Dynamic correction method for digital twin model parameters of line equipment of MPC

Abstract

The invention discloses a dynamic correction method for digital twinborn model parameters of line equipment of MPC, which belongs to the technical field of power system automation, and comprises the following steps: acquiring original working condition data through distributed edge nodes, extracting load fluctuation, environmental disturbance and mechanical spectrum key features, forming chaotic feature vectors, and carrying out dynamic correction on the chaotic feature vectors; according to the method, the complex interaction is quantified through the multi-physics field coupling entropy source formula, the equipment state evolution process can be described more accurately, the sensitivity of the system to the initial disturbance is further evaluated through introduction of the Lyapunov index, the potential instability risk is warned in advance, and in the power transformer overheating fault prediction process, the method has the advantages of being high in reliability and high in reliability. According to the method, thermal-electric coupling abnormity can be detected several hours ahead of time, sudden shutdown is avoided, the safety and reliability of equipment operation are remarkably improved, real-time self-adaptive correction of parameters is achieved through double-channel parallel calculation, and the adaptability to complex working conditions is remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of power system automation technology, specifically referring to a dynamic correction method for parameters of a digital twin model of line equipment in MPC. Background Technology

[0002] With the development of smart grids, the operating environment of power line equipment is becoming increasingly complex, facing challenges such as multi-source disturbances, nonlinear coupling, and time-varying parameters. Although traditional model predictive control has the ability to feedforward control and constraint processing.

[0003] However, existing methods for dynamic parameter correction of digital twin models of line equipment still have certain shortcomings. In the MPC framework, the model is usually assumed to be a single physical field, which leads to an incomplete description of the system state evolution. Unmodeled dynamic behaviors and parameter drift in actual equipment can cause the model to mismatch with the state of the real system. Using single-channel linear filtering or fixed threshold strategies to handle operating condition disturbances lacks the ability to identify high-frequency chaotic disturbances in real time. Relying on low-pass filters to extract trend signals cannot capture instantaneous impacts or high-frequency resonance modulation signals, resulting in lag in adjustment of correction parameters and inability to dynamically adapt to complex operating conditions. In parameter correction, empirical rules or static thresholds are often relied upon, lacking a quantitative mechanism for dynamic matching of historical parameters with current operating conditions. Therefore, a dynamic parameter correction method for digital twin models of line equipment using MPC is proposed. Summary of the Invention

[0004] The purpose of this invention is to provide a method for dynamic correction of parameters of a digital twin model of MPC line equipment, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for dynamic correction of parameters of a digital twin model of MPC line equipment, comprising the following steps:

[0006] S1. Collect raw operating condition data through distributed edge nodes, extract key features of load fluctuation, environmental disturbance and mechanical spectrum, and form a chaotic feature vector;

[0007] S2. Based on the output chaotic feature vector, the entropy rate is calculated in real time through multi-physics partial differential coupling and Lyapunov exponent evaluation.

[0008] S3. Based on the generated entropy change rate, the chaotic quantization factor and pre-correction parameters are generated through dual-channel parallel computation.

[0009] S4. Integrate the pre-corrected parameters and historical parameters, evaluate the fitness using the Cauchy-Schwarz inequality, and output the optimal parameters;

[0010] S5. Map the optimized parameters to nanosecond-level pulse sequences, and use quantum pulse sequences to perform dynamic compensation and correction of FPGA device parameters;

[0011] S6. Monitor the mutual information of the twin entities during execution. When the credibility is lower than the threshold, trigger the entropy model reconstruction to form a closed-loop optimization.

[0012] Preferably, in step S1, multi-node edge sensors are deployed around the key equipment to collect raw operating condition data such as vibration, temperature, current, and rotation speed in real time. The multi-node data is aligned by timestamps, noise and invalid signals are removed, and indicators such as peak value, RMS, and kurtosis of the signal are calculated.

[0013] This can be understood as quantifying the instantaneous impact and long-term trend of load fluctuations, combining external parameters such as temperature and humidity to extract environmental interference patterns related to equipment operating status, extracting spectral peaks and harmonic components through Fourier transform, and capturing high-frequency resonant modulation signals using envelope spectrum analysis to identify local mechanical damage.

[0014] Based on phase space reconstruction, chaotic feature quantities such as Lyapunov exponent and fractal dimension are calculated, and time-domain and frequency-domain features are concatenated with chaotic features to form a high-dimensional feature vector, thus forming a chaotic feature vector containing linear and nonlinear information.

