A multi-loop electrical safety synchronous sampling and protection method

Abstract

The present application relates to a kind of multi-loop electrical safety synchronous sampling and protection method, belong to low-voltage distribution electrical safety protection technical field.The method includes the synchronous sampling of multiple electrical circuits in distribution system, calculates each loop electrical parameter, extracts the cooperativity feature between each loop and calculates global cooperativity index, inputs single-loop electrical parameter and cooperativity feature into safety risk assessment model, outputs risk probability value, according to risk probability value and its change rate executes hierarchical active protection strategy, and monitors feedback after protection action to determine whether to upgrade protection.The present application deeply integrates synchronous sampling, cross-loop cooperativity feature extraction and hierarchical active protection, realizes from single-loop threshold judgment to multi-loop cooperativity analysis, from post-passivity trip to hierarchical active protection, from static offline model to online self-evolution model Comprehensive upgrade, significantly improves the electrical safety protection level of multi-loop distribution system, while maximum degree guarantees power supply continuity.

Description

Technical Field

[0001] This invention relates to the field of low-voltage power distribution electrical safety protection technology, specifically a method for synchronous sampling and protection of multi-circuit electrical safety. Background Technology

[0002] With the rapid development of industrial automation and intelligent buildings, the number of electrical circuits in low-voltage power distribution systems is increasing, and the power supply capacity and load types are becoming more complex, placing higher demands on the reliability, timeliness, and intelligence of electrical safety protection. Currently, electrical safety monitoring of multi-circuit power distribution systems mainly adopts two technical approaches: First, a distributed protection scheme based on independent detectors, where independent protection devices such as residual current devices, thermal relays, and arc fault detection devices are installed in each feeder circuit. Each device operates independently, without communication with others, and judges and acts according to its own preset thresholds. Second, a centralized monitoring scheme based on multi-circuit data acquisition devices, which uses multi-circuit meters or data acquisition terminals to poll and centrally display electrical parameters such as voltage and current in each circuit. Some schemes have the communication function to upload monitoring data to a background monitoring system. At the standards and specifications level, the GB / T 18216 series of standards put forward clear technical requirements for the measurement and monitoring equipment of electrical safety protection measures in low-voltage power distribution systems. While the above-mentioned existing schemes meet basic electrical safety monitoring needs to a certain extent, the functional modules are relatively independent, lacking deep data fusion and collaborative linkage.

[0003] However, the aforementioned existing technical solutions still have the following significant shortcomings. First, the sampling and protection functions are disconnected: data acquisition in existing systems primarily serves power metering or operational monitoring, while safety protection functions are performed by independent protection devices. The data from both is not effectively integrated, and the rich fault precursor information contained in the synchronously sampled data (such as the phase relationship of voltage and current, and the synchronous change trend of electrical parameters in each circuit) is not fully utilized by the protection system. Second, the protection method is passive and lagging: existing protection schemes are mostly based on a threshold triggering mechanism for a single electrical quantity—tripping or alarming only occurs after the residual current, temperature, or current amplitude reaches a preset threshold. This "post-event" response mode lacks the ability to identify and warn of progressive faults such as slowly increasing contact resistance and gradual aging of insulation materials; often, by the time the threshold is triggered, the fault has already developed to a certain extent. Secondly, there is a lack of multi-circuit collaborative analysis: existing technologies mainly focus on the independent monitoring and protection of single circuits, lacking systematic analysis methods for the collaborative characteristics between circuits (such as the synchronous changing trend of current imbalance in each circuit, the simultaneous rise of residual current in multiple circuits, and the phase correlation of harmonic distortion rates). This makes it difficult to identify systemic risks that single-circuit monitoring cannot detect, such as overall aging of the insulation system and bus-side anomalies. Furthermore, the protection strategies are simplistic: existing solutions typically only have two states, "normal" and "tripped," lacking the ability to differentiate responses based on risk levels, making it difficult to ensure power supply continuity while guaranteeing safety. These problems hinder further improvements in the electrical safety protection level of multi-circuit power distribution systems. Summary of the Invention

[0004] In order to solve the problems of the prior art, the present invention provides a method for synchronous sampling and protection of multi-circuit electrical safety.

[0005] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: Firstly, a method for synchronous sampling and protection of multi-circuit electrical safety, comprising the steps of synchronously sampling multiple electrical circuits in a power distribution system and calculating the electrical parameters of each circuit, and further comprising the following steps: S1: Using a unified sampling clock and sampling frequency, the power distribution system... The voltage and current signals of each electrical circuit are sampled synchronously to obtain the instantaneous voltage and current values ​​of each circuit at the same moment. The sampling frequency is dynamically adjusted according to the current system risk level. When the risk level is lower than the first preset level, the first sampling frequency is used. When the risk level reaches the first preset level, the sampling frequency will switch to the second sampling frequency. , , Not less than 64 times the fundamental frequency of the power grid. Not less than 128 times the fundamental frequency of the power grid; S2: Based on the synchronous sampling data of each circuit obtained in step S1, calculate the electrical parameters of each circuit. The electrical parameters include the effective value of voltage, the effective value of current, the power factor, active power, reactive power, harmonic distortion rate, three-phase current imbalance, zero-sequence current, and residual current. S3: Construct a multi-circuit parameter matrix from the electrical parameters of each circuit calculated in step S2, and extract the coordination characteristic indicators between each circuit. These indicators include the correlation coefficient of the synchronous change trend of the current imbalance of each circuit, the phase correlation of the harmonic distortion rate of each circuit, and the deviation between the vector sum and each component of the residual current of each circuit. Calculate the global coordination index based on these indicators. , ,in For the first loop and the first Pearson correlation coefficient of the loop parameter sequence; S4: Input the single-circuit electrical parameters obtained in step S2 and the cross-circuit coordination characteristic indicators obtained in step S3 into the safety risk assessment model, and output the risk probability values ​​of each circuit and the power distribution system. And risk level; the security risk assessment model is a classification model pre-trained based on machine learning; S5: Based on the risk probability value output in step S4 and its rate of change over time Implement a tiered proactive protection strategy: when Below the first threshold At the same time, maintain normal monitoring status; when achieve And below the second threshold ,and Below the first rate threshold In such cases, an early warning signal is issued and abnormal data is recorded; when achieve And below the third threshold ,or Below but Exceed When necessary, execute the load reduction operation command or switch to the backup circuit; when achieve Or after implementing load reduction operation Still more At that time, selective isolation protection actions are executed to cut off the faulty circuit with the highest risk; S6: After the active protection action in step S5 is executed, continue executing steps S1 to S4, monitoring the changes in electrical parameters and coordination characteristic indicators of each circuit after the protection action. If the risk probability value... Not down to The protection strategy will then be upgraded step by step.

[0006] In one specific implementation of the first aspect, the synchronous sampling in step S1 is achieved in the following manner: A unified second pulse signal is generated by a GPS or BeiDou satellite timing module, and each sampling channel synchronizes its sampling time based on the second pulse signal. When the external time synchronization signal is lost or abnormal, it automatically switches to self-synchronization mode, using the zero-crossing point of the voltage signal as a reference, and uses a software phase-locked loop algorithm to generate a synchronization sampling trigger signal to maintain the sampling synchronization of each channel.

