequence, the simulation parameters are updated based on at least two parameter update strategies to obtain the updated simulation parameters.

[0067] Furthermore, determining whether a trend drift occurs based on the parameter deviation indicator sequence includes: Calculate the ratio between the total number of parameter deviation indicators in the parameter deviation indicator sequence that are greater than the set threshold and the total number of parameter deviation indicators in the parameter deviation indicator sequence; Determine if the ratio is greater than the set ratio threshold. If so, it indicates that a trend of drift has occurred.

[0068] Trend drift refers to a continuous change in parameter deviation in a certain direction (such as a gradual increase in processing time), which may indicate equipment performance degradation. Identifying trend drift can distinguish between random fluctuations and system evolution, avoiding unnecessary updates for temporary deviations; it can provide timely warnings of changes in equipment status, such as early detection of problems like tool wear and bearing aging in aircraft component production; and it can trigger parameter update mechanisms, initiating parameter updates only when a trend change is detected, thus improving system stability and update efficiency.

[0069] Furthermore, the ratio threshold can be a value between 0.6 and 0.8, preferably 0.7. If the threshold is set to 0.7, it means that when the deviation exceeds 70% and remains consistently high or low, the system determines it as a trend drift.

[0070] For example, the first parameter update strategy includes: Calculate the mean δmean and standard deviation δstd of the data in the parametric deviation indicator series; Calculate the update amount using the following formula: Δθ1=α δmean θcurr α is a configurable learning rate (0 < α ≤ 1) used to control the magnitude of a single update and avoid oscillations. δmean is the arithmetic mean of the most recent N consecutive δi that triggered the evolution.

[0071] Calculate the updated parameters using the following formula: θnew1=θcurr+Δθ1 θnew1 is the result of the first parameter update strategy. θcurr is the current parameter.

[0072] For example, the second parameter update strategy includes: Updates are performed using a distributed update strategy based on Bayesian inference.

[0073] The following example uses the updated failure rate λj: The first step is to treat the failure interval as following an exponential distribution Exp(λj). We treat λj itself as a random variable and assume that its prior distribution is a conjugate gamma distribution Gamma(αprior, βprior).

[0074] αprior is the prior value of the shape parameter of the gamma distribution, reflecting the initial cognitive strength of the failure rate λj.

[0075] βprior is the prior value of the rate parameter of the gamma distribution, which is related to the expected time scale of the fault occurrence.

[0076] In Bayesian updates, these two parameters together constitute the prior distribution Gamma(αprior,βprior) of λj, which can be set based on historical data or expert experience.

[0077] The second step is to update the posterior distribution of λj according to Bayes' theorem after observing a series of fault interval time data {Tbf1,...,TbfK}: The posterior distribution remains a gamma distribution, and the parameters are updated as follows: αposterior = αprior + K βposterior=βprior+Σ(Tbfk) αposterior is the updated gamma distribution shape parameter. K is the number of observed fault interval time data points. βposterior is the updated gamma distribution rate parameter.

[0078] Tbfk: Time between failures (k-th failure).

[0079] The third step is point estimation: take the mean of the posterior distribution as the new parameter estimate. λjnew=αposterior / βposterior.

[0080] Preferably, such as Figure 2 The simulation parameters are updated based on at least two parameter update strategies to obtain the updated simulation parameters, including: The first step is to calculate the adaptive weights of the calculation results for each parameter update strategy: In the previous update, the model parameters in the digital twin were updated using the model parameters calculated by each parameter update strategy to obtain multiple digital twins; Input the planned work orders within the time interval from the previous update to the current update into each digital twin, run the simulation, and obtain multiple predicted outputs. ; After the actual production line operation plan work order is completed, the actual output is calculated. ; Calculate the error ratio between predicted output and actual output. :

[0081] Calculate the adaptive weights using the following formula:

[0082] is the adaptive weight of the calculation result of the z-th parameter update strategy; np is the total number of parameter update strategies. The maximum value of the error proportion for updating the policy with np parameters. The error ratio corresponding to the z-th parameter update strategy.

[0083] The second step involves fusing the calculation results obtained from various parameter update strategies based on adaptive weights to obtain the updated simulation parameters, including:

[0084] For the updated simulation parameters, The calculation result for the z-th parameter update strategy.

[0085] The update strategy of the present invention has the following advantages: It effectively improves the robustness of updates. Different strategies have their own focuses (such as mean adjustment vs. Bayesian distribution update), and the fusion can take into account both the central tendency and distribution characteristics of the data.

[0086] It can perform adaptive weight adjustment, dynamically adjusting the weights based on the prediction accuracy of each strategy in historical periods, so that the system can automatically select a more reliable update method.

[0087] It is more suitable for highly mixed production environments. In aircraft parts production lines, the equipment status is affected by multiple factors. A single strategy may not be suitable for all scenarios. The integration of multiple strategies enhances the system's adaptability.

[0088] Preferably, it also includes a model update module; The model update module is used to update the digital twin based on the updated simulation parameters to obtain the digital twin with updated parameters.

