A power stealing prevention twin system for metering cabinet state mapping

By collecting and processing electrical signals in the metering cabinet state mapping system, and utilizing consistency arbitration and feedback compensation mechanisms, the divergence problem in the state mapping between virtual space and physical entity is solved, achieving stable synchronization and autonomous repair, and improving the effectiveness of anti-theft control.

CN123067232APending Publication Date: 2026-11-10ZHENGZHOU KAIBEITE TRANSFORMER CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202611020869.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-09
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

In existing technologies, the state mapping model between virtual space and physical entities in power distribution network may diverge or deadlock due to the input or tampering of non-real data from physical data sources. Traditional control methods cannot effectively isolate transient anomalies, leading to synchronization control failure.

Method used

The metering cabinet signal is acquired by an electrical data acquisition unit, and an initial feature vector is generated by low-pass filtering. Combined with topology state mapping, data processing and consistency arbitration units, the projection residual value is calculated. The discrete event-driven state machine and feedback compensation mechanism are used to block the data flow and autonomously generate a virtual feature repair vector to ensure the stability of the virtual space.

Benefits of technology

Stable synchronization of metering cabinet state mapping under unsteady disturbances is achieved, preventing virtual model divergence, enhancing the system's self-strain and fault tolerance capabilities, and improving the accuracy and stability of anti-theft control.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123067232A_ABST
    Figure CN123067232A_ABST
Patent Text Reader

Abstract

This disclosure presents an anti-theft electricity twin system for metering cabinet state mapping. The system includes an electrical data acquisition unit, a topology state mapping unit, a data processing unit, and a data consistency arbitration unit. It generates a projection estimation vector by multiplying the state feature autocorrelation matrix (calculated from historical virtual feature vectors) with the initial physical state feature vector. The transient projection residual value is calculated based on the absolute value of the algebraic difference between the two. Continuous exceeding of limits by the transient projection residual value triggers discrete events that drive state machine flips, thereby blocking contaminated data flow channels. This implementation establishes a truth-feedforward arbitration isolation mechanism within the virtual information space, effectively eliminating model back-pollution and topology deadlock caused by the input of non-real business data, and maintaining steady-state convergence of the virtual-real space state mapping.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] Embodiments of this disclosure relate to the fields of digital twin and industrial process control technology, and specifically to an anti-theft electricity twin system for metering cabinet status mapping. Background Technology

[0002] Currently, in the operation and management of power distribution networks, a high-dimensional virtual space topology model is constructed using industrial process control and digital twin systems to dynamically track the physical state of field metering cabinets, forming the basis for maintaining the synchronization of power supply and distribution network control. This virtual-physical space state mapping strategy often relies on discrete-time step electrical acquisition signals obtained from the physical acquisition end, using them as initial feature vectors input to the state transition matrix to calculate and generate a power topology map model in the high-dimensional virtual space. Traditional closed-loop feedback regulation mechanisms have long assumed that physical sampling data is an absolutely reliable truth benchmark, relying on periodically calculating the difference residual between the virtual projection vector and the actual physical acquisition vector to adjust the matrix parameter weight distribution. Maintaining steady-state synchronization between the virtual space and physical entity layer of a digital twin control system is challenging. However, under real power distribution network operation conditions, physical data sources are often in a state of data insecurity due to the entry of non-real business data or illegal tampering with power parameters. The electrical waveforms generated by such malicious interference usually maintain a fake mathematical self-consistent equilibrium in local time series. This causes traditional control loops to lack an independent data truth audit path, driving the topology mapping engine to approximate the physical sampling input that has undergone nonlinear mutations. This results in impedance mismatch between virtual and real control flows, and the distorted contaminated parameters are back-transmitted into the high-dimensional virtual space, causing local divergence in the overall state mapping model, deadlock, or model oscillation.

[0003] To address the model variations caused by data misrepresentation, conventional linear improvement approaches typically involve adding contact-type tamper-proof accessories to the casing or deploying time-window smoothing filtering algorithms in the receiver's data processor. However, this introduces significant phase lag and dynamic backoff in the control flow, masking high-frequency discrete anomalies in the sampled data. Besides limitations at the hardware level, such as physical tamper-proof components, there are also shortcomings at the software and data control method levels. For example, Chinese invention patent application CN121499873A discloses an anti-theft detection system for electricity metering boxes, which constructs a comprehensive reliable [system] based on communication quality factors and terminal behavior deviation. The concept drift index of the main artificial intelligence model is evaluated using a physical reference model based on Kirchhoff's laws. However, this type of control method suffers from fundamental mismatch when facing complex power distribution conditions. The quantification of the drift index is highly dependent on the overall statistics of the regional line loss balance model. The underlying mechanism requires the accumulation of long-term and multi-node topological power data, which results in computational time lag. It is impossible to accurately identify transient anomalies in a single metering cabinet. When malicious interference signals maintain a fake local self-consistent balance in a very short time series, the verification mechanism based on the overall line loss balance will fail to detect due to spatial coupling and time lag, resulting in contaminated data back-infiltrating the virtual space.

[0004] Therefore, the technical problem to be solved by this invention is how to construct a feedforward data consistency arbitration mechanism independent of the physical sampling stream during the state synchronization process, cut off the matrix parameter update chain that has been abnormally contaminated without delay, and autonomously deduce and generate the topological state self-repair vector for the period of loss of trust based on the conservation control flow equation inside the virtual space. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure propose an anti-theft electricity twin system for metering cabinet status mapping to address the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide an anti-theft twin system for metering cabinet state mapping. The system includes: an electrical data acquisition unit for acquiring time-series electrical signals from the outgoing side of the metering cabinet and generating an initial physical state feature vector using a low-pass filtering algorithm; a topology state mapping unit connected to the electrical data acquisition unit for storing a discrete state transition matrix and calculating the product of the discrete state transition matrix and the initial physical state feature vector to generate a virtual twin state vector; and a data processing unit for using the continuous-time-step historical virtual feature vector latched by an internal shift register to calculate and generate state feature autocorrelation moments using an orthogonalization algorithm. The system includes a data consistency arbitration unit, which is connected to both the topology state mapping unit and the data processing unit. It calculates the product of the initial physical state feature vector and the state feature autocorrelation matrix to generate a projection estimation vector, and calculates the absolute value of the algebraic difference between the initial physical state feature vector and the projection estimation vector as the transient projection residual value. The data consistency arbitration unit embeds a discrete event-driven state machine. When the transient projection residual value exceeds the safety threshold of 1.0 for three consecutive sampling periods, the discrete event-driven state machine flips its state and outputs a data blocking command to the electrical data acquisition unit to close the data channel from the electrical data acquisition unit to the topology state mapping unit.