[0015] Preferably, step S2, the multi-physics partial differential coupling step, includes: obtaining chaotic feature vectors; constructing a physical field partial differential coupling equation model based on the actual physical characteristics of the equipment; where a single physical field includes a mechanical field and an electrical field; and the mechanical parameters are defined as follows: Its electrical parameters The gradient is This indicates the sensitivity of the mechanical state to electrical changes;

[0016] Electrical parameters are Its effect on thermodynamic parameters The gradient is , representing the sensitivity of thermodynamic changes to electrical state, the realization formula of the multiphysics partial differential coupling entropy source is:

[0017] ,

[0018] In the formula, This represents a multiphysics partial differential coupling entropy source. This represents the coupling weight coefficient of the i-th physical field. Indicates electrical parameters, Represents thermodynamic parameters, This represents the gradient of mechanical parameters with respect to electrical parameters. This represents the gradient of electrical parameters with respect to thermodynamic parameters. Represents the magnitude of the vector. This represents a weighted summation of the coupling terms over all n physical fields. This represents the amplification factor of chaotic features on coupling entropy. This represents the chaotic eigenvector.

[0019] Preferably, step S2, the Lyapunov exponent step, includes: perturbing the initial state of the multiphysics partial differential coupled entropy source to generate two adjacent orbits, a main orbit and a perturbed orbit. Specific steps include:

[0020] The current stable operating state of the multiphysics partial differential coupled entropy source is obtained as the initial state of the main orbit. The stable operating state contains the parameter values ​​of all physical fields. A very small perturbation value (perturbation amount) is added to each parameter dimension of the initial state to ensure that the system after perturbation is still in the linear response region and will not cause a fundamental change in the system behavior.

[0021] Adding perturbations to each parameter of the initial state creates a new initial state point that is very close to, but not exactly the same as, the original initial state point, thus forming the starting point of the perturbation trajectory.

[0022] Record the state differences between the two tracks at the initial moment, and set the time step and number of iterations.

[0023] The evolution of a multiphysics partial differential coupled entropy source is simulated by numerical integration. The state changes of the main orbit and the perturbation orbit are tracked in real time, the distance changes of the two orbits within the time step are calculated, and the divergence velocity is evaluated.

[0024] The Lyapunov exponent is calculated based on the logarithmic average growth rate of orbital divergence, using the following formula:

[0025] ,

[0026] In the formula, The Lyapunov index represents the Lyapunov index. Indicates the time step. Let represent the magnitude of the system state perturbation vector at time t. This represents the magnitude of the perturbation vector at the initial moment. Represents the natural logarithm. This represents the correction coefficient of the multiphysics partial differential coupling entropy source to the Lyapunov exponent, reflecting the enhancing effect of the multiphysics partial differential coupling entropy source on chaotic behavior.

[0027] Preferably, step S2, the real-time entropy rate change step, includes: calculating the entropy rate change in real time based on the multiphysics partial differential coupled entropy source and the Lyapunov exponent, with the formula as follows:

[0028] ,

[0029] In the formula, Represents the rate of entropy change. Represents the base coefficient for entropy rate of change. This represents an exponential function, used to amplify the effect of the Lyapunov exponent on the rate of entropy change. It represents the amplification factor of the entropy rate of the multiphysics partial differential coupling entropy source, reflecting the nonlinear enhancement effect of the entropy rate on the entropy change by the multiphysics partial differential coupling entropy source.

[0030] Preferably, step S3, the chaotic quantization factor generation step, includes: obtaining the entropy change rate; splitting the entropy change rate into channel A and channel B according to the time series; calculating the chaotic quantization factor for channel A and calculating the pre-correction parameter for channel B; extracting high-frequency components from the entropy change rate in channel A through nonlinear filtering; generating a chaotic intensity index through dynamic threshold comparison; and mapping the chaotic intensity index to the chaotic quantization factor. The formula is as follows:

[0031] ,

[0032] In the formula, This indicates the rate of entropy change within a time window. The moving average within, This represents the nonlinear amplification factor, controlling the sensitivity of the entropy rate of change to the CQF. This represents the dynamic threshold offset, which adjusts the baseline for chaos determination.