[0007] In one specific implementation of the first aspect, the sampling frequency in step S1 is dynamically adjusted according to the current system risk level in the following way: When the risk probability value Below At that time, the first sampling frequency is used. ,in It is the fundamental frequency of the power grid; when achieve And lower than At that time, the sampling frequency was increased to ; when achieve At that time, the sampling frequency was increased to ; After the sampling frequency is adjusted, step S4 recalculates the risk probability value based on the updated high-precision sampling data. ,like If the sampling frequency decreases by more than a preset percentage compared to before the adjustment, the sampling frequency will be gradually reduced.

[0008] In one specific implementation of the first aspect, the global synergy index mentioned in step S3 It is also used to trigger preparatory protection: When the global synergy index In continuous The rate of decrease within a sampling period exceeds a preset rate threshold. At that time, among them Before the risk assessment results are output in step S4, the preparatory protection action in step S5 is triggered. The preparatory protection action includes issuing an early warning signal, pre-charging the backup circuit, and increasing the sampling density of the faulty circuit. After performing selective isolation protection in step S5, steps S1 to S3 are re-executed to calculate the global coordination index after isolation. ,like ,in The preset recovery coefficient and If so, then maintain the current protection status; if Then, further isolation is performed on the loops with the second highest risk.

[0009] In one specific implementation of the first aspect, the security risk assessment model described in step S4 employs an online incremental learning mechanism: The feature vector before each active protection action in step S5 is executed. Feedback results on the protective effect of step S6 Composition of online training samples ,in This indicates whether the protective measures successfully eliminated the risk; When the accumulated number of online training samples reaches the preset batch size When this occurs, an incremental update of the security risk assessment model is triggered, and the model parameters are fine-tuned using new samples. ; The incrementally updated model verifies the risk assessment results of subsequent sampled data in real time. If the deviation between the risk probability value output by the model and the actual protection effect feedback in step S6 continues to increase beyond the preset deviation threshold, full retraining is triggered.

[0010] In one specific implementation of the first aspect, the selective isolation protection action in step S5 specifically includes: Based on the risk probability values ​​of each loop output in step S4, the loops are ranked by risk. The probability value of identifying the risk is the highest and exceeds the third threshold. The circuit is used as a fault circuit; Send a trip command to the trip unit of the circuit breaker corresponding to the faulty circuit to disconnect the faulty circuit; After disconnecting the faulty circuit, repeat steps S1 to S4 to verify whether the remaining circuits have returned to normal operation.

[0011] In one specific embodiment of the first aspect, the residual current in step S2 is calculated in the following manner: The instantaneous residual current value of the circuit is obtained by summing the instantaneous values ​​of the three-phase current sampled synchronously from each circuit: ; The effective value of the instantaneous residual current is calculated to obtain the effective value of the residual current; The effective value of the residual current obtained from each circuit Input step S3 to calculate the correlation coefficient matrix between the residual current sequences of each loop. Consistency index with time series trends ; When the residual current of all circuits does not exceed their respective preset leakage current thresholds And the correlation coefficient matrix of the residual current sequence of each circuit The correlation coefficient between any two loops is greater than 0.7, and the residual current in each loop is continuously increasing. When the trend shows a monotonically increasing trend in each sampling period The system was deemed to be at risk of overall aging of the insulation system, and the leakage current threshold of each circuit was uniformly lowered to the original threshold. times, of which It also generates preventative maintenance early warning signals.

[0012] In one specific implementation of the first aspect, a temperature parameter fusion step is also included: In step S1, the temperature signals of each circuit terminal or cable sheath are collected synchronously. ; Calculate each loop within the time window internal temperature change rate and rate of change of current ; when Exceeding the first temperature change threshold and When the current change rate is below the first threshold, the circuit is determined to have abnormal contact resistance and is marked as a contact resistance risk circuit. For circuits marked as having contact resistance risk, the sampling frequency of the circuit is automatically increased to the highest level, the temperature warning threshold of the circuit is lowered to 80% of the normal threshold, and the temperature characteristics of the circuit are assigned a weighting coefficient 1.5 times higher than the normal weight in the risk assessment in step S4. Once the ratio of the temperature change rate to the current change rate of the circuit returns to the normal range, the above operation will be automatically terminated.

[0013] Secondly, an active predictive electrical safety protection system based on multi-circuit synchronous sampling includes: Multi-loop synchronous sampling unit is used to sample data in the power distribution system using a unified sampling clock and sampling frequency. The voltage and current signals of each electrical circuit are synchronously sampled, and the digital sampled values ​​of each circuit at the same moment are output; the multi-circuit synchronous sampling unit includes a voltage signal conditioning circuit, a current signal conditioning circuit, a sample and hold circuit, an analog-to-digital conversion module, a synchronous triggering module, and a central processing unit; An electrical parameter calculation unit, connected to the multi-circuit synchronous sampling unit, is used to calculate the electrical parameters of each circuit based on the digital sampling values. The electrical parameters include the effective value of voltage, the effective value of current, power factor, active power, reactive power, harmonic distortion rate, three-phase current imbalance, zero-sequence current, and residual current. A cross-circuit collaborative feature extraction unit, connected to the electrical parameter calculation unit, is used to construct a multi-circuit parameter matrix based on the electrical parameters of each circuit, extract collaborative feature indicators between circuits, and calculate a global collaborative index. ; The safety risk assessment unit, connected to the electrical parameter calculation unit and the cross-circuit collaborative feature extraction unit, has a built-in pre-trained safety risk assessment model, which is used to output the risk probability value of each circuit and the system based on the single-circuit electrical parameters and cross-circuit collaborative feature indicators. and risk level; The active protection execution unit, connected to the security risk assessment unit, is used to perform actions based on the risk probability value. and its rate of change over time Implement a tiered active protection strategy, including early warning output, load reduction control command output, and selective isolation protection command output; The protection effect feedback unit is connected to the active protection execution unit and the multi-loop synchronous sampling unit. It is used to monitor the changes in electrical parameters and coordination characteristics of each loop after the active protection action is executed, and to decide whether to upgrade the protection strategy based on the monitoring results.

[0014] In one specific implementation of the second aspect, the active protection execution unit includes: The graded early warning module is used to output different levels of audible and visual alarm signals or remote notifications according to the risk level; The load reduction control module is used to output load reduction commands to the power regulation device of the power distribution system; The selective isolation module is used to output tripping commands to the trip units of the circuit breakers corresponding to each circuit, thereby achieving selective disconnection of a single circuit. The graded early warning module, load reduction control module, and selective isolation module are activated sequentially or in combination according to the risk level. The system also includes a human-machine interaction unit and a communication unit. The human-machine interaction unit is used to display the electrical parameters, coordination characteristic indicators, risk probability values ​​and risk levels of each circuit in real time, and to receive user configuration input for protection strategy thresholds. The communication unit is used to upload the electrical parameters, coordination characteristic indicators, risk probability values ​​and risk levels of each circuit to a cloud server or a higher-level monitoring system, and to receive model update parameters or threshold configuration instructions issued by the cloud server.