[0008] Preferably, the system further includes a feedback compensation unit; when the data consistency arbitration unit outputs a data blocking command, it triggers the feedback compensation unit to close the transmission channel of the initial physical state feature vector to the update interface of the topology state mapping unit; the feedback compensation unit stops updating the discrete state transition matrix and locks the current parameters of the discrete state transition matrix at the fixed values ​​of the previous sampling period, using the fixed values ​​of the parameters to maintain the parameter stability of the topology state mapping unit.

[0009] Preferably, the system further includes a data extrapolation unit; the data extrapolation unit is connected to the data consistency arbitration unit and the data processing unit, and is used to extract the continuous time step historical virtual feature vector latched in the shift register inside the data processing unit in response to the data blocking instruction, and substitute the continuous time step historical virtual feature vector as the boundary constraint value into the preset power balance state equation for recursive iteration, generate a virtual feature repair vector and output it to the output end of the topology state mapping unit to replace the current virtual twin state vector.

[0010] Preferably, the system further includes a dynamic gating unit; the dynamic gating unit is connected to the input of the data consistency arbitration unit, and is used to collect the initial physical state feature vector and calculate the slope of change between adjacent sampling periods. When the slope of change exceeds the set slope threshold, a circuit breaker signal is generated internally, and the current parameters of the discrete state transition matrix are locked in response to the circuit breaker signal to suspend the iterative control of the data processing unit.

[0011] Preferably, the system further includes a sliding window filtering unit; the sliding window filtering unit is connected in series to the front input of the data consistency arbitration unit, and is used to sort the transient projection residual values ​​and extract the median within a continuous 50ms sliding time window, and generate smooth residual values ​​to output to the discrete event-driven state machine.

[0012] Preferably, the system further includes a state locking unit; the state locking unit is connected to the data processing unit and is used to fix the current parameters of the discrete state transition matrix to the values ​​of the previous sampling period when a network communication interruption signal is detected.

[0013] Preferably, the system further includes a trend analysis unit; the trend analysis unit is connected between the data processing unit and the feedback compensation unit, and is used to continuously cache the transient projection residual values ​​within ten consecutive sampling periods to construct a discrete difference matrix. When the parameters of the discrete difference matrix show a monotonically increasing trend and the cumulative difference value exceeds a preset trend threshold, an abnormal state signal is output to the feedback compensation unit, triggering the feedback compensation unit to superimpose parameter corrections onto the discrete state transition matrix through a sliding mode control algorithm.

[0014] Preferably, the system further includes a topology reconstruction unit; the topology reconstruction unit is connected to the topology state mapping unit and is used to split the topology model in the topology state mapping unit into multiple sub-region models and independently calculate the local residual value of each sub-region model. When the deviation between the local residual value of a certain sub-region model and the local residual value of the adjacent sub-region model exceeds a set ratio, it is determined that the branch node inside the metering cabinet corresponding to the sub-region model is in an abnormal state. The parameters of the branch node corresponding to the abnormal state are set to zero, and the virtual twin state vectors of the adjacent sub-region models are extracted for interpolation calculation to update the topology model.

[0015] Preferably, the sampling period of the electrical data acquisition unit is 50ms, and the initial physical state feature vector includes the node current component, voltage phase change rate, and transient active power deviation under the same sampling period.

[0016] The above-described embodiments of this disclosure have the following beneficial effects: 1. In the anti-theft twin of metering cabinet status mapping, the physical edge decoupling unit collects signals from multiple time steps to construct an initial physical sampling vector. The processor retrieves the virtual feature vector latched by the shift register and constructs an autocorrelation constraint field matrix containing historical behavior features through the orthogonalization operator. The consistency arbitration unit calculates the transient projection residual value of the initial physical sampling vector on the orthogonal complement space of the constraint field matrix. Through this interaction of time-series features, a data truth auditing mechanism is established in the virtual information space to solve the problem of the control loop approximating the erroneous truth value when the physical input data is spatiotemporally forged, and to ensure the stability of the system's synchronous state under non-steady-state disturbances.

[0017] 2. When the transient projection residual value exceeds the set safety threshold for three consecutive sampling periods, the discrete event-driven state machine configured inside the consistency arbitration unit undergoes a state transition, generating a physical data loss control instruction. The processor responds to the instruction by cutting off the control chain of the data flow to the parameter update interface of the topology state image engine, blocking the contaminated data flow, changing the traditional feedback loop hysteresis adjustment mechanism, introducing feedforward isolation action, realizing the transformation from hysteresis residual repair to dynamic self-healing control, eliminating the virtual topology model divergence and local deadlock caused by the input of non-real business data, and maintaining the continuous convergence of the high-dimensional space state mapping.