[0033] Preferably, step S3, the pre-calibration parameter generation step in channel B, includes: generating pre-calibration parameters based on the chaotic quantization factor in channel A. The formula is as follows:

[0034] ,

[0035] In the formula, Indicates the reference correction factor. This represents the enhancement factor of the entropy change rate on the correction parameter, which will affect channel A. With channel B Alignment is performed on the time dimension, and PCP is weighted and fused to achieve the desired effect. As a weighting factor, the correction magnitude of the pre-calibration parameters is dynamically adjusted: the higher the CQF value, the larger the correction magnitude; the lower the CQF value, the smaller the correction magnitude, resulting in the final chaotic quantization factor output. and pre-calibration parameters .

[0036] Preferably, step S4, the optimal parameter step, includes: obtaining pre-correction parameters and historical parameter sequences, aligning the pre-correction parameters and historical parameter sequences by timestamps, standardizing the parameter vectors, and constructing parameter vector pairs. , j represents the historical parameter index.

[0037] The pre-calibration parameters and historical parameter sequences are quantized using the Cauchy-Schwarz inequality, and the inner product is calculated for each pair of parameter vectors. Calculate the norms of the pre-correction parameter and historical parameter sequences respectively. and The quantized value of the Cauchy-Schwarz inequality is , The closer the value is to 1, the more similar the current pre-calibration parameters are to historical parameters. The smaller the value, the greater the deviation.

[0038] The optimal parameter combination is selected based on the quantized values ​​of the Cauchy-Schwarz inequality: if the objective is stability, then choose... The largest parameter pair; if the goal is exploratory, then choose... The smallest parameter pair, for multiple historical parameters The results are weighted and averaged to obtain the overall fitness value. If the overall fitness is lower than the preset threshold, the parameter correction mechanism is triggered, and the optimal parameter combination is selected based on the fitness results.

[0039] The objectives refer to the optimization direction of the system's operating strategy: stability objectives and exploratory objectives;

[0040] Stability objective: The system is currently operating smoothly (e.g., key indicators such as equipment temperature and vibration are within safe ranges), retaining historically proven effective strategies and avoiding significant parameter adjustments;

[0041] Exploratory objectives: When system performance declines (such as reduced equipment efficiency or increased failure rate), new operating conditions are encountered, or it is necessary to break through the current performance bottleneck, try new strategies.

[0042] The steps to determine whether to choose a stable objective or an exploratory objective include:

[0043] System operation status assessment: Monitor whether the operating parameters of the monitoring equipment are within a safe range. If the fluctuation is small and the status is stable, select the stability target; if the fluctuation is large and close to the critical value, select the exploratory target.

[0044] Overall fitness value judgment: If the overall fitness value is high, it means that the current parameter combination is effective, so a stable target is selected; if the overall fitness value is low, it means that the current parameter combination is ineffective, so an exploratory target is selected.

[0045] Preferably, in step S5, the optimal parameters are obtained, the numerical range of the optimal parameters is mapped to the encoding range of the pulse sequence, and the quantized parameter values ​​are converted into pulse sequences of corresponding frequencies through Poisson pulse coding, and the pulse sequence is divided into nanosecond-level time windows of fixed length.

[0046] By calling the pulse generation module inside the FPGA, nanosecond-level pulses are generated based on the encoded parameters. The optimized pulse parameters are loaded into the FPGA's memory in real time through an external interface, and the generated nanosecond-level pulse sequence is injected into the target device through a quantum control interface.

[0047] Based on the feedback signal of the quantum pulse, the pulse parameters are dynamically adjusted to compensate for device drift, and the parameters are adaptively corrected through real-time feedback from the FPGA.

[0048] Preferably, in step S6, when deploying the digital twin of the line equipment, an initial digital twin model is constructed based on the equipment design drawings, historical operating data and physical model, real-time operating data of the twin physical equipment is obtained, the physical equipment data and the output data of the digital twin model are aligned by timestamp to form a time series pair, and the mutual information value is dynamically calculated through the sliding window statistical method.

[0049] The mutual information is compared with a preset credibility threshold. If the mutual information is lower than the threshold for N consecutive time windows, the model is deemed to have insufficient credibility, triggering a reconstruction. The reconstructed model is then deployed to the digital twin system to replace the original model.

[0050] Compared with the prior art, the beneficial effects of the present invention are:

[0051] 1. This invention quantifies the entropy source formula through multi-physics field coupling, which can more accurately describe the state evolution process of equipment. The introduction of the Lyapunov exponent further evaluates the sensitivity of the system to initial disturbances, thereby providing early warning of potential instability risks. In the prediction of overheating faults in power transformers, thermal-electric coupling anomalies can be detected several hours in advance, avoiding sudden shutdowns and significantly improving the safety and reliability of equipment operation.

