Electricity larceny prevention intelligent monitoring control method and system for electric energy metering box

Abstract

The invention relates to the technical field of electric energy metering protection, and discloses an electric energy metering box electricity larceny prevention intelligent monitoring control method and system. The system comprises an electricity larceny prevention feature recognition module, a security policy matching module, a dynamic parameter adjustment module and an intervention control execution module. The electricity larceny prevention feature recognition module collects voltage ripple and current ripple features of the metering box in real time and outputs an electricity larceny prevention feature data set. A security policy matching module inquires a preset electricity stealing mode feature library according to the preset electricity stealing mode feature library and outputs a target security policy identifier and an initial security policy parameter; the dynamic parameter adjustment module corrects the initial parameter based on the feature data set and generates a real-time security policy parameter; and the intervention control execution module executes hierarchical intervention actions according to the real-time parameters. The system can comprehensively capture electric energy parameter changes, accurately identify electricity stealing behaviors, adapt to different electricity utilization scenes, and realize a complete electricity stealing prevention process from monitoring to intervention.

Description

Technical Field

[0001] This invention relates to the field of electricity metering and protection technology, specifically to an intelligent monitoring and control method and system for preventing electricity theft from electricity metering boxes. Background Technology

[0002] In the power supply network, the electricity metering box serves as the core carrier for measuring electricity consumption. Its operational safety and metering accuracy directly affect the economic benefits of power companies and the stable operation of the power system. With the continuous growth in the number of electricity users and the increasing complexity of electricity consumption scenarios, electricity theft is showing a diversified, concealed, and intelligent development trend. Traditional protection methods for electricity metering boxes are no longer sufficient to meet the actual needs of current anti-theft work. Currently, common protective measures for electricity metering boxes on the market mainly consist of physical locks, periodic manual inspections, and simple current and voltage anomaly alarms. Physical locks only restrict external opening of the metering box and cannot prevent covert electricity theft operations such as internal wiring tampering and meter bypassing. Furthermore, the locks themselves are easily damaged or copied, resulting in low reliability. Periodic manual inspections are limited by inspection cycles, the number of personnel, and geographical scope, making real-time monitoring of a large number of dispersed metering boxes difficult. Often, electricity theft is only discovered after a considerable period, leading to significant economic losses for power companies. While simple current and voltage anomaly alarm systems can sound an alarm when parameters exceed set thresholds, these systems typically use fixed threshold judgment methods and cannot flexibly adjust to normal parameter fluctuations under different power consumption scenarios and at different times. For example, during peak industrial power consumption periods, the normal current and voltage fluctuation range will significantly increase. If the system still judges based on the thresholds for the normal period, it is very likely to generate a large number of false alarms, which not only increases the workload of maintenance personnel but may also cause genuine electricity theft alarms to be ignored. In addition, such systems can only realize alarm functions and lack the ability to further identify, analyze, and actively intervene in electricity theft, failing to form a complete closed loop of electricity theft prevention from monitoring to handling, and thus failing to effectively curb electricity theft. With the continuous upgrading of electricity theft technology, new methods of electricity theft, such as using power electronic devices to tamper with voltage and current waveforms and interfering with the normal operation of metering chips through wireless signals, are constantly emerging. Traditional anti-electricity theft systems are significantly insufficient in terms of the comprehensiveness and accuracy of electricity theft feature identification, and cannot effectively capture the subtle parameter changes caused by these new electricity theft behaviors, leading to a passive situation in anti-electricity theft work. An anti-electricity theft monitoring system with real-time, intelligent and adaptive capabilities is needed to solve the above problems. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent monitoring and control method and system for preventing electricity theft in electricity metering boxes, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, this invention provides an intelligent monitoring system for preventing electricity theft in electricity metering boxes. The system includes: an anti-theft feature identification module, a security strategy matching module, a dynamic parameter adjustment module, and an intervention control execution module. The anti-theft feature identification module is used to collect voltage ripple characteristics and current ripple characteristics of the electricity metering box in real time and output an anti-theft feature dataset. The security strategy matching module is connected to the anti-theft feature identification module and is used to query a preset electricity theft pattern feature library based on the anti-theft feature dataset, outputting a target security strategy identifier and corresponding initial security strategy parameters. The dynamic parameter adjustment module is connected to the security strategy matching module and is used to dynamically correct the initial security strategy parameters based on the anti-theft feature dataset to generate real-time security strategy parameters. The intervention control execution module is connected to the dynamic parameter adjustment module and is used to execute tiered intervention actions based on the real-time security strategy parameters.

[0005] Preferably, the anti-electricity theft feature recognition module generates the anti-electricity theft feature dataset in the following manner: extracting the abrupt frequency feature and distortion amplitude feature from the voltage ripple feature; extracting the phase shift feature and abnormal harmonic feature from the current ripple feature; and combining the abrupt frequency feature, distortion amplitude feature, phase shift feature and abnormal harmonic feature into a multi-dimensional feature vector as the anti-electricity theft feature dataset.

