A method for controlling a ring main unit

By acquiring voltage and wall current event sequences, calculating residual conductivity timing and return voltage hazard index, and combining space charge relaxation time proxy, a barrier function is constructed to generate a minimum energy control command. This solves the problem of insufficient reliability in risk assessment during the switching process of ring main unit, and reduces arc reignition rate and equipment wear.

CN123121980APending Publication Date: 2026-11-17CEEC JIANGSU ELECTRIC POWER DESIGN INST CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611240355.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-17
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In the current ring main unit switchgear, the reliability of risk assessment is insufficient, and it cannot adaptively output the optimal energy control according to the instantaneous degree of danger, resulting in a high arc reignition rate and severe equipment wear.

Method used

By acquiring voltage and wall current event sequences, calculating residual conductivity timing and return voltage hazard index, and combining space charge relaxation time surrogate quantity, a barrier function is constructed to generate a minimum energy control command, thereby enabling the assessment and intervention of transient risks during interruption.

Benefits of technology

It effectively reduces the arc reignition rate, reduces the control energy consumption and equipment wear of the ring main unit, and improves the safety and reliability of the switchgear.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123121980A_ABST
    Figure CN123121980A_ABST
Patent Text Reader

Abstract

This invention discloses a ring main unit (RNB) control method in the field of power grid equipment control technology. The method includes: acquiring voltage event sequences and wall current event sequences during the opening and closing process of the target RNB; obtaining residual conductivity timing and return voltage hazard index based on the voltage event sequences; obtaining space charge relaxation time surrogate quantity based on the wall current event sequences; correcting the return voltage hazard index based on the space charge relaxation time surrogate quantity to obtain a standard hazard index; constructing a barrier function based on the standard hazard index, residual conductivity timing, and space charge relaxation time surrogate quantity; generating a minimum energy control command based on the barrier function and inequality constraints; and controlling the actuator of the target RNB through the minimum energy control command. This invention achieves assessment and intervention of transient risks during opening and closing through dual-path disturbance-free observation and energy shaping control with safety certificates, effectively reducing the arc reignition rate, while also reducing control energy consumption and equipment wear.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to a ring main unit control method, belonging to the field of power grid equipment control technology. Background Technology

[0002] As a critical node device in medium-voltage power distribution networks, the reliability of ring main units (RMMs) directly affects the safety, stability, and power supply quality of the regional power grid. During grid load switching and fault clearing operations, the switching equipment within the RMM needs to frequently interrupt current. Within tens to hundreds of microseconds after the current crosses zero, the switch gap is in an extremely vulnerable critical window. During this period, on the one hand, a rapidly recovering transient voltage (TRV) continuously impacts the gap; on the other hand, the high-temperature plasma remaining from the arc combustion has not yet completely deionized, and the insulation strength of the gap has not yet recovered to its rated level. If the application rate or amplitude of the transient recovery voltage exceeds the recovery rate of the gap's dielectric insulation strength, arc reignition will occur. This can lead to shortened switch life and equipment damage, or even trigger cascading faults such as phase-to-phase short circuits and system oscillations, posing a serious threat to the safety of the power system.

[0003] Currently, various technical solutions have been developed for the control and protection of ring main unit switches during the switching process. At the condition monitoring level, a common method is to use a high-speed analog-to-digital converter to continuously sample the break voltage at the MHz level, and then calculate indicators such as the voltage change rate to determine the risk of reignition. Another method focuses on monitoring the physical state of the switchgear itself and indirectly assessing the insulation margin of the gap by combining empirical thresholds or static compensation models. At the control execution level, traditional control strategies are mostly preset, open-loop modes, while some solutions attempt to link monitoring and control.

[0004] However, existing solutions struggle to penetrate superficial phenomena like voltage waveforms at the transient risk observation level, lacking unobstructed, causal observation of intrinsic physical states such as residual conductivity that determine whether reignition occurs. Due to over-reliance on a single voltage measurement path, their assessment results are highly susceptible to equipment parameter drift, and the lack of independent physical channels for cross-validation and calibration leads to insufficient reliability in risk assessment. This limitation in observation directly results in blind control strategies, failing to adaptively output energy-optimal control actions based on the instantaneous level of danger, and generally exhibiting energy redundancy or insufficient control power. Summary of the Invention

[0005] This invention provides a ring main unit control method that solves the defects of insufficient reliability of risk assessment during the switching process of ring main unit and the inability to adaptively output optimal energy control according to the instantaneous degree of danger.

[0006] To achieve the above objectives, the present invention is implemented using the following technical solution.

[0007] This invention proposes a ring main unit control method, comprising:

[0008] Obtain the voltage event sequence and wall current event sequence during the opening and closing process of the target ring main unit;

[0009] The residual conductivity timing and return voltage hazard index are obtained based on the voltage event sequence;

[0010] The space charge relaxation time surrogate quantity is obtained based on the wall current event sequence;

[0011] Based on the space charge relaxation time proxy, the return voltage hazard index is corrected to obtain a standard hazard index;

[0012] A barrier function is constructed based on the standard hazard index, the residual conductivity timing, and the space charge relaxation time surrogate quantity. A minimum energy control command is generated based on the barrier function, and the actuator of the target ring main unit is controlled by the minimum energy control command.

[0013] Further, obtaining the residual conductance timing and return voltage hazard index based on the voltage event sequence includes:

[0014] Local least squares fitting is performed on the time difference of the out-of-bounds voltage divider thresholds contained in the voltage event sequence to obtain the gap voltage slope sequence and the gap voltage sequence;

[0015] Based on the physical equation of the return voltage, the gap voltage slope sequence, and the gap voltage sequence, the residual conductance timing is recursively obtained;

[0016] Based on the residual conductance timing sequence and the gap voltage sequence, the return voltage hazard index is calculated; the formula for calculating the return voltage hazard index is:

[0017] ;

[0018] in, To return the voltage hazard index, Sampling time, For the integration time window, for The residual conductance timing at time 10:00. for Time-gap voltage sequence.

[0019] Furthermore, the recursive derivation of the residual conductance timing based on the return voltage physical equation, the gap voltage slope sequence, and the gap voltage sequence includes:

[0020] Based on the gap voltage slope sequence and the gap voltage sequence, a residual sequence based on the returned voltage physical equation is constructed;

[0021] Establish a constraint set that includes nonnegativity constraints and monotonic decay constraints, so that the residual sequence satisfies the constraint set while being minimized, and obtain the residual conductance time series.

[0022] Furthermore, the formula for the physical equation of the returned voltage is:

[0023] ;

[0024] in, This is the gap equivalent parasitic capacitance. Sampling time The slope of the gap voltage, The equivalent source voltage of the upstream power grid. The equivalent source resistance of the upstream power grid, Sampling time The residual conductance timing;

[0025] The formula for the residual sequence is:

[0026] ;

[0027] in, For discrete sampling point index, For the first Residual values ​​of discrete sampling points The slope of the gap voltage is in the th... The value of each discrete sampling point For the first Residual conductivity at each discrete sampling point For the gap voltage sequence in the first The values ​​of each discrete sampling point.

[0028] Further, obtaining the space charge relaxation time surrogate quantity based on the wall current event sequence includes:

[0029] Extract wall current pulse segments from the wall current event sequence;

[0030] The space charge relaxation time surrogate quantity is obtained by fitting the exponential decay characteristics of the wall current pulse segment.