[0052] 2. This invention uses dual-channel parallel computing. In power grid frequency fluctuation correction, channel A quickly identifies high-frequency chaotic disturbances through nonlinear filtering and dynamic threshold comparison, and maps them to chaotic quantization factors. Meanwhile, channel B dynamically adjusts the pre-correction parameters according to the chaotic quantization factors, realizing real-time adaptive correction of parameters, which significantly improves the adaptability to complex working conditions.

[0053] 3. This invention quantifies the similarity between pre-calibration parameters and historical parameters using the Cauchy-Schwarz inequality, dynamically calculates the inner product and norm, and selects the parameter combination that best matches the current operating conditions, thereby avoiding overcalibration or undercalibration. If the goal is stability, parameter pairs with high similarity are selected; if the goal is exploration, parameter pairs with large differences are selected. This supports flexible strategy switching under different scenarios and significantly improves the long-term stability of the equipment.

[0054] 4. This invention maps optimal parameters into nanosecond-level pulse sequences through Poisson pulse coding and implements hardware-level real-time correction through FPGA. By leveraging the parallel computing capabilities of FPGA, pulse generation and injection are completed at the nanosecond level, ensuring precise control of device parameters. The real-time feedback mechanism of FPGA can dynamically adjust parameters based on the feedback signal of quantum pulses, forming a closed-loop compensation, effectively addressing device drift and significantly improving dynamic performance. Attached Figure Description

[0055] Figure 1 The present invention describes the operation flow of a dynamic correction method for digital twin model parameters of MPC line equipment. Figure 1 ;

[0056] Figure 2 The present invention describes the operation flow of a dynamic correction method for digital twin model parameters of MPC line equipment. Figure 2 ;

[0057] Figure 3 The present invention describes the operation flow of a dynamic correction method for digital twin model parameters of MPC line equipment. Figure 3 ;

[0058] Figure 4 The present invention describes the operation flow of a dynamic correction method for digital twin model parameters of MPC line equipment. Figure 4 . Detailed Implementation

[0059] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] Example:

[0061] Please see Figures 1-4 As shown, the present invention provides a technical solution comprising the following steps:

[0062] S1. Collect raw operating condition data through distributed edge nodes, extract key features of load fluctuation, environmental disturbance and mechanical spectrum, and form a chaotic feature vector;

[0063] S2. Based on the output chaotic feature vector, the entropy rate is calculated in real time through multi-physics partial differential coupling and Lyapunov exponent evaluation.

[0064] S3. Based on the generated entropy change rate, the chaotic quantization factor and pre-correction parameters are generated through dual-channel parallel computation.

[0065] S4. Integrate the pre-corrected parameters and historical parameters, evaluate the fitness using the Cauchy-Schwarz inequality, and output the optimal parameters;

[0066] S5. Map the optimized parameters to nanosecond-level pulse sequences, and use quantum pulse sequences to perform dynamic compensation and correction of FPGA device parameters;

[0067] S6. Monitor the mutual information of the twin entities during execution. When the credibility is lower than the threshold, trigger the entropy model reconstruction to form a closed-loop optimization.

[0068] In this embodiment, step S1 involves deploying multi-node edge sensors around key equipment, including but not limited to transformers, circuit breakers, transmission lines, current transformers, voltage transformers, and power electronic equipment. These sensors collect raw operating condition data such as vibration, temperature, current, and rotational speed in real time. The multi-node data is then aligned using timestamps to remove noise and invalid signals, and the peak value, RMS, kurtosis, and other indicators of the signals are calculated.

[0069] This can be understood as quantifying the instantaneous impact and long-term trend of load fluctuations, combining external parameters such as temperature and humidity to extract environmental interference patterns related to equipment operating status, extracting spectral peaks and harmonic components through Fourier transform, and capturing high-frequency resonant modulation signals using envelope spectrum analysis to identify local mechanical damage.

[0070] Based on phase space reconstruction, chaotic feature quantities such as Lyapunov exponent and fractal dimension are calculated, and time-domain and frequency-domain features are concatenated with chaotic features to form a high-dimensional feature vector, thus forming a chaotic feature vector containing linear and nonlinear information.