[0015] The beneficial effects of this invention are as follows: 1. This invention uses a multi-loop synchronous sampling unit to synchronously acquire voltage, current, and temperature signals from each loop with a unified sampling clock and sampling frequency. This fundamentally ensures the phase consistency of electrical parameter calculations and vector synthesis for each loop, effectively eliminating power calculation errors and residual current vector deviations caused by asynchronous sampling, and providing a precise data foundation for subsequent fault identification. Simultaneously, this invention establishes a cross-loop collaborative feature extraction mechanism. By constructing a multi-loop parameter matrix and calculating the correlation coefficient matrix, it quantitatively extracts the collaborative features between loops, thereby obtaining a global collaborative index reflecting the overall operating status of the system. This mechanism breaks through the limitations of traditional technologies that only focus on single-loop electrical parameter threshold judgments, expanding the perception dimension from "point" to "surface," enabling the system to capture system-level anomalies that cannot be detected by single-loop monitoring from the degree of synchronous consistency of parameter changes in each loop. Based on this, when the rate of decrease of the global collaborative index exceeds a preset threshold, the system triggers preparatory protection actions before the risk assessment model outputs results, achieving a leap from "passive response" to "advance prediction," effectively reducing the time delay of fault response. In addition, the mechanism of adaptive adjustment of sampling frequency according to system risk level enables the system to operate at a lower sampling frequency under normal conditions to reduce data throughput and computational burden, and automatically increase the sampling frequency when the risk increases to obtain more refined monitoring data, thus achieving a dynamic balance between system resource consumption and monitoring accuracy requirements. 2. This invention constructs a complete intelligent closed loop of "risk assessment—tiered protection—effect feedback—model evolution". The safety risk assessment model integrates multi-source information such as single-circuit electrical parameters, cross-circuit collaborative characteristics, and temperature for comprehensive judgment. Its decision-making basis is more comprehensive and accurate than traditional single electrical quantity threshold judgment. The model is continuously updated through an online incremental learning mechanism using the actual effect feedback of each protection action, enabling the model to adapt to long-term drift factors such as equipment aging, environmental changes, and load characteristic evolution. The judgment accuracy continues to improve after long-term operation. The tiered active protection strategy executes differentiated protection actions based on the risk probability value and its rate of change. In the low-risk stage, only a warning signal is output. When the risk continues to rise, load reduction or backup switching is performed. Selective isolation is only performed when the risk reaches the highest level or load reduction is ineffective. This mechanism ensures electrical safety while maximizing the maintenance of power supply continuity, avoiding unnecessary power outages caused by the "one-touch trip" of traditional single threshold protection. After selective isolation is implemented, the global synergy index is recalculated and compared with that before isolation to quantitatively verify the isolation effect. If synergy is not restored, the next highest risk circuit is isolated, forming a closed-loop control of "decision execution - effect verification - strategy correction," effectively preventing the risk of accident escalation due to single ineffective isolation. Furthermore, the residual current joint early warning mechanism, by analyzing the correlation coefficient matrix and trend consistency of residual currents in multiple circuits, can identify the overall aging risk of the insulation system in advance when the residual current of each circuit does not exceed the leakage threshold of a single circuit. It also adaptively lowers the leakage threshold of each circuit to improve monitoring sensitivity, transforming the traditional single-circuit threshold-triggered "post-tripping" into a multi-circuit trend-based "pre-maintenance." The temperature parameter fusion and current change rate cross-verification mechanism can identify contact resistance faults through abnormal temperature rises when there is no significant change in current, effectively compensating for the blind spot of mechanical contact faults that cannot be detected by single electrical quantity monitoring. At the same time, the adaptive adjustment of temperature feature weights and the dynamic lowering of temperature thresholds enable high-risk circuits to obtain differentiated monitoring sensitivity, further improving the system's sensing accuracy and protection reliability. The synergistic effect of the aforementioned multi-level linkage mechanism enables the present invention to form a complete technical closed loop from perception, decision-making, execution to evolution, which improves electrical safety while taking into account power supply reliability and long-term system adaptability. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall system architecture of the present invention.

[0017] Figure 2 This is a schematic diagram of the overall process of the method of the present invention.

[0018] Figure 3 This is a schematic diagram of the collaborative feature extraction and preparatory protection triggering process of the present invention.

[0019] Figure 4 This is a schematic diagram of the hierarchical active protection strategy of the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figures 1 to 4 This paper presents a method for synchronous sampling and protection of multi-circuit electrical safety.

[0022] I. System Overall Architecture The active predictive electrical safety protection system based on multi-circuit synchronous sampling provided by this invention has an overall architecture that mainly includes a multi-circuit synchronous sampling unit, an electrical parameter calculation unit, a cross-circuit collaborative feature extraction unit, a safety risk assessment unit, an active protection execution unit, and a protection effect feedback unit.

[0023] Multi-circuit synchronous sampling unit for power distribution system One electrical circuit ( The voltage and current signals are sampled synchronously. Each electrical circuit can be a three-phase four-wire circuit, a three-phase three-wire circuit, or a single-phase circuit. For a three-phase circuit, the three-phase voltage (V / C / D) signals of each circuit need to be sampled synchronously. ) and three-phase current ( ), and neutral current ( For single-phase circuits, phase voltage and phase current need to be collected.

[0024] As an optional implementation, the multi-loop synchronous sampling unit can employ multiple chips. Parallel sampling architecture. This architecture consists of 3 chips. Sampling chip, main control , outward expansion Memory and voltage-to-current conversion circuit. 3 pieces The sampling chips are all connected to the voltage-to-current conversion circuit, and the main control... pass Interface and each chip Sampling chip connection, Number of interfaces and The sampling chips are the same number and connected in a one-to-one correspondence. One of them... The interface is connected in host mode, the rest The interface connects in slave mode. Each chip... The sampling chip can simultaneously sample voltage or current values ​​from eight channels. During sampling, the main controller... Through the same The interface sends commands to the three chips. The sampling chip, then through 3 The interface can receive three chips simultaneously. The data received by the sampling chip. The sampling completion pin of the sampling chip ( (feet) and interrupt pin ( (Interrupted pins) connected, when all After all sampling chips have completed sampling, they are triggered simultaneously. External interrupt, Immediately passed Host mode interface to each chip The sampling chip simultaneously sends a command to read the data. Simultaneously activate each The interface receives data from each chip. The sampling data from the sampling chip.

[0025] As another optional implementation method, for scenarios with a large number of loops (such as...) (Single) can be used This architecture is designed for multi-stage analog switching. It includes voltage loop signal conditioning circuitry, current loop signal conditioning circuitry, a two-stage switching circuit, an analog-to-digital converter, and... The two-stage switching circuit includes a first-stage switching switch and a second-stage switching switch. The first-stage switching switch is composed of... indivual select Composed of multiple analog switches, the second-stage switching switch consists of... indivual Select 1 multi-channel analog switch. Taking 96 current inputs as an example, it can be composed of... The configuration consists of 12 eight-to-one analog multiplexers forming the first stage, and 3 four-to-one analog multiplexers forming the second stage. Under the control of the CPU's channel switching control signal, the two-stage switching circuit selects 3 outputs from 96 current inputs to the analog-to-digital converter each time. The analog-to-digital converter supports multi-channel synchronous sampling, sampling multiple channel signals simultaneously after each sampling starts.