[0031] Furthermore, the standard hazard index is obtained by correcting the return voltage hazard index based on the space charge relaxation time proxy, including:

[0032] Based on a preset mapping relationship, the charge relaxation time proxy is mapped to a consistency calibration factor.

[0033] The return voltage hazard index is scaled and corrected using the consistency calibration factor to obtain a standard hazard index.

[0034] Further, the step of constructing a barrier function based on the standard hazard index, the residual conductivity timing, and the space charge relaxation time surrogate, and generating a minimum energy control command based on the barrier function, includes:

[0035] A state vector is constructed based on the standard hazard index, the residual conductivity timing, the space charge relaxation time proxy, and the pre-acquired position signal of the target ring main unit actuator.

[0036] A barrier function is constructed based on the state vector, and the barrier function value and the time derivative estimate of the barrier function are calculated; the formula for the barrier function is:

[0037] ;

[0038] in, This is the value of the barrier function; , , , These are the first weighting coefficient, the second weighting coefficient, the third weighting coefficient, and the fourth weighting coefficient, respectively. , , , All are positive numbers; As a standard hazard indicator, For residual conductance timing, This is a space charge relaxation time surrogate quantity. The preset safety threshold; The indicator function is set to 1 when the position signal of the actuator is in an intermediate transient state, and 0 otherwise.

[0039] Based on the barrier function value and the time derivative of the barrier function, the minimum energy control command of the target ring main unit is determined through preset inequality constraints.

[0040] The inequality constraint is:

[0041] ;

[0042] in, For the estimation of the time derivative of the barrier function, The convergence coefficient is positive.

[0043] Furthermore, the energy minimization command includes coil current timing shaping, and the determination process of the coil current timing shaping includes:

[0044] Based on the state vector, the sensitivity of the barrier function to the coil current is estimated;

[0045] Based on the barrier function value, the time derivative estimate of the barrier function, and the sensitivity, the minimum current increment that satisfies the inequality constraint is determined.

[0046] The minimum current increment is converted into coil current timing shaping to improve the tripping speed of the target ring main unit or suppress mechanical backflow.

[0047] Furthermore, the energy minimization command also includes a dynamic absorption branch gating command, the determination process of which includes:

[0048] When the barrier function value approaches the preset safety domain boundary, a reignition signature is detected from the voltage event sequence; the reignition signature refers to the reignition precursor characteristics indicated by the electrical parameters of the ring main unit, and the safety domain is defined as satisfying the barrier function value. The state region ≤0;

[0049] When the reignition signature is detected, a dynamic absorption branch gating command is generated based on the reignition signature and the preset energy budget constraint to suppress the return voltage peak of the target ring main unit and absorb energy.

[0050] Furthermore, the energy minimization command also includes a fast secondary break trigger command, the determination process of which includes:

[0051] When the barrier function value remains at or near the preset safety domain boundary, the voltage event sequence of the target ring main unit is continuously monitored; the safety domain is defined as satisfying the barrier function value. The state region ≤0;

[0052] When a reignition mode is identified from the voltage event sequence, a rapid secondary disconnection trigger command is generated to enable the actuator of the target ring main unit to perform a tripping, closing, and tripping operation sequentially at the fastest speed. The reignition mode refers to one or more of the following characteristics: a sharp reversal of the return voltage slope, an unexpected rebound of the main circuit current, and a short-term local collapse of the gap voltage. The short-term local collapse of the gap voltage refers to the phenomenon that the recovering voltage suddenly drops rapidly and slightly after reaching a preset voltage value, and then recovers and rises again.

[0053] The beneficial effects achieved by this invention are as follows:

[0054] (1) By introducing the residual conductivity timing and return voltage hazard index, instead of injecting excitation signals into the main gap of the ring network cabinet, the return voltage process that occurs naturally after the switch is turned off is used for passive observation and estimation, thus avoiding the disturbance of the residual plasma state in the main gap by external excitation and ensuring the non-destructive nature and causal closure of the observation.

[0055] (2) Introduce the space charge relaxation time surrogate quantity and correct the return voltage danger index based on the space charge relaxation time surrogate quantity to prevent the parasitic parameter drift problem existing in a single observation path and reduce systematic estimation bias.

[0056] (3) By constructing barrier functions and inequality constraints to generate minimum energy control commands, the complex safety problem of switching equipment is transformed into a constraint optimization problem, which realizes the assessment and intervention of the transient risk of interruption, thereby effectively reducing the arc reignition rate and reducing the control energy consumption and equipment wear of the ring main unit. Attached Figure Description

[0057] Figure 1 A flowchart of a ring main unit control method provided by the present invention. Detailed Implementation

[0058] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0059] Example 1, as Figure 1 As shown, this invention discloses a ring main unit control method, comprising:

[0060] Obtain the voltage event sequence and wall current event sequence during the opening and closing process of the target ring main unit;

[0061] The residual conductivity timing and return voltage hazard index are obtained based on the voltage event sequence;

[0062] The space charge relaxation time surrogate quantity is obtained based on the wall current event sequence;

[0063] Based on the space charge relaxation time proxy, the return voltage hazard index is corrected to obtain a standard hazard index;

[0064] A barrier function is constructed based on the standard hazard index, the residual conductivity timing, and the space charge relaxation time surrogate quantity. A minimum energy control command is generated based on the barrier function, and the actuator of the target ring main unit is controlled by the minimum energy control command.

[0065] (1) The voltage event sequence is acquired using an event-driven method. Multiple voltage comparators with different amplitudes are set on the secondary side of the voltage transformer or voltage divider of the ring main unit, and the outputs of each voltage comparator are connected to a high-precision time-to-digital converter (TDC). After the ring main unit is disconnected, when the transient recovery voltage rises rapidly in the gap and crosses the preset multi-level voltage thresholds in sequence, the logic level of each level comparator will flip. The high-precision time-to-digital converter captures and records the absolute timestamp of each flip event, forming an original voltage event sequence consisting of data pairs including timestamps and corresponding voltage levels. Subsequently, multi-threshold consistency verification, nonlinear median filtering, and high-precision time reference alignment based on PTP or IRIG-B codes are performed on the original voltage event sequence to eliminate abnormal data and suppress single-point random jitter, and finally output a standardized voltage event sequence with a highly unified time axis. Compared to continuous sampling schemes that rely on MHz-level high-speed analog-to-digital converters (ADCs), the method adopted in this embodiment efficiently captures the slope of the voltage waveform using an event-driven approach. This reduces the system's data processing burden, hardware costs, and sensitivity to severe electromagnetic interference (EMI) environments, thereby improving the overall feasibility and operational reliability of the solution.

[0066] The acquisition of wall current event sequences utilizes the weak displacement current generated during the space charge dissipation process. Rogowski coils or high-frequency current sensors (CTs) are installed on the insulating wall or grounded outer shell of the ring main unit switch chamber to non-invasively couple and induce the dielectric induced current caused by the dramatic change in electric field during the switching process. Within the acquisition window, the system uses the natural zero-crossing point of the main circuit current or the steep rising edge of the return voltage as the trigger anchor point. First, the original wall current signal undergoes baseline correction and detrending processing to eliminate the influence of temperature drift. Then, based on the signal's energy density characteristics and pulse interval, the waveform is divided into multiple independent wall current pulse segments. These pulse segments themselves constitute the wall current event sequence. In subsequent processing, the algorithm performs exponential decay nonlinear fitting on the decay tail of each pulse segment, thereby further extracting the space charge relaxation time surrogate, providing an independent data basis for physical consistency calibration with the voltage observation path.

