A method and device for automatic calibration of electric energy error

Abstract

The present invention discloses an automatic calibration method and device for electric energy errors, and relates to the technical field of electric energy metering. The method comprises: summarizing the error sources of electric energy meters; traversing the error sources of electric energy meters, retrieving and calling calibration records with big data, and mining the underlying calibration logic; performing positive and negative polarization segmentation and mapping on the underlying calibration logic, and supervising the training of an adaptive calibration module, wherein the adaptive calibration module comprises a positive error calibration branch and a negative error calibration branch of a reverse twin architecture; receiving the electric data stream of the electric energy meter operation, introducing an analog calculator for electric energy correction, combining the adaptive calibration module for branch activation and calibration decision, and generating a cascade calibration strategy; based on the cascade calibration strategy, calibrating and regulating the electric energy error. The method solves the technical problem in the prior art that the electric energy meter is difficult to adapt to the diversified error sources, resulting in inaccurate error calibration and low efficiency. By training the adaptive calibration module, the technical effect of accurate calibration of the electric energy meter error is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric energy measurement, and in particular to an automatic calibration method and device for electric energy errors. Background Art

[0002] As the cornerstone of modern society, the measurement accuracy of electric energy is directly related to many key areas such as energy distribution, economic transactions, and energy conservation and emission reduction. As the core equipment for electric energy measurement, the measurement accuracy of electric energy meters directly affects the fairness of electricity transactions and the rights and interests of users. However, in actual applications, electric energy meters will produce errors due to various factors. These error sources include main line errors caused by manufacturing processes, material selection, etc., as well as additional errors such as environmental factors and load changes. Traditional electric energy meter error calibration methods often rely on manual intervention, with long calibration cycles, low efficiency, and difficulty in fully covering various error types. Especially in the face of complex and changeable electric energy usage scenarios, the calibration effect is often unsatisfactory. Summary of the invention

[0003] The present application provides a method and device for automatic calibration of electric energy errors, which solves the technical problem in the prior art that electric energy meters are difficult to adapt to diverse error sources, resulting in inaccurate error calibration and low efficiency.

[0004] In view of the above problems, the present application provides a method and device for automatic calibration of electric energy errors.

[0005] In a first aspect of the present application, a method for automatic calibration of electric energy errors is provided, the method comprising:

[0006] Summarize the error sources of the energy meter, wherein the error sources of the energy meter include main line errors and additional errors; traverse the error sources of the energy meter, retrieve and call calibration records from big data, and mine the underlying calibration logic, wherein the underlying calibration logic includes linear error characteristics and measure logic; perform positive and negative polarization segmentation and mapping on the underlying calibration logic, and supervise the training of an adaptive calibration module, wherein the adaptive calibration module includes a positive error calibration branch and a negative error calibration branch of a reverse twin architecture; receive the electrical data stream of the energy meter operation, introduce an analog calculator to perform electric energy correction, combine the adaptive calibration module to perform branch activation and calibration decision, and generate a cascade calibration strategy; based on the cascade calibration strategy, perform calibration and regulation of the energy error.

[0007] A second aspect of the present application provides an automatic calibration device for electric energy errors, the device comprising:

[0008] An error source induction component, the error source induction component is used to summarize the error sources of the electric energy meter, wherein the error sources of the electric energy meter include main line errors and additional errors; a mining component, the mining component is used to traverse the error sources of the electric energy meter, retrieve and call calibration records from big data, and mine the underlying calibration logic, wherein the underlying calibration logic includes linear error characteristics and measure logic; a training component, the training component is used to perform positive and negative polarization segmentation and mapping on the underlying calibration logic, and supervise the training of an adaptive calibration module, wherein the adaptive calibration module includes a positive error calibration branch and a negative error calibration branch of an inverse twin architecture; a strategy generation component, the strategy generation component is used to receive the electrical data stream of the electric energy meter operation, introduce an analog calculator to perform electric energy correction, combine the adaptive calibration module to perform branch activation and calibration decision, and generate a cascade calibration strategy; a calibration component, the calibration component is used to perform calibration and regulation of electric energy errors based on the cascade calibration strategy.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] First, summarize the error sources of the energy meter, where the error sources of the energy meter include main line errors and additional errors. Next, traverse the error sources of the energy meter, retrieve and call calibration records from big data, and mine the underlying calibration logic, where the underlying calibration logic includes linear error characteristics and measure logic. Then, perform positive and negative polarization segmentation and mapping on the underlying calibration logic, and supervise the training of the adaptive calibration module, where the adaptive calibration module includes the positive error calibration branch and the negative error calibration branch of the reverse twin architecture. Next, receive the electrical data stream of the energy meter operation, introduce an analog calculator for energy correction, combine the adaptive calibration module for branch activation and calibration decision, and generate a cascade calibration strategy. Finally, based on the cascade calibration strategy, perform calibration and regulation of the energy error. The technical problem that the energy meter in the prior art is difficult to adapt to the diverse error sources, resulting in inaccurate error calibration and low efficiency is solved. By training the adaptive calibration module, the technical effect of accurate calibration of the energy meter error is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A schematic flow chart of an automatic calibration method for electric energy errors provided in an embodiment of the present application.