[0026] The reference value f0 for the sampling frequency is determined based on the fundamental frequency of the power grid. For a 50Hz power grid, For a 60Hz power frequency grid, First sampling frequency (That is, 3200Hz under a 50Hz power grid, 64 sampling points per cycle), second sampling frequency (128 sampling points per cycle), third sampling frequency (256 sampling points per cycle). The three sampling frequencies correspond to the three operating modes: normal monitoring, early warning monitoring, and fault monitoring.

[0027] II. Specific Implementation Methods of Synchronous Sampling The core of synchronous sampling is to ensure that all sampling channels sample the signal at the same time to guarantee the accuracy of subsequent electrical parameter calculations (especially power calculations and vector synthesis calculations).

[0028] (a) External clock synchronization mode The system passes Alternatively, the BeiDou satellite timing module receives satellite signals and generates a unified second pulse signal. (One pulse per second). Each sampling channel synchronizes its sampling time based on this second pulse signal. Specifically, the second pulse signal output by the timing module is connected to each... The chip's synchronous trigger input, all The chip initiates a sampling operation simultaneously with the rising edge of the second pulse. At the same time, the timing module transmits data via serial port to... Output a timecode containing absolute time information (compliant with IRIG-B or NMEA0183 protocol). Each sampled data is bound to an absolute time stamp to achieve data synchronization based on the absolute time stamp.

[0029] (ii) Self-synchronization mode When the external time synchronization signal is lost or abnormal (such as satellite signal obstruction, time synchronization module failure, etc.), the system automatically switches to self-synchronization mode. Self-synchronization mode uses the zero-crossing point of the voltage signal as a reference and utilizes a software phase-locked loop (PLL). The algorithm generates a synchronous sampling trigger signal.

[0030] The specific implementation of the software phase-locked loop is as follows: First, for Zero-crossing detection is performed on the phase voltage signal. When a zero-crossing point is detected where the voltage changes from negative to positive, the sampling point number at that moment is recorded. The time interval between two adjacent positive zero-crossing points is one power grid cycle. The phase-locked loop is based on the measured period. Real-time calculation of power grid frequency The frequency of the sampling trigger signal is adjusted accordingly to ensure that the sampling frequency always tracks the changes in the power grid frequency, maintaining a constant number of sampling points per cycle. When the deviation between the periodic measurements of three consecutive cycles is less than ±0.5%, the phase-locked loop (PLL) enters a locked state, at which point the system recognizes that synchronization has been established. The adjustment step size of the PLL does not exceed 1 / 10 of a sampling interval.

[0031] After the phase-locked loop is locked, each cycle is divided into equal parts, with the positive zero-crossing point as the starting time of each cycle. Each sampling interval ( The number of sampling points per cycle (64, 128, or 256 depending on the current sampling frequency level) is used to sample simultaneously through each sampling channel under the control of a unified sampling trigger signal.

[0032] (iii) Dynamic adjustment of sampling frequency The sampling frequency dynamically switches between three levels based on the current system risk level. The specific switching logic is as follows: When the risk probability value output by the safety risk assessment unit Below the first threshold At that time, the system was in a low-risk operating state and used the first sampling frequency. This mode has the smallest data volume. The computational burden is the lightest.

[0033] when achieve And lower than At that time, the system entered a medium-risk state, and the sampling frequency was increased to [missing information]. .

[0034] when achieve At that time, the system enters a high-risk state, and the sampling frequency is increased to [a higher level]. .

[0035] After the sampling frequency is increased, the safety risk assessment unit recalculates the risk probability value based on the updated high-precision sampling data. .like A decrease of more than 20% compared to before the adjustment indicates that the actual risk level is lower than the assessed value, and the sampling frequency can be gradually increased (from...). Callback to , or from Callback to ).

[0036] III. Specific Implementation Methods for Electrical Parameter Calculation The electrical parameter calculation unit receives the instantaneous voltage and current values ​​of each circuit at the same moment from the multi-circuit synchronous sampling unit, and calculates the electrical parameters of each circuit according to the following method.

[0037] (a) Effective values ​​of voltage and current For a given circuit, data is collected within one cycle. Instantaneous voltage value and Instantaneous current value RMS voltage and the effective value of current Calculate using the following formulas respectively:

[0038]

[0039] For a three-phase circuit, calculate the effective value of the voltage in each phase separately. and the effective value of each phase current .

[0040] (ii) Active power, reactive power and power factor For a given circuit, the active power Calculate using the following formula:

[0041] Apparent power

[0042] Power factor (when hour) For a three-phase circuit, the total active power Total apparent power Total power factor Reactive power according to calculate.

[0043] (iii) Harmonic distortion rate The instantaneous voltage and current value sequences obtained from the sampling are subjected to Fast Fourier Transform (FFT) to obtain the amplitude of each harmonic. The number of sampling points for FFT is an integer power of 2, which is consistent with the number of sampling points per cycle corresponding to the current sampling frequency level (64, 128, or 256). If it is less than one cycle, zeros are padded to the next integer power of 2.

[0044] The total harmonic distortion (THD) of voltage (THD_U) and the total harmonic distortion (THD_I) of current (THD_I) are calculated using the following formulas:

[0045]

[0046] in, For the first RMS value of subharmonic voltage For the first RMS value of subharmonic current This is the effective value of the fundamental voltage. This is the effective value of the fundamental current. The highest harmonic order, .

[0047] (iv) Three-phase current imbalance Three-phase current imbalance Calculate using the following formula:

[0048] in, It is the maximum value among the effective values ​​of the three-phase current. It is the minimum value among the effective values ​​of the three-phase current. This represents the average value of the three-phase current's effective value.

[0049] (v) Zero-sequence current For a three-phase circuit, the zero-sequence current Calculate using the following formula:

[0050] in These are the phasor values ​​of the three-phase currents. Based on synchronous sampling, the phase relationship of each phase current is accurately maintained, thus allowing for precise calculation of the amplitude and phase of the zero-sequence current.

[0051] (vi) Residual current Residual current (also known as leakage current) is an important parameter reflecting the insulation condition of a circuit. For a three-phase four-wire circuit, the instantaneous value of the residual current... Calculate using the following formula:

[0052] in They are respectively Mutually, Mutually, The instantaneous current values ​​of the phase and neutral lines at the same moment.

[0053] RMS value of residual current Calculate using the following formula:

[0054] For a three-phase three-wire circuit (without a neutral line), the instantaneous value of the residual current. .

[0055] When the effective value of the residual current in a certain circuit Exceeding the preset leakage current threshold of the circuit When this occurs, the circuit is marked as a leakage risk circuit.

[0056] IV. Specific Implementation Methods of Cross-Loop Collaborative Feature Extraction The cross-circuit collaborative feature extraction unit constructs a multi-circuit parameter matrix from the electrical parameters of each circuit and extracts the collaborative feature indicators between each circuit.