[0006] Preferably, the security policy matching module outputs the target security policy identifier in the following manner: calculating the similarity weight value between the anti-electricity theft feature dataset and each historical electricity theft feature template in the electricity theft pattern feature library; selecting all historical electricity theft feature templates whose similarity weight value exceeds a preset threshold as the matching feature template set; and determining the target security policy identifier based on the policy priority coefficient corresponding to the matching feature template set.

[0007] Preferably, the dynamic parameter adjustment module generates the real-time security strategy parameters in the following manner: acquiring the ripple mutation ratio feature and phase shift trend feature from the anti-theft feature dataset; calculating the voltage compensation weight based on the ripple mutation ratio feature; calculating the current compensation weight based on the phase shift trend feature; and using the voltage compensation weight and current compensation weight to perform weighted fusion on the threshold parameters in the initial security strategy parameters to obtain real-time threshold parameters as a component of the real-time security strategy parameters.

[0008] Preferably, the system further includes a system delay compensation module; the system delay compensation module is used to call a pre-stored metering response delay duration when the anti-electricity theft feature identification module is working; to perform time-domain compensation on the voltage ripple feature and current ripple feature according to the metering response delay duration, and to output the compensated ripple feature to the anti-electricity theft feature identification module.

[0009] Preferably, the intervention control execution module performs the graded intervention actions in the following ways: when the real-time safety strategy parameters include a primary intervention indicator, the electromagnetic locking device of the metering box protective shell is activated; when the real-time safety strategy parameters include an intermediate intervention indicator, the disconnection control circuit of the metering circuit is triggered; when the real-time safety strategy parameters include a high-level intervention indicator, the wireless alarm unit of the positioning and tracking device is activated.

[0010] Preferably, the security policy matching module is further configured to track the rate of change of the anti-electricity theft feature dataset in real time; update the policy priority coefficient of the matching feature template set according to the rate of change; and feed back the updated policy priority coefficient to the dynamic parameter adjustment module.

[0011] Preferably, the intervention control execution module is further configured to continuously collect the vibration frequency characteristics of the metering box protective shell when performing the graded intervention action; when the vibration frequency characteristics exceed a preset safety threshold, the intermediate intervention indicator is upgraded to the advanced intervention indicator.

[0012] Preferably, the dynamic parameter adjustment module is further configured to obtain the current operating status parameters of the power metering box before generating the real-time security strategy parameters; when the difference between the current operating status parameters and the initial security strategy parameters exceeds a preset tolerance, the parameter difference verification process is initiated. The parameter difference verification process includes: retrieving a baseline parameter set under historical normal operating conditions; calculating the deviation coefficient between the current operating status parameter and the baseline parameter set; and when the deviation coefficient is less than a preset anomaly judgment value, replacing the initial security policy parameter with the baseline parameter set as the input baseline for the real-time security policy parameter.

[0013] Preferably, the present invention also includes an intelligent monitoring and control method for preventing electricity theft in an electricity metering box, the method comprising all the modules and method flow of the intelligent monitoring and control system for preventing electricity theft in an electricity metering box as described above.

[0014] Compared with the prior art, the beneficial effects of the present invention are: The anti-theft feature identification module collects voltage and current ripple characteristics of the electricity metering box in real time and outputs an anti-theft feature dataset. Compared with the traditional method of only monitoring the magnitude of current and voltage values, it can capture parameter changes during the electricity metering process more comprehensively and meticulously. The voltage and current ripple characteristics contain the subtle fluctuation patterns of the electrical signal in the time dimension. Even if new electricity theft methods such as waveform tampering and interference with metering chips are used, unique abnormal traces will be left on these ripple characteristics. This provides richer and more accurate basic data for subsequent electricity theft identification, helping to discover covert electricity theft operations that are difficult to detect with traditional monitoring methods. The security policy matching module is connected to the anti-electricity theft feature recognition module. Based on the anti-electricity theft feature dataset, it queries a pre-set electricity theft pattern feature library and outputs the target security policy identifier and corresponding initial security policy parameters. This enables rapid classification of electricity theft behavior and matching of preliminary response plans. The pre-set electricity theft pattern feature library integrates the characteristic patterns corresponding to various known electricity theft behaviors. When the anti-electricity theft feature dataset collected by the system matches the features of a certain electricity theft pattern in the library, it can quickly locate the possible type of electricity theft and call the initial security policy parameters for that type of electricity theft behavior. This avoids the response delay caused by the need for re-analysis and judgment when facing different electricity theft behaviors in traditional systems, thus improving the timeliness of handling electricity theft behavior. The dynamic parameter adjustment module dynamically corrects the initial security strategy parameters based on the anti-theft feature dataset, generating real-time security strategy parameters. This effectively solves the problems of poor adaptability and high false alarm rate caused by traditional systems using fixed thresholds or fixed strategies. Under different electricity consumption scenarios, the normal voltage ripple characteristics and current ripple characteristics of the electricity metering box will differ. For example, the ripple fluctuation patterns of residential electricity consumption and industrial electricity consumption are different, and the ripple characteristics of the same user's electricity consumption will also change at different times. The dynamic parameter adjustment module can flexibly adjust the initial security strategy parameters based on the real-time collected feature data and the actual situation of the current electricity consumption scenario. This ensures that the security strategy always matches the current normal electricity consumption parameter fluctuation patterns, reducing false alarms or missed alarms caused by improper parameter settings and improving the system's adaptability to complex electricity environments. The intervention control execution module executes tiered intervention actions based on real-time security policy parameters, overcoming the limitations of traditional systems that can only issue alarms but cannot proactively address the issue. This creates a complete anti-theft process from monitoring, identification, analysis to intervention. Tiered intervention actions can be differentiated based on the severity, scope of impact, and potential risks of the theft. For example, for minor, suspected theft anomalies, mild intervention actions such as data recording and anomaly marking can be performed first to further observe parameter trends. For clear, serious theft, moderate intervention actions such as cutting off the auxiliary power supply to the metering box and locking the data upload function of the metering chip can be executed to prevent the theft from continuing. For theft involving malicious damage to metering equipment and causing significant power loss, the power operation and maintenance platform can be linked to send on-site handling instructions and initiate higher-level intervention measures. This tiered intervention approach avoids excessive intervention that could impact normal electricity use while taking effective countermeasures based on the actual situation of the theft, minimizing economic losses for power companies and maintaining the stability of the power supply order. Attached Figure Description

