Self-healing method of primary and secondary fusion ring network box based on intelligent decision

Abstract

The application discloses a self-healing method of primary and secondary fusion ring network box based on intelligent decision, and belongs to the technical field of power distribution network fault self-healing. The method comprises the following steps: collecting electrical operation parameters and switch state sequences of each node in the ring network box. By analyzing the statistical difference of the parameters in the sliding window, the steady-state deviation factor is calculated; according to the transition consistency of the switch state in the continuous window, the action reliability factor is evaluated; the fault tendency coefficient is determined by comprehensively considering the two factors. The nodes with the coefficient exceeding the threshold value are divided into the fault influence area and the boundary node according to the topological connection, the isolation priority is determined according to the boundary parameter propagation delay, and the switch action time sequence is adjusted to complete the isolation. Subsequently, the recovery path and the closing sequence are determined based on the power supply capacity and the load demand of the non-fault area, and the corresponding switch is controlled to close to restore power supply. The application realizes accurate fault positioning, orderly isolation and rapid recovery.

Description

Technical Field

[0001] This invention relates to the field of distribution network fault self-healing technology, specifically a self-healing method for primary and secondary integrated ring network boxes based on intelligent decision-making. Background Technology

[0002] The self-healing control of integrated primary and secondary ring main units in distribution networks relies on accurate fault perception and rapid response. Existing technologies typically determine faults and initiate isolation and recovery processes based solely on threshold exceedances of electrical quantities or changes in a single type of characteristic. For example, faults are identified by monitoring whether voltage dips or current surges exceed preset thresholds, and isolation and transfer operations are performed using preset logic tables or simple topology searches. In complex operating conditions involving both steady-state disturbances and transient impacts, this method is prone to misjudging normal load fluctuations or electrical quantity changes caused by normal switch operations as permanent faults, leading to unnecessary expansion of power outages and insufficient accuracy and selectivity in self-healing actions. Furthermore, when determining the urgency of fault isolation, existing technologies often only consider the electrical distance to the fault point or a preset priority list, failing to effectively quantify the actual propagation dynamics of fault characteristics within the power grid. This results in isolation sequences relying solely on static rules, making it difficult to prioritize the isolation of fault areas with rapid propagation speeds and missing the optimal opportunity to suppress the impact of the fault. Meanwhile, most existing switch action strategies employ fixed timing sequences or simple time-separation coordination, failing to establish a correlation with the dynamic characteristics of fault propagation and the consistency of the switch's own actions. This leads to a disconnect between the isolation process and the actual development of the fault. During the fault recovery phase, existing technologies typically close the circuit in a fixed sequence or only consider a single recovery path, without fully considering the real-time phase synchronization between the power supply side and the load side along the recovery path. Direct closing may generate inrush circulating currents, endangering equipment safety and causing protection malfunctions. Therefore, an intelligent self-healing method is needed that can integrate switch action characteristics and electrical steady-state deviation to quantify fault tendency, dynamically adjust the isolation sequence based on fault propagation delay, and then combine phase synchronization conditions to complete safe recovery. Summary of the Invention

[0003] This paper presents a self-healing method for a primary and secondary integrated ring main unit based on intelligent decision-making. It quantifies the fault tendency coefficient by integrating the steady-state deviation factor of electrical nodes and the action reliability factor of switch states, thereby improving the accuracy of fault identification. It determines the fault isolation priority by analyzing the difference in propagation delay between electrical nodes in the area, thereby improving the matching degree between isolation actions and fault development trends. It also achieves safe and smooth load transfer by evaluating the phase difference between the power supply side and the load side before closing the power supply path.

[0004] To achieve the above objectives, this invention provides the following technical solution: This invention provides a self-healing method for a primary and secondary integrated ring main unit based on intelligent decision-making. By collecting the electrical operating parameter sequences and switch state sequences of each electrical node within the ring main unit, and utilizing the statistical characteristic differences and state transition consistency within a sliding window, the fault tendency of nodes is comprehensively evaluated. Based on the distribution network topology and parameter propagation delay, the fault-affected area and isolation priority are divided. After fault isolation, a recovery path and closing sequence are planned according to the remaining power capacity and load demand to achieve rapid power restoration. This method can accurately identify potential fault nodes from two dimensions: parameter steady-state deviation and switch action reliability. It also adaptively determines the fault-affected range and boundary nodes by combining electrical connection relationships, effectively preventing fault expansion.

[0005] As a technical solution of this invention, when determining the steady-state deviation factor, for each electrical node, parameter values ​​at multiple sampling times within the current sliding window are obtained. A symmetry measure and a discrete measure relative to this symmetry measure are calculated, and these are compared with the baseline symmetry measure and baseline discrete measure under historical normal operating conditions. The difference measures are then weighted and summed to obtain the steady-state deviation factor. This factor can sensitively reflect the degree of deviation of the electrical node's operating state from the normal steady state, providing a stable basis for fault tendency judgment.