[0067] (2) Based on the voltage event sequence, obtain the residual conductance timing and return voltage hazard index, including:

[0068] Local least squares fitting is performed on the time difference of the out-of-bounds voltage divider thresholds contained in the voltage event sequence to obtain the gap voltage slope sequence and the gap voltage sequence;

[0069] Based on the physical equation of the return voltage, the gap voltage slope sequence, and the gap voltage sequence, the residual conductance timing is recursively obtained;

[0070] Based on the residual conductance timing sequence and the gap voltage sequence, the return voltage hazard index is calculated; the formula for calculating the return voltage hazard index is:

[0071] ;

[0072] in, To return the voltage hazard index, Sampling time, For the integration time window, for The residual conductance timing at time 10:00. for Time-gap voltage sequence.

[0073] The return voltage hazard index is an energy-weighted index used to quantify the risk of reignition energy in the break gap of a ring main unit. This step uses the return voltage process that occurs naturally after the switch is turned off for passive observation and estimation, avoiding external excitation from disturbing the residual plasma state in the main gap, ensuring the non-destructive nature and causal closure of the observation, and thus more realistically reflecting the inherent deionization characteristics of the gap.

[0074] Based on the physical equation of the return voltage, the gap voltage slope sequence, and the gap voltage sequence, the residual conductance timing is recursively obtained, including:

[0075] Based on the gap voltage slope sequence and the gap voltage sequence, a residual sequence based on the returned voltage physical equation is constructed;

[0076] Establish a constraint set that includes nonnegativity constraints and monotonic decay constraints, so that the residual sequence satisfies the constraint set while being minimized, and obtain the residual conductance time series.

[0077] In the above steps, the formula for the returned voltage physical equation is:

[0078] ;

[0079] in, This is the gap equivalent parasitic capacitance. Sampling time The slope of the gap voltage, The equivalent source voltage of the upstream power grid. The equivalent source resistance of the upstream power grid, Sampling time The residual conductance timing;

[0080] The formula for the residual sequence is:

[0081] ;

[0082] in, For discrete sampling point index, For the first Residual values ​​of discrete sampling points The slope of the gap voltage is in the th... The value of each discrete sampling point For the first Residual conductivity at each discrete sampling point For the gap voltage sequence in the first The values ​​of each discrete sampling point.

[0083] (3) Based on the wall current event sequence, the space charge relaxation time surrogate quantity is obtained, including:

[0084] Extract wall current pulse segments from the wall current event sequence;

[0085] The exponential decay characteristics of the wall current pulse segment are fitted to obtain the space charge relaxation time surrogate quantity, which indirectly reflects the speed of gap medium recovery.

[0086] (4) Based on the space charge relaxation time surrogate quantity, the return voltage hazard index is corrected to obtain a standard hazard index, including:

[0087] Based on a preset mapping relationship, the charge relaxation time proxy quantity is mapped to a consistency calibration factor; the mapping relationship can be based on the factory specification or the field calibration curve.

[0088] The return voltage hazard index is scaled and corrected using the consistency calibration factor to obtain a standard hazard index.

[0089] Because a single observation path may introduce systematic estimation bias due to drift in parasitic parameters caused by factors such as equipment aging and changes in environmental temperature and humidity, this method suppresses such bias through dual-path physical consistency calibration, resulting in reliable hazard assessment results.

[0090] (5) Constructing a barrier function based on the standard hazard index, the residual conductivity timing, and the space charge relaxation time surrogate, and generating a minimum energy control command based on the barrier function, including:

[0091] A state vector is constructed based on the standard hazard index, the residual conductivity timing, the space charge relaxation time proxy, and the pre-acquired position signal of the target ring main unit actuator.

[0092] A barrier function is constructed based on the state vector, and the barrier function value and the time derivative estimate of the barrier function are calculated; the formula for the barrier function is:

[0093] ;

[0094] in, This is the value of the barrier function; , , , These are the first weighting coefficient, the second weighting coefficient, the third weighting coefficient, and the fourth weighting coefficient, respectively. , , , All are positive numbers; As a standard hazard indicator, For residual conductance timing, This is a space charge relaxation time surrogate quantity. The preset safety threshold; The indicator function is set to 1 when the position signal of the actuator is in an intermediate transient state, and 0 otherwise.

[0095] The energy shaping control method based on barrier function (BF-LES) determines the minimum energy control command of the target ring main unit based on the barrier function value and the time derivative of the barrier function, through preset inequality constraints.

[0096] The inequality constraint is:

[0097] ;

[0098] in, For the estimation of the time derivative of the barrier function, The convergence coefficient is positive.

[0099] 1) The energy minimization command includes coil current timing shaping, and the determination process of the coil current timing shaping includes:

[0100] Based on the state vector, the sensitivity of the barrier function to the coil current is estimated;

[0101] Based on the barrier function value, the time derivative estimate of the barrier function, and the sensitivity, the minimum current increment that satisfies the inequality constraint is determined.

[0102] The minimum current increment is converted into coil current timing shaping to improve the tripping speed of the target ring main unit or suppress mechanical backflow.

[0103] 2) The energy minimization command further includes a dynamic absorption branch gating command, the determination process of which includes:

[0104] When the barrier function value approaches the preset safety domain boundary, a reignition signature is detected from the voltage event sequence; the reignition signature refers to the reignition precursor characteristics indicated by the electrical parameters of the ring main unit.

[0105] When the reignition signature is detected, a dynamic absorption branch gating command is generated based on the reignition signature and the preset energy budget constraint to suppress the return voltage peak of the target ring main unit and absorb energy.

[0106] 3) The energy minimization command further includes a fast secondary break trigger command, and the process for determining the fast secondary break trigger command includes:

[0107] When the barrier function value remains at or near the preset safety domain boundary, the voltage event sequence of the target ring main unit is continuously monitored;

[0108] When a reignition mode is identified from the voltage event sequence, a rapid secondary disconnection trigger command is generated to enable the actuator of the target ring main unit to perform a tripping, closing, and tripping operation sequentially at the fastest speed. The reignition mode refers to one or more of the following characteristics: a sharp reversal of the return voltage slope, an unexpected rebound of the main circuit current, and a short-term local collapse of the gap voltage. The short-term local collapse of the gap voltage refers to the phenomenon that the recovering voltage suddenly drops rapidly and slightly after reaching a preset voltage value, and then recovers and rises again.

[0109] Example 2, based on the same inventive concept as Example 1, provides a ring main unit control method, the specific steps of which are as follows.

[0110] Step 1: Measure the time stamps of multiple voltage thresholds using a time-to-digital converter (TDC) to generate a voltage event sequence. Then, use a Rogowski coil or a high-frequency current sensor (CT) to couple and induce the dielectric induced current caused by the dramatic change in the electric field during the switching process to generate a wall current event sequence.

[0111] A unified time axis alignment is performed on the voltage event sequence and the wall current event sequence. The moment when the current naturally crosses zero or the controller issues an interruption command is used as the global time anchor point. Through time base calibration and necessary resampling or interpolation operations, the timestamps of all events are converted into relative times relative to the anchor point, forming a time axis aligned sequence that is friendly to subsequent estimation algorithms. This provides a common time reference for subsequent multi-source information fusion and causal relationship analysis.