[0015] Figure 1 This is a timing diagram of the intelligent monitoring system for preventing electricity theft in the electricity metering box described in this invention; Figure 2 A flowchart for generating a dataset of features for preventing electricity theft; Figure 3 A flowchart for generating real-time security policy parameters; Figure 4 A flowchart for updating strategy priority coefficients; Figure 5 This is a flowchart for verifying parameter differences. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 This invention provides an intelligent monitoring system for preventing electricity theft in electricity metering boxes. The system includes: an anti-theft feature identification module, a security strategy matching module, a dynamic parameter adjustment module, and an intervention control execution module. Specific implementation methods are as follows: The anti-theft feature identification module collects voltage and current ripple characteristics of the electricity metering box in real time and outputs an anti-theft feature dataset. The security policy matching module, connected to the anti-theft feature identification module, queries a pre-defined electricity theft pattern feature library based on the anti-theft feature dataset and outputs the target security policy identifier and corresponding initial security policy parameters. The dynamic parameter adjustment module, connected to the security policy matching module, dynamically corrects the initial security policy parameters based on the anti-theft feature dataset, generating real-time security policy parameters. The intervention control execution module, connected to the dynamic parameter adjustment module, executes tiered intervention actions based on the real-time security policy parameters. Through the coordinated operation of these modules, the system achieves real-time monitoring and dynamic protection of the electricity metering box.

[0018] Example 1: See Figure 2 In the process of generating the anti-theft feature dataset, the anti-theft feature identification module performs multi-dimensional analysis of voltage ripple and current ripple features. Voltage ripple features are captured in real-time by a high-precision sampling circuit, and the signal is input to a digital signal processor after analog-to-digital conversion. In the voltage ripple analysis unit, the system first detects the time-domain waveform of the voltage signal and identifies instantaneous jump points in the waveform. By statistically analyzing the number of jump points occurring per unit time, a sudden change frequency feature value is quantified. This feature value reflects the intensity of abnormal fluctuations in the voltage signal. Simultaneously, the voltage waveform is compared point-by-point with a standard sinusoidal reference signal, and the root mean square value of the amplitude deviation at each sampling point is calculated to generate a distortion amplitude feature. This feature characterizes the severity of voltage waveform distortion.

[0019] Current ripple characteristics are acquired by a Rogowski coil sensor, amplified through isolation, and then fed into the feature extraction channel. Phase offset characteristics are obtained using a dual-channel correlation analysis method: the current signal and a reference voltage signal are input into a phase detection circuit, and the real-time phase angle is calculated by measuring the zero-crossing time difference. The system records the phase angle change over ten consecutive power frequency cycles, and uses its standard deviation as the phase offset characteristic value. Abnormal harmonic characteristic analysis employs Fast Fourier Transform (FFT) technology to perform spectral decomposition on the current signal. The system pays particular attention to spectral components other than the 3rd, 5th, and 7th integer harmonics, calculating the percentage of non-integer harmonic energy in the total harmonic energy to generate abnormal harmonic characteristic values. These four characteristic values ​​are integrated into a four-dimensional feature vector by a data encapsulation unit, forming an anti-theft feature dataset. This dataset uses a fixed-length data frame structure, with each frame containing a timestamp, a feature value array, and a checksum, and is transmitted to the security policy matching module via a serial communication interface.

[0020] After receiving the anti-electricity theft feature dataset, the security policy matching module starts the feature matching engine. This engine retrieves historical electricity theft feature templates stored in the electricity theft pattern feature library. Each template contains a standardized feature vector and an associated policy identifier. The similarity weight value is calculated using a vector space model: the real-time feature vector is multiplied by the vector of each historical template, and then divided by the product of the vector moduli to obtain a cosine similarity coefficient. This coefficient is multiplied by a preset feature weight matrix, ultimately outputting a similarity weight value in the range of 0 to 1.