[0089] Specifically, during updates, the updated simulation parameters can be used to replace the corresponding parameters of the running digital twin through the API provided by the digital twin platform.

[0090] Preferably, it also includes a simulation optimization module; The simulation optimization module is used to input the planned work order into the updated digital twin to obtain high-fidelity simulation results.

[0091] Specifically, simulations can include: Capacity forecasting: Input the planned work order {Cworkorder} for a future period of time, run the simulation, and obtain the predicted output Qforecast and the completion time Tforecast.

[0092] Bottleneck analysis: During simulation, the utilization rate Uj and average queue length Lqj of each device j are statistically analyzed. Bottleneck devices are defined as those where Uj > Uthreshold and Lqj continuously increases. This analysis is more accurate due to parameter evolution.

[0093] Scheduling scheme testing: Rapidly simulate different scheduling rules (such as SPT and EDD) in a digital twin, evaluate their impact on Qforecast and Tforecast, and select the best scheme.

[0094] Uthreshold represents the threshold for equipment utilization, used for bottleneck equipment identification. When equipment utilization consistently exceeds this threshold and the queue grows, it is identified as a bottleneck. This value can be set according to the production line design goals, and is generally between 0.85 and 0.95, indicating that equipment utilization exceeding 85% to 95% may constitute a bottleneck.

[0095] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An intelligent simulation system for the operating status of key equipment in a production line, characterized in that, It includes a preprocessing module, a slicing module, a calculation module, and a parameter update module; The preprocessing module is used to preprocess the data obtained by the sensor, obtain preprocessed data, and fuse the preprocessed data with the business context data to obtain a standard twin data package; The slicing module is used to perform operating condition slicing on standard twin data packets obtained in a continuous time period to obtain a data judgment set divided according to the operating condition cycle. The calculation module is used to obtain the parameter deviation indicator sequence based on the data judgment set; The parameter update module is used to update the simulation parameters according to the parameter deviation indication sequence to obtain the updated simulation parameters.

2. The intelligent simulation system for the operating status of key equipment in a production line according to claim 1, characterized in that, It also includes a model update module; The model update module is used to update the digital twin based on the updated simulation parameters to obtain the digital twin with updated parameters.

3. The intelligent simulation system for the operating status of key equipment in a production line according to claim 2, characterized in that, It also includes a simulation optimization module; The simulation optimization module is used to input the planned work order into the updated digital twin to obtain high-fidelity simulation results.

4. The intelligent simulation system for the operating status of key equipment in a production line according to claim 1, characterized in that, The data obtained by the sensor includes vibration frequency, current, rotational speed, and voltage.

5. The intelligent simulation system for the operating status of key equipment in a production line according to claim 4, characterized in that, The data acquired by the sensor is preprocessed to obtain preprocessed data, including: The data obtained from the sensor are sequentially processed by clock synchronization, outlier removal, and missing value imputation to obtain preprocessed data.

6. The intelligent simulation system for the operating status of key equipment in a production line according to claim 1, characterized in that, Business context data includes work order information, process recipes, and status words.

7. The intelligent simulation system for the operating status of key equipment in a production line according to claim 6, characterized in that, The standard twin data packets obtained within a continuous time period are processed by operating condition slicing to obtain a data judgment set divided according to the operating condition cycle, including: Get the time tstarti when the status word changes from idle to running; Get the status word `tendi`, which indicates the time it takes for the status word to change from running to idle. Store the standard twin data packets obtained within the time interval tstarti to tendi into a temporary set; If the process recipes of the standard twin data packets in the temporary set have the same operation code and the status word is "running", then the temporary set will be used as the data judgment set.

8. The intelligent simulation system for the operating status of key equipment in a production line according to claim 1, characterized in that, The parameter deviation indicator sequence is obtained based on the data set, including: Obtain the preprocessed data from each standard twin data packet in the data judgment set; Extract the first feature value from the preprocessed data; Using a digital twin with unupdated parameters, simulate and run a process with the same process code as the process recipe in the data judgment set, and obtain the second feature value after the simulation is completed; Calculate the relative deviation between the first eigenvalue and the second eigenvalue to obtain the parameter deviation indication; The parameter deviation indication values ​​for the same operating cycle are stored in the parameter deviation indication value sequence.

9. The intelligent simulation system for the operating status of key equipment in a production line according to claim 1, characterized in that, The simulation parameters are updated based on the parameter deviation indicator sequence to obtain the updated simulation parameters, including: Obtain the simulation parameters of the digital twin whose parameters have not been updated; If a trend drift is detected based on the parameter deviation indicator sequence, the simulation parameters are updated based on at least two parameter update strategies to obtain the updated simulation parameters.

10. The intelligent simulation system for the operating status of key equipment in a production line according to claim 1, characterized in that, The simulation parameters are updated based on at least two parameter update strategies to obtain the updated simulation parameters, including: Calculate the adaptive weights of the calculation results for each parameter update strategy; The calculation results obtained by various parameter update strategies based on adaptive weights are fused to obtain the updated simulation parameters.

Images