[0112] Before aligning to the same timeline, the quality of the original event sequence can be improved through one or more normalization steps, including normalization methods such as high-precision clock synchronization, multi-threshold consistency verification, and timestamp smoothing.

[0113] The specific steps of high-precision clock synchronization include: the local clock of the acquisition unit is periodically synchronized with the high-precision clock reference of the station control layer or GPS through the Network Time Protocol (PTP) or IRIG-B code, thereby estimating and compensating for the frequency drift of the local crystal oscillator caused by factors such as temperature, and ensuring the long-term accuracy of all timestamps.

[0114] The specific steps of multi-threshold consistency verification include using the physical characteristics of fixed high and low threshold voltages in the voltage divider circuit to perform logical legality verification on the order of the collected out-of-bounds time series. If the out-of-bounds time of the low voltage threshold is later than that of the high voltage threshold, it can be determined as an erroneous acquisition and discarded or marked, thereby correcting the discrete bias caused by interference.

[0115] The specific steps of timestamp smoothing include processing the original timestamp sequence with a nonlinear median filter with a window width of 3 or 5 to effectively filter out isolated outliers; on this basis, an adaptive window smoothing algorithm based on event occurrence density can be combined to more finely suppress jitter in dense event regions, ultimately outputting a smoother and more reliable event sequence.

[0116] Step 2: Process the voltage event sequence related to the ring main unit's switching process and estimate the return voltage hazard parameters.

[0117] (1) Based on the voltage event sequence related to the ring main unit opening process, local least squares fitting is performed on the time difference of the multiple voltage divider thresholds contained in the voltage event sequence to determine the gap voltage slope sequence and the gap voltage sequence.

[0118] This process establishes a precise mapping relationship between the threshold index and the actual gap voltage based on the voltage out-of-bounds event sequence aligned with the time axis and the voltage divider parameters preset in the parameter library, thereby initially calculating the local voltage slope within this time period.

[0119] To overcome the interference of quantization error in the time-to-digital converter and minor noise in the signal on the slope calculation, this embodiment adopts local least squares fitting. A time window containing 3 to 5 consecutive event points is selected, and linear regression is performed on the (time, voltage) data pairs within this window. The slope of the fitted line is used as the estimated value of the gap voltage slope at the center time of the window, and the voltage value corresponding to that time is used as the estimated value of the slope.

[0120] To enhance robustness to strong noise and outliers, local least squares fitting can employ the Huber loss function, which is insensitive to outliers, instead of the traditional quadratic loss function for optimization. Before performing the fitting, nonlinear filtering methods such as rank filtering can be applied to the event data within the window to pre-remove or suppress obvious outlier data points. Through these processes, a smooth and reliable gap voltage slope sequence and gap voltage sequence can be obtained, providing high-quality input for subsequent physical model recursion.

[0121] (2) Based on the physical equation of the return voltage, combined with the gap voltage slope sequence and the gap voltage sequence, the residual conductance time sequence is recursively estimated.

[0122] The physical equation for the return voltage is:

[0123] ;

[0124] in, This is the gap equivalent parasitic capacitance. Sampling time The slope of the gap voltage, The equivalent source voltage of the upstream power grid. The equivalent source resistance of the upstream power grid, Sampling time The timing of residual conductance.

[0125] Based on the gap voltage slope sequence and the gap voltage sequence, a residual sequence based on the physical equation of the returned voltage is constructed. The physical equation is discretized at each discrete event sampling point k, and terms are rearranged to obtain the residual. The formula:

[0126] ;

[0127] in, For discrete sampling point index, For the first Residual values ​​of discrete sampling points The slope of the gap voltage is in the th... The value of each discrete sampling point For the first Residual conductivity at each discrete sampling point For the gap voltage sequence in the first The values ​​of each discrete sampling point.

[0128] Ideally, if If the estimate is accurate, then the residuals It should be close to zero. Therefore, the objective of the recursive estimation is transformed into finding the value at each discrete sampling point. Looking for one Make Minimize the square of .

[0129] Directly performing unconstrained least squares calculation on the residuals may yield physically unreasonable results, such as negative conductance or increased conductance during deionization. To avoid this, this embodiment introduces a set of constraints based on prior physical knowledge, which includes: 1) 1) Because conductance cannot be negative; 2) Monotonic decay constraint ,in, Sampling time point The residual conductivity, Sampling points The residual conductivity, To represent the positive increment of noise and model error, this constraint implies that the conductivity of the residual plasma generally decays over time without continuous energy injection, and a small positive increment is allowed to tolerate computational noise and model error.

[0130] By projecting the unconstrained least squares estimate onto the constraint set in each iteration cycle, the residual conductivity time series is updated and obtained while minimizing the residual sequence and satisfying the constraint set.

[0131] At every moment Find the unconstrained least squares solution. ,examine Check if the constraint set is satisfied; if not, project it onto the boundary of the constraint set. If <0, then the final estimated value Set to 0; if > ,but Set as Through iteration, a residual conductance time series that minimizes both the model residuals and conforms to physical laws is obtained. .

[0132] In an optional implementation, to ensure the stable start of the recursive algorithm, an initial value for the residual conductivity can be set before the recursion begins. A specific estimate is required. Within a very short time window after the zero-point of the break, Very small, return to the voltage equation The influence of the term is also relatively small, so it can be approximated. ,in It is a low voltage protection limit set to prevent the denominator from being zero.

[0133] (3) The first danger index is calculated based on the residual conductivity time series and the gap voltage series.

[0134] Within a preset time window, the product of the residual conductivity time sequence and the square of the gap voltage sequence is integrated to generate the return voltage hazard index, which is calculated using the following formula: ;in, To return the voltage hazard index, Sampling time, For the integration time window, for The residual conductance timing at time 10:00. for Time-gap voltage sequence.

[0135] This represents the instantaneous power consumed in the residual arc channel at time t; The physical meaning is the critical time window after the break. The total energy injected into the residual arc channel, if exceeding the deionization threshold that the medium can withstand, may lead to arc reignition. As an energy-weighted hazard indicator, it comprehensively quantifies the reignition energy risk of the gap compared to single indicators such as voltage slope or amplitude. In its digital implementation, the above integration is achieved through numerical integration methods. Within one control cycle, the trapezoidal rule is used to accumulate the corresponding products of the residual conductivity time sequence and the gap voltage sequence to obtain the result. Discrete calculated values.

[0136] Step 3: Process the wall current event sequence to generate a space charge relaxation time surrogate quantity.

[0137] (1) Extracting wall current pulse segments from the wall current event sequence. The process of extracting wall current pulse segments is usually triggered by voltage or current events in the main circuit, for example, using the moment when the current crosses zero or the return voltage shows the first steep rising edge as the starting point of the analysis window. Within this window, baseline correction is performed on the wall current signal, for example, by using a local polynomial fitting method to remove slow drift caused by factors such as temperature, resulting in a detrended signal. Based on the signal's energy density or pulse interval, the signal is divided into one or more independent pulse segments, each pulse segment corresponding to one space charge injection and relaxation process.

