Capacitive voltage transformer online monitoring device calibration instrument and calibration method

Abstract

The invention relates to the technical field of online monitoring of capacitor voltage transformers, in particular to a calibrator and method for an online monitoring device of a capacitor voltage transformer, the calibrator comprises a difference adding module, an algorithm module and a judgment module, the difference adding module generates a plurality of groups of differential voltage signals, inputs the differential voltage signals into the online monitoring device and forms parameter vectors to be estimated; the algorithm module obtains a related voltage amplitude calculation theoretical ratio error, and constructs a hybrid optimization algorithm in combination with a Markov chain Monte Carlo method and a simulated annealing algorithm to calculate a posterior probability density so as to obtain the uncertainty of an error measurement result; and the judgment module combines the historical average uncertainty, the real-time temperature and the use year of the online monitoring device to calculate a verification parameter, compares the verification parameter with a preset threshold value and outputs an abnormal judgment result. According to the invention, non-power-cut verification can be realized, the verification efficiency and accuracy are improved, and the reliability of power supply of a power grid is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of online monitoring technology for capacitive voltage transformers, and specifically to a calibrator and calibration method for online monitoring devices of capacitive voltage transformers. Background Technology

[0002] With the continuous development of power systems and the increasing demands for power supply reliability, capacitive voltage transformers, as key primary equipment, play a crucial role in power grids with voltage levels of 110kV and above. To ensure the accuracy and stability of their metering performance, online monitoring devices for transformer metering performance have been widely used in these high-voltage scenarios. While these devices can achieve real-time monitoring of the operating status of capacitive voltage transformers, facilitate lean management and rapid fault handling, and reduce equipment downtime and on-site operational risks, existing online monitoring devices still have significant limitations in practical applications. Factors such as bumps during transportation, temperature and humidity fluctuations and electromagnetic interference in complex on-site environments, as well as component aging and software failures after long-term operation, can easily lead to data deviations in the monitoring devices, directly affecting the accuracy and reliability of the monitoring data and making it difficult to ensure long-term stable metering calibration accuracy.

[0003] In the existing technology, there are significant deficiencies in the calibration methods for online monitoring devices for the metrological performance of instrument transformers: there is a lack of specially designed portable calibrators at engineering sites, making it impossible to quickly verify the reliability of the monitoring devices; some calibration equipment requires external power supply, which is difficult to adapt to scenarios without stable power supply, such as field or temporary operations, resulting in calibration work being greatly restricted by the environment and unable to respond to on-site calibration needs in a timely manner, thereby affecting the continuous and efficient operation of the monitoring devices.

[0004] Furthermore, existing verification schemes lack a comprehensive calibration system: even those few verification methods for monitoring devices often fail to consider the impact of environmental factors on the verification results and do not establish calibration logic adapted to the usage status and environmental parameters of the monitoring devices, resulting in limited reference value of the verification results; at the same time, the lack of systematic integration and analysis of verification data makes it impossible to provide data support for the maintenance and upgrading of monitoring devices, making it difficult to further improve the maintainability and scalability of monitoring devices, and failing to meet the needs of modern power systems for refined and convenient equipment calibration. Summary of the Invention

[0005] This invention provides a calibration instrument and calibration method for an online monitoring device of a capacitive voltage transformer, which overcomes the shortcomings of the prior art. It can effectively solve the problems that existing online monitoring devices of transformers are easily affected by internal and external factors, resulting in data deviation, and lack portable and adaptable calibration methods, making it difficult to quickly and accurately verify their reliability.

[0006] One of the technical solutions of the present invention is achieved through the following measures: a calibrator for an online monitoring device of a capacitive voltage transformer, comprising: a differential module, an algorithm module, and a judgment module;

[0007] The differential module is used to realize digital differential voltage, generate multiple sets of micro differential voltage signals and input them to the online monitoring device of capacitive voltage transformer. The micro differential voltage signals constitute the vector of parameters to be estimated.

[0008] The algorithm module is used to obtain the voltage amplitude of the input channel of the calibrator and the primary voltage amplitude measured by the capacitive voltage transformer, calculate the theoretical ratio error of the online monitoring device, and construct a hybrid optimization algorithm by combining the Markov chain Monte Carlo method and the simulated annealing algorithm. Based on the prior probability density of the parameter vector to be estimated and the likelihood function of the measurement model, the posterior probability density is calculated to obtain the uncertainty of the error measurement result.