[0071] In this embodiment, step S2, multi-physics partial differential coupling, includes: obtaining chaotic feature vectors; constructing a physical field partial differential coupling equation model based on the actual physical characteristics of the equipment; where a single physical field includes a mechanical field and an electrical field; and the mechanical parameters are defined as follows: Its electrical parameters The gradient is This indicates the sensitivity of the mechanical state to electrical changes;

[0072] Electrical parameters are Its effect on thermodynamic parameters The gradient is , representing the sensitivity of thermodynamic changes to electrical state, the realization formula of the multiphysics partial differential coupling entropy source is:

[0073] ,

[0074] In the formula, This represents a multiphysics partial differential coupling entropy source. This represents the coupling weight coefficient of the i-th physical field. Indicates electrical parameters, Represents thermodynamic parameters, This represents the gradient of mechanical parameters with respect to electrical parameters. This represents the gradient of electrical parameters with respect to thermodynamic parameters. Represents the magnitude of the vector. This represents a weighted summation of the coupling terms over all n physical fields. This represents the amplification factor of chaotic features on coupling entropy. This represents the chaotic eigenvector.

[0075] In this embodiment, step S2, the Lyapunov exponent step, includes: perturbing the initial state of the multiphysics partial differential coupling entropy source to generate two adjacent orbits, a main orbit and a perturbed orbit. Specific steps include:

[0076] The current stable operating state of the multiphysics partial differential coupled entropy source is obtained as the initial state of the main orbit. The stable operating state contains the parameter values ​​of all physics fields. A very small perturbation value (perturbation quantity) is added to each parameter dimension of the initial state, typically 0. to The magnitude is used to ensure that the system remains in the linear response region after the disturbance, and that the system behavior does not undergo a fundamental change.

[0077] Adding perturbations to each parameter of the initial state creates a new initial state point that is very close to, but not exactly the same as, the original initial state point, thus forming the starting point of the perturbation trajectory.

[0078] Record the state differences between the two tracks at the initial moment, and set the time step and number of iterations.

[0079] In this embodiment, the evolution process of the multiphysics partial differential coupled entropy source is simulated by numerical integration, the state changes of the main orbit and the perturbation orbit are tracked in real time, the distance change of the two orbits within the time step is calculated, and the divergence velocity is evaluated.

[0080] The Lyapunov exponent is calculated based on the logarithmic average growth rate of orbital divergence, using the following formula:

[0081] ,

[0082] In the formula, The Lyapunov index represents the Lyapunov index. Indicates the time step. Let represent the magnitude of the system state perturbation vector at time t. This represents the magnitude of the perturbation vector at the initial moment. Represents the natural logarithm. This represents the correction coefficient of the multiphysics partial differential coupling entropy source to the Lyapunov exponent, reflecting the enhancing effect of the multiphysics partial differential coupling entropy source on chaotic behavior.

[0083] In this embodiment, step S2, the real-time entropy rate change step, includes: calculating the entropy rate change in real time based on the multiphysics partial differential coupling entropy source and the Lyapunov exponent, with the formula as follows:

[0084] ,

[0085] In the formula, Represents the rate of entropy change. Represents the base coefficient for entropy rate of change. This represents an exponential function, used to amplify the effect of the Lyapunov exponent on the rate of entropy change. It represents the amplification factor of the entropy rate of the multiphysics partial differential coupling entropy source, reflecting the nonlinear enhancement effect of the entropy rate on the entropy change by the multiphysics partial differential coupling entropy source.

[0086] In this embodiment, step S3, the chaotic quantization factor generation step, includes: obtaining the entropy change rate, splitting the entropy change rate into channel A and channel B according to the time series, calculating the chaotic quantization factor for channel A, and calculating the pre-correction parameter for channel B; and extracting high-frequency components from the entropy change rate in channel A through nonlinear filtering.

[0087] In this embodiment, a chaos intensity index is generated through dynamic threshold comparison, and then mapped to a chaos quantization factor. The formula is as follows:

[0088] ,

[0089] In the formula, This indicates the rate of entropy change within a time window. The moving average within, This represents the nonlinear amplification factor, controlling the sensitivity of the entropy rate of change to the CQF. This represents the dynamic threshold offset, which adjusts the baseline for chaos determination.