[0013] Figure 2A schematic diagram of the structure of an automatic calibration device for electric energy errors provided in an embodiment of the present application.

[0014] Explanation of the reference numerals: error source summarization component 11 , mining component 12 , training component 13 , strategy generation component 14 , calibration component 15 . DETAILED DESCRIPTION

[0015] The present application provides a method and device for automatic calibration of electric energy errors, thereby solving the technical problem in the prior art that electric energy meters are difficult to adapt to diverse error sources, resulting in inaccurate error calibration and low efficiency.

[0016] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0017] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or are inherent to these processes, methods, products or devices.

[0018] Embodiment 1, as Figure 1 As shown, the present application provides a method for automatic calibration of electric energy errors, wherein the method comprises:

[0019] The error sources of the electric energy meter are summarized, wherein the error sources of the electric energy meter include main line error and additional error.

[0020] The electric energy metering device consists of three parts: electric energy meter, mutual inductor and secondary wiring. Its error is mainly caused by these three parts, namely, the error of electric energy meter, the synthetic error of mutual inductor and the error caused by the voltage drop of secondary conductor of voltage transformer.

[0021] The error sources of electric energy meters can be summarized as main line errors and additional errors; the main line errors are mainly caused by the manufacturing process, design defects, material selection, and aging of components of the electric energy meter itself. They are inherent in the production and use of the electric energy meter and usually manifest as fixed or predictable deviations; additional errors refer to errors caused by external factors (such as changes in ambient temperature, humidity, electromagnetic interference, changes in load characteristics, and instability of grid voltage and current, etc.), load changes, and other external conditions; additional errors are usually dynamic and will change with changes in external conditions. In short, the error sources of electric energy meters mainly include main line errors caused by internal factors and additional errors caused by external factors.

[0022] Furthermore, the error sources of the electric energy meter include main line error and additional error, including:

[0023] A layer of error sources is generated with the electric energy meter error, the mutual inductor synthesis error, and the conductor voltage drop error; the layer of error sources is traversed to perform multi-level error source subdivision and tracing to generate X layers of error sources; an inter-layer mapping from the layer of error sources to the X layers of error sources is established as the main line error; the additional error is an external influence error.

[0024] The errors of the electric energy metering device are mainly the electric energy meter error, the mutual inductor synthesis error, and the wire voltage drop error. These error sources constitute a layer of error sources. Among them, the mutual inductor synthesis error refers to the error generated by the mutual inductor during the conversion process of current and voltage, and the wire voltage drop error refers to the voltage drop generated when the current passes through the wire, resulting in inaccurate electric energy measurement.

[0025] On the basis of one layer of error sources, further multi-level segmentation and tracing are performed to identify more subtle and specific error sources, which may involve the internal circuit of the electric energy meter, signal processing algorithm, magnetic field distribution inside the mutual inductor, and other aspects, thus forming X layers of error sources, where X represents the number of levels of error source segmentation. By establishing an inter-layer mapping from one layer of error sources to X layers of error sources, it is possible to obtain how the errors propagate and accumulate in each layer of the electric energy meter, thereby providing guidance for subsequent error calibration.