[0009] The judgment module is used to calculate the verification parameters by combining the historical average uncertainty, real-time temperature and the years of use of the online monitoring device, and compares the verification parameters with the preset verification threshold to output the abnormal judgment result of the online monitoring device of the capacitive voltage transformer.

[0010] The following are further optimizations and / or improvements to one of the above-mentioned technical solutions:

[0011] The aforementioned online monitoring device calibrator for capacitive voltage transformers may also include a front panel assembly, which may include a display unit, an indicator unit, an operation unit, and a power control unit.

[0012] The display unit is used to show verification data and the operation interface;

[0013] The indicator unit is used to provide feedback on the power supply status, operating status, time synchronization status, alarm status, and power amplifier operating status of the calibrator;

[0014] The operation unit is used to input operation commands and parameters;

[0015] The power control unit is used to enable the calibrator to go into standby and wake up.

[0016] The second technical solution of the present invention is achieved through the following measures: a verification method for an online monitoring device for a capacitive voltage transformer, comprising the following steps:

[0017] Step 1: Generate multiple sets of differential voltage signals and input them into the online monitoring device for the capacitive voltage transformer, and construct the vector of parameters to be estimated from the multiple sets of differential voltage signals.

[0018] Step 2: Obtain the voltage amplitude input to the online monitoring device and the primary voltage amplitude measured by the capacitive voltage transformer, and calculate the theoretical ratio error of the online monitoring device;

[0019] Step 3: Combine the Markov chain Monte Carlo method and the simulated annealing algorithm to construct a hybrid optimization algorithm. Based on the prior probability density of the parameter vector to be estimated and the likelihood function of the measurement model, the hybrid optimization algorithm is used to calculate the posterior probability density and obtain the uncertainty of the error measurement result.

[0020] Step 4: Calculate the verification parameters by combining the historical average uncertainty, real-time temperature, and the years of use of the online monitoring device. Compare the verification parameters with the preset verification threshold and output the judgment result of whether the online monitoring device has any abnormalities.

[0021] The following are further optimizations and / or improvements to the second technical solution of the above invention:

[0022] In step 1 above, the number of multiple sets of differential voltage signals can be 3-10 sets, and the amplitude of each set of differential voltage signals is 0.01%-0.5% of the rated voltage of the capacitive voltage transformer.

[0023] In step 3 above, the prior probability density can be a normal distribution with a mean of 0 and a standard deviation of 0.001%-0.01%.

[0024] The likelihood function of the measurement model in step 3 above can be constructed based on the historical measurement error data of the online monitoring device and satisfies the Gaussian white noise distribution characteristics.

[0025] The execution process of the hybrid optimization algorithm in step 3 above can be as follows: first, the initial value of the global optimal solution is obtained through the simulated annealing algorithm, and then the initial value is iteratively optimized through the Markov chain Monte Carlo method, with the number of iterations being 1000-5000.

[0026] The initial temperature of the simulated annealing algorithm described above can be 100-500K, the cooling coefficient can be 0.85-0.95, and the termination temperature can be 1-5K.

[0027] It may also include step 5: if it is determined that there is an abnormality in the online monitoring device, generate a report that includes the type of abnormality, the error value, and a recommended maintenance plan.

[0028] In step 1 above, the differential voltage signal can be generated by digital synthesis, and the signal frequency is consistent with the operating frequency of the capacitive voltage transformer.

[0029] This invention effectively addresses the limitations of existing verification methods, providing reliable support for ensuring the metering performance of critical equipment in high-voltage power grids. Through the collaborative design of its core modules and verification process, it generates suitable differential voltage signals and constructs a vector of parameters to be estimated. Combined with a hybrid optimization algorithm, it accurately calculates the uncertainty of error measurement results. Simultaneously, it incorporates key influencing factors such as historical data, ambient temperature, and equipment age, enabling a scientific assessment of the online monitoring device's operating status. This significantly improves the accuracy and reliability of verification results, avoiding verification deviations caused by a single algorithm or neglecting environmental interference. Furthermore, the accompanying front panel components optimize the human-machine interface, facilitating real-time access to verification data, monitoring equipment status, and convenient command input for operators. The optional anomaly report generation function provides clear guidance for equipment maintenance, helping to quickly locate problems and implement targeted measures. Moreover, the entire verification process does not interrupt the normal power supply to the grid, effectively avoiding the power supply risks and economic losses caused by power outages. It also reduces on-site labor costs and safety hazards, further adapting to the lean management needs of modern power systems and ensuring the long-term stable operation of capacitive voltage transformers, laying a solid foundation for the safe and efficient power supply of the power grid. Attached Figure Description

[0030] Appendix Figure 1 This is a schematic diagram of the principle of the online monitoring device calibrator for capacitive voltage transformers according to an embodiment of the present invention.