[0138] (2) The exponential decay characteristics of the wall current pulse segment are fitted to obtain the space charge relaxation time surrogate quantity. From a physical perspective, the pulse decay process is mainly dominated by the recombination, diffusion, and migration of space charge near the gap, and its macroscopic performance is close to first-order exponential decay. Therefore, nonlinear least-squares fitting can be performed on the data points at the tail end of the pulse segment. The space charge relaxation time surrogate quantity directly reflects the speed of space charge dissipation, thus indirectly characterizing the recovery speed of the gap dielectric insulation strength.

[0139] Optionally, to improve the reliability of the space charge relaxation time surrogate quantity estimation, multiple pulse segments or multiple small discharge pulses in the same interruption process can be fitted separately to obtain a candidate set. Simultaneously, a quality score can be assigned to each pulse segment, based on signal parameters such as signal-to-noise ratio and pulse amplitude. By weighting the effective values ​​in the candidate set using a median or weighted average, where the weights are derived from the quality score, a more stable and reliable space charge relaxation time surrogate quantity is obtained as the final output of this step.

[0140] Step four: Based on the space charge relaxation time surrogate quantity, the return voltage hazard index is corrected to obtain the standard hazard index.

[0141] (1) The space charge relaxation time surrogate is converted into a dimensionless consistency calibration factor through a preset mapping relationship. This mapping relationship can be a lookup table, a piecewise linear function, or a more complex nonlinear function. Its specific form can be based on: 1) the equipment's factory specification data, i.e., the relationship between the space charge relaxation time surrogate and the actual dielectric recovery characteristics established by the manufacturer through experiments under different insulation states; 2) on-site offline calibration, i.e., during the equipment commissioning stage, by actively applying different severity levels of interruption conditions, the space charge relaxation time surrogate and the actual deionization time constant measured by high-precision instruments are recorded simultaneously, thereby regressing the mapping model. For example, a space charge relaxation time surrogate of tens of microseconds means that the space charge dissipates quickly and the dielectric recovery capability is strong, which should correspond to a consistency calibration factor less than 1, such as 0.9, indicating that the danger level obtained from the voltage observation path needs to be corrected downward. Conversely, a longer space charge relaxation time surrogate corresponds to a consistency calibration factor greater than 1.

[0142] (2) The residual conductivity timing and return voltage hazard index are scaled or corrected using the consistency calibration factor to obtain the standard hazard index. In order to self-verify the effectiveness of the calibration process, a consistency verification step can also be introduced: After the return voltage hazard index is initially calibrated, the mutual information or Spearman rank correlation coefficient between its time evolution trend and the wall current decay envelope after time alignment can be calculated. If the correlation index is lower than the preset threshold, it indicates that the evolution trends of the two independent physical processes are inconsistent. The system can fine-tune the value of the consistency calibration factor accordingly, or generate an alarm to indicate that there may be a sensor failure or a more complex physical process.

[0143] Step 5: Construct a barrier function based on standard hazard indicators, residual conductivity timing, and space charge relaxation time surrogate quantity, and generate a minimum energy control command based on the barrier function.

[0144] (1) A state vector is constructed based on the standard hazard index, the residual conductivity timing, the space charge relaxation time surrogate quantity, and the pre-acquired position signal of the target ring main unit actuator. The position signal of the ring main unit actuator includes the closing position, the opening position, and the intermediate transient position signal. Before constructing the state vector, each state component is normalized to eliminate dimensional differences.

[0145] (2) Construct a barrier function based on the state vector to obtain the barrier function value and the time derivative estimate of the barrier function. The purpose of designing the barrier function is to divide the system's state space into a safe region and a dangerous region. The safe region is defined as satisfying the inequality... The state region ≤0.

[0146] The barrier function is:

[0147] ;

[0148] in, This is the value of the barrier function; , , , These are the first weighting coefficient, the second weighting coefficient, the third weighting coefficient, and the fourth weighting coefficient, respectively. , , , All are positive numbers; As a standard hazard indicator, For residual conductance timing, This is a space charge relaxation time surrogate quantity. The preset safety threshold; The indicator function is set to 1 when the position signal of the actuator is in an intermediate transient state, and 0 otherwise.

[0149] In each control cycle, the current state vector is substituted into the above formula to calculate the current barrier function value. Simultaneously, by performing forward differencing on the current and historical barrier function value sequences or applying a Kalman filter, the time derivative of the barrier function can be estimated.

[0150] (3) Under the constraints of preset inequality constraints, determine the minimum energy control command. The inequality constraints ensure that the barrier function value converges within the defined safety region. The system state of the ring main unit is ensured to converge within the safety region through inequality constraints, which are expressed as follows: ,in For the estimation of the time derivative of the barrier function, The convergence coefficient is positive. This inequality means that the controller's task is to apply a control action that makes the rate of change of the barrier function... It must be negative enough to ensure that even if the current barrier function value is in the danger region, it can be pulled back to the safe region at a rate not less than the exponential decay rate α; if the barrier function value is already in the safe region, it must be ensured that it will not drift toward the danger boundary, which is B=0.

[0151] This step generates various control commands that minimize energy consumption:

[0152] 1) Coil current timing shaping, the determination process includes: estimating the sensitivity of the barrier function to the coil current based on the state vector; combining the barrier function value, the time derivative estimate of the barrier function, and the barrier function sensitivity, solving for the minimum current increment that satisfies the inequality constraints; and synthesizing the coil current timing shape based on the minimum current increment. The control effect of this instruction on the ring main unit is reflected in the control of the trip coil current. On the adjustment.

[0153] The total derivative of the barrier function with respect to the control input can be approximated as:

[0154] ;

[0155] in, For the estimation of the time derivative of the barrier function, The time derivative of the barrier function. For the sensitivity of the barrier function, This represents the minimum current increment.

[0156] The barrier function sensitivity indicates the extent to which changes in coil current affect the barrier function. It can be obtained by combining an offline identified mechanistic model or data model with online correction.

[0157] At this point, the inequality constraint becomes:

[0158] ;

[0159] in, This is the value of the barrier function;

[0160] The control objective is to minimize control energy while satisfying this constraint, i.e., minimize [the current energy required for the current control cycle]. This is achieved by solving inequalities to obtain the minimum current change rate or current increment required for the current cycle. The current increment is then converted into an executable coil current timing sequence through PWM synthesis or current loop setting.

[0161] This control command enables the controller to hardly operate when the safety margin is large; only when the system state approaches the danger boundary will the controller precisely output the minimum energy sufficient to maintain safety, in order to increase the tripping speed or suppress mechanical backlash.

[0162] (2) Dynamic absorption branch gating command, the determination process of which includes: when the barrier function value indicates that the state is close to the safety domain boundary, detecting the reignition signature from the voltage event sequence; in response to the detection of the reignition signature, and under the constraint of the preset energy budget, generating a dynamic absorption branch gating command.

[0163] When the barrier margin approaches zero and typical pre-reignition characteristics are detected in the voltage event sequence, such as a sudden reversal of the voltage slope or a short-term voltage drop followed by a rebound, the ring main unit is deemed to have an extremely high risk of reignition. A dynamic absorption branch gating command is generated to instantaneously activate a parallel RC or metal oxide varistor branch to quickly clamp the return voltage peak and absorb energy. The command generation is constrained by a dynamically updated energy budget, ensuring that the total energy input to the absorption branch does not exceed an upper limit throughout the entire interruption process, thus avoiding the energy waste and component aging problems of traditional fixed absorption schemes.