[0026] Additional errors are mainly caused by external factors, which may include changes in ambient temperature, fluctuations in humidity, interference from electromagnetic fields, changes in load characteristics, and instability in grid voltage and current, etc. These external factors will have a dynamic impact on the metering accuracy of the energy meter, causing a deviation between the meter reading and the actual value.

[0027] The error sources of the electric energy meter are traversed, calibration records are retrieved and called by big data, and the underlying calibration logic is mined, wherein the underlying calibration logic includes linear error characteristics and measure logic.

[0028] By traversing the error sources of the energy meter, including main line errors (such as energy meter errors, mutual inductor synthesis errors, wire voltage drop errors, etc.) and additional errors (such as external influences such as ambient temperature changes and humidity fluctuations), the calibration records related to these error sources can be retrieved and called using big data retrieval technology. These records may come from past calibration experiments, field test data, user feedback, etc. Through the screening and analysis of big data, the calibration records most relevant to the current error source can be quickly located, thereby obtaining useful calibration information and experience.

[0029] After obtaining the calibration records, it is necessary to dig deep into the underlying calibration logic, which includes linear error characteristics and measure logic; linear error refers to the error with a linear relationship between the error and the input quantity (such as current, voltage), which can be fitted to the error curve through mathematical methods such as linear regression and least squares method to find the calibration method; measure logic refers to the calibration measures and steps taken for specific error sources, which may include adjusting the internal parameters of the energy meter, replacing the mutual inductor, improving the wire connection, etc. By traversing the error sources of the energy meter, using big data to retrieve the calibration records, and digging into the underlying calibration logic, comprehensive and systematic support can be provided for the calibration of the energy meter.

[0030] Further, the underlying calibration logic includes linear error characteristics, including:

[0031] Identify a first error source based on the error source of the electric energy meter; if the first error source is a linear error source, take the first error source as an independent variable and the metering error as a dependent variable, and mine a first linear function based on a first calibration record; add the first linear function into the linear error characteristic.

[0032] From all the error sources of the electric energy meter, an error source to be processed is identified as the first error source, and it is determined whether the first error source is a linear error source. The characteristic of the linear error source is that the metering error caused by it is linearly related to the input quantity (such as current, voltage); if the first error source is a linear error source, the first error source is used as the independent variable, the metering error is used as the dependent variable, and the first linear function is mined in combination with the first calibration record. The first linear function can be fitted through linear regression and least squares method. The first linear function describes the linear relationship between the first error source and the metering error; the mined first linear function is added to the linear error characteristic. In the subsequent calibration process, the first linear function can be used to adjust the reading of the electric energy meter to eliminate or reduce the metering error caused by the first error source.

[0033] Furthermore, the method further comprises:

[0034] If the first error source is nonlinear, a discrete analysis is performed on the first calibration record to determine a discrete distribution graph; the discrete distribution graph is traversed to generate an approximate proxy function as the first linear function.

[0035] If the first error source is a nonlinear error source, a discrete analysis is performed on the first calibration record, that is, the data points in the first calibration record can be grouped according to a certain rule (such as the size of the input quantity), and statistics such as the error mean or median in each group are calculated; through these statistics, a discrete distribution graph can be generated, which describes the error distribution of the first error source under different input quantities; the discrete distribution graph is traversed to find an approximate proxy function that can approximately describe the error distribution. The approximate proxy function can be a polynomial function, exponential function, logarithmic function or other type of nonlinear function; by continuously adjusting the parameters of the proxy function, it can be made to fit the error data points in the discrete distribution graph as accurately as possible. Ultimately, the approximate proxy function will be used as the first linear function for processing nonlinear error sources.

[0036] Furthermore, after adding the first linear function to the linear error characteristic, the method comprises:

[0037] Mining N linear functions mapped to the error sources of the electric energy meter; performing coupling analysis on the error sources of the electric energy meter to determine source coupling relationships, wherein the source coupling relationships are collaboratively determined based on at least two error sources of the electric energy meter; based on the source coupling relationships, cross-linking the N linear functions to determine multivariate cross-linking relationships, and adding the linear error characteristics.

[0038] By identifying and determining all error sources that have a significant impact on the measurement accuracy of the electric energy meter, and finding a corresponding linear function (or approximate linear function) for each error source to describe the error caused by it.