[0031] Appendix Figure 2 This is a flowchart illustrating the verification method of the online monitoring device for capacitive voltage transformers according to an embodiment of the present invention. Detailed Implementation

[0032] The present invention is not limited to the following embodiments, and the specific implementation can be determined according to the technical solution of the present invention and the actual situation.

[0033] The present invention will be further described below with reference to embodiments:

[0034] Example 1: As shown in the attached document Figure 1 , 2 As shown, the online monitoring device calibrator for capacitive voltage transformers includes: a differential module, an algorithm module, and a judgment module;

[0035] The differential module is used to realize digital differential voltage, generate multiple sets of micro differential voltage signals and input them to the online monitoring device of capacitive voltage transformer. The micro differential voltage signals constitute the vector of parameters to be estimated.

[0036] The algorithm module is used to obtain the voltage amplitude of the input channel of the calibrator and the primary voltage amplitude measured by the capacitive voltage transformer, calculate the theoretical ratio error of the online monitoring device, and construct a hybrid optimization algorithm by combining the Markov chain Monte Carlo method and the simulated annealing algorithm. Based on the prior probability density of the parameter vector to be estimated and the likelihood function of the measurement model, the posterior probability density is calculated to obtain the uncertainty of the error measurement result.

[0037] The judgment module is used to calculate the verification parameters by combining the historical average uncertainty, real-time temperature and the years of use of the online monitoring device, and compares the verification parameters with the preset verification threshold to output the abnormal judgment result of the online monitoring device of the capacitive voltage transformer.

[0038] The differential voltage module generates multiple sets of differential voltage signals through digital voltage differential calculation. These signals are input into the online monitoring device while simultaneously integrating them into a vector of parameters to be estimated, providing fundamental data for subsequent error analysis. The algorithm module first collects the voltage amplitude from the calibrator's input channel and the primary voltage amplitude measured by the capacitive voltage transformer. It calculates the theoretical ratio error based on the relationship between the two, then integrates the probabilistic statistical capabilities of the Markov chain Monte Carlo method with the global search capability of the simulated annealing algorithm to construct a hybrid optimization algorithm. Using the prior probability density of the parameter vector to be estimated and the likelihood function of the measurement model, the algorithm calculates the posterior probability density, thereby determining the uncertainty of the error measurement result. The judgment module collects historical average uncertainty, real-time temperature data, and the years of use of the online monitoring device. These factors are incorporated into the calculation process of the calibration parameters. The calculated calibration parameters are then compared with a preset calibration threshold. Based on the comparison result, the module determines whether the online monitoring device is abnormal and outputs the result. This allows for accurate acquisition of error information and uncertainty of the online monitoring device, scientifically judging its operating status, providing core hardware support for the calibration of the capacitive voltage transformer online monitoring device, and ensuring a reliable equipment foundation for calibration. The error in this patent refers to the ratio error.

[0039] In this embodiment, the online monitoring device calibrator for capacitive voltage transformers also includes a front panel assembly, which includes a display unit, an indicator unit, an operation unit, and a power control unit.

[0040] The display unit is used to show verification data and the operation interface;

[0041] The indicator unit is used to provide feedback on the power supply status, operating status, time synchronization status, alarm status, and power amplifier operating status of the calibrator;

[0042] The operation unit is used to input operation commands and parameters;

[0043] The power control unit is used to enable the calibrator to go into standby and wake up.

[0044] Specifically, the online monitoring device calibrator for capacitive voltage transformers consists of an indicator light panel, an LCD screen, and buttons on the front panel. (See attached...) Figure 1 In the center, the power source provides stable power for the overall operation of the calibrator or its core modules. It includes an 11-inch LCD screen with resistive touch; five indicator lights, including a power light, a running light, a time synchronization light, an alarm light, and a power amplifier running light; digital input keys, directional navigation keys, and multiple dedicated function keys; and a standby / wake-up power button. The display unit receives various data (such as error values, uncertainties, etc.) and operation interface information generated during the verification process, presenting them to the operator in a visual form. The indicator unit uses different status indicators (such as light on / off, color changes, etc.) to provide feedback on the current power supply status of the calibrator, whether it is in operation, whether time synchronization was successful, whether an alarm has occurred, and the operating status of the power amplifier. The operation unit receives operation commands (such as start verification, stop verification) and related parameters (such as preset verification thresholds) input by the operator and transmits this information to the core module of the calibrator. The power control unit controls the calibrator to enter standby mode to save energy or wakes it up from standby mode to resume operation according to the operator's operation or preset rules. This optimizes the human-machine interaction experience, allowing operators to intuitively obtain verification information, monitor the equipment status in real time, conveniently control the calibrator, ensure the smooth progress of the verification process, and improve operational efficiency.