[0021] The system sets a dynamic similarity threshold, which is automatically adjusted based on the grid load rate. When the similarity weight value exceeds the current threshold, the corresponding historical template is selected into the matching feature template set. After the template set is generated, the strategy selection unit retrieves the strategy priority coefficient associated with each template. This coefficient is maintained by the historical data analysis subsystem and dynamically updated based on the strategy execution success rate and timeliness. The system adopts the maximum priority principle, selecting the strategy identifier corresponding to the template with the highest coefficient as the target security strategy identifier, and simultaneously outputs the initial security strategy parameter set for that strategy.

[0022] During feature matching, the system continuously monitors the stability of the feature vectors. When the change in the feature vectors over three consecutive sampling periods falls below a set value, the similarity threshold requirement is automatically increased to enhance the strictness of the matching. After the matching result is generated, the system encapsulates the target security policy identifier and initial security policy parameters into a policy instruction package, which is then sent to the dynamic parameter adjustment module via a parallel data bus.

[0023] The entire implementation process adopts a pipelined architecture, with voltage and current signal acquisition, feature extraction, and policy matching operating in parallel. The feature extraction unit employs a double-buffering mechanism to ensure continuous data processing. The policy matching engine implements multi-threaded queries, supporting simultaneous similarity calculations on 256 historical templates. Data transmission utilizes CRC checksum and retransmission mechanisms to ensure the reliability of command transmission. The system completes a full feature recognition and policy matching cycle every 200 milliseconds, achieving real-time monitoring and response.

[0024] Example 2: See Figure 3 The dynamic parameter adjustment module performs a dynamic correction process for the initial security policy parameters during the generation of real-time security policy parameters. After receiving the target security policy identifier and initial security policy parameter set from the security policy matching module, this module first extracts key feature components from the anti-theft feature dataset. The ripple abrupt change ratio feature is obtained through the voltage waveform analysis unit: the system sets a 200-millisecond sliding time window and counts the number of abrupt change points where the voltage sample value exceeds ±2% of the rated voltage threshold within the window. The ratio of the number of abrupt change points to the total number of sample points within the window is calculated to generate a ripple abrupt change ratio feature value in the range of 0 to 1. This feature value is updated every 50 milliseconds and recorded in a circular buffer.

[0025] The phase shift trend feature is extracted using phase sequence tracking technology. The current phase detection unit continuously outputs phase shift data, and the system retains the phase shift records for the most recent 20 power frequency cycles. The phase change between adjacent cycles is calculated using a first-order difference algorithm, and then a five-cycle moving average is applied to the difference sequence to generate a feature value characterizing the phase change trend. This feature value is expressed in degrees per second; a positive value indicates a continuous phase lag, and a negative value indicates a continuous phase lead.

[0026] The voltage compensation weight is calculated based on the ripple abrupt change ratio characteristic. The system pre-sets a ratio-weight mapping table, dividing the ripple abrupt change ratio characteristic value into five intervals: 0-0.2 corresponds to a weight coefficient of 0.3, 0.2-0.4 to 0.5, 0.4-0.6 to 0.7, 0.6-0.8 to 0.9, and 0.8-1.0 to 1.0. When the characteristic value is at the interval boundary, linear interpolation is used to determine the final weight. This mapping relationship is stored in a programmable read-only memory and supports parameter updates via a configuration interface. The current compensation weight is generated based on the phase offset trend characteristic. The system maintains a phase trend historical queue of length 10 and uses an exponentially weighted moving average algorithm to process historical data. The latest characteristic value is assigned a weight coefficient of 0.5, and the weights of historical data decrease successively with a decay factor of 0.8. The weighted calculation result is input into an S-curve conversion function, which outputs the current compensation weight for the interval from 0 to 1. The conversion function sets a saturation range of ±15 degrees / second. When the input value exceeds this range, the weight is automatically locked to the extreme value.

[0027] The generation of real-time threshold parameters involves a weighted fusion operation, extracting three basic parameters—upper voltage threshold, lower voltage threshold, and current threshold—from the initial safety policy parameter set. Voltage compensation weights are applied to the voltage threshold parameters, multiplying the initial upper and lower voltage thresholds by (1 + 0.5 × voltage compensation weight) and (1 - 0.5 × voltage compensation weight), respectively. Current compensation weights are applied to the current threshold parameters, multiplying the initial current threshold by (1 + 0.6 × current compensation weight). The fused parameters are then constrained using a convex combination algorithm: when the calculation result exceeds the safe operating range of the equipment, it is automatically truncated to the boundary value; when the parameter change exceeds 20% of the previous value, a gradual adjustment mechanism is initiated, gradually transitioning to the target value over three calculation cycles. The final generated real-time safety policy parameter set includes the corrected upper and lower voltage thresholds, current thresholds, and intervention level parameters. This parameter set uses a binary encoding format, containing fields such as version number, parameter value, and checksum, and is transmitted to the intervention control execution module via a high-speed data bus. The system executes a complete parameter adjustment cycle every 100 milliseconds, implementing a dedicated computational pipeline within the FPGA chip: a feature extraction unit, a weight calculation unit, and a parameter fusion unit operate in parallel, with the output of the previous stage serving as the input for the next. A double-buffering mechanism is implemented in the data path to ensure uninterrupted continuous output. The parameter version management unit records the parameter sequence number for each adjustment, supporting parameter rollback and operation traceability.