[0039] There is often a coupling effect between multiple error sources of the electric energy meter. This coupling relationship may lead to the superposition or mutual cancellation of errors, thus affecting the final measurement accuracy. The error sources of the electric energy meter are coupled to determine the source coupling relationship between them. Optionally, a corresponding mathematical model or physical model is established for each error source to describe the error caused by it. After experiments, the measurement error data of each error source acting alone and multiple error sources acting together are collected, including the measured values ​​of key variables such as current, voltage, and temperature and the corresponding measurement error values. The collected data are analyzed using statistical methods to determine the coupling relationship between the error sources. N linear functions are associated according to the source coupling relationship, that is, according to the interaction between different error sources, the parameters or form of each linear function are adjusted to ensure that the final calibration model can accurately describe the error characteristics of the electric energy meter. Specifically, new interaction terms can be introduced into the original linear function to describe the interaction between different error sources. These interaction terms may be the product, ratio, or other form of combination of two or more error sources; the parameters of each linear function are adjusted according to the strength and direction of the coupling relationship. Through the cross-linking process, a multivariate cross-linking relationship is obtained, and the multivariate cross-linking relationship is added into the linear error characteristic.

[0040] The underlying calibration logic is segmented and mapped into positive and negative polarizations, and an adaptive calibration module is supervised and trained, wherein the adaptive calibration module includes a positive error calibration branch and a negative error calibration branch of an inverse twin architecture.

[0041] The error data in the underlying calibration logic is segmented into positive and negative polarizations, that is, the error data is divided into positive error (i.e., the measured value is higher than the true value) and negative error (i.e., the measured value is lower than the true value). A mapping relationship is established between the segmented positive and negative polarization error data and the calibration parameters so that these parameters can be adjusted in the subsequent training process to reduce the error.

[0042] The adaptive calibration module contains the positive error calibration branch and the negative error calibration branch of the inverse twin architecture. This design allows the module to handle positive and negative errors at the same time and calibrate them separately. The inverse twin architecture means that the positive error calibration branch and the negative error calibration branch have similar structures but different parameters. They are trained and optimized for positive errors and negative errors respectively. During the supervised training process, the adaptive calibration module is trained using the known error data between the true value and the measured value. The training goal is to adjust the parameters of the positive error calibration branch and the negative error calibration branch so that the output of the module (i.e., the calibrated measured value) is as close to the true value as possible.

[0043] Furthermore, supervised training of the adaptive calibration module includes:

[0044] Based on the electric energy error state, a mapping between the linear error characteristic and the measure logic is established to determine an error mapping group, wherein the error mapping group corresponds one-to-one to the electric energy error state; the error mapping group is traversed, each mapping group is divided into positive and negative polarizations, and a polarization mapping group is determined, wherein the polarization mapping is a mapping between a positive error direction and a negative error direction; based on the polarization mapping group, the adaptive calibration module is supervised for training.

[0045] Specifically, the error states in the electric energy meter are identified and analyzed, which may include different types of errors (such as systematic errors, random errors, drift errors, etc.); a mapping between linear error characteristics and measure logic is established, that is, the error characteristic parameters are associated with the correction parameters in the measure logic; according to the above mapping, a corresponding error mapping group is determined for each electric energy error state. These error mapping groups contain linear models that describe the error characteristics under this state and their corresponding correction measures. Each error mapping group is unique and corresponds to a specific electric energy error state one by one; all error mapping groups are traversed, and each error mapping group is divided into positive and negative polarizations, that is, the error data is divided into positive errors (measured values ​​are higher than the true values) and negative errors (measured values ​​are lower than the true values). The value is lower than the true value); after positive and negative polarization segmentation, polarization mapping groups are obtained, which contain the mapping relationship between the positive error direction (i.e., the set of all positive errors) and the negative error direction (i.e., the set of all negative errors); an adaptive calibration module is designed, which includes a positive error calibration branch and a negative error calibration branch, which are used to handle the calibration tasks of positive errors and negative errors, respectively; the data in the polarization mapping group is used to supervise the training of the adaptive calibration module, that is, the positive error direction and the negative error direction are input into the positive error calibration branch and the negative error calibration branch, respectively, and the parameters of the module are adjusted to minimize the output error; the training process may require multiple iterations until the calibration performance of the module reaches a satisfactory level.