[0045] During operation, the differential voltage module of the calibrator first generates multiple sets of differential voltage signals that meet the quantity and amplitude requirements. Digital synthesis ensures the signal frequency matches the operating frequency of the capacitive voltage transformer. The signals are then input into the online monitoring device to form a vector of parameters to be estimated. Next, the algorithm module obtains the relevant voltage amplitude and calculates the theoretical ratio error. Combining the set normal distribution prior probability density and the likelihood function of Gaussian white noise distribution constructed based on historical data, a simulated annealing algorithm (with corresponding initial temperature, cooling coefficient, and termination temperature) is used to obtain the initial value of the global optimal solution. Then, iterative optimization using the Markov chain Monte Carlo method yields the posterior probability density and error uncertainty. The judgment module calculates the verification parameters based on the historical average uncertainty, real-time temperature, and equipment usage year, compares them with a threshold, and outputs anomaly results. If an anomaly is found, a report containing anomaly information and maintenance plan is generated. The front panel components simultaneously display data, feedback status, receive operations, and control the power supply. The overall technical effect is to achieve accurate and efficient verification of the online monitoring device for capacitive voltage transformers, which can be completed without power outages, improving verification convenience and reliability, ensuring the stable operation of key power grid equipment, and reducing maintenance costs and power supply risks.

[0046] This invention combines the Markov Chain Monte Carlo (MCMC) method with the Simulated Annealing (SA) algorithm to propose a hybrid optimization algorithm to approximate the posterior probability density, thereby using Bayesian theory to calculate the uncertainty and determine the reliability of the online monitoring device. This invention achieves the following technical effects: (1) Uninterrupted verification: Traditional verification of capacitive voltage transformers usually requires a power outage, which not only affects the normal operation of the power system but may also cause certain economic losses. However, this invention can verify the online monitoring device of capacitive voltage transformers under uninterrupted power conditions, greatly improving the operating efficiency and reliability of the power system. (2) High efficiency and convenience: By adopting advanced algorithm technology, this invention realizes that the verification process can be completed by software algorithm alone, without the need for complex hardware adjustments or physical connection changes. This design greatly simplifies the verification process, making the operation simpler and faster, and reducing labor costs and technical requirements. (3) Improved accuracy and reliability: By using precise algorithm analysis, this invention can more accurately identify the working status of the online monitoring device of capacitive voltage transformers and promptly discover potential problems. This helps improve the accuracy and reliability of voltage monitoring throughout the power system, providing a strong guarantee for the safe and stable operation of the power grid.

[0047] Example 2: As shown in the attached document Figure 1 , 2 As shown in the figure, this embodiment discloses a verification method for an online monitoring device for a capacitive voltage transformer, including the following steps:

[0048] Step 1: Generate multiple sets of differential voltage signals and input them into the online monitoring device for the capacitive voltage transformer, and construct the vector of parameters to be estimated from the multiple sets of differential voltage signals.

[0049] Step 2: Obtain the voltage amplitude input to the online monitoring device and the primary voltage amplitude measured by the capacitive voltage transformer, and calculate the theoretical ratio error of the online monitoring device;

[0050] Step 3: Combine the Markov chain Monte Carlo method and the simulated annealing algorithm to construct a hybrid optimization algorithm. Based on the prior probability density of the parameter vector to be estimated and the likelihood function of the measurement model, the hybrid optimization algorithm is used to calculate the posterior probability density and obtain the uncertainty of the error measurement result.

[0051] Step 4: Calculate the verification parameters by combining the historical average uncertainty, real-time temperature, and the years of use of the online monitoring device. Compare the verification parameters with the preset verification threshold and output the judgment result of whether the online monitoring device has any abnormalities.