[0028] In handling special operating conditions, when the ripple abrupt change ratio characteristic value exceeds 0.9 five times consecutively, the system automatically switches to emergency correction mode. In this mode, the voltage compensation weight is directly set to the maximum value of 1.0, and the parameter update cycle is shortened to 20 milliseconds. When the grid frequency fluctuation exceeds ±0.5Hz, the phase offset trend characteristic calculation automatically switches to frequency adaptive mode, recalculating the phase offset using a dynamic reference frequency. The historical records of all correction parameters are stored in non-volatile memory, forming a parameter adjustment log for diagnostic analysis.

[0029] Example 3: The system delay compensation module initiates the compensation process when the anti-theft feature identification module is working. This module retrieves the pre-stored metering response delay duration from non-volatile memory. This duration, determined through laboratory calibration and field testing, includes three components: signal transmission delay, sample-and-hold delay, and data processing delay. The metering response delay duration is stored in milliseconds, achieving microsecond-level accuracy. Time-domain compensation for voltage and current ripple characteristics is implemented using digital signal processing technology. The system establishes a sampling data queue, the queue length of which is dynamically adjusted according to the delay duration. For each newly acquired voltage and current sampling point, the compensation algorithm calculates its correct position on the time axis.

[0030] The time-domain compensation process uses the following compensation formula: in: Indicates the index of the current sampling point. It is the compensated signal value. These are historical sampling point values. It is an integer delay component. It is the filter order. These are the window function coefficients. This formula reconstructs the signal using a weighted moving average. The window function employs a Kaiser window design, and the main lobe width and side lobe attenuation are optimized based on the signal characteristics.

[0031] After compensation, the ripple characteristics are output to the anti-theft feature identification module. After voltage signal compensation, waveform phase distortion is reduced, and the accuracy of abrupt change point timing is improved. After current signal compensation, the phase relationship of harmonic components is corrected, enhancing the reliability of non-integer harmonic detection. The compensation module performs a delay calibration every 8 milliseconds, automatically adjusting the delay parameters according to changes in ambient temperature. A temperature sensor monitors the circuit board temperature in real time, and the temperature-delay correction curve is stored in a lookup table.

[0032] After receiving the real-time safety policy parameters, the intervention control execution module parses the intervention level identifier within the parameters. When the parameters contain a primary intervention identifier, the module sends an activation command to the electromagnetic locking device of the metering box's protective housing. The electromagnetic locking device includes a drive circuit and a mechanical lock body. The drive circuit generates a strong magnetic field upon receiving a 12V DC pulse signal, pushing the latch into the locking slot. The locking status is detected by a Hall sensor, and the feedback signal is returned to the execution module. The locking device is designed to be self-locking when powered off, releasing only upon receiving a specific unlocking command. When the real-time safety policy parameters contain a mid-level intervention identifier, the module triggers the disconnection control circuit of the metering circuit. The disconnection control circuit adopts a dual-redundancy design, with a high-capacity magnetic latching relay in the main circuit and a solid-state switch in the backup circuit. The relay coil drive voltage is 24V, and the pull-in time is less than 10 milliseconds. The solid-state switch uses IGBT devices with a rated current of 100A and a disconnection time of less than 500 microseconds. The disconnection command is sent to both the main and backup circuits simultaneously, and the status monitoring circuit monitors the disconnection execution status in real time. After successful disconnection, the circuit impedance monitoring unit verifies the disconnection status to prevent loose connections or sticking.

[0033] When the real-time security policy parameters include an advanced intervention flag, the module activates the wireless alarm unit of the location tracking device. The wireless alarm unit includes a cellular communication module and a GPS positioning chip. The cellular module supports 4G LTE, has a built-in SIM card, and transmits alarm information as data packets in a specific format via the mobile network. The GPS chip acquires latitude and longitude coordinates with meter-level accuracy. After the alarm unit is activated, it first establishes a network connection and then sends alarm data containing the device number, location information, timestamp, and event type. The alarm data is protected by AES encryption, and integrity verification is enabled during transmission.

[0034] During the execution of graded intervention actions, the system maintains status monitoring. The operating current of the electromagnetic locking device is sampled in real time, and a retry mechanism is activated when the current is abnormal. The breaking status of the disconnection control circuit is monitored by a voltage probe, and the line voltage should drop to a safe range after disconnection. The transmission status of the wireless alarm unit is confirmed through network response, and the backup frequency band is automatically switched when transmission fails. All executed actions are logged in detail, including parameters such as instruction transmission time, execution result, and equipment status. Log data is stored cyclically in ferroelectric memory for a retention period of no less than 30 days. The system is equipped with an interlocking mechanism for intervention actions, with a 100-millisecond delay between primary and intermediate interventions to prevent action conflicts. When a high-level intervention is triggered, the cancellation function of other intervention actions is automatically disabled. The execution module adopts a multi-core processor architecture, with instruction parsing, action execution, and status monitoring handled by dedicated cores. The communication interface adopts an opto-isolation design to enhance anti-interference capabilities. The power management system provides multiple independent power supplies to ensure stable control circuit voltage when performing high-current disconnection actions.