[0018] 3. While blocking the external control chain, the processor simultaneously activates the inertial extrapolation state machine inside the virtual space. It is decoupled from the physical sampling source input, uses the historical steady-state characteristics latched in the shift register as boundary conditions, injects the Kirchhoff power balance control flow equations preset in the virtual space, and generates the virtual feature self-repair vector of the current step through forward extrapolation. The self-repair vector covers the actual output of the topology state image engine in situ. It spontaneously maintains the evolution continuity of the virtual and real spaces by relying on the law of control flow conservation. This solves the problem of spatial image model divergence caused by the lack of real reference under non-ideal working conditions in traditional monitoring methods, and enhances the self-strain and fault tolerance of the overall control system. Attached Figure Description

[0019] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0020] Figure 1 This is a data flow diagram of the anti-electricity theft twin system of the present invention; Figure 2 This is a connection diagram of the anti-electricity theft twin system of the present invention; Figure 3 This is a residual curve diagram of the anti-electricity theft twin system of the present invention. Detailed Implementation

[0021] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0022] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0023] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0024] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0025] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0026] Before performing any of the operations involving the collection, storage, processing, or use of the target images disclosed herein, the relevant organizations or individuals shall fulfill their obligations, including conducting information security impact assessments, informing the information subjects, and obtaining prior authorization and consent from the information subjects.

[0027] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0028] refer to Figure 1 The flowchart illustrates some embodiments of an anti-theft electricity twin system for metering cabinet state mapping according to this disclosure. An anti-theft electricity twin system for metering cabinet state mapping includes the following steps: The electrical data acquisition unit is used to acquire the timing electrical signals on the outgoing side of the metering cabinet and to calculate and generate the initial physical state feature vector through a low-pass filtering algorithm. The topology state mapping unit, connected to the electrical data acquisition unit, is used to store the discrete state transition matrix and calculate the product of the discrete state transition matrix and the initial physical state feature vector to generate a virtual twin state vector. The data processing unit is used to calculate and generate the state feature autocorrelation matrix by utilizing the continuous time step historical virtual feature vectors latched by the internal shift register and through an orthogonalization algorithm. The data consistency arbitration unit is connected to the topology state mapping unit and the data processing unit, respectively. It is used to calculate the product of the initial physical state feature vector and the state feature autocorrelation matrix to generate the projection estimation vector, and to calculate the absolute value of the algebraic difference between the initial physical state feature vector and the projection estimation vector as the transient projection residual value. The data consistency arbitration unit has an embedded discrete event-driven state machine. When the transient projection residual value exceeds the safety threshold of 1.0 for three consecutive sampling periods, the discrete event-driven state machine will flip its state and output a data blocking command to the electrical data acquisition unit to close the data channel from the electrical data acquisition unit to the topology state mapping unit.

[0029] Preferably, the system further includes a feedback compensation unit; when the data consistency arbitration unit outputs a data blocking command, it triggers the feedback compensation unit to close the transmission channel of the initial physical state feature vector to the update interface of the topology state mapping unit; the feedback compensation unit stops updating the discrete state transition matrix and locks the current parameters of the discrete state transition matrix at the fixed values ​​of the previous sampling period, using the fixed values ​​of the parameters to maintain the parameter stability of the topology state mapping unit.

[0030] Preferably, the system further includes a data extrapolation unit; the data extrapolation unit is connected to the data consistency arbitration unit and the data processing unit, and is used to extract the continuous time step historical virtual feature vector latched in the shift register inside the data processing unit in response to the data blocking instruction, and substitute the continuous time step historical virtual feature vector as the boundary constraint value into the preset power balance state equation for recursive iteration, generate a virtual feature repair vector and output it to the output end of the topology state mapping unit to replace the current virtual twin state vector.

[0031] Preferably, the system further includes a dynamic gating unit; the dynamic gating unit is connected to the input of the data consistency arbitration unit, and is used to collect the initial physical state feature vector and calculate the slope of change between adjacent sampling periods. When the slope of change exceeds the set slope threshold, a circuit breaker signal is generated internally, and the current parameters of the discrete state transition matrix are locked in response to the circuit breaker signal to suspend the iterative control of the data processing unit.

[0032] Preferably, the system further includes a sliding window filtering unit; the sliding window filtering unit is connected in series to the front input of the data consistency arbitration unit, and is used to sort the transient projection residual values ​​and extract the median within a continuous 50ms sliding time window, and generate smooth residual values ​​to output to the discrete event-driven state machine.

[0033] Preferably, the system further includes a state locking unit; the state locking unit is connected to the data processing unit and is used to fix the current parameters of the discrete state transition matrix to the values ​​of the previous sampling period when a network communication interruption signal is detected.

[0034] Preferably, the system further includes a trend analysis unit; the trend analysis unit is connected between the data processing unit and the feedback compensation unit, and is used to continuously cache the transient projection residual values ​​within ten consecutive sampling periods to construct a discrete difference matrix. When the parameters of the discrete difference matrix show a monotonically increasing trend and the cumulative difference value exceeds a preset trend threshold, an abnormal state signal is output to the feedback compensation unit, triggering the feedback compensation unit to superimpose parameter corrections onto the discrete state transition matrix through a sliding mode control algorithm.

[0035] Preferably, the system further includes a topology reconstruction unit; the topology reconstruction unit is connected to the topology state mapping unit and is used to split the topology model in the topology state mapping unit into multiple sub-region models and independently calculate the local residual value of each sub-region model. When the deviation between the local residual value of a certain sub-region model and the local residual value of the adjacent sub-region model exceeds a set ratio, it is determined that the branch node inside the metering cabinet corresponding to the sub-region model is in an abnormal state. The parameters of the branch node corresponding to the abnormal state are set to zero, and the virtual twin state vectors of the adjacent sub-region models are extracted for interpolation calculation to update the topology model.

[0036] Preferably, the sampling period of the electrical data acquisition unit is 50ms, and the initial physical state feature vector includes the node current component, voltage phase change rate, and transient active power deviation under the same sampling period.