[0052] Step 1 generates multiple sets of differential voltage signals using a specific method. After inputting these signals into the online monitoring device, they are integrated into a parameter vector to be estimated, providing data support for subsequent error calculation. Step 2 collects the voltage amplitude input to the online monitoring device and the primary voltage amplitude measured by the capacitive voltage transformer. Based on the mathematical relationship between the two, the theoretical ratio error of the online monitoring device is calculated to preliminarily determine the error situation. Step 3 integrates the advantages of two algorithms to construct a hybrid optimization algorithm. Based on the prior probability density of the parameter vector to be estimated and the likelihood function of the measurement model, the algorithm calculates the posterior probability density to determine the uncertainty of the error measurement result. Step 4 collects historical average uncertainty, real-time temperature, and the years of use of the online monitoring device. These factors are quantified and incorporated into the calculation of the verification parameters. The verification parameters are then compared with a preset threshold to output the anomaly judgment result. This process forms a complete and orderly verification procedure, progressing step by step from data preparation to result output, providing a standardized and systematic method for judging the reliability of online monitoring devices and ensuring that the verification process is systematic and traceable.

[0053] In this embodiment, the number of differential voltage signals in step 1 is 3-10, and the amplitude of each differential voltage signal is 0.01%-0.5% of the rated voltage of the capacitive voltage transformer. When generating differential voltage signals in step 1, the number of control signals is kept within the range of 3-10, while the amplitude of each signal is limited to 0.01%-0.5% of the rated voltage of the capacitive voltage transformer. This ensures that the generated signals can cover the needs of different verification scenarios without causing redundancy or inaccuracy in the verification data due to excessive quantity or excessively large / small amplitudes, providing suitable sample data for subsequent steps. This ensures that the differential voltage signal parameters input to the online monitoring device are reasonable, avoiding interference with the verification results due to improper signal parameters, improving the effectiveness and reliability of the verification data, and laying the foundation for accurate error calculation.

[0054] In this embodiment, the prior probability density in step 3 is a normal distribution with a mean of 0 and a standard deviation of 0.001%-0.01%. Before calculating the posterior probability density in step 3, the prior probability density of the parameter vector to be estimated is set to a normal distribution with a mean of 0 and a standard deviation in the range of 0.001%-0.01%. This distribution conforms to the common statistical laws of error data and can reflect the probability distribution characteristics of the parameter to be estimated without additional information. This provides a scientific initial statistical basis for the subsequent calculation of the posterior probability density using the likelihood function. In this way, a reasonable statistical basis can be provided for the operation of the hybrid optimization algorithm, reducing the calculation deviation caused by improper setting of the prior probability density, improving the accuracy of the posterior probability density calculation, and thus ensuring the reliability of the uncertainty of the error measurement results.

[0055] In this embodiment, the likelihood function of the measurement model in step 3 is constructed based on the historical measurement error data of the online monitoring device and satisfies the characteristics of Gaussian white noise distribution. When constructing the likelihood function of the measurement model in step 3, past measurement error data of the online monitoring device is collected, and the data patterns are analyzed to construct the likelihood function based on these patterns. Simultaneously, it is ensured that the likelihood function conforms to the characteristics of Gaussian white noise distribution, because Gaussian white noise is a common noise distribution in actual measurement scenarios and can accurately describe the probabilistic characteristics of random errors during the measurement process. This allows the likelihood function to better fit the actual measurement situation of the online monitoring device, accurately reflect the probabilistic characteristics of the measurement model, improve the accuracy of the hybrid optimization algorithm in calculating the posterior probability density, and provide a reliable guarantee for the accurate calculation of error uncertainty.

[0056] In this embodiment, the execution process of the hybrid optimization algorithm in step 3 is as follows: First, the initial value of the global optimal solution is obtained through simulated annealing algorithm, and then the initial value is iteratively optimized through Markov chain Monte Carlo method, with 1000-5000 iterations. When executing the hybrid optimization algorithm in step 3, the simulated annealing algorithm is started first, using its global search capability to traverse the solution space and avoid getting trapped in local optima, thereby obtaining the initial value of the global optimal solution; then the Markov chain Monte Carlo method is started, starting from the obtained initial value, and 1000-5000 iterations are performed to gradually optimize the accuracy of the solution, so that the result continuously approaches the true optimal solution; in this way, the global search capability and local optimization accuracy of the algorithm can be taken into account, avoiding the calculation deviation caused by improper initial value, and controlling the calculation efficiency while ensuring calculation accuracy through a reasonable number of iterations, thereby improving the overall performance of the hybrid optimization algorithm.