[0164] (3) The determination process of the fast secondary interruption trigger command includes: continuously monitoring the voltage event sequence to identify the reignition mode when the barrier function value indicates that the state is continuously at or near the safety domain boundary; and generating a fast secondary interruption trigger in response to the identification of the reignition mode. If the barrier function remains in the danger domain after the dynamic absorption branch is put into operation, or if the system clearly identifies a reignition that has occurred, the fast secondary interruption trigger flag is immediately set. This flag will instruct the operating mechanism to perform a break-close-break operation at the fastest speed in order to successfully interrupt the circuit under a new, better phase or a longer opening distance.

[0165] Example 3, based on the same inventive concept as other examples, describes the complete process of the present invention from data acquisition to final control command generation through a specific scenario with set initial conditions and input data.

[0166] Suppose a 10kV ring main unit is performing a load disconnection operation. The relevant physical parameters of the system have been pre-calibrated and stored in the parameter library, as follows:

[0167] Gap equivalent parasitic capacitance C gap =50pF, upstream grid equivalent source resistance R th =10kΩ, upstream grid equivalent source voltage E s =10kV (peak value), barrier function convergence coefficient α=0.5s -1 .

[0168] During a critical window in the switching process, voltage event sequences and wall current event sequences were acquired; for the voltage event sequences, the system monitored three voltage thresholds: u level [1] = 100V, u level [2] = 200V, ulevel [3]=300V, the time-to-digital converter TDC recorded the out-of-bounds timestamps as follows: t[1]=10.5μs, t[2]=12.0μs, t[3]=13.0μs; For the wall current event sequence, the system detected a clear wall current pulse, the decay tail of which can be fitted by an exponential model.

[0169] Based on the voltage event sequence, the average voltage slope during the time interval from t[1] to t[3] was calculated. Local least squares fitting was used to fit the three (time, voltage) data points {(10.5,100),(12.0,200),(13.0,300)} to obtain the slope estimate. The slope at time t[2] was approximated using the central difference method, resulting in a slope of 8.0 x 10⁻⁶. 7 V / s. The gap voltage at this time is u. level [2], i.e., 200V;

[0170] Using the data between t[1] and t[2], the initial recursive value was estimated based on the methods described in Examples 1 and 2. Subsequently, a constrained recursive estimation process was initiated. After several recursive steps, within the critical integration window Δt from t=50μs to t=150μs, the average value of the residual conductivity timing obtained was approximately 5.0x10-5S (Siemens), and the root mean square value of the reconstructed gap voltage was approximately 4.0kV;

[0171] according to The formula calculates the return voltage hazard index to be 80J.

[0172] Process the wall current event sequence and perform exponential decay fitting on the decay tail of the acquired wall current pulses: The space charge relaxation time surrogate quantity was obtained as 85 μs.

[0173] Based on the space charge relaxation time surrogate, the return voltage hazard index is corrected to obtain a standard hazard index: The space charge relaxation time surrogate is substituted into a preset mapping function or lookup table to obtain a consistency calibration factor. Assuming the mapping relationship indicates that the relaxation time is slightly longer than normal, an upward correction calibration factor is applied. For example, if the calibration factor is 1.1, then… Average residual conductivity after calibration The calibrated hazard index H RV_star =88J, which will serve as the basis for the final control decision.

[0174] Finally, the minimum energy control command is generated, following these steps:

[0175] Perform state vector assembly and barrier function evaluation. Assemble the current state vector, assuming the actuator is currently in a safe vacancy state, and specify the indicator function. If the value is 0, then the key component of the state vector is ;

[0176] Barrier function Assuming the weights are normalized as follows: =1.0J -1 The contributions of other items are temporarily ignored, and a safety threshold is set. =90. The calculated barrier function value is -2.0, which indicates that the system is within the safe region. The time derivative of the barrier function estimated from historical data is +10.0s. -1 This indicates that the danger index is rising rapidly, and the system state is quickly approaching the danger boundary.

[0177] We perform a minimum current increment solution, assuming the control sensitivity obtained from the offline model and online correction is at this point. =-0.4A -1, This indicates that increasing the coil current helps reduce the danger, i.e., reducing the B value; to simplify calculations, ... Approximately Where ΔT is the control period. The inequality constraint becomes Substituting... That is, minimum current increment

[0178] The control command output calculation results show that, in order to ensure that the system state still converges within the safe domain in the next control cycle, the controller must generate a command to increase the current of the trip coil by at least 22.5A. Therefore, the controller will synthesize a specific PWM waveform or current loop setpoint based on this minimum increment and in combination with the physical limitations of the actuator, such as the maximum current and the maximum rate of change of current, as the final energy-minimizing control command. This command precisely provides the minimum necessary energy to suppress dangerous trends, avoiding unnecessary overdrive.

[0179] Example 4, based on the same inventive concept as other examples, introduces a ring main unit control method. The difference between this example and other examples is that, in order to deal with problems such as electromagnetic interference, signal noise, and model mismatch in real industrial environments, this example designs a series of robustness protection mechanisms and further introduces more refined fidelity improvement and verification measures.

[0180] (1) For voltage event sequences, a multi-threshold consistency verification mechanism is introduced. For a monotonically increasing return voltage waveform, it crosses multiple voltage thresholds u arranged from low to high. levelThe timestamp t[k] of [k] must also be monotonically increasing. Therefore, after data acquisition, the system can automatically verify the logical relationship t[k+1]>t[k]. If a violation is found, it can be determined that the data points in that group have been subjected to strong electromagnetic interference. The system can mark the abnormal point or the data in the small window it contains as unreliable and remove it. This mechanism can effectively prevent erroneous data caused by occasional strong interference from contaminating subsequent slope calculation stages.

[0181] Furthermore, to suppress random jitter in timestamps, this embodiment employs a two-stage jitter suppression strategy. In the first stage, a nonlinear median filter with a window size of 3 is used to process the original timestamp sequence. This filter effectively removes independent, high-amplitude pulse noise without blurring the edge information of abrupt voltage waveform changes, unlike linear filters. In the second stage, an adaptive window smoothing algorithm based on event density is used. A smaller smoothing window is used in regions with drastic voltage changes and dense event points to preserve details, while a larger smoothing window is used in regions with gentle voltage changes and sparse event points to enhance noise suppression. This adaptive strategy achieves a better balance between noise suppression and signal fidelity compared to a fixed-window smoothing algorithm.

[0182] (2) To ensure that the recursive estimation algorithm for the return voltage hazard index converges stably under various operating conditions, a local regression algorithm based on the Huber loss function is used instead of the traditional least squares fitting in the reconstruction step of the gap voltage slope. The traditional least squares method minimizes the square of the error, and a single outlier can have a huge impact, which may cause the fitting result to deviate significantly from the true slope. The Huber loss function is a hybrid function that uses squared loss for small errors and linear loss for large errors. This characteristic makes the regression process insensitive to outliers, so that a stable and reliable slope estimate can still be obtained even in the case of data contamination.