[0057] (I) Construction of the multi-loop parameter matrix Assume the power distribution system has a total of There are 1 loop, and each loop is at time 1. extract Several electrical parameters (including RMS current, power factor, harmonic distortion rate, three-phase current imbalance, and residual current) are used to construct a multi-loop parameter matrix. :

[0058] in For the first One loop dimensional parameter vector, , .

[0059] (ii) Calculation of the correlation coefficient matrix For any two loops and Select a certain electrical parameter (e.g., current imbalance), within the time window Internal collection Parameter values ​​of each sampling point ( (The number of sampling points corresponding to 10 to 100 cycles) are used to construct two time series:

[0060]

[0061] Pearson correlation coefficient Calculate using the following formula:

[0062] in and These are the means of the two sequences, respectively. The range of values ​​is Indicates a positive correlation. This indicates a negative correlation.

[0063] right Calculate the correlation coefficient for each loop pairwise to obtain... Correlation coefficient matrix :

[0064] in .

[0065] (III) Global Coordination Index Calculation Global synergy index It reflects the degree of synchronization and consistency of changes in electrical parameters among the various circuits of the entire power distribution system, and is calculated using the following formula:

[0066] The range of values ​​is . A decrease in the value indicates that the changes in parameters of each loop are diverging, which may indicate potential safety hazards in the system.

[0067] (iv) Preparatory protection triggering based on synergy index Global synergy index The real-time trend of changes is used to trigger preparatory protective actions. The system monitors using a sliding time window method. Changes in value. Taking the past. One consecutive sampling period ( (range 3 to 10) value sequence , The least squares method is used to analyze this. Linear fitting is performed on each point to obtain... rate of change of value .

[0068] when The absolute value exceeds the preset rate threshold. When the frequency is / cycle (adjustable range 0.01~0.05 / cycle) and the direction is negative, it indicates that the coordination between the various loops of the system is deteriorating rapidly. At this time, before the safety risk assessment result of step S4 is output, a preparatory protection action is triggered, including: (1) issuing an early warning signal; (2) pre-charging the backup loop; (3) temporarily increasing the sampling frequency of the suspected faulty loop to the highest level. .

[0069] After the selective isolation protection action is executed, the system re-executes steps S1 to S3 to calculate the global coordination index after isolation. .like (Range range: 0.80~0.95), indicating that the isolation action effectively eliminated the factors leading to the deterioration of system synergy and maintained the current protection status. If This indicates that the isolation action failed to effectively restore system synergy, and further isolation should be implemented for the loop with the second highest risk.

[0070] V. Specific Implementation Methods of the Safety Risk Assessment Model The safety risk assessment unit has a built-in pre-trained safety risk assessment model, which is used to output the risk probability value R and risk level of each circuit and system based on the single-circuit electrical parameters and cross-circuit coordination characteristic indicators.

[0071] (a) Model input feature vector The model's input feature vector It consists of the following three parts: (1) Characteristics of electrical parameters of a single circuit: the effective value of voltage, effective value of current, power factor, active power, reactive power, harmonic distortion rate (2~50th), three-phase current imbalance, zero-sequence current, and residual current of each circuit, totaling about 60 dimensions / circuit.

[0072] (2) Cross-loop synergy characteristics: global synergy index Correlation coefficient matrix between each loop The statistical characteristics (mean, variance, maximum, minimum), and vector deviation of the residual current in each circuit, totaling approximately ( )dimension.

[0073] (3) Temperature characteristics: Temperature values ​​and temperature change rates of each circuit terminal or cable sheath, totaling 2N dimensions (optional).

[0074] The total feature dimension is approximately .

[0075] (II) Model Training Methods The model is trained using a combination of offline training and online incremental learning.

[0076] Offline training phase: Historical synchronous sampling data of the power distribution system under normal and various fault conditions are collected to construct a training dataset. Fault conditions include: single-phase grounding fault, two-phase short-circuit fault, three-phase short-circuit fault, overload operation, harmonic exceedance, insulation aging (slow increase in leakage current), and abnormal contact resistance (abnormal temperature rise). The training dataset consists of: 500 normal state samples and 200 samples for each type of fault condition (7 types of faults, totaling 1400 samples), for a total of 1900 samples, meeting the baseline requirement of at least 2000 samples. A feature vector X is extracted from each sample, and the corresponding safety status label Y (normal / warning / fault / serious fault) is assigned.

[0077] K-fold cross-validation was used. or The model is trained and validated. The dataset is randomly divided into... Divide into equal parts, and take turns putting them into the same place. One set is used as the training set, and one set is used as the validation set. This is repeated. Next, the model parameters with the highest average classification accuracy were selected as the final model.

[0078] The model type can be selected from any of the following: Support Vector Machine (SVM) Random Forest Gradient boosting decision tree ( Convolutional Neural Networks (CNNs) )or For applications with high real-time requirements, random forests or gradient boosting decision trees are preferred; for applications with extremely high accuracy requirements, random forests or gradient boosting decision trees are preferred. network.

[0079] (III) Online Incremental Learning Mechanism of the Model Before each active protection action is executed, the system records the feature vector at that time. After the protective action is performed, the feedback result is obtained through the protective effect feedback in step S6. This indicates that the protective measures have successfully eliminated the risk. (This indicates that the protective measures failed to eliminate the risk). ( ) to form online training samples.

[0080] When the accumulated number of online training samples reaches the preset batch size When the value is 10~200, the incremental update of the model is triggered. The incremental update method is as follows: For the random forest model, a new decision tree is generated based on the original decision tree using new samples and added to the forest; for the gradient boosting decision tree model, the model parameters are updated iteratively using new samples; for the neural network model, 3~5 rounds of fine-tuning training are performed using new samples, with the learning rate being 1 / 10 of the initial learning rate.

[0081] After the incremental update, the model verifies the risk assessment results of subsequent sampled data in real time. If the deviation between the risk probability value output by the model and the actual protection effect feedback continues to increase, and the deviation exceeds 0.2 in 10 consecutive assessments, a full retraining is triggered, and the model is retrained from scratch using the complete dataset that combines historical data and online data.

[0082] VI. Specific Implementation Methods of the Tiered Active Protection Strategy The active protection execution unit is based on the risk probability value. and its rate of change over time Implement a tiered proactive protection strategy.

[0083] (a) Calculation of risk probability value R and threshold setting The safety risk assessment model outputs the risk probability value for each loop. and the overall system risk probability value . and The range of values ​​is The three thresholds are set as follows: (Adjustable range 0.15~0.45) (Adjustable range 0.45~0.75), R3=0.85 (adjustable range 0.70~0.95).

[0084] (ii) Risk Change Rate Calculation Take the current time and before Each sampling time ( A risk probability value sequence (ranging from 3 to 10). , Linear fitting is performed using the least squares method, and the resulting slope is... . This indicates that the risk is rising. This indicates that the risk is decreasing.

[0085] (III) Level 4 Protection Strategy according to Value and The combination of these measures will implement the following four-level protection strategy: Level 1 – Normal Monitoring: When At that time, the system maintains normal monitoring status, with a sampling frequency of [missing information]. Only runtime data is recorded.

[0086] Level Two – Warning: When and When the frequency range is 0.02~0.10 / s, the system issues an early warning signal, including local audible and visual alarms and remote notifications, and records detailed waveform data for 10 cycles before and after the warning time.