[0057] In this embodiment, the initial temperature of the simulated annealing algorithm is 100-500K, the cooling coefficient is 0.85-0.95, and the termination temperature is 1-5K. When using the simulated annealing algorithm in step 3, setting the initial temperature within the range of 100-500K ensures that the algorithm has sufficient energy to broadly search the solution space in the early stages, avoiding premature limitation to local regions. Setting a cooling coefficient of 0.85-0.95 controls the rate of temperature decrease, allowing the algorithm to gradually focus on high-quality solution regions during the search process without missing the global optimum due to excessively rapid cooling. Setting the termination temperature to 1-5K ensures that the algorithm eventually converges stably to the vicinity of the optimal solution, ending the search process. This optimizes the operating parameters of the simulated annealing algorithm, ensures its global search effect, improves the quality of the initial value of the global optimum, provides a good foundation for subsequent iterative optimization of the Markov chain Monte Carlo method, and further guarantees the computational effect of the hybrid optimization algorithm.

[0058] In this embodiment, step 5 is also included: if an anomaly is determined in the online monitoring device, a report containing the anomaly type, error value, and suggested maintenance plan is generated. After the judgment result that the online monitoring device is abnormal is output in step 4, the execution of step 5 is triggered. The system automatically collects anomaly-related information, determines the type of anomaly (such as error exceeding the standard, parameter drift, etc.), extracts the corresponding error value, and then combines common fault handling experience and equipment characteristics to generate a targeted suggested maintenance plan. This information is integrated into a report and output. In this way, after an anomaly is discovered, not only is the key information of the anomaly clearly identified, but also specific handling guidance is provided to maintenance personnel, reducing maintenance and troubleshooting time, improving equipment maintenance efficiency, and helping the online monitoring device quickly return to normal operation.

[0059] In this embodiment, the differential voltage signal in step 1 is generated through digital synthesis, and the signal frequency is consistent with the operating frequency of the capacitive voltage transformer. When generating the differential voltage signal in step 1, digital synthesis technology is used to construct the signal. This technology can precisely control the signal parameters, ensuring the stability and accuracy of the signal. Simultaneously, the frequency of the generated signal is adjusted to match the operating frequency of the capacitive voltage transformer, avoiding abnormal signal reception by the online monitoring device due to frequency mismatch, which would affect the validity of the verification data. This improves the generation quality of the differential voltage signal and its compatibility with the transformer, reduces the interference of signal defects on the verification results, and further ensures the accuracy and reliability of the verification process.

[0060] In this invention, the Markov chain Monte Carlo method is a random sampling algorithm based on Markov chains. Its core logic is to construct a Markov chain with a stationary distribution consistent with the target posterior probability density, and to estimate the parameters of complex probability distributions through extensive sampling. In this invention, it is used for iterative optimization (1000-5000 iterations) after obtaining the initial global optimal solution output by the simulated annealing algorithm. By sampling and quantifying the probability distribution of the parameter vector to be estimated, the uncertainty of the error measurement result is finally accurately calculated. The simulated annealing algorithm simulates the "heating-cooling" process of metal annealing in physics, achieving optimization by gradually decreasing the "temperature" parameter. The initial temperature (100-500K) ensures the algorithm can search extensively in the solution space, avoiding getting trapped in local optima; the cooling coefficient (0.85-0.95) controls the cooling rate, balancing search breadth and efficiency; the termination temperature (1-5K) allows the algorithm to converge to near the optimal solution. In this invention, this algorithm is used to obtain the initial global optimal solution required by the hybrid optimization algorithm, laying the foundation for subsequent iterative optimization. This hybrid optimization algorithm combines the probabilistic and statistical advantages of the Markov Chain Monte Carlo method with the global search advantage of the Simulated Annealing algorithm. During execution, the initial value of the global optimum is first obtained through simulated annealing, addressing the problem of single algorithms easily getting trapped in local optima. Then, the Markov Chain Monte Carlo method iteratively optimizes the initial value, improving the accuracy of the solution. This algorithm balances search breadth and accuracy, enabling more precise calculation of the posterior probability density, improving the reliability of the uncertainty in error measurement results, and meeting verification requirements.

[0061] Example 3: As shown in the attached document Figure 1 , 2 As shown in the figure, this embodiment discloses the specific steps of a verification method for an online monitoring device for a capacitive voltage transformer, including:

[0062] The voltage is digitally differentially amplified by the differential module of the calibrator, and used as the voltage amplitude signal input. This signal is then fed into the online monitoring device for the capacitive voltage transformer. The set of k random differential amplifications is denoted as U. d ,Right now:

[0063] U D ={U1,U2,...,U k}

[0064] By definition, the theoretical specific difference y of an online monitoring device is...