[0037] Example 1: In the operating environment of an intelligent power distribution system containing multiple high-voltage metering cabinets, the system uses a hierarchical network topology to dynamically track the power status of the high-voltage metering cabinets. When the hierarchical network topology faces disturbances such as bypass shunting or the entry of non-real business data, the static structure using a time window for smooth adjustment suffers from phase lag, leading to topology synchronization deviations. This causes the anti-theft control loop to fail to converge accurately. The electrical data acquisition unit collects the timing electrical signals from the outgoing side of the high-voltage metering cabinet at a sampling period of 50ms and calculates an initial physical state feature vector containing node current components, voltage phase change rate, and transient active power deviation using a low-pass filtering algorithm. The topological state mapping unit stores the discrete state transition matrix. And calculate the discrete state transition matrix. With the initial physical state feature vector The product of these is used to generate a virtual twin state vector. The conversion process satisfies Formula 1: ,in, For virtual twin state vectors, The discrete state transition matrix, The initial physical state feature vector is used; the data processing unit uses an internal shift register to latch the historical virtual feature vectors of 5 consecutive time steps, and calculates the state feature autocorrelation matrix through an orthogonalization algorithm. The initial physical state feature vector is calculated by the data consistency arbitration unit. Autocorrelation matrix of state features The product of these is used to generate the projection estimation vector, and based on the initial physical state feature vector... The absolute value of the algebraic difference between the projection estimation vector and the standard deviation of the preset time step, and the benchmark constant. The transient projection residual value was calculated. The conversion process satisfies Formula 2: ,in, The value represents the transient projection residual, with the subscript res indicating the residual. This is the initial physical state feature vector, where the subscript ini indicates the initial value. This is the state feature autocorrelation matrix, where the subscript "field" indicates the feature field. The standard deviation of the time step is the baseline constant, and the subscript std indicates the standard deviation.

[0038] When the discrete event-driven state machine set in the data consistency arbitration unit detects the transient projected residual value When the data flow exceeds the preset threshold of 1.0 for three consecutive sampling periods, the discrete event-driven state machine changes its control state and outputs a data blocking signal to the electrical data acquisition unit to close the data channel to the topology state mapping unit. At this time, the state feature autocorrelation matrix output by the data processing unit... The initial physical state feature vector output by the electrical data acquisition unit The combined effect, through the state feature autocorrelation matrix Provides a topological feature reference based on historical time steps for the current sampling time, and an initial physical state feature vector after low-pass filtering. Improving the topological contrast stability of the orthogonal projection space enables isolation actions, reduces phase lag caused by smoothing filtering in a single topology architecture, and improves the response speed for anomaly state capture; simultaneously, the system inputs the initial physical state feature vector To the state feature autocorrelation matrix The orthogonal complement space is projected and transformed to convert the data anti-tampering verification on the high-voltage metering cabinet side into the characteristic matrix autocorrelation audit and signal channel adjustment inside the data consistency arbitration unit. This limits the diffusion of abnormal disturbances to the update parameters of the topology state mapping unit and suppresses the divergence tendency of the virtual synchronization model.

[0039] When the data channel is blocked and the topology state mapping unit performs extrapolation calculations using the power balance control equation, the topology state mapping unit generates a virtual feature repair vector to cover the output of the current step. This ensures that the high-voltage metering cabinet's image model remains in a pre-defined topology boundary convergence state during periods of data anomalies. Furthermore, the hierarchical network topology utilizes existing computing units to maintain tracking of the power state, allowing the virtual and real space state mapping to recover to a stable equilibrium state of follow-up tracking after disturbance elimination. In the specific extrapolation and iterative process, the core mathematical logic of the aforementioned power balance state equation is based on the inflow and outflow of power into the distribution network nodes. Beyond the algebraic constraint of active power conservation, when the data channel is cut off, the topology state mapping unit extracts the historical steady-state feature vectors of the first 5 time steps latched in the internal shift register as boundary constraints. The recursive iteration process specifically includes the following steps: substituting the node voltage estimate and node current estimate of the previous sampling period into the power conservation equation to calculate the initial power residual value of each node; secondly, using the grid optimization operator, with the average rate of change of the historical steady-state feature vector as the initial iteration step size, a first-order linear approximation is performed on the missing voltage phase change rate and transient active power deviation in the current time period.

[0040] Check whether the calculated algebraic sum of active power is less than the preset convergence threshold of 0.001 watts. If not, continue to update the node feature parameters along the negative gradient direction until the absolute value of the power deviation in three consecutive recursive iterations is lower than the convergence threshold. The feature vector of the final converged output is used as the virtual feature repair vector and directly replaces the current output in situ. Thus, the continuously converged spatial topology state can be autonomously deduced when the physical sampling signal is removed.

[0041] Example 2: On a physical verification platform simulating the operating characteristics of a high-voltage metering cabinet, the platform limits the current measurement range to 0A to 1000A, the voltage measurement range to 0V to 10000V, the system resolution to 24 bits, the measurement accuracy to 0.05%, and the basic sampling rate to 10kHz. The acquired timing electrical signals are used to determine the core control parameter in the control loop, namely the sampling period. The specific value depends on the real-time requirements of electrical data acquisition and the upper limit of the computational load of the data processing unit. When the spectral bandwidth of the timing electrical signal is broadened due to external harmonic interference, in order to ensure that the signal acquisition satisfies the sampling theorem to avoid aliasing, and to ensure that the data consistency arbitration unit completes the state comparison within 3 sampling periods, the sampling period is determined as follows: The adjustment range is set to 10ms to 100ms, and under the current operating conditions, it is set to 50ms. Gaussian white noise with a signal-to-noise ratio of 20dB is superimposed on the timing electrical signal, and power frequency interference harmonics with a frequency of 50Hz and an amplitude of 5% of the rated voltage are introduced to construct a test dataset containing abnormal fluctuations.

[0042] The physical verification platform uses test datasets to establish a multi-dimensional comparison system, which includes a control group that uses only a time window smoothing algorithm without an autocorrelation arbitration mechanism, and an experimental group consisting of a complete technical solution. Non-real business data input interference generated by bypass shunting is set on the outgoing line of the high-voltage metering cabinet. The proportion of bypass shunting interference is set at a low intensity of 5%, a medium intensity of 15%, and a high intensity absolute upper limit of 35%. Under the unprocessed raw data input state, the control group generates a fluctuation component with an amplitude of 12.4% of the reference voltage when facing a 5% proportion of bypass shunting interference, and generates a phase lag of 125.6ms. When the bypass shunting interference increases to the absolute upper limit of 35%, the synchronization error of the virtual model parameters of the control group exceeds 150% within 3 sampling periods and stops converging.