[0087] Level 3 – Load Reduction or Standby Switchover: When Sometimes, or but When this occurs, the system executes a load reduction command or switches to a standby circuit. The load reduction command reduces the load current of the faulty circuit to 60% to 80% of its rated value.

[0088] Level 4 – Selective Isolation: When At that time, or after implementing load reduction operation still At that time, the system performs selective isolation protection actions.

[0089] (iv) Specific steps of selective isolation and protection actions The first step is to determine the risk probability value for each loop. Sort all loops in descending order to obtain the risk sequence. .

[0090] The second step is to identify risks with the highest probability values ​​that exceed the third threshold. The circuit is considered a faulty circuit. If multiple circuits exceed the limit simultaneously... They are handled in descending order of risk, with only one loop isolated at a time.

[0091] The third step is to send a trip command to the trip unit of the circuit breaker corresponding to the faulty circuit, with a pulse width of 100ms~200ms.

[0092] Fourth step: After disconnecting the faulty circuit, repeat steps S1 to S4 to verify whether the remaining circuits have returned to normal operation.

[0093] VII. Specific Implementation Methods of Joint Early Warning for Residual Current The system calculates the effective value of the residual current for each circuit. Input the cross-loop collaborative feature extraction unit to calculate the correlation coefficient matrix between the residual current sequences of each loop. . Matrix, the first Line number Column elements Indicates the first loop and the first Pearson correlation coefficient of the residual current sequence in the loop.

[0094] Simultaneously, calculate the time series trend consistency index of the residual current in each circuit. For each loop, take the past... Each sampling period ( The residual current value (ranging from 6 to 20) is used to determine whether it exhibits a monotonically increasing trend using the Mann-Kendall trend test.

[0095] The system is considered to be at risk of overall aging of the insulation system when the following three conditions are met: (1) The residual current of all circuits did not exceed their respective preset leakage current thresholds. ; (2) Correlation coefficient matrix of residual current sequence of each loop In this study, the correlation coefficient between any two loops was greater than 0.7. (3) The residual current in all circuits is continuous The trend showed a monotonic increase in all sampling periods.

[0096] After determining that the insulation system is at risk of overall aging, the system automatically performs the following linkage operations: (1) uniformly lowers the leakage current threshold of each circuit to the original threshold. times ( (2) Generate preventive maintenance early warning signals and recommend that the insulation resistance of the power distribution system be tested; (3) Input the judgment results as additional features into the safety risk assessment model to improve the overall risk level of the system.

[0097] VIII. Specific Implementation Methods of Temperature Parameter Fusion In step S1, the temperature parameter fusion step synchronously acquires the temperature signals of each circuit terminal or cable sheath. The temperature sensor uses a PT100 platinum resistance temperature sensor or an NTC thermistor sensor, and is installed at the incoming and outgoing terminals of each circuit breaker or on the cable sheath. Temperature signal sampling is performed synchronously with voltage and current signal sampling.

[0098] The system calculates each loop within the time window. internal temperature change rate and rate of change of current . Set the time to 30 seconds (range 10~60 seconds).

[0099] when Exceeding the first temperature change rate threshold (Range 0.3~0.8℃ / s) and Below the first current change rate threshold When the rated current is 3% to 8% per second (range 3% to 8% per second), the circuit is determined to have abnormal contact resistance and is marked as a contact resistance risk circuit.

[0100] For circuits marked as having a contact resistance risk, the system automatically performs the following linked operations: (1) Increases the sampling frequency of the circuit to the highest level. (2) Lower the temperature warning threshold of the circuit to 80% of the normal threshold; (3) In the safety risk assessment, assign a weighting coefficient to the temperature characteristics of the circuit that is 1.5 times higher than the normal weight.

[0101] Once the ratio of the temperature change rate to the current change rate of the circuit returns to the normal range, the system automatically deactivates the aforementioned linkage operation and returns to normal monitoring status.

[0102] IX. Specific Composition and Connection Relationships of Each Unit in the System The multi-loop synchronous sampling unit includes a voltage signal conditioning circuit, a current signal conditioning circuit, a sample-and-hold circuit, an analog-to-digital converter (ADC) module, a synchronization trigger module, and a central processing unit (CPU). The voltage signal conditioning circuit isolates and transforms the voltage input signals of each loop using voltage transformers, then filters and adjusts their amplitude to match the ADC's input range (0~5V or ±10V). The current signal conditioning circuit isolates and transforms the current input signals of each loop using current transformers, then filters and adjusts their amplitude. The sample-and-hold circuit simultaneously holds the instantaneous signal values ​​of all channels under the control of the sampling trigger signal. The ADC module converts the analog signal output from the sample-and-hold circuit into a digital signal with a resolution of at least 16 bits and a sampling rate of at least 200kSPS. The synchronization trigger module generates a unified sampling trigger signal, derived from an external clock source or an internal self-synchronizing phase-locked loop (PLL). The CPU is connected to the ADC module and the synchronization trigger module.

[0103] The electrical parameter calculation unit is connected to the multi-circuit synchronous sampling unit and is implemented by the central processing unit through software algorithms.

[0104] The cross-loop collaborative feature extraction unit is connected to the electrical parameter calculation unit and is implemented by the central processing unit through software algorithms.

[0105] The safety risk assessment unit is connected to the electrical parameter calculation unit and the cross-circuit collaborative feature extraction unit, and has a built-in pre-trained safety risk assessment model.

[0106] The active protection execution unit is connected to the safety risk assessment unit and includes a graded early warning module, a load reduction control module, and a selective isolation module. The graded early warning module outputs different levels of audible and visual alarm signals or remote notifications based on the risk level. The load reduction control module outputs load reduction commands to the power regulation devices of the power distribution system. The selective isolation module outputs tripping commands to the trip units of the circuit breakers corresponding to each circuit.

[0107] The protection effect feedback unit is connected to the active protection execution unit and the multi-loop synchronous sampling unit.

[0108] The human-machine interface unit includes an LCD screen and an input keyboard, which are used to display the electrical parameters, coordination characteristics, risk probability values ​​and risk levels of each circuit in real time, and to receive user configuration input for protection strategy thresholds.

[0109] The communication unit includes an RS485 interface (supporting Modbus-RTU protocol), an Ethernet interface (supporting Modbus-TCP or IEC 61850 protocol), and a 4G / 5G wireless communication module, which are used to upload data to the cloud server or the upper-level monitoring system, and receive model update parameters or threshold configuration instructions issued by the cloud server.

[0110] 10. Examples This embodiment uses a power distribution system in an industrial park as an example to provide a detailed description of the method and system provided by the present invention.

[0111] The industrial park's power distribution system is powered by a 1600kVA dry-type transformer (10kV / 0.4kV), with 12 feeder circuits branching off from the 0.4kV busbar on the low-voltage side. The circuits consist of: 4 power circuits (rated current 400A), 4 lighting circuits (rated current 200A), 2 air conditioning circuits (rated current 315A), and 2 standby circuits (rated current 400A). Each circuit is equipped with a smart circuit breaker.