[0065]

[0066] U D For the superimposed differential amplitude, U C U is the primary voltage amplitude measured by the CVT. N Input channel voltage amplitude to the calibrator. Vector U D Let X be a random variable with components x.i y is an indirect measurement.

[0067] In this patent, the parameter vector to be estimated is set as X = (x1, x2, x3, ..., x k ), the i-th random variable x i The prior probability density is p i (x i ), is by U d If the information is transformed, then the prior probability density p x for:

[0068]

[0069] Assuming the measurement model is e, used to estimate the device's deviation, the measurement model deals with multidimensional parameters to be estimated. Statistically, y follows a distribution with mean u and standard deviation σ. The likelihood function L is then calculated.

[0070]

[0071] Indirect measurement models involve multidimensional parameters to be estimated, making the actual solution of the posterior probability density quite complex and difficult to derive a clear analytical expression. Furthermore, the computational cost of numerical integration of the best estimate and uncertainty increases exponentially with increasing dimension. Therefore, it is necessary to design other improved methods to solve the posterior probability density.

[0072] This patent proposes a hybrid optimization algorithm to approximate the posterior probability density by combining the Markov chain Monte Carlo (MCMC) method with the Simulated Annealing (SA) algorithm. The MCMC method is used to construct the Markov chain, while the simulated annealing algorithm is used to assist in the search, improving sampling efficiency and accuracy.

[0073] Choose the initial value of parameter vector X as X (0) The posterior probability density to be calculated is:

[0074] h(x|y)=p(x)L(y|x)

[0075] Represented by π, that is

[0076] π(x)=h(x|y)

[0077] Set the simulated annealing parameters: initial temperature T0, dehumidification coefficient α, and final temperature T. end The proposal distribution is automatically set to q(x'|x), and its distribution is as follows:

[0078] q(x'|x)=N(x',x,σ 2 I)

[0079] Where σ is the standard deviation of the proposed distribution, and I is the identity matrix. A large maximum number of iterations Q is set (preferably Q = 200), and then the simulated annealing iteration begins:

[0080] (1) From the current state x (t) Generate a candidate state x' (where t is the iteration number), which follows the proposal distribution q(x'|x).

[0081] (2) Calculate the MCMC acceptance probability α MCMC

[0082]

[0083] Based on the symmetric proposal distribution q(x'|x (t) )=q(x (t) |x'), the above equation can be simplified to

[0084]

[0085] (3) Design an improved annealing acceptance probability algorithm to calculate the simulated annealing acceptance probability α. SA :

[0086]

[0087] T represents the current temperature. This formula achieves the following: in the early stages of the search, the former is close to 1, allowing for a larger acceptance probability and encouraging exploration; while as time progresses, it tends to choose a better solution, which helps the algorithm converge. The larger the sample k, the greater the acceptance probability is required.

[0088] (4) Calculate the overall acceptance probability α

[0089] α = max(α) MCMC α SA )

[0090] (5) Generate a uniformly distributed random number u ~ U(0,1)

[0091] If u≤α, then accept candidate state x', i.e., x (t+1) =x', otherwise, reject the candidate state and keep the current state unchanged, i.e., x'. (t+1) =x (t)

[0092] (6) Update the temperature T = αT0 when the number of iterations is greater than Q, or when the temperature T drops to the termination temperature T. end The iteration stops when the Markov chain has converged. Let α be the posterior probability density h(x|y).

[0093] Calculate uncertainty

[0094]

[0095] Statistical historical average uncertainty Record the real-time temperature as C, in degrees Celsius, and the online calibration device's usage year as R (R≥0). Calculate the calibration parameter W:

[0096]

[0097] The verification threshold W0 is set by expert experience (preferably, W0 = 0.001%). When W ≥ W0, the calibrator outputs that the online monitoring device has an abnormality, and when W ≤ W0, the calibrator outputs that the online monitoring device does not have an abnormality.

[0098] The above technical features constitute the embodiments of the present invention, which have strong adaptability and implementation effect. Unnecessary technical features can be added or removed according to actual needs to meet the needs of different situations.