[0046] Based on the power error state, a mapping between linear error characteristics and measure logic is established, and the error mapping group is determined accordingly. Then, these mapping groups are traversed and positive and negative polarization are split to determine the polarization mapping group. Finally, the adaptive calibration module is supervised and trained based on the polarization mapping group to improve the calibration accuracy and adaptability of the power meter.

[0047] Furthermore, based on the polarization mapping group, supervised training of the adaptive calibration module includes:

[0048] Based on the polarization mapping group, a first error direction is determined, and a first error calibration branch is supervised and trained, wherein the first error direction is a positive error direction or a negative error direction; reverse twinning is performed on the first error calibration branch to generate a second error calibration branch; the first error calibration branch and the second error calibration branch are performed in parallel to identify positive and negative directions, and the adaptive calibration module is determined.

[0049] An error direction is selected from the polarization mapping group as a first error direction, which can be a positive error direction or a negative error direction; the data in the selected first error direction is used to supervise the training of a first error calibration branch, which is a neural network model, and its goal is to learn how to map the input positive error or negative error to a more accurate calibration value; a reverse twinning operation is performed on the first error calibration branch to generate a second error calibration branch, and the reverse twinning means that the second branch has a similar structure to the first branch (such as the number of layers, the number of neurons, etc.), but the parameters are independent and initialized to different values, that is, to create a calibration branch that can handle errors of the opposite type to the first branch (that is, if the first branch handles positive errors, the second branch handles negative errors, and vice versa); during training During the training process, the first error calibration branch and the second error calibration branch are processed in parallel, which means that they receive input data (i.e., positive error or negative error) at the same time and output the calibrated results respectively; in order to ensure that the adaptive calibration module can correctly distinguish between positive errors and negative errors and make targeted adjustments during the training process, it is necessary to add positive and negative identifications to the input data, which can be achieved by adding an additional label or feature to the data set, which indicates whether the input data is a positive error or a negative error; after completing the training of the first error calibration branch and the second error calibration branch, they are combined into a complete adaptive calibration module, which can process positive errors and negative errors at the same time, and select the corresponding calibration branch for output according to the positive and negative identification of the input data.

[0050] The electric data stream of the electric energy meter is received, an analog calculator is introduced to perform electric energy deviation correction, and the adaptive calibration module is combined to perform branch activation and calibration decision to generate a cascade calibration strategy.

[0051] During the operation of the electric energy meter, the electric data stream generated by it is received, and an analog calculator is introduced to correct the electric energy. The analog calculator is equivalent to a calculation unit. It comprehensively analyzes and judges the error of the distributed and irregular electric data stream transmitted back, determines the error direction and vector characteristics, and corrects the electric energy value in the electric energy meter according to the output results of the analog calculator to ensure that the reading of the electric energy meter is more accurate; at the same time, the adaptive calibration module is combined to perform branch activation and calibration decisions to generate a cascade calibration strategy. Cascade calibration is a multi-level calibration method, in which each level is further calibrated based on the output results of the previous level. In the calibration of the electric energy meter, cascade calibration can ensure that errors are corrected at multiple levels, thereby improving the accuracy and stability of the calibration. Combining the electric energy correction results of the analog calculator and the calibration decisions of the adaptive calibration module, a cascade calibration strategy can be generated. The cascade calibration strategy may include multiple calibration steps, each of which uses a different calibration method or model to reduce the error. By gradually applying these calibration steps, the reading of the electric energy meter can be ensured to be more accurate and reliable.

[0052] Furthermore, the generating of the cascade calibration strategy includes:

[0053] Calibration decisions are made to determine direct calibration parameters and indirect constraint relationships, wherein the indirect constraint relationship is a relationship between unit adjustment and error response; the direct calibration parameters are used as a strategy inner loop, and the indirect constraint relationship is used as a strategy outer loop to generate the cascade calibration strategy, wherein the strategy outer loop constrains the strategy inner loop.