[0112] The system hardware configuration is as follows: The multi-loop synchronous sampling unit uses four ADS131M08 8-channel synchronous sampling ADC chips; the main control MCU uses an STM32H743 Cortex-M7 processor with a main frequency of 400MHz. The synchronous trigger module uses a UBLOXNEO-M8N GPS / BeiDou dual-mode timing module. The temperature sensors use PT100 platinum resistance temperature sensors, two per loop, for a total of 24. The human-machine interface unit uses a 7-inch TFT color touchscreen. The communication unit is equipped with an RS485 interface and a 4G wireless communication module.

[0113] After the system powers on, it executes the initialization process: the MCU initializes all peripherals; the GPS / BeiDou timing module acquires a valid PPS output; each ADC chip is configured to continuous sampling mode, and the sampling frequency is initially set to... The security risk assessment model is loaded into the MCU's internal Flash memory. The model uses the random forest algorithm and contains 200 decision trees.

[0114] Under normal system conditions, the total load current of the 12 circuits is approximately 1800A, and the transformer load rate is approximately 75%. The electrical parameters of each circuit are within the normal range, and the global coordination index C remains stable between 0.85 and 0.92.

[0115] At 14:23:17 on a certain day, the system detected the following anomalies: (1) The current imbalance of the No. 2 power circuit gradually increased from the normal 2.3% to 8.7%; (2) The harmonic distortion rate of the No. 5 lighting circuit increased from the normal 3.1% to 9.8%; (3) The global synergy index C continuously decreased from 0.89, dropping to 0.76 at 14:23:17. The C value sequence of the past 5 cycles is as follows: Fitting slope / cycle, exceeding cycle.

[0116] Because the rate of decrease of the C value exceeds the threshold, the system triggers preparatory protection actions: (1) issuing an early warning signal; (2) reducing the sampling frequency of power circuit No. 2 and lighting circuit No. 5 from Upgraded to (3) Precharge the backup circuit 1.

[0117] Subsequently, the safety risk assessment unit recalculated the risk probability values ​​of each loop based on the updated high-precision sampling data: Power Loop No. 2 Lighting circuit No. 5 The remaining loops all have a voltage rating below 0.2, indicating an overall system risk. .because ,but near and The system executes the third level of protection—sending a load reduction command to the inverter of power circuit No. 2 to reduce the load current from 85% to 65%.

[0118] Thirty seconds after the load reduction command was executed (14:23:47), monitoring data showed that the current imbalance of power circuit No. 2 continued to rise from 8.7% to 12.3%; the terminal temperature rose from 42℃ to 58℃ (temperature rise rate). ); Rate of change of current lower than The ratio of temperature change rate to current change rate is significantly abnormal, indicating abnormal contact resistance. It rose from 0.52 to 0.78. It still surpasses V1.

[0119] Due to the implementation of load reduction Still more The system executes the fourth level of protection – selective isolation protection action: after prioritizing the risks of each circuit, it sends a tripping command to the intelligent circuit breaker of power circuit 2 with a pulse width of 150ms; the circuit breaker successfully trips; at the same time, it sends a closing command to backup circuit 1 with a switching time of less than 100ms.

[0120] After isolation was completed, the system re-executed steps S1 to S4: Power circuit No. 2 was disconnected; the electrical parameters of the remaining 11 circuits returned to normal; the global coordination index... The recalculation result is 0.88; The risk probability values ​​for the remaining loops are all below 0.2, and the system has returned to normal monitoring status.

[0121] Subsequently, maintenance personnel inspected power circuit 2 and found that the screws on the B-phase terminals of the circuit breaker were loose, causing excessive contact resistance and resulting in localized overheating. Additionally, there was poor contact in the B-phase current transformer wiring. After tightening the screws and rewiring, power circuit 2 returned to normal. After confirming the fault was resolved via the human-machine interface, the system switched the load from backup circuit 1 back to power circuit 2.

[0122] The entire process, from the occurrence of the anomaly to the completion of isolation, took approximately 40 seconds, with the time from determining isolation to completing the switchover being less than 200ms. The feature vectors collected during this event... Feedback results on protective effectiveness The online training samples are compiled and added to the online sample library for subsequent incremental updates of the model.

[0123] 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. A method for synchronous sampling and protection of multi-circuit electrical safety, comprising the steps of synchronously sampling multiple electrical circuits in a power distribution system and calculating the electrical parameters of each circuit, characterized in that, It also includes the following steps: S1: Using a unified sampling clock and sampling frequency, the power distribution system... The voltage and current signals of each electrical circuit are sampled synchronously to obtain the instantaneous voltage and current values ​​of each circuit at the same moment. ; The sampling frequency is dynamically adjusted according to the current system risk level. When the risk level is lower than the first preset level, the first sampling frequency is used. When the risk level reaches the first preset level, the sampling frequency will switch to the second sampling frequency. , , Not less than 64 times the fundamental frequency of the power grid. Not less than 128 times the fundamental frequency of the power grid; S2: Based on the synchronous sampling data of each circuit obtained in step S1, calculate the electrical parameters of each circuit. The electrical parameters include the effective value of voltage, the effective value of current, the power factor, active power, reactive power, harmonic distortion rate, three-phase current imbalance, zero-sequence current, and residual current. S3: Construct a multi-circuit parameter matrix from the electrical parameters of each circuit calculated in step S2, and extract the coordination characteristic indicators between each circuit. These indicators include the correlation coefficient of the synchronous change trend of the current imbalance of each circuit, the phase correlation of the harmonic distortion rate of each circuit, and the deviation between the vector sum and each component of the residual current of each circuit. Calculate the global coordination index based on these indicators. , ,in For the first loop and the first Pearson correlation coefficient of the loop parameter sequence; S4: Input the single-circuit electrical parameters obtained in step S2 and the cross-circuit coordination characteristic indicators obtained in step S3 into the safety risk assessment model, and output the risk probability values ​​of each circuit and the power distribution system. And risk level; the security risk assessment model is a classification model pre-trained based on machine learning; S5: Based on the risk probability value output in step S4 and its rate of change over time Implement a tiered proactive protection strategy: when Below the first threshold At the same time, maintain normal monitoring status; when achieve And below the second threshold ,and Below the first rate threshold In such cases, an early warning signal is issued and abnormal data is recorded; when achieve And below the third threshold ,or Below but Exceed When necessary, execute the load reduction operation command or switch to the backup circuit; when achieve Or after implementing load reduction operation Still more At that time, selective isolation protection actions are executed to cut off the faulty circuit with the highest risk; S6: After the active protection action in step S5 is executed, continue executing steps S1 to S4, monitoring the changes in electrical parameters and coordination characteristic indicators of each circuit after the protection action. If the risk probability value... Not down to The protection strategy will then be upgraded step by step.

2. The method according to claim 1, characterized in that, The synchronous sampling described in step S1 is achieved in the following way: A unified second pulse signal is generated by a GPS or BeiDou satellite timing module, and each sampling channel synchronizes its sampling time based on the second pulse signal. When the external time synchronization signal is lost or abnormal, it automatically switches to self-synchronization mode, using the zero-crossing point of the voltage signal as a reference, and uses a software phase-locked loop algorithm to generate a synchronization sampling trigger signal to maintain the sampling synchronization of each channel.