Claims

1. A calibrator for an online monitoring device of a capacitive voltage transformer, characterized in that, include: The module consists of a difference module, an algorithm module, and a judgment module. The differential module is used to realize digital differential voltage, generate multiple sets of micro differential voltage signals and input them to the online monitoring device of capacitive voltage transformer. The micro differential voltage signals constitute the vector of parameters to be estimated. The algorithm module is used to obtain the voltage amplitude of the input channel of the calibrator and the primary voltage amplitude measured by the capacitive voltage transformer, calculate the theoretical ratio error of the online monitoring device, and construct a hybrid optimization algorithm by combining the Markov chain Monte Carlo method and the simulated annealing algorithm. Based on the prior probability density of the parameter vector to be estimated and the likelihood function of the measurement model, the posterior probability density is calculated to obtain the uncertainty of the error measurement result. The judgment module is used to calculate the verification parameters by combining the historical average uncertainty, real-time temperature and the years of use of the online monitoring device, and compares the verification parameters with the preset verification threshold to output the abnormal judgment result of the online monitoring device of the capacitive voltage transformer.

2. The online monitoring device calibrator for capacitive voltage transformers according to claim 1, characterized in that, It also includes a front panel assembly, which includes a display unit, an indicator unit, an operation unit, and a power control unit; The display unit is used to show verification data and the operation interface; The indicator unit is used to provide feedback on the power supply status, operating status, time synchronization status, alarm status, and power amplifier operating status of the calibrator; The operation unit is used to input operation commands and parameters; The power control unit is used to enable the calibrator to go into standby and wake up.

3. A verification method for an online monitoring device for a capacitive voltage transformer, characterized in that, Includes the following steps: Step 1: Generate multiple sets of differential voltage signals and input them into the online monitoring device for the capacitive voltage transformer, and construct the vector of parameters to be estimated from the multiple sets of differential voltage signals. Step 2: Obtain the voltage amplitude input to the online monitoring device and the primary voltage amplitude measured by the capacitive voltage transformer, and calculate the theoretical ratio error of the online monitoring device; Step 3: Combine the Markov chain Monte Carlo method and the simulated annealing algorithm to construct a hybrid optimization algorithm. Based on the prior probability density of the parameter vector to be estimated and the likelihood function of the measurement model, the hybrid optimization algorithm is used to calculate the posterior probability density and obtain the uncertainty of the error measurement result. Step 4: Calculate the verification parameters by combining the historical average uncertainty, real-time temperature, and the years of use of the online monitoring device. Compare the verification parameters with the preset verification threshold and output the judgment result of whether the online monitoring device has any abnormalities.

4. The verification method for the online monitoring device of the capacitive voltage transformer according to claim 3, characterized in that, In step 1, the number of multiple sets of differential voltage signals is 3-10 sets, and the amplitude of each set of differential voltage signals is 0.01%-0.5% of the rated voltage of the capacitive voltage transformer.

5. The verification method for the online monitoring device of the capacitive voltage transformer according to claim 3 or 4, characterized in that, In step 3, the prior probability density follows a normal distribution with a mean of 0 and a standard deviation of 0.001%–0.01%.

6. The verification method for the online monitoring device of the capacitive voltage transformer according to claim 3 or 4, characterized in that, In step 3, the likelihood function of the measurement model is constructed based on the historical measurement error data of the online monitoring device and satisfies the characteristics of Gaussian white noise distribution.

7. The verification method for the online monitoring device of the capacitive voltage transformer according to claim 3 or 4, characterized in that, The execution process of the hybrid optimization algorithm in step 3 is as follows: first, the initial value of the global optimal solution is obtained through the simulated annealing algorithm, and then the initial value is iteratively optimized through the Markov chain Monte Carlo method, with the number of iterations being 1000-5000.

8. The verification method for the online monitoring device of the capacitive voltage transformer according to claim 7, characterized in that, The simulated annealing algorithm has an initial temperature of 100-500K, a cooling coefficient of 0.85-0.95, and a termination temperature of 1-5K.

9. The verification method for the online monitoring device of a capacitive voltage transformer according to claim 3, 4, or 8, characterized in that, It also includes step 5: If it is determined that there is an abnormality in the online monitoring device, a report is generated that includes the type of abnormality, the error value, and a recommended maintenance plan.

10. The verification method for the online monitoring device of a capacitive voltage transformer according to claim 3, 4, or 8, characterized in that, In step 1, the differential voltage signal is generated by digital synthesis, and the signal frequency is consistent with the operating frequency of the capacitive voltage transformer.

Images