[0035] The execution module also features a self-testing function, periodically checking parameters such as the resistance value of the electromagnetic locking device, the contact resistance of the disconnecting circuit, and the wireless signal strength daily. The self-test results are compared with historical data, and a warning message is generated when deviations exceed limits. The maintenance interface supports parameter configuration and action testing. In test mode, intervention actions are displayed as indicator lights, without actual mechanical operation. All configuration changes require dual authentication, and operation logs are permanently saved.

[0036] Example 4: See Figure 4 After completing the initial feature matching, the security policy matching module initiates a real-time tracking mechanism to continuously monitor the rate of change of the anti-theft feature dataset. This feature is obtained by calculating the change in Euclidean distance of the feature vectors within a continuous sampling period. The system uses a sliding window of length 5, calculating the average rate of change of the feature vectors within the window every 100 milliseconds. The rate of change feature value is divided into five levels, corresponding to different policy adjustment modes. When a continuous increase in the rate of change feature value is detected, the module initiates a dynamic update procedure for the policy priority coefficient.

[0037] The strategy priority coefficients are updated based on the correlation analysis between the rate of change feature value and the historical matching template. The system maintains a priority coefficient mapping table, recording the coefficient adjustment range corresponding to different rate of change intervals. This mapping table is continuously optimized based on field operation data. The mapping relationship of the latest version is shown in Table 1.

[0038] Table 1: Mapping relationship between change rate characteristics and strategy priority coefficient adjustment.

[0039] The updated policy priority coefficients are fed back to the dynamic parameter adjustment module in real time via a dedicated data channel. The feedback data includes metadata such as coefficient values, update timestamps, and validity periods, and differential transmission is used to reduce data volume. The dynamic parameter adjustment module recalculates the weight parameters based on the new priority coefficients, achieving coordinated updates of security policy parameters.

[0040] During the execution of graded intervention actions, the intervention control execution module simultaneously activates the vibration monitoring function. A three-axis MEMS accelerometer continuously acquires vibration signals from the protective shell of the metering box at a sampling frequency of 2kHz. Vibration frequency feature extraction employs digital filtering technology; the signal is passed through a 0.1-500Hz bandpass filter for frequency domain feature analysis. The system calculates the dominant frequency component of the vibration signal every 50 milliseconds and simultaneously statistically analyzes the energy distribution within the 0-100Hz frequency band.

[0041] The vibration safety threshold is set using a multi-level configuration. The initial threshold is set to an acceleration amplitude of 0.5 m / s² for general abnormal vibrations; the intermediate threshold is set to 2.0 m / s² for significantly destructive vibrations; and the advanced threshold is set to 5.0 m / s² for severely destructive behavior. When the vibration frequency characteristic value exceeds the currently set safety threshold, the system initiates an intervention level escalation procedure. The intervention flag escalation process follows a strict logical judgment process. When the vibration characteristic value is detected to exceed the intermediate threshold for three consecutive sampling cycles, the system automatically escalates the intermediate intervention flag to an advanced intervention flag. After the escalation command is generated, the currently executing intermediate intervention action is first paused, and then the execution sequence is reinitialized according to the advanced intervention procedure. During the escalation process, the mechanical lock remains continuously effective to avoid a safety protection gap.

[0042] Vibration monitoring data is correlated with intervention execution status. The system records the time, amplitude, frequency characteristics, and corresponding intervention type for each vibration exceeding the limit event. This data is used to optimize vibration threshold parameters and establish a correspondence between vibration patterns and electricity theft through machine learning algorithms. Long-term operational data shows a high correlation between vibration signals within a specific frequency range and illegal opening of metering boxes. The system employs a dual verification mechanism to prevent erroneous upgrades. When vibration characteristics exceed limits, spatial consistency conditions must be met simultaneously: acceleration measurements along the three axes must maintain a specific proportional relationship to eliminate random vibration interference. Before executing an upgrade command, the system performs a brief 20-millisecond delay for secondary verification to confirm the persistence and regularity of vibration characteristics. Detailed logs of upgrade operations are recorded, including vibration data snapshots, decision basis, and execution results. The vibration monitoring unit has a self-calibration function. A zero-point calibration procedure is automatically executed daily at midnight to eliminate sensor drift errors. Sensitivity calibration is performed quarterly using a standard vibration source to verify measurement accuracy. Calibration data is stored in an independent memory, and calibration history is traceable. When a sensor anomaly is detected, the system automatically switches to a backup sensor and generates a maintenance alarm.