[0054] Direct calibration parameters are those that can be directly adjusted to affect the meter reading, such as voltage gain, current offset, etc. These parameters are usually determined by experiments or simulations to ensure that they can accurately correct the errors in the meter. Indirect constraints are the relationships between parameters or conditions that are not directly adjusted but affect the calibration effect. In meter calibration, a typical indirect constraint is the relationship between unit adjustment and error response. Unit adjustment refers to a small adjustment to a calibration parameter, while error response refers to the effect of this adjustment on the meter reading error.

[0055] In the cascade calibration strategy, direct calibration parameters are usually used as the inner loop of the strategy because they can be directly controlled and adjusted; indirect constraints are used as the outer loop of the strategy because they do not directly participate in the adjustment of the calibration parameters, but will constrain and affect the calibration process. The outer loop of the strategy limits the adjustment range or method of the direct calibration parameters in the inner loop of the strategy by defining a series of constraints. These constraints may include the threshold of the error response, the limit of the adjustment step, etc.; in this way, the outer loop of the strategy can ensure that the calibration process will neither be over-adjusted to cause instability nor fail to achieve the expected calibration effect due to insufficient adjustment. Based on the above decisions and configurations, a complete cascade calibration strategy can be generated. This strategy will include a series of ordered calibration steps, each of which involves the adjustment of direct calibration parameters and the consideration of indirect constraints. By gradually applying these steps, it can be ensured that the reading error of the electric energy meter is gradually reduced, and finally the expected calibration accuracy is achieved.

[0056] Based on the cascade calibration strategy, calibration and regulation of electric energy errors are performed.

[0057] The calibration and regulation of electric energy errors based on the cascade calibration strategy aims to gradually reduce the reading error of the electric energy meter through multi-level control and adjustment to achieve the expected calibration accuracy. Specifically, according to the model, specification and error characteristics of the electric energy meter, the initial values ​​of the direct calibration parameters (such as voltage gain, current offset, etc.) are initialized; the initial conditions and ranges of the indirect constraint relationships (such as the relationship between unit adjustment and error response) are determined; according to the initial calibration parameters and the current reading error of the electric energy meter, the inner loop begins to adjust, and the adjustment process may involve fine-tuning of parameters such as voltage and current to gradually reduce the error; while the inner loop is adjusting, the outer loop continuously evaluates the impact of the adjustment result on the error response. If the adjustment result exceeds the preset error response threshold or violates other constraints, the outer loop will issue a warning and limit further adjustments of the inner loop. By continuously optimizing and adjusting the calibration parameters and constraints, the reading error of the electric energy meter can be gradually reduced and its measurement accuracy and stability can be improved.

[0058] In summary, the embodiments of the present application have at least the following technical effects:

[0059] First, summarize the error sources of the energy meter, where the error sources of the energy meter include main line errors and additional errors. Next, traverse the error sources of the energy meter, retrieve and call calibration records from big data, and mine the underlying calibration logic, where the underlying calibration logic includes linear error characteristics and measure logic. Then, perform positive and negative polarization segmentation and mapping on the underlying calibration logic, and supervise the training of the adaptive calibration module, where the adaptive calibration module includes the positive error calibration branch and the negative error calibration branch of the reverse twin architecture. Next, receive the electrical data stream of the energy meter operation, introduce an analog calculator for energy correction, combine the adaptive calibration module for branch activation and calibration decision, and generate a cascade calibration strategy. Finally, based on the cascade calibration strategy, perform calibration and regulation of the energy error. The technical problem that the energy meter in the prior art is difficult to adapt to the diverse error sources, resulting in inaccurate error calibration and low efficiency is solved. By training the adaptive calibration module, the technical effect of accurate calibration of the energy meter error is achieved.