[0090] In this embodiment, step S3, generating pre-calibration parameters in channel B, includes: generating pre-calibration parameters based on the chaotic quantization factor in channel A. The formula is as follows:

[0091] ,

[0092] In the formula, Indicates the reference correction factor. This represents the enhancement factor of the entropy change rate on the correction parameter, which will affect channel A. With channel B Alignment is performed on the time dimension, and PCP is weighted and fused to achieve the desired effect. As a weighting factor, the correction magnitude of the pre-calibration parameters is dynamically adjusted: the higher the CQF value, the larger the correction magnitude; the lower the CQF value, the smaller the correction magnitude, resulting in the final chaotic quantization factor output. and pre-calibration parameters ,when When, i.e., the current rate of entropy change equals the rate of entropy change within the time window The moving average within a certain range indicates that the system is in a stable state without abnormal fluctuations. When, the correction amplitude increases; when If so, the correction amplitude will decrease.

[0093] In this embodiment, step S4, the optimal parameter step, includes: obtaining pre-correction parameters and historical parameter sequences, aligning the pre-correction parameters and historical parameter sequences by timestamps, standardizing the parameter vectors, and constructing parameter vector pairs. , j represents the historical parameter index.

[0094] The pre-calibration parameters and historical parameter sequences are quantized using the Cauchy-Schwarz inequality, and the inner product is calculated for each pair of parameter vectors. Calculate the norms of the pre-correction parameter and historical parameter sequences respectively. and The quantized value of the Cauchy-Schwarz inequality is , The closer the value is to 1, the more similar the current pre-calibration parameters are to historical parameters. The smaller the value, the greater the deviation.

[0095] In this embodiment, the optimal parameter combination is selected based on the quantization value of the Cauchy-Schwarz inequality: if the objective is stability, then select... The largest parameter pair; if the goal is exploratory, then choose... The smallest parameter pair, for multiple historical parameters The results are weighted and averaged to obtain the overall fitness value. If the overall fitness is lower than the preset threshold, the parameter correction mechanism is triggered, and the optimal parameter combination is selected based on the fitness results.

[0096] The objectives refer to the optimization direction of the system operation strategy: stability objectives and exploratory objectives.

[0097] Stability objective: The system is currently operating smoothly (e.g., key indicators such as equipment temperature and vibration are within safe ranges), retaining historically proven effective strategies and avoiding significant parameter adjustments.

[0098] Exploratory objectives: When system performance declines (such as reduced equipment efficiency or increased failure rate), new operating conditions are encountered, or it is necessary to break through the current performance bottleneck, try new strategies.

[0099] The steps to determine whether to choose a stable objective or an exploratory objective include:

[0100] System operation status assessment: Monitor whether the operating parameters of the monitoring equipment are within a safe range. If the fluctuation is small and the status is stable, select the stability target; if the fluctuation is large and close to the critical value, select the exploratory target.

[0101] Overall fitness value judgment: If the overall fitness value is high, it means that the current parameter combination is effective, so a stable target is selected; if the overall fitness value is low, it means that the current parameter combination is ineffective, so an exploratory target is selected.

[0102] In this embodiment, step S5 involves obtaining the optimal parameters, mapping the numerical range of the optimal parameters to the encoding range of the pulse sequence, converting the quantized parameter values ​​into a pulse sequence of the corresponding frequency through Poisson pulse coding, and dividing the pulse sequence into fixed-length nanosecond-level time windows.

[0103] In this embodiment, a nanosecond-level pulse is generated by calling the pulse generation module inside the FPGA based on the encoded parameters. The optimized pulse parameters are loaded into the FPGA's memory in real time through an external interface, and the generated nanosecond-level pulse sequence is injected into the target device through a quantum control interface.

[0104] Based on the feedback signal of the quantum pulse, the pulse parameters are dynamically adjusted to compensate for device drift, and the parameters are adaptively corrected through real-time feedback from the FPGA.

[0105] In this embodiment, during step S6, when deploying the digital twin of the line equipment, an initial digital twin model is constructed based on the equipment design drawings, historical operating data, and physical model. Real-time operating data of the twin physical equipment is obtained, and the physical equipment data and the output data of the digital twin model are aligned by timestamp to form a time series pair. The mutual information value is dynamically calculated using a sliding window statistical method.

[0106] The mutual information is compared with a preset credibility threshold. If the mutual information is lower than the threshold for N consecutive time windows, the model is deemed to have insufficient credibility, triggering a reconstruction. The reconstructed model is then deployed to the digital twin system to replace the original model.