[0043] After the intervention level was upgraded, the system entered enhanced monitoring mode, increasing the vibration sampling frequency to 5kHz and shortening the data analysis window to 20 milliseconds. The wireless alarm unit started continuous transmission mode, increasing the location information update frequency from once per minute to once every 10 seconds. At the same time, the peripheral device linkage function was activated, sending early warning signals to adjacent metering boxes to form a regional protection network.

[0044] Example 5: See Figure 5 The dynamic parameter adjustment module performs an operational status verification process before generating real-time safety policy parameters. This module acquires current operational status parameters through the monitoring unit built into the metering box. These parameters include RMS voltage, RMS current, active power, reactive power, power factor, and frequency measurements. RMS voltage measurement uses a true RMS converter chip with a sampling period of 10 milliseconds and a measurement accuracy of 0.2%. RMS current is acquired through a Hall sensor array; the sensor output is isolated and amplified before entering a 16-bit analog-to-digital converter. The power calculation unit uses a time-division multiplier principle to calculate active and reactive power in real time. All operational status parameters are updated every 20 milliseconds and transmitted to the dynamic parameter adjustment module in the form of data packets.

[0045] The parameter difference calculation employs a multi-dimensional comparison algorithm, where the system compares the current operating status parameters with their corresponding items in the initial safety policy parameters item by item. Voltage difference is calculated as the percentage deviation between the current effective voltage value and the initial voltage threshold. Current difference is calculated as the percentage deviation between the current effective current value and the initial current threshold. Power factor difference is calculated as the degree of deviation between the current power factor value and the initial power factor range. The system sets a preset tolerance parameter group, including voltage tolerance percentage, current tolerance percentage, and power factor tolerance value. When any difference exceeds its corresponding preset tolerance, the parameter difference verification process is triggered.

[0046] After the parameter difference verification process is initiated, the system accesses the distributed database to retrieve the baseline parameter set under historical normal operating conditions. The baseline parameter set contains statistical values ​​of operating parameters under the same time period and load conditions over the past 30 days, including the average voltage range, average current range, typical power factor value, and its allowable fluctuation range. Data retrieval is based on multi-dimensional indexes such as timestamp, load type, and ambient temperature to ensure the applicability of the baseline parameters. The baseline parameter set is stored in a compressed data format and is restored to its complete parameter structure after decompression.

[0047] The system treats the current operating state parameter set and the reference parameter set as two points in a multi-dimensional space, calculating their relative distance in the multi-dimensional feature space. This calculation considers the different dimensions and importance of each parameter, assigning higher weighting factors to key parameters such as voltage and current. The calculation also considers the correlation between parameters to avoid repeatedly calculating the influence of interdependent parameters. The deviation coefficient output is a dimensionless value between 0 and 1; the smaller the value, the closer it is to the reference state.

[0048] The preset anomaly judgment value is determined based on power grid operation experience and historical data analysis. This judgment value has multiple levels: below 0.1 indicates a completely normal state, 0.1 to 0.3 indicates a slight deviation state, 0.3 to 0.5 indicates a moderate deviation state, and above 0.5 indicates a significant anomaly state. When the calculated deviation coefficient is less than 0.3, the system determines that the current operating state belongs to normal operating condition fluctuations, rather than an anomaly caused by electricity theft. At this time, the system replaces the initial security policy parameters with the benchmark parameter set as the input benchmark for the real-time security policy parameters.

[0049] The parameter replacement operation employs a smooth transition approach. The system does not immediately and completely switch parameters, but gradually adjusts the parameter values ​​from their initial values ​​to the baseline values ​​over five calculation cycles. Each cycle adjusts the difference by 20% to avoid system oscillations caused by sudden parameter changes. During the parameter replacement process, the system continuously monitors the changing trends of the operating status parameters. If a new deviation is detected between the replaced parameters and the current operating status, the replacement will be paused and re-evaluated.

[0050] The benchmark parameter set is managed using a rolling update mechanism. Every day at midnight, the system automatically updates the benchmark parameter set, incorporating normal operation data from the same time period over the past 30 days and removing the earliest data. A weighted average algorithm is used during the update process, assigning higher weights to recent data to ensure the benchmark parameters reflect the latest operational status. All benchmark parameter update records are stored in the audit log, including detailed information such as update time, parameter values ​​before the update, and parameter values ​​after the update. The system also has a benchmark parameter validity verification procedure. Before each use of benchmark parameters, their statistical significance is checked, requiring the sample size of the benchmark parameter set to meet the minimum statistical requirements and the parameter distribution to conform to a normal distribution. If the benchmark parameter set does not meet the usage conditions, the system will revert to using the initial security policy parameters and generate an alert message indicating insufficient benchmark data. The alert message is sent to the maintenance center, prompting the need to supplement normal operation data for that period.

[0051] The entire parameter verification process employs a redundant computing architecture. The main processor executes the primary computational tasks, while coprocessors perform the same computations in parallel for result verification. The computation results from the two processors are compared in real time, and a third-party arbitration procedure is initiated when the difference exceeds the allowable range. Detailed process logs are maintained for all computations, including input parameters, intermediate results, and final conclusions. These logs are stored in a circular buffer for post-processing analysis and auditing.