[0183] In the recursive estimation of residual conductance, a residual monitoring and backoff mechanism is introduced to ensure that the algorithm does not diverge. While performing the recursion, the system monitors in real time the statistical characteristics of the residual sequence constructed based on the physical equation of the returned voltage, such as its variance or mean within the sliding window. The system also counts the frequency of constraint set triggering. If the residual variance increases sharply in a short period, or the constraints are frequently triggered, the system determines that the estimation algorithm may have entered an unstable state, for example, due to severe model parameter mismatch or persistent abnormal input data. In this case, the system will trigger a backoff action: actively discarding the estimation results of the most recent few cycles, rolling back the state of the residual conductance to the last considered stable and reliable value, and restarting the recursion from that point. After the backoff, the positive increment in the monotonic decay constraint can be temporarily relaxed. This gives the algorithm greater freedom to escape the current unstable region.

[0184] (3) To reliably extract the space charge relaxation time surrogate from the wall current signal with a complex noise background, this embodiment incorporates an anti-interference design for wall current feature extraction, introducing a pulse segment quality assessment and abnormal segment rejection mechanism. After dividing the wall current signal into multiple pulse segments, the system calculates a quality score for each pulse segment. This score is a comprehensive index, including signal-to-noise ratio, peak amplitude, pulse width, etc. After performing exponential decay fitting on each pulse segment, the system checks two conditions: whether the quality score of the pulse segment itself is higher than a preset threshold, and whether the root mean square error of the fitting result is lower than a preset threshold. Only fitting results that simultaneously meet both conditions are considered valid. This method can effectively filter out pulse segments whose fitting results are unreliable due to weak signals, severe noise, or poor morphology, ensuring the purity of the candidate values ​​of the second index used for final aggregation calculation.

[0185] (4) This implementation also systematically optimizes the recursive starting point of residual conductivity and proposes a systematic initial value estimation and evaluation method to replace simple approximate calculation.

[0186] Within a very short initial window after the zero-point break, the initial voltage and slope are reconstructed. Then, the initial value of the residual conductance is calculated using an initial value approximation formula. The formula is expressed as:

[0187] ;

[0188] Next, an initial value deviation bound assessment is performed, taking into account the values ​​ignored within the initial window. The impact of this item is determined by the system based on the equivalent source voltage of the upstream power grid. and the equivalent source resistance of the upstream power grid The possible range is estimated, and the maximum positive and negative deviations caused by the neglected term to the initial value are calculated. This yields an initial value package containing the initial value of residual conductivity and its confidence interval, providing a starting point for the recursion and also providing tighter constraints.

[0189] (5) To improve physical consistency calibration, this embodiment introduces a deep verification and adaptive calibration mechanism based on information theory. After obtaining the initial calibration factor through the mapping function and calculating the calibrated residual conductivity and hazard index, a consistency verification is performed. The decay sequence of the calibrated residual conductivity and the decay envelope sequence of the wall current pulse are compared, and the mutual information or Spearman rank correlation coefficient of these two time series within the aligned window is calculated. These two indicators can measure the nonlinear correlation strength between two variables. If the calculated correlation is lower than a preset threshold, it indicates that the two physical processes that are considered to be highly correlated exhibit inconsistency under the current calibration factor. For low consistency detection, the system triggers adaptive calibration. For example, the system can perform a small-step search around the current value of the calibration factor, recalculate the indicators under different calibration factors, and find the calibration factor that maximizes the mutual information or rank correlation coefficient, and use the optimized calibration factor as the final calibration factor.

[0190] (6) This embodiment also incorporates online identification and adaptation steps for the sensitivity of the control model. During the equipment manufacturing or commissioning phase, a series of switching experiments are conducted to systematically change the driving waveform of the coil current and record the changes in the corresponding state vector and barrier function. A nonlinear mapping model is trained using machine learning methods, and the baseline predicted value of the sensitivity is directly output based on the input current state vector. In the actual operation of the ring main unit, after the controller issues a control command in each control cycle, it observes the actual change in the barrier function. Using this input-output pair, the output of the offline model can be corrected in real time and with small amplitude through recursive least squares or Kalman filters. That is, the controller is constantly learning and calibrating its control model while executing control.

[0191] (7) In order to achieve the minimum necessary energy input for controlling resources, when triggering high-cost intervention measures, an adaptive identification logic based on specific physical phenomena is adopted, rather than a static threshold that solely relies on the barrier function value. The decision to input the dynamic absorption branch is made when the barrier function value indicates that the state is close to the boundary of the safe domain. This embodiment further limits its triggering conditions: the system not only requires the barrier margin to be less than the preset threshold, but also requires that a clear reignition signature be detected from a high-frequency voltage or current event sequence through a pattern recognition module. The reignition signature is direct physical evidence that the arc is about to reignite or has already reignited at the microsecond level.

[0192] The pattern recognition module can be configured to identify one or more of the following typical features: a sharp reversal of the return voltage slope, an unexpected rebound in the main circuit current, and a short-term local collapse of the gap voltage. This short-term local collapse refers to a sudden, rapid, and small drop in the recovering voltage after reaching a relatively high value, followed by a possible recovery and rise. This is typically interpreted as a micro-discharge or streamer process that fails to develop into breakdown. By logically ANDing the macroscopic state assessment of the barrier function with the accurate identification of these microscopic physical events, the controller can improve the precision of interventions and avoid malfunctions caused by temporary reductions in barrier margin due to model errors or uncertainties in state estimation.

[0193] (8) In order to ensure the sustainability of control actions and the security of hardware, the present invention performs dynamic and closed-loop management of limited control resources, especially the energy capacity of the dynamic absorption branch, and introduces an energy budget, which is a state variable that is dynamically updated during each on / off operation.

[0194] When the controller receives a tripping command, the energy budget is initialized to its maximum allowable value, which is preset based on the thermal capacity and single-cycle energy tolerance specifications of the core components of the absorption branch. Each time the dynamic absorption branch is activated by a gating command, the controller calculates the energy absorbed during activation in real time by integrating the measurements from the voltage and current sensors on the absorption branch. This consumed energy is then subtracted from the current budget. Before each gating command is generated, a pre-check is included in the control logic. If the budget is exhausted, the generation of the command will be suppressed even if all triggering conditions are met, and the system can generate an alarm indicating that the absorbed energy has exceeded the limit. This dynamic closed-loop management mechanism not only ensures that the power components will not overheat and be damaged due to repeated operations but also provides greater flexibility to the control strategy, allowing for multiple, short-duration absorption operations to cope with complex reignition processes, provided that the total energy limit is not exceeded.

[0195] After each critical operation, such as successful interruption, detection of a dangerous event, or execution of a secondary interruption, the system generates and stores a structured safety log. This log includes, but is not limited to: energy consumption records (total energy consumption of coil current shaping and total absorbed energy of the absorption branch during the operation); danger window statistics (peak values ​​of danger indicators, duration of the barrier function being positive, and number of times approaching the danger boundary); control action sequences (the issuance time, duration, and key parameters of all control commands); and event context (the complete state vector when a critical control action is triggered). These logs are stored long-term in local non-volatile memory, constituting a so-called data deposition. This data deposition provides the foundation for the long-term evolution of the system, namely parameter refeeding and model optimization.

[0196] These safety logs can be periodically uploaded to the backend main station or cloud analysis platform via the communication network. By statistically analyzing large amounts of log data from a single or multiple ring main units of the same type, the following can be achieved:

[0197] 1) Offline optimization of global parameters. For example, if the logs of a specific region or season are analyzed and a systematic bias is found in the estimated values ​​of dangerous indicators, the weight coefficients of the basic parameters or barrier functions in the parameter library can be uniformly optimized, and the optimized parameter package can be fed back into the device through remote firmware upgrade.