[0043] The test group using this technical solution exhibited monotonically convergent synchronization characteristics under the same electromagnetic noise and power frequency interference harmonic environment. When faced with a 5% bypass shunt interference, the initial physical state feature vector acquired by the electrical data acquisition unit... Including a 20dB noise component, the data consistency arbitration unit will use the initial physical state feature vector The state feature autocorrelation matrix constructed from historical time step features using orthogonal operators Perform a projection transformation on the orthogonal complement space and calculate the output transient projection residual value. The background fluctuation level remained at 0.12 during normal periods, but jumped to 2.45 during shunt abrupt changes. The measured transient projection residual values ​​were obtained when the bypass shunt interference increased to 15% and 35% respectively. The values ​​monotonically increase to 4.62 and 9.18 respectively, and the discrete event-driven state machines complete the state machine transition and output data blocking signals within one sampling period. In actual operation, the aforementioned sliding window filter unit is equipped with a first-in-first-out temporary storage queue with a depth of 10 data bits. Although the sampling period of the electrical data acquisition unit is fixed at 50 milliseconds, the so-called 50-millisecond sliding time window means that the sliding window filter unit slides forward every 50 milliseconds of sampling step and triggers a median calculation. At each trigger moment, the sliding window filter unit does not only read the residual value of the current single period, but extracts the transient projection residual values ​​of 10 consecutive sampling periods pre-buried in the temporary storage queue to form a discrete value set. The system internally arranges these 10 historical residual values ​​in ascending order and extracts the arithmetic mean of the two values ​​in the middle position as the current smooth residual value output, thereby ensuring that sudden non-transient glitches are filtered out under the 50-millisecond discrete step size.

[0044] During the sampling period In an environment with a 50ms timeframe and defined parameter configuration baseline, when the bypass shunt interference is reduced to 1%, the calculated transient projection residual value is lower than the random disturbance caused by 20dB background noise because the signal abrupt change amplitude is less than that caused by 20dB background noise. The value remains at 0.85, not exceeding the preset threshold of 1.0. When the bypass shunt interference exceeds 35% to 40%, the voltage phase change rate and the amplitude of the transient active power deviation tend to level off because the magnetic circuit inside the high-voltage metering cabinet enters the nonlinear saturation region. The measured transient projection residual value... The value is 9.50. Further increasing the interference ratio results in a transient projection residual value. The monotonically increasing value range of 5% to 35% confirms that this range constitutes a working window for maintaining a monotonically responding state. At a test point with a bypass shunt interference of 15%, the original transient active power deviation acquired by the electrical data acquisition unit was 45.2W, which, after low-pass filtering, was included in the initial physical state feature vector. The topology state mapping unit will generate the initial physical state feature vector. Multiply by the stored discrete state transition matrix To generate virtual twin state vectors It satisfies Formula 1: ,in, This is the virtual twin state vector, where the subscript "twin" represents the virtual twin value. This is a discrete state transition matrix, where the subscript *trans* denotes a state transition. This is the initial physical state feature vector, with the subscript ini indicating the initial value.

[0045] The data processing unit uses an internal shift register to latch historical virtual feature vectors from five consecutive time steps and obtains the state feature autocorrelation matrix using an orthogonalization algorithm. Perform the algebraic difference absolute value operation according to Formula 2: ,in, The value represents the transient projection residual, with the subscript res indicating the residual. This is the initial physical state feature vector, where the subscript ini indicates the initial value. This is the state feature autocorrelation matrix, where the subscript "field" indicates the feature field. The standard deviation of the time step is the reference constant, and the subscript std indicates the standard deviation. Its numerator is... The intermediate calculation result was 4.62, which was then divided by the time step standard deviation benchmark constant with a value of 1.0 by the data consistency arbitration unit. Calculate the transient projection residual value of the output. It equals 4.62.

[0046] When the data consistency arbitration unit detects transient projection residual values When the data transmission exceeds the preset threshold of 1.0 for three consecutive sampling periods, the discrete event-driven state machine outputs a data blocking signal to the electrical data acquisition unit to cut off the data channel to the topology state mapping unit. The topology state mapping unit calls the internal Kirchhoff power balance control equation to perform extrapolation calculations and generates a virtual feature repair vector to cover the output. During the sampling period when the bypass interference continues, the hierarchical network topology utilizes the stored discrete state transition matrix. The system uses fixed parameters to maintain tracking of the power state, keeping the output of the topology state mapping unit at the preset topology boundary convergence state, and resuming tracking of the high-voltage metering cabinet state after the data blocking signal is released. Furthermore, under actual operating variant conditions, when the aforementioned trend analysis unit outputs an abnormal state signal to the feedback compensation unit, the system initiates a sliding mode control mechanism based on a switching hyperplane to dynamically reconstruct the matrix characteristics. The preset sliding mode switching surface function in the feedback compensation unit is defined as the difference between the current norm scalar of the discrete difference matrix and the preset trend threshold. The sliding mode control algorithm uses an exponential reaching law function, with the input control quantity being the current cumulative residual change rate and the output being the scalar control law parameters.

[0047] To distribute the scalar control law to each parameter bit of the discrete state transition matrix, a fixed weight mapping sensitivity coefficient matrix is ​​established internally. Each corresponding element of the matrix represents the contribution ratio of the baseline impedance of the corresponding topological branch to the overall residual fluctuation. The feedback compensation unit generates the corresponding two-dimensional matrix parameter correction by directly multiplying the calculated scalar control law by the corresponding row and column coefficients of the sensitivity coefficient matrix, and then superimposes it in place onto the discrete state transition matrix of the previous sampling period, so that the topological parameters of the virtual space model converge rapidly to the steady state under the constraint of the sliding surface.