[0060] Embodiment 2, based on the same inventive concept as the method for automatic calibration of electric energy error in the above embodiment, Figure 2 As shown, the present application provides an automatic calibration device for electric energy errors, wherein the device comprises:

[0061] An error source induction component 11, the error source induction component 11 is used to summarize the error sources of the electric energy meter, wherein the error sources of the electric energy meter include main line errors and additional errors; a mining component 12, the mining component 12 is used to traverse the error sources of the electric energy meter, retrieve and call calibration records from big data, and mine the underlying calibration logic, wherein the underlying calibration logic includes linear error characteristics and measure logic; a training component 13, the training component 13 is used to perform positive and negative polarization segmentation and mapping on the underlying calibration logic, and supervise the training of an adaptive calibration module, wherein the adaptive calibration module includes a positive error calibration branch and a negative error calibration branch of an inverse twin architecture; a strategy generation component 14, the strategy generation component 14 is used to receive the electrical data stream of the electric energy meter operation, introduce an analog calculator to perform electric energy correction, combine the adaptive calibration module to perform branch activation and calibration decision, and generate a cascade calibration strategy; a calibration component 15, the calibration component 15 is used to perform calibration and regulation of the electric energy error based on the cascade calibration strategy.

[0062] Furthermore, the error source induction component 11 is used to perform the following method:

[0063] A layer of error sources is generated with the electric energy meter error, the mutual inductor synthesis error, and the conductor voltage drop error; the layer of error sources is traversed to perform multi-level error source subdivision and tracing to generate X layers of error sources; an inter-layer mapping from the layer of error sources to the X layers of error sources is established as the main line error; the additional error is an external influence error.

[0064] Furthermore, the mining component 12 is used to perform the following method:

[0065] Identify a first error source based on the error source of the electric energy meter; if the first error source is a linear error source, take the first error source as an independent variable and the metering error as a dependent variable, and mine a first linear function based on a first calibration record; add the first linear function into the linear error characteristic.

[0066] Furthermore, the mining component 12 is used to perform the following method:

[0067] If the first error source is nonlinear, a discrete analysis is performed on the first calibration record to determine a discrete distribution graph; the discrete distribution graph is traversed to generate an approximate proxy function as the first linear function.

[0068] Furthermore, the mining component 12 is used to perform the following method:

[0069] Mining N linear functions mapped to the error sources of the electric energy meter; performing coupling analysis on the error sources of the electric energy meter to determine source coupling relationships, wherein the source coupling relationships are collaboratively determined based on at least two error sources of the electric energy meter; based on the source coupling relationships, cross-linking the N linear functions to determine multivariate cross-linking relationships, and adding the linear error characteristics.

[0070] Furthermore, the training component 13 is used to perform the following method:

[0071] Based on the electric energy error state, a mapping between the linear error characteristic and the measure logic is established to determine an error mapping group, wherein the error mapping group corresponds one-to-one to the electric energy error state; the error mapping group is traversed, each mapping group is divided into positive and negative polarizations, and a polarization mapping group is determined, wherein the polarization mapping is a mapping between a positive error direction and a negative error direction; based on the polarization mapping group, the adaptive calibration module is supervised for training.

[0072] Furthermore, the training component 13 is used to perform the following method:

[0073] Based on the polarization mapping group, a first error direction is determined, and a first error calibration branch is supervised and trained, wherein the first error direction is a positive error direction or a negative error direction; reverse twinning is performed on the first error calibration branch to generate a second error calibration branch; the first error calibration branch and the second error calibration branch are performed in parallel to identify positive and negative directions, and the adaptive calibration module is determined.

[0074] Furthermore, the strategy generation component 14 is used to execute the following method:

[0075] Calibration decisions are made to determine direct calibration parameters and indirect constraint relationships, wherein the indirect constraint relationship is a relationship between unit adjustment and error response; the direct calibration parameters are used as a strategy inner loop, and the indirect constraint relationship is used as a strategy outer loop to generate the cascade calibration strategy, wherein the strategy outer loop constrains the strategy inner loop.