[0198] 2) Iterative enhancement of the control model: The offline model used to estimate control sensitivity can be periodically retrained and iterated using the massive amount of accumulated real {input-state-output} data, making its predictions increasingly accurate.

[0199] 3) Predictive health management, through longitudinal time-series analysis of the safety logs of individual devices, can identify performance degradation trends. For example, if the average return voltage hazard index of a switching unit continues to rise slowly with the increase in the number of operations, or if the call to the absorption branch becomes more and more frequent, it can provide an early warning of potential health problems in its internal vacuum interrupter or operating mechanism, thereby realizing the transformation from post-fault maintenance to condition-based / predictive maintenance.

[0200] This log-deposition-recharge mechanism constitutes a complete, long-cycle learning loop, enabling the control system disclosed in this invention to have the ability to self-evolve and continuously optimize.

[0201] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0202] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0203] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0204] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0205] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for controlling a ring main unit, characterized by, include: Obtain the voltage event sequence and wall current event sequence during the opening and closing process of the target ring main unit; The residual conductivity timing and return voltage hazard index are obtained based on the voltage event sequence; The space charge relaxation time surrogate quantity is obtained based on the wall current event sequence; Based on the space charge relaxation time proxy, the return voltage hazard index is corrected to obtain a standard hazard index; A barrier function is constructed based on the standard hazard index, the residual conductivity timing, and the space charge relaxation time surrogate quantity. A minimum energy control command is generated according to the barrier function and the preset inequality constraints. The actuator of the target ring main unit is controlled by the minimum energy control command.

2. The control method of the ring main unit according to claim 1, wherein The process of obtaining the residual conductance timing and return voltage hazard index based on the voltage event sequence includes: Local least squares fitting is performed on the time difference of the out-of-bounds voltage divider thresholds contained in the voltage event sequence to obtain the gap voltage slope sequence and the gap voltage sequence; Based on the physical equation of the return voltage, the gap voltage slope sequence, and the gap voltage sequence, the residual conductance timing is recursively obtained; Based on the residual conductance timing sequence and the gap voltage sequence, the return voltage hazard index is calculated; the formula for calculating the return voltage hazard index is: ; wherein, is a return voltage danger index, is a sampling time, is an integration time window, is a residual on-state conductivity timing at the time instant, is a gap voltage sequence at the time instant.

3. The ring main unit control method according to claim 2, characterized in that, The residual conductance timing sequence is recursively obtained based on the physical equation of the return voltage, the gap voltage slope sequence, and the gap voltage sequence, including: Based on the gap voltage slope sequence and the gap voltage sequence, a residual sequence based on the returned voltage physical equation is constructed; Establish a constraint set that includes nonnegativity constraints and monotonic decay constraints, so that the residual sequence satisfies the constraint set while being minimized, and obtain the residual conductance time series.

4. The ring main unit control method according to claim 3, characterized in that, The formula for the physical equation of the returned voltage is: ; in, This is the gap equivalent parasitic capacitance. Sampling time The slope of the gap voltage, The equivalent source voltage of the upstream power grid. The equivalent source resistance of the upstream power grid, Sampling time The residual conductance timing; The formula for the residual sequence is: ; in, For discrete sampling point index, For the first Residual values ​​of discrete sampling points The slope of the gap voltage is in the th... The value of each discrete sampling point For the first Residual conductivity at each discrete sampling point For the gap voltage sequence in the first The values ​​of each discrete sampling point.

5. The ring main unit control method according to claim 1, characterized in that, The process of obtaining the space charge relaxation time surrogate based on the wall current event sequence includes: Extract wall current pulse segments from the wall current event sequence; The exponential decay characteristics of the wall current pulse segment are fitted to obtain the space charge relaxation time surrogate quantity.

6. The ring main unit control method according to claim 1, characterized in that, The standard hazard index is obtained by correcting the return voltage hazard index based on the space charge relaxation time proxy, including: Based on a preset mapping relationship, the charge relaxation time proxy is mapped to a consistency calibration factor. The return voltage hazard index is scaled and corrected using the consistency calibration factor to obtain a standard hazard index.

7. The ring main unit control method according to claim 1, characterized in that, The process of constructing a barrier function based on the standard hazard index, the residual conductivity timing, and the space charge relaxation time surrogate, and generating a minimum energy control command based on the barrier function and preset inequality constraints, includes: A state vector is constructed based on the standard hazard index, the residual conductivity timing, the space charge relaxation time proxy, and the pre-acquired position signal of the target ring main unit actuator. A barrier function is constructed based on the state vector, and the barrier function value and the time derivative estimate of the barrier function are calculated; the formula for the barrier function is: ; in, This is the value of the barrier function; , , , These are the first weighting coefficient, the second weighting coefficient, the third weighting coefficient, and the fourth weighting coefficient, respectively. , , , All are positive numbers; As a standard hazard indicator, For residual conductance timing, This is a space charge relaxation time surrogate quantity. The preset safety threshold; The indicator function is set to 1 when the position signal of the actuator is in an intermediate transient state, and 0 otherwise. Based on the barrier function value and the time derivative of the barrier function, the minimum energy control command of the target ring main unit is determined through preset inequality constraints. The inequality constraint is: ; in, For the estimation of the time derivative of the barrier function, The convergence coefficient is positive.

8. The ring main unit control method according to claim 7, characterized in that, The energy minimization command includes coil current timing shaping, and the process of determining the coil current timing shaping includes: Based on the state vector, the sensitivity of the barrier function to the coil current is estimated; Based on the barrier function value, the time derivative estimate of the barrier function, and the sensitivity, the minimum current increment that satisfies the inequality constraint is determined. The minimum current increment is converted into coil current timing shaping to improve the tripping speed of the target ring main unit or suppress mechanical backflow.

9. The ring main unit control method according to claim 7, characterized in that, The energy minimization command also includes a dynamic absorption branch gating command, the determination process of which includes: When the barrier function value approaches the preset safety domain boundary, a reignition signature is detected from the voltage event sequence; the reignition signature refers to the reignition precursor characteristics indicated by the electrical parameters of the ring main unit, and the safety domain is defined as satisfying the barrier function value. The state region ≤0; When the reignition signature is detected, a dynamic absorption branch gating command is generated based on the reignition signature and the preset energy budget constraint to suppress the return voltage peak of the target ring main unit and absorb energy.

10. The ring main unit control method according to claim 7, characterized in that, The energy minimization command also includes a fast secondary break trigger command, the determination process of which includes: When the barrier function value remains at or near the preset safety domain boundary, the voltage event sequence of the target ring main unit is continuously monitored; the safety domain is defined as satisfying the barrier function value. The state region ≤0; When a reignition mode is identified from the voltage event sequence, a rapid secondary disconnection trigger command is generated to enable the actuator of the target ring main unit to perform a tripping, closing, and tripping operation sequentially at the fastest speed. The reignition mode refers to one or more of the following characteristics: a sharp reversal of the return voltage slope, an unexpected rebound of the main circuit current, and a short-term local collapse of the gap voltage. The short-term local collapse of the gap voltage refers to the phenomenon that the recovering voltage suddenly drops rapidly and slightly after reaching a preset voltage value, and then recovers and rises again.