[0048] Example 3: This example combines Figures 1 to 2 This describes an anti-theft electricity twin system for mapping the status of metering cabinets, such as... Figure 1As shown, the outgoing side of the timing electrical signal metering cabinet is connected to the input terminal of the low-pass filtering algorithm of the electrical data acquisition unit. The output terminal of the low-pass filtering algorithm of the electrical data acquisition unit is connected to the initial physical state feature vector. The initial physical state feature vector is connected to the discrete state transition matrix stored in the topology state mapping unit and the discrete event-driven state machine embedded in the data consistency arbitration unit, respectively, to calculate the absolute value of the algebraic difference and generate the transient projection residual value. The output terminal of the discrete state transition matrix stored in the topology state mapping unit is connected to the virtual twin state vector. The virtual twin state vector is connected to the orthogonalization algorithm of the shift register latch inside the data processing unit through the continuous time step historical virtual feature vector. The output terminal of the orthogonalization algorithm of the shift register latch inside the data processing unit is connected to the state feature autocorrelation matrix. The output terminal of the state feature autocorrelation matrix is ​​connected to the discrete event-driven state machine embedded in the data consistency arbitration unit, to calculate the absolute value of the algebraic difference and generate the transient projection residual value. The discrete event-driven state machine embedded in the data consistency arbitration unit calculates the absolute value of the algebraic difference and generates the transient projection residual value, and then outputs a data blocking command to close the data channel flowing to the imaging unit and feeds back to the low-pass filtering algorithm of the electrical data acquisition unit.

[0049] like Figure 2 As shown, the output of the electrical data acquisition unit is connected to the topology state mapping unit and the data consistency arbitration unit respectively, the output of the data processing unit is connected to the data consistency arbitration unit, and the output of the data consistency arbitration unit is fed back to the topology state mapping unit.

[0050] like Figure 3 As shown, the graph objectively presents the dimensionless technical correlation between the bypass shunt interference ratio (%) and the transient projection residual value. It includes the input independent variable scales of 0, 5, 10, 15, 20, 25, 30, 35, 40, and 45, and the output dependent variable scales of 0.00, 2.00, 4.00, 6.00, 8.00, and 10.00. It also includes the experimental group residual curve and the safety threshold boundary line in situ, which are used to collaboratively characterize the residual response law and threshold limit of the anti-theft twin system under the corresponding interference ratio.

[0051] Example 4: When the system faces the dynamic drift of baseline impedance caused by the parallel operation of multiple power distribution lines in the field, the initial timing electrical signal flowing through the high-voltage metering cabinet will generate non-stationary long amplitude fluctuations. Due to the change in the electromagnetic environment at the field, the power spectral density of the measurement noise will fluctuate. The fixed parameter mapping structure will have difficulty distinguishing between the transient jumps caused by normal substation amplitude changes and the signal amplitude changes caused by the diversion of non-real business data entry. This will cause the data consistency arbitration unit to make missed judgments in the environmental fluctuation period, which will limit the tracking response accuracy of the high-voltage metering cabinet status mapping system in the long-term substation operation environment.

[0052] To determine the reference constant for the standard deviation of the time step The numerical source is the parameter calibration process run by the data processing unit during the initialization and readiness phase before the system is put into operation. The independent variable input to the parameter calibration process is the initial physical state feature vector of 1000 sampling periods continuously captured under standard steady-state operating conditions without bypass shunt interference. The sequence, comprising the following ordered computational steps: First, write the vectors for these 1000 sampling periods sequentially into a circular non-volatile memory array; second, calculate the projection variance sequence of these 1000 discrete feature vectors after transformation in the orthogonal projection phase space; third, calculate the arithmetic square root of the projection variance sequence to obtain the discrete estimate standard deviation, and assign the arithmetic square root as the calibration output value to the time step standard deviation benchmark constant. This allows for the determination of the time step standard deviation baseline constant. In the actual calibration procedure, the transformation matrix used for the aforementioned orthogonal projection phase space transformation is a dimensionless orthogonal projection operator matrix constructed from the first three principal component eigenvectors extracted after singular value decomposition of the basic network admittance matrix acquired during system initialization. This transformation matrix linearly maps 1000 discrete eigenvectors, eliminating the correlation between physical quantities and thus calculating the pure projection variance. Based on this calibration procedure, the time step standard deviation benchmark constant reflects the natural white noise background fluctuations of the system under completely interference-free conditions. The aforementioned preset safety threshold is precisely set to 1.0. In engineering and physical terms, this means that the transient projection residual value has reached 1.0 times the standard deviation of the calibrated background noise. From a statistical and engineering fault tolerance perspective, residual values ​​below 1.0 are completely within the reasonable allowable range of random electromagnetic noise or small normal drift of baseline impedance in the field, and the system will not make a misjudgment. However, once the residual value continuously exceeds the physical critical boundary of 1.0, it means that non-random bypass shunting or tampering data must be superimposed in the electrical waveform, and the system will determine that the physical input data is unreliable.