3. The method according to claim 1, characterized in that, The specific method for dynamically adjusting the sampling frequency according to the current system risk level in step S1 is as follows: When the risk probability value Below At that time, the first sampling frequency is used. ,in It is the fundamental frequency of the power grid; when achieve And lower than At that time, the sampling frequency was increased to ; when achieve At that time, the sampling frequency was increased to ; After the sampling frequency is adjusted, step S4 recalculates the risk probability value based on the updated high-precision sampling data. ,like If the sampling frequency decreases by more than a preset percentage compared to before the adjustment, the sampling frequency will be gradually reduced.

4. The method according to claim 1, characterized in that, The global synergy index mentioned in step S3 It is also used to trigger preparatory protection: When the global synergy index In continuous The rate of decrease within a sampling period exceeds a preset rate threshold. At that time, among them Before the risk assessment results are output in step S4, the preparatory protection action in step S5 is triggered. The preparatory protection action includes issuing an early warning signal, pre-charging the backup circuit, and increasing the sampling density of the faulty circuit. After performing selective isolation protection in step S5, steps S1 to S3 are re-executed to calculate the global coordination index after isolation. ,like ,in The preset recovery coefficient and If so, then maintain the current protection status; if Then, further isolation is performed on the loops with the second highest risk.

5. The method according to claim 1, characterized in that, The security risk assessment model described in step S4 employs an online incremental learning mechanism: The feature vector before each active protection action in step S5 is executed. Feedback results on the protective effect of step S6 Composition of online training samples ,in This indicates whether the protective measures successfully eliminated the risk; When the accumulated number of online training samples reaches the preset batch size When this occurs, an incremental update of the security risk assessment model is triggered, and the model parameters are fine-tuned using new samples. ; The incrementally updated model verifies the risk assessment results of subsequent sampled data in real time. If the deviation between the risk probability value output by the model and the actual protection effect feedback in step S6 continues to increase beyond the preset deviation threshold, full retraining is triggered.

6. The method according to claim 1, characterized in that, The selective isolation protection action described in step S5 specifically includes: Based on the risk probability values ​​of each loop output in step S4, the loops are ranked by risk. The probability value of identifying the risk is the highest and exceeds the third threshold. The circuit is used as a fault circuit; Send a trip command to the trip unit of the circuit breaker corresponding to the faulty circuit to disconnect the faulty circuit; After disconnecting the faulty circuit, repeat steps S1 to S4 to verify whether the remaining circuits have returned to normal operation.

7. The method according to claim 1, characterized in that, The residual current mentioned in step S2 is calculated in the following way: The instantaneous residual current value of the circuit is obtained by summing the instantaneous values ​​of the three-phase current sampled synchronously from each circuit: ; The effective value of the instantaneous residual current is calculated to obtain the effective value of the residual current; The effective value of the residual current obtained from each circuit Input step S3 to calculate the correlation coefficient matrix between the residual current sequences of each loop. Consistency index with time series trends ; When the residual current of all circuits does not exceed their respective preset leakage current thresholds And the correlation coefficient matrix of the residual current sequence of each circuit The correlation coefficient between any two loops is greater than 0.7, and the residual current in each loop is continuously increasing. When the trend shows a monotonically increasing trend in each sampling period The system was deemed to be at risk of overall aging of the insulation system, and the leakage current threshold of each circuit was uniformly lowered to the original threshold. times, of which It also generates preventative maintenance early warning signals.

8. The method according to claim 1, characterized in that, It also includes a temperature parameter fusion step: In step S1, the temperature signals of each circuit terminal or cable sheath are collected synchronously. ; Calculate each loop within the time window internal temperature change rate and rate of change of current ; when Exceeding the first temperature change threshold and When the current change rate is below the first threshold, the circuit is determined to have abnormal contact resistance and is marked as a contact resistance risk circuit. For circuits marked as having contact resistance risk, the sampling frequency of the circuit is automatically increased to the highest level, the temperature warning threshold of the circuit is lowered to 80% of the normal threshold, and the temperature characteristics of the circuit are assigned a weighting coefficient 1.5 times higher than the normal weight in the risk assessment in step S4. Once the ratio of the temperature change rate to the current change rate of the circuit returns to the normal range, the above operation will be automatically terminated.

9. An active predictive electrical safety protection system based on multi-circuit synchronous sampling, characterized in that, include: A multi-loop synchronous sampling unit is used to synchronously sample the voltage and current signals of N electrical circuits in a power distribution system with a unified sampling clock and sampling frequency, and output the digital sampled values ​​of each circuit at the same time. The multi-loop synchronous sampling unit includes a voltage signal conditioning circuit, a current signal conditioning circuit, a sample-and-hold circuit, an analog-to-digital conversion module, a synchronous triggering module, and a central processing unit. An electrical parameter calculation unit, connected to the multi-circuit synchronous sampling unit, is used to calculate the electrical parameters of each circuit based on the digital sampling values. The electrical parameters include the effective value of voltage, the effective value of current, power factor, active power, reactive power, harmonic distortion rate, three-phase current imbalance, zero-sequence current, and residual current. A cross-circuit collaborative feature extraction unit, connected to the electrical parameter calculation unit, is used to construct a multi-circuit parameter matrix based on the electrical parameters of each circuit, extract collaborative feature indicators between circuits, and calculate a global collaborative index. ; The safety risk assessment unit is connected to the electrical parameter calculation unit and the cross-circuit collaborative feature extraction unit. It has a built-in pre-trained safety risk assessment model, which is used to output the risk probability value R and risk level of each circuit and system based on the single-circuit electrical parameters and cross-circuit collaborative feature indicators. The active protection execution unit, connected to the security risk assessment unit, is used to perform actions based on the risk probability value. and its rate of change over time Implement a tiered active protection strategy, including early warning output, load reduction control command output, and selective isolation protection command output; The protection effect feedback unit is connected to the active protection execution unit and the multi-loop synchronous sampling unit. It is used to monitor the changes in electrical parameters and coordination characteristics of each loop after the active protection action is executed, and to decide whether to upgrade the protection strategy based on the monitoring results.

10. The system according to claim 9, characterized in that, The active protection execution unit includes: The graded early warning module is used to output different levels of audible and visual alarm signals or remote notifications according to the risk level; The load reduction control module is used to output load reduction commands to the power regulation device of the power distribution system; The selective isolation module is used to output tripping commands to the trip units of the circuit breakers corresponding to each circuit, thereby achieving selective disconnection of a single circuit. The graded early warning module, load reduction control module, and selective isolation module are activated sequentially or in combination according to the risk level. The system also includes a human-machine interaction unit and a communication unit. The human-machine interaction unit is used to display the electrical parameters, coordination characteristic indicators, risk probability values ​​and risk levels of each circuit in real time, and to receive user configuration input for protection strategy thresholds. The communication unit is used to upload the electrical parameters, coordination characteristic indicators, risk probability values ​​and risk levels of each circuit to a cloud server or a higher-level monitoring system, and to receive model update parameters or threshold configuration instructions issued by the cloud server.

Images