[0107] Working principle: By deploying multi-node edge sensors around key equipment, raw operating condition data such as vibration, temperature, current, and rotation speed are collected in real time. Time deviation of multi-source data is eliminated by timestamp alignment. Key features of load fluctuation, environmental disturbance and mechanical spectrum are extracted by signal processing technology to quantify the changing trend of equipment operating status. On this basis, chaotic feature quantities such as Lyapunov exponent and fractal dimension are calculated by phase space reconstruction and nonlinear dynamic analysis. Time domain, frequency domain and nonlinear features are fused into a high-dimensional chaotic feature vector.

[0108] Based on the physical characteristics of the device, a multi-physics partial differential coupling equation model is constructed to quantify the interaction effects between various physical fields. By introducing chaotic eigenvectors as input, the multi-physics coupled entropy source is calculated to reflect the nonlinear changes in the system's energy distribution. Furthermore, the evolution process of the entropy source is simulated through numerical integration to evaluate the system's sensitivity to initial disturbances. Combined with the nonlinear enhancement effect of the entropy source, the entropy change rate is calculated in real time. The real-time calculated entropy change rate is processed in parallel through two channels. Channel A extracts high-frequency chaotic components through nonlinear filtering, generates a chaotic intensity index by combining dynamic threshold comparison, and maps it to a chaotic quantization factor. Channel B allocates pre-correction parameters according to the chaotic quantization factor. Through time-dimensional alignment and weighted fusion, the weights of the pre-correction parameters are dynamically adjusted. After standardization by aligning the pre-correction parameters with historical parameter sequences, the Cauchy-Schwarz inequality is used for quantization. Based on the target requirements, the parameter combinations with the highest or lowest similarity are selected, and a comprehensive fitness value is generated by weighted averaging. If the fitness value is lower than the threshold, a parameter correction mechanism is triggered to dynamically adjust the correction strategy. The optimized parameters are mapped to nanosecond-level pulse sequences, and the parameter values ​​are converted into pulse signals of corresponding frequencies through Poisson pulse coding. The FPGA hardware module generates control signals in real time based on the encoded pulse sequences and injects them into the target device through a quantum control interface to dynamically compensate for device drift. The mutual information between the physical device and the digital twin model is dynamically calculated using a sliding window statistical method to evaluate the consistency of the two data. When the mutual information is continuously lower than a preset threshold, the model is deemed to have insufficient credibility, triggering a model reconstruction mechanism. The reconstruction process updates the digital twin model based on the latest operating data and chaotic feature vectors and redeploys it to the system.

[0109] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their likenesses.

[0110] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A method for dynamic correction of parameters in a digital twin model of MPC line equipment, characterized in that, Includes the following steps: S1. Collect raw operating condition data through distributed edge nodes, extract key features of load fluctuation, environmental disturbance and mechanical spectrum, and form a chaotic feature vector; S2. Based on the output chaotic feature vector, the entropy rate is calculated in real time through multi-physics partial differential coupling and Lyapunov exponent evaluation. S3. Based on the generated entropy change rate, the chaotic quantization factor and pre-correction parameters are generated through dual-channel parallel computation. S4. Integrate the pre-corrected parameters and historical parameters, evaluate the fitness using the Cauchy-Schwarz inequality, and output the optimal parameters; S5. Map the optimized parameters to nanosecond-level pulse sequences, and use quantum pulse sequences to perform dynamic compensation and correction of FPGA device parameters; S6. Monitor the mutual information of the twin entities during execution. When the credibility is lower than the threshold, trigger the entropy model reconstruction to form a closed-loop optimization.

2. The method for dynamic correction of parameters of a digital twin model of MPC line equipment according to claim 1, characterized in that: The S2 step, multi-physics partial differential coupling, includes: obtaining chaotic feature vectors; constructing a physical field partial differential coupling equation model based on the actual physical characteristics of the equipment; where a single physical field includes a mechanical field and an electrical field; and the mechanical parameters are defined as follows: Electrical parameters are The formula for realizing the entropy source of multiphysics partial differential coupling is: ， In the formula, This represents a multiphysics partial differential coupling entropy source. This represents the coupling weight coefficient of the i-th physical field. Indicates electrical parameters, Represents thermodynamic parameters, This represents the gradient of mechanical parameters with respect to electrical parameters. This represents the gradient of electrical parameters with respect to thermodynamic parameters. Represents the magnitude of the vector. This represents a weighted summation of the coupling terms over all n physical fields. This represents the amplification factor of chaotic features on coupling entropy. This represents the chaotic eigenvector.