[0053] To make the state feature autocorrelation matrix As a reference for topological features, the data processing unit arranges the five consecutive time-step historical virtual feature vectors latched in the internal shift register into a matrix containing five column vectors. It then calculates mutually orthogonal basis vector sets using the Gram-Schmidt orthogonalization method, calculates the outer product of each basis vector in the basis vector set, and sums the matrix of all outer products to output the state feature autocorrelation matrix. In a specific set of computational runs, when the time step standard deviation benchmark constant is calculated through the parameter calibration process... When the value is equal to 1.25, if the electrical data acquisition unit outputs fluctuating initial physical state characteristic vectors due to line baseline impedance drift. The data consistency arbitration unit retrieves the orthogonally calculated state feature autocorrelation matrix. For residual verification, the data consistency arbitration unit uses the initial physical state feature vector. Autocorrelation matrix of state features After multiplication and subtraction, the absolute difference in the numerator is calculated to be 5.75. This absolute difference is then divided by the calibrated time step standard deviation reference constant of 1.25. Based on Formula 3, the transient projection residual value for the current sampling period is then calculated. Equals 4.60: ,in, Here, represents the transient projection residual value, with the subscript 'res' indicating the residual. The value 5.75 is the absolute value of the estimated algebraic difference generated by multiplying the initial physical state eigenvector and the state eigenvalue autocorrelation matrix, and the value 1.25 is the benchmark constant for the time step standard deviation determined through the parameter calibration process. The corresponding numerical value, with the subscript std representing the standard deviation, is used. Since the calculated transient projection residual value of 4.60 exceeds the preset judgment threshold of 1.0, the discrete event-driven state machine set in the data consistency arbitration unit changes the control state and sends a data blocking signal to the electrical data acquisition unit to close the data channel flowing to the topology state mapping unit. The topology state mapping unit calls the Kirchhoff power balance control flow equation stored internally to complete the extrapolation calculation and outputs a virtual feature repair vector to cover the output end.

[0054] Because the data updates affected by interference are blocked, the update parameters of the topology state mapping unit are controlled within the preset topology boundary convergence region. This allows the virtual twin state vector to be restored to the topology convergence state when faced with interference scenarios such as distribution line baseline drift. To address the issue that the physical dimensions and mathematical dimensions of the node current component, voltage phase change rate, and transient active power deviation contained in the initial physical state feature vector are different and cannot be directly multiplied, the data processing unit incorporates a normalization conversion interface before performing orthogonalization calculations. Specifically, the system pre-sets current reference values, phase change rate reference values, and power reference values, and at each sampling time, divides the measured amperes, radians per second, and watts by the corresponding reference values. The initial physical state feature vector is transformed into a dimensionless scalar value. After normalization, it is transformed into a standard one-dimensional 3x1 mathematical vector. Meanwhile, the historical state feature autocorrelation matrix calculated by the Gram-Schmidt orthogonalization method is a dimensionless 3x3 orthogonal matrix. Based on this, the data consistency arbitration unit performs a multiplication operation between the state feature autocorrelation matrix and the initial physical state feature vector, following the standard matrix multiplication rules, that is, multiplying each row element of the matrix with the corresponding component of the one-dimensional column vector and summing the products. Finally, a projection estimation vector with the same 3x1 dimensionless structure is output, thus achieving seamless integration of multidimensional heterogeneous features at the underlying mathematical logic and physical dimensions.

[0055] Example 5: When the system faces the condition of a newly installed high-voltage metering cabinet being connected and a sudden change in the physical bus admittance benchmark, the hierarchical network topology lacks initial baseline parameters, making it difficult to establish virtual and real control flow alignment, causing false alarms in the data consistency arbitration unit during pre-estimation. To establish a definite baseline alignment before field deployment, the system uses a zero-input self-collimation procedure to correct the discrete state transition matrix. Its input requires the high-voltage metering cabinet to be in a physical zero state with no load current flowing through it. The functional specifications of the enabling environment limit the injection signal generator to output a standard voltage signal with a frequency of 50Hz and an effective value of 100V to the metering induction winding. The data processing unit calculates the admittance trajectory of the voltage signal on the complex impedance plane to obtain the reference adjustment operator vector used to correct the matrix parameters. The conversion process satisfies the following equation: ,in, This is a discrete state transition matrix, where the subscript *trans* denotes a state transition. The operator vector is adjusted based on the reference, and the subscript cal indicates calibration. The base state transition matrix is ​​used, with the subscript `base` indicating the base value. Specifically, the multiplication operation between the aforementioned reference adjustment operator vector and the base state transition matrix employs a row-corresponding scalar scaling alignment mapping rule at the physical level. The base state transition matrix is ​​defined as a 3x3 square matrix, while the reference adjustment operator vector is defined as a column vector containing three real components. During matrix parameter calibration, the data processing unit does not perform traditional matrix inner product multiplication but instead uses the first component of the reference adjustment operator vector as a scaling factor, directly multiplying it with the base state transition matrix. Multiply all elements in the first row of the matrix to obtain the elements in the first row of the calibrated discrete state transition matrix. Similarly, multiply the second and third components of the reference adjustment operator vector with all elements in the second and third rows of the basic state transition matrix to reconstruct the elements in the second and third rows of the discrete state transition matrix. Through this row-by-row scalar product mapping mechanism, the admittance calibration trajectory of the one-dimensional vector is completely injected into each corresponding parameter bit of the two-dimensional state transition matrix, eliminating the dimension mismatch problem when the vector and matrix are directly multiplied.

[0056] During the sampling period The baseline adjustment operator vector is 50ms. During a fixed load-bearing operation phase, when the physical side experiences a micro-ohmic level increase in resistance due to heating, inducing a response deviation, the initial physical state characteristic vector... The local attenuation residuals of amplitude variation are superimposed; during the numerical calculation process, the reference adjustment operator vector is measured. The component is 0.98, which is multiplied by the fundamental state transition matrix. Obtain the calibrated discrete state transition matrix Numerical, the topological state mapping unit calculates based on calibration parameters, causing the absolute value of the estimated difference obtained by multiplying the initial physical state feature vector and the state feature autocorrelation matrix to converge to zero and decrease to 0.08, in the transient projection residual value When the data remains within a safe zone below the preset threshold of 1.0, the data consistency arbitration unit closes the blocking loop, and the hierarchical network topology continues to track the power state. When specifically monitoring the differential state, the monotonically increasing trend of the parameters of the aforementioned discrete difference matrix is ​​quantified at the algorithm level by calculating the Frobenius norm of the two-dimensional matrix. Specifically, the trend analysis unit continuously caches the transient projection residual values ​​within 10 sampling periods, and fills a 10-row, 10-column discrete difference matrix by calculating the algebraic difference between the residuals of adjacent periods. This process transforms the multidimensional matrix into a single matrix that can be directly compared in magnitude and monotonicity. For a single scalar value, the system performs norm dimensionality reduction mapping on the discrete difference matrix in each sampling period. This involves squaring all parameter elements in each row and column of the matrix, summing all the squared values, and finally calculating the square root of the sum to output a dimensionless matrix norm scalar. The trend analysis unit compares the magnitudes of the norm scalar output over multiple consecutive periods. If the norm scalar shows a monotonically increasing value over consecutive time steps, and the cumulative increase exceeds a preset trend threshold, it is determined that the parameters of the discrete difference matrix exhibit a monotonically increasing trend, thus triggering subsequent control.