[0076] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0077] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0078] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A method for automatic calibration of electric energy error, characterized in that: The method comprises: Summarize the error sources of the electric energy meter, wherein the error sources of the electric energy meter include main line error and additional error; Traversing the error sources of the electric energy meter, retrieving and calling calibration records through big data, and mining the underlying calibration logic, wherein the underlying calibration logic includes linear error characteristics and measure logic; Performing positive and negative polarization segmentation and mapping on the underlying calibration logic, and supervising the training of an adaptive calibration module, wherein the adaptive calibration module includes a positive error calibration branch and a negative error calibration branch of an inverse twin architecture; Receiving the electric data stream of the electric energy meter, introducing the analog calculator to perform electric energy deviation correction, combining the adaptive calibration module to perform branch activation and calibration decision, and generating a cascade calibration strategy; Based on the cascade calibration strategy, calibration and regulation of electric energy errors are performed; Supervised training of adaptive calibration modules, including: Based on the electric energy error state, establishing a mapping between the linear error characteristic and the measure logic, and determining an error mapping group, wherein the error mapping group corresponds to the electric energy error state one by one; Traversing the error mapping groups, performing positive and negative polarization segmentation on each mapping group, and determining a polarization mapping group, wherein the polarization mapping is a mapping of a positive error direction and a negative error direction; Based on the polarization mapping group, supervise the training of the adaptive calibration module; Based on the polarization mapping group, supervised training of the adaptive calibration module includes: Based on the polarization mapping group, determining a first error direction, and supervising training of a first error calibration branch, wherein the first error direction is a positive error direction or a negative error direction; Performing reverse twinning on the first error calibration branch to generate a second error calibration branch; The first error calibration branch and the second error calibration branch are operated in parallel, positive and negative directions are marked, and the adaptive calibration module is determined; The generating of the cascade calibration strategy comprises: Making a calibration decision to determine a direct calibration parameter and an indirect constraint relationship, wherein the indirect constraint relationship is a relationship between a unit calibration parameter and an error response; The cascade calibration strategy is generated by taking the direct calibration parameters as a strategy inner loop and taking the indirect constraint relationship as a strategy outer loop, wherein the strategy outer loop constrains the strategy inner loop.

2. The method for automatic calibration of electric energy error according to claim 1, characterized in that: The error sources of the electric energy meter include main line error and additional error, including: A layer of error sources is generated by using the electric energy meter error, mutual inductor synthesis error and conductor voltage drop error; Traversing the one-layer error source, performing multi-level error source segmentation and tracing, and generating X-layer error sources; Establishing an inter-layer mapping from the one-layer error source to the X-layer error source as the main line error; The additional error is an external influence error.

3. The method for automatic calibration of electric energy error according to claim 1, characterized in that: The underlying calibration logic contains linearity error characteristics, including: identifying a first error source based on the electric energy meter error source; If the first error source is a linear error source, taking the first error source as an independent variable and the measurement error as a dependent variable, mining a first linear function based on the first calibration record; The first linear function is added to the linear error characteristic.

4. The method for automatic calibration of electric energy error according to claim 3, characterized in that: The method further comprises: If the first error source is nonlinear, performing a discrete analysis on the first calibration record to determine a discrete distribution graph; The discrete distribution graph is traversed to generate an approximate proxy function as the first linear function.

5. The method for automatic calibration of electric energy error according to claim 3, characterized in that: After adding the first linear function to the linear error characteristic, the method comprises: Mining N linear functions mapped to the error sources of the electric energy meter; Performing coupling analysis on the error sources of the electric energy meter to determine a source coupling relationship, wherein the source coupling relationship is collaboratively determined based on at least two error sources of the electric energy meter; Based on the source coupling relationship, the N linear functions are cross-linked to determine a multivariate cross-linking relationship, and the linear error characteristic is added.

6. An automatic calibration device for electric energy error, characterized in that: The device is used to implement the method for automatic calibration of electric energy error according to any one of claims 1 to 5, comprising: An error source summarizing component, the error source summarizing component is used to summarize the error sources of the electric energy meter, wherein the error sources of the electric energy meter include main line errors and additional errors; A mining component, the mining component is used to traverse the error source of the electric energy meter, retrieve and call the calibration record of the big data, and mine the underlying calibration logic, wherein the underlying calibration logic includes linear error characteristics and measure logic; A training component, the training component is used to perform positive and negative polarization segmentation and mapping on the underlying calibration logic, and supervise the training of an adaptive calibration module, wherein the adaptive calibration module includes a positive error calibration branch and a negative error calibration branch of an inverse twin architecture; A strategy generation component, the strategy generation component is used to receive the electric data stream of the electric energy meter, introduce an analog calculator to perform electric energy deviation correction, combine the adaptive calibration module to perform branch activation and calibration decision, and generate a cascade calibration strategy; A calibration component is used to calibrate and regulate the electric energy error based on the cascade calibration strategy.

Images