[0052] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0053] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent monitoring system for preventing electricity theft in electricity metering boxes, characterized in that, include: The system comprises an anti-theft feature identification module, a security strategy matching module, a dynamic parameter adjustment module, and an intervention control execution module. The anti-theft feature identification module collects voltage and current ripple characteristics of the electricity metering box in real time and outputs an anti-theft feature dataset. The security strategy matching module, connected to the anti-theft feature identification module, queries a preset electricity theft pattern feature library based on the anti-theft feature dataset and outputs a target security strategy identifier and corresponding initial security strategy parameters. The dynamic parameter adjustment module, connected to the security strategy matching module, dynamically corrects the initial security strategy parameters based on the anti-theft feature dataset to generate real-time security strategy parameters. The intervention control execution module is connected to the dynamic parameter adjustment module and is used to execute graded intervention actions according to the real-time security policy parameters.

2. The intelligent monitoring system for preventing electricity theft in electricity metering boxes according to claim 1, characterized in that, The anti-electricity theft feature recognition module generates the anti-electricity theft feature dataset in the following way: extracting the abrupt frequency feature and distortion amplitude feature from the voltage ripple feature; extracting the phase shift feature and abnormal harmonic feature from the current ripple feature; and combining the abrupt frequency feature, distortion amplitude feature, phase shift feature and abnormal harmonic feature into a multi-dimensional feature vector as the anti-electricity theft feature dataset.

3. The intelligent monitoring system for preventing electricity theft in electricity metering boxes according to claim 1, characterized in that, The security policy matching module outputs the target security policy identifier in the following manner: calculating the similarity weight value between the anti-electricity theft feature dataset and each historical electricity theft feature template in the electricity theft pattern feature library; selecting all historical electricity theft feature templates whose similarity weight value exceeds a preset threshold as the matching feature template set; and determining the target security policy identifier based on the policy priority coefficient corresponding to the matching feature template set.

4. The intelligent monitoring system for preventing electricity theft in electricity metering boxes according to claim 1, characterized in that, The dynamic parameter adjustment module generates the real-time security strategy parameters by acquiring the ripple mutation ratio and phase offset trend features from the anti-electricity theft feature dataset. The voltage compensation weight is calculated based on the ripple abrupt change ratio characteristic; the current compensation weight is calculated based on the phase offset trend characteristic; the threshold parameters in the initial security strategy parameters are weighted and fused using the voltage compensation weight and the current compensation weight to obtain real-time threshold parameters as a component of the real-time security strategy parameters.

5. The intelligent monitoring system for preventing electricity theft in electricity metering boxes according to claim 1, characterized in that, Also includes: The system delay compensation module is used to call the pre-stored metering response delay duration when the anti-electricity theft feature identification module is working; to perform time-domain compensation on the voltage ripple feature and current ripple feature according to the metering response delay duration, and to output the compensated ripple feature to the anti-electricity theft feature identification module.

6. The intelligent monitoring system for preventing electricity theft in electricity metering boxes according to claim 1, characterized in that, The intervention control execution module performs the graded intervention actions in the following ways: when the real-time safety policy parameters include a primary intervention indicator, the electromagnetic locking device of the metering box protective shell is activated; when the real-time safety policy parameters include an intermediate intervention indicator, the disconnection control circuit of the metering circuit is triggered; when the real-time safety policy parameters include a high-level intervention indicator, the wireless alarm unit of the positioning and tracking device is activated.

7. The intelligent monitoring system for preventing electricity theft in electricity metering boxes according to claim 3, characterized in that, The security policy matching module is also used to track the rate of change of the anti-electricity theft feature dataset in real time; update the policy priority coefficient of the matching feature template set according to the rate of change; and feed back the updated policy priority coefficient to the dynamic parameter adjustment module.

8. The intelligent monitoring system for preventing electricity theft in electricity metering boxes according to claim 6, characterized in that, The intervention control execution module is also used to continuously collect the vibration frequency characteristics of the metering box protective shell when performing the graded intervention action; when the vibration frequency characteristics exceed the preset safety threshold, the intermediate intervention indicator is upgraded to the advanced intervention indicator.

9. The intelligent monitoring system for preventing electricity theft in electricity metering boxes according to claim 1, characterized in that, The dynamic parameter adjustment module is also used to obtain the current operating status parameters of the power metering box before generating the real-time security strategy parameters; when the difference between the current operating status parameters and the initial security strategy parameters exceeds the preset tolerance, the parameter difference verification process is initiated. The parameter difference verification process includes: retrieving a baseline parameter set under historical normal operating conditions; calculating the deviation coefficient between the current operating status parameter and the baseline parameter set; and when the deviation coefficient is less than a preset anomaly judgment value, replacing the initial security policy parameter with the baseline parameter set as the input baseline for the real-time security policy parameter.

10. A smart monitoring and control method for preventing electricity theft in an electricity metering box, characterized in that, The invention includes all modules and method flows of the intelligent monitoring and control system for preventing electricity theft in the electricity metering box as described in any one of claims 1 to 9.

Images