[0057] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A digital twin system for anti-theft electricity twinning used for metering cabinet status mapping, characterized in that, include: The electrical data acquisition unit is used to acquire the timing electrical signals on the outgoing side of the metering cabinet and to calculate and generate the initial physical state feature vector through a low-pass filtering algorithm. The topology state mapping unit, connected to the electrical data acquisition unit, is used to store the discrete state transition matrix and calculate the product of the discrete state transition matrix and the initial physical state feature vector to generate a virtual twin state vector. The data processing unit is used to calculate and generate the state feature autocorrelation matrix by utilizing the continuous time step historical virtual feature vectors latched by the internal shift register and through an orthogonalization algorithm. The data consistency arbitration unit is connected to the topology state mapping unit and the data processing unit, respectively. It is used to calculate the product of the initial physical state feature vector and the state feature autocorrelation matrix to generate the projection estimation vector, and to calculate the absolute value of the algebraic difference between the initial physical state feature vector and the projection estimation vector as the transient projection residual value. The data consistency arbitration unit has an embedded discrete event-driven state machine. When the transient projection residual value exceeds the safety threshold of 1.0 for three consecutive sampling periods, the discrete event-driven state machine will flip its state and output a data blocking command to the electrical data acquisition unit to close the data channel from the electrical data acquisition unit to the topology state mapping unit.

2. The anti-theft electricity twin system for metering cabinet status mapping according to claim 1, characterized in that, The system also includes a feedback compensation unit; when the data consistency arbitration unit outputs a data blocking command, it triggers the feedback compensation unit to close the transmission channel of the initial physical state feature vector to the update interface of the topology state mapping unit. The feedback compensation unit stops updating the discrete state transition matrix and locks the current parameters of the discrete state transition matrix at the fixed values ​​of the previous sampling period, using the fixed values ​​of the parameters to maintain the parameter stability of the topology state mapping unit.

3. The anti-theft electricity twin system for metering cabinet status mapping according to claim 1, characterized in that, The system also includes a data extrapolation unit; the data extrapolation unit is connected to the data consistency arbitration unit and the data processing unit, and is used to extract the continuous time step historical virtual feature vector latched in the shift register inside the data processing unit in response to the data blocking instruction, and substitute the continuous time step historical virtual feature vector as the boundary constraint value into the preset power balance state equation for recursive iteration, generate a virtual feature repair vector and output it to the output end of the topology state mapping unit to replace the current virtual twin state vector.

4. The anti-theft electricity twin system for metering cabinet status mapping according to claim 1, characterized in that, The system also includes a dynamic gating unit; the dynamic gating unit is connected to the input of the data consistency arbitration unit, and is used to collect the initial physical state feature vector and calculate the slope of change between adjacent sampling periods. When the slope of change exceeds the set slope threshold, a circuit breaker signal is generated internally, and the current parameters of the discrete state transition matrix are locked in response to the circuit breaker signal to suspend the iterative control of the data processing unit.

5. The anti-theft electricity twin system for metering cabinet status mapping according to claim 1, characterized in that, The system also includes a sliding window filtering unit; the sliding window filtering unit is connected in series to the front input of the data consistency arbitration unit, and is used to sort the transient projection residual values ​​and extract the median within a continuous 50ms sliding time window, and generate smooth residual values ​​to output to the discrete event-driven state machine.

6. The anti-theft electricity twin system for metering cabinet status mapping according to claim 1, characterized in that, The system also includes a state locking unit; the state locking unit is connected to the data processing unit and is used to fix the current parameters of the discrete state transition matrix to the values ​​of the previous sampling period when a network communication interruption signal is detected.

7. The anti-theft electricity twin system for metering cabinet status mapping according to claim 1, characterized in that, The system also includes a trend analysis unit; the trend analysis unit is connected between the data processing unit and the feedback compensation unit, and is used to continuously cache the transient projection residual values ​​within ten consecutive sampling periods to construct a discrete difference matrix. When the parameters of the discrete difference matrix show a monotonically increasing trend and the cumulative difference value exceeds the preset trend threshold, an abnormal state signal is output to the feedback compensation unit, which triggers the feedback compensation unit to superimpose parameter corrections onto the discrete state transition matrix through a sliding mode control algorithm.

8. The anti-theft electricity twin system for metering cabinet status mapping according to claim 1, characterized in that, The system also includes a topology reconstruction unit; the topology reconstruction unit is connected to the topology state mapping unit and is used to split the topology model in the topology state mapping unit into multiple sub-region models and independently calculate the local residual value of each sub-region model. When the deviation between the local residual value of a certain sub-region model and the local residual value of the adjacent sub-region model exceeds a set ratio, it is determined that the branch node inside the metering cabinet corresponding to the sub-region model is in an abnormal state. The parameters of the branch node corresponding to the abnormal state are reset to zero, and the virtual twin state vectors of the adjacent sub-region models are extracted for interpolation calculation to update the topology model.

9. The anti-theft electricity twin system for metering cabinet status mapping according to claim 1, characterized in that, The sampling period of the electrical data acquisition unit is 50ms, and the initial physical state feature vector includes the node current component, voltage phase change rate, and transient active power deviation under the same sampling period.

Citation Information

Patent Citations

  • Electricity larceny prevention detection system for electric energy metering box

    CN121499873A