3. The method for dynamic correction of parameters of a digital twin model of MPC line equipment according to claim 2, characterized in that: The S2 step, the Lyapunov exponent step, includes: perturbing the initial state of the multiphysics partial differential coupled entropy source to generate two adjacent orbits, recording the state difference between the two orbits at the initial moment, setting the time step and the number of iterations, and calculating the Lyapunov exponent, with the formula as follows: ， In the formula, The Lyapunov index represents the Lyapunov index. Indicates the time step. Let represent the magnitude of the system state perturbation vector at time t. This represents the magnitude of the perturbation vector at the initial moment. Represents the natural logarithm. This represents the correction coefficient of the Lyapunov exponent to the multiphysics partial differential coupling entropy source.

4. The method for dynamic correction of parameters of a digital twin model of MPC line equipment according to claim 3, characterized in that: The S2 step, the real-time entropy rate change step, includes: calculating the entropy rate change in real time based on the multi-physics partial differential coupling entropy source and the Lyapunov exponent, with the formula as follows: ， In the formula, Represents the rate of entropy change. Represents the base coefficient for entropy rate of change. Represents an exponential function. It represents the amplification factor of the entropy rate of a multi-physics partial differential coupling entropy source.

5. The method for dynamic correction of parameters of a digital twin model of MPC line equipment according to claim 1, characterized in that: The S3 step of generating the chaos quantization factor includes: obtaining the entropy change rate; splitting the entropy change rate into channel A and channel B according to the time series; extracting high-frequency components from the entropy change rate in channel A through nonlinear filtering; generating a chaos intensity index through dynamic threshold comparison; and mapping the chaos intensity index to a chaos quantization factor. .

6. The method for dynamic correction of parameters of a digital twin model of MPC line equipment according to claim 5, characterized in that: The step S3, generating pre-calibration parameters in channel B, includes: generating pre-calibration parameters based on the chaotic quantization factor in channel A. The formula is as follows: ， In the formula, Indicates the reference correction factor. This represents the enhancement factor of the entropy change rate on the correction parameter.

7. The method for dynamic correction of parameters of a digital twin model of MPC line equipment according to claim 1, characterized in that: The optimal parameter step in S4 includes: obtaining the pre-correction parameters and historical parameter sequences, constructing parameter vector pairs, quantizing the pre-correction parameters and historical parameter sequences using the Cauchy-Schwarz inequality, and calculating the inner product for each parameter vector pair. The norms of the pre-correction parameter and historical parameter sequences are calculated separately, and the Cauchy-Schwarz inequality quantization value is... ; The optimal parameter combination is selected based on the quantized values ​​of the Cauchy-Schwarz inequality: if the objective is stability, then choose... The largest parameter pair; if the goal is exploratory, then choose... The smallest parameter pair, for multiple historical parameters The results are weighted and averaged to obtain the overall fitness value. If the overall fitness is lower than the preset threshold, the parameter correction mechanism is triggered, and the optimal parameter combination is selected based on the fitness results.

8. The method for dynamic correction of parameters of a digital twin model of MPC line equipment according to claim 1, characterized in that: In step S5, the optimal parameters are obtained, and the numerical range of the optimal parameters is mapped to the encoding range of the pulse sequence. Through Poisson pulse coding, the quantized parameter values ​​are converted into pulse sequences of corresponding frequencies. The pulse sequence is divided into fixed-length nanosecond-level time windows. By calling the pulse generation module inside the FPGA, nanosecond-level pulses are generated according to the encoded parameters. The optimized pulse parameters are loaded into the FPGA's memory in real time through an external interface. The generated nanosecond-level pulse sequence is injected into the target device through a quantum control interface. Based on the feedback signal of the quantum pulse, the pulse parameters are dynamically adjusted to compensate for device drift. Parameter adaptive correction is achieved through real-time feedback from the FPGA.

9. The method for dynamic correction of parameters of a digital twin model of MPC line equipment according to claim 1, characterized in that: In step S6, real-time operating data of the twin physical device is acquired, and the physical device data and the output data of the digital twin model are aligned by timestamp to form a time series pair. The mutual information value is dynamically calculated by the sliding window statistical method, and the mutual information value is compared with the preset credibility threshold. If the mutual information value is lower than the threshold for N consecutive time windows, it is determined that the model credibility is insufficient, and reconstruction is triggered. The reconstructed model is deployed to the digital twin system for updating.

Images