[0006] Preferably, when determining the operational reliability factor, the state code value of the switch state sequence of each electrical node within each sliding window is obtained. The number of transitions where the state code value changes between adjacent sampling times and the total number of sampling times where changes occur are counted. The state consistency rate is determined by the ratio between the number of transitions and the total number of changes. Then, the difference between the state consistency rate of the current window and the baseline state consistency rate under historical normal operating conditions is calculated to obtain the operational reliability factor. This factor can characterize the consistency and reliability changes of switch operation behavior, and help determine whether there is a risk of maloperation or failure to operate.

[0007] Based on this, the steady-state deviation factor and the action reliability factor are amplified according to preset steady-state influence weights and action influence weights, respectively, to obtain steady-state contribution values ​​and action contribution values, which are then superimposed to form a fault tendency coefficient. When the fault tendency coefficient exceeds a preset initiation threshold, the corresponding electrical node is marked as the fault initiation node. A neighborhood search is performed along the electrical connection path in the distribution network topology, and the initial impact area is adjusted based on the comparison result of the fault tendency coefficient and the expansion threshold. Finally, the fault impact area and the boundary nodes of the areas inside and outside the directly connected areas are determined. This area division method can accurately define the fault spread range and avoid over-isolation or under-isolation.

[0008] To determine the urgency of fault isolation, electrical operating parameter sequences of regional boundary nodes and corresponding fault initiation nodes are obtained. Similarity comparisons are performed at each sampling time step to determine the feature propagation delay of fault characteristics from the fault initiation node to each boundary node. Boundary nodes with feature propagation delays less than a preset delay threshold are marked as fast boundary nodes, and the rest are marked as slow boundary nodes. The ratio of the number of fast boundary nodes to the number of slow boundary nodes represents the fault isolation priority; a higher ratio indicates a faster fault propagation rate and a higher isolation priority. In this way, isolation operations can be prioritized for rapidly spreading fault areas to suppress fault propagation.

[0009] The fault isolation sequence is formed by prioritizing faults from high to low. For the area currently affected by the fault, the expected operating time of each boundary node switch is determined based on the real-time trend of the electrical operating parameter sequence. The corresponding switches are then controlled to trip in ascending order of expected operating times until all boundary node switches have tripped before the next affected area is addressed. This process dynamically adjusts the switch action sequence to ensure timely action at critical boundary nodes, minimizing the impact of the fault on healthy areas.

[0010] After fault isolation is completed, the available power supply capacity of the power nodes and the required capacity of the load nodes in the remaining non-faulty areas are obtained. Load nodes with available power supply capacity greater than required capacity are marked as recoverable nodes. The shortest power supply path from each recoverable node to a power node is searched as a candidate recovery path, and these paths are sorted in ascending order of length. The closing sequence is consistent with the path sorting. This method can utilize limited power capacity to prioritize the recovery of more important loads, improving power restoration efficiency.

[0011] During the closing process, the current phase value of the power supply node at the beginning of the current recovery path and the current phase value of the load node at the end are obtained, and the phase difference is calculated. If the phase difference is less than a preset phase synchronization threshold, the switches on the path are closed sequentially according to the closing sequence. If the phase difference is greater than or equal to the preset phase synchronization threshold, the voltage output of the power supply node at the beginning is delayed until the phase difference meets the synchronization condition before the closing action is performed. Preferably, the delay time is dynamically calculated according to the difference between the phase difference and the preset phase synchronization threshold using a proportional-integral control law. By judging phase synchronization, the inrush current caused by excessive phase difference at the closing moment can be effectively avoided, protecting the equipment and system stability.

[0012] In the above technical solution, the electrical operating parameter sequence can be obtained by collecting instantaneous voltage and current value sequences from voltage transformers and current transformers configured in the primary and secondary integrated ring network box, respectively. The switch status sequence is obtained by collecting the switch opening and closing indication signal sequence from the switch auxiliary contacts. The two are aligned according to the sampling time and cached in a first-in-first-out manner for use in self-healing decision-making. The overall method forms a complete closed-loop self-healing decision chain from parameter perception, fault identification, area determination, isolation sequence optimization to recovery path planning and closing control, significantly enhancing the intelligent handling capability and power supply reliability of the distribution network in response to faults.

[0013] The technical effects and advantages provided by the present invention in the above technical solution are as follows: By constructing steady-state deviation factors and operational reliability factors for electrical nodes separately, and then superimposing them to form a fault tendency coefficient, this approach changes the previous model that relied solely on a single electrical quantity threshold for judgment. The steady-state deviation factor, by comparing the central tendency and dispersion of parameter values ​​within a sliding window with historical normal benchmarks, can accurately distinguish whether a node's continuous abnormal fluctuations are caused by a real fault or transient deviations caused by short-term disturbances. The operational reliability factor, by statistically analyzing the consistency between the number of transitions of switch status codes between adjacent sampling points and the total number of change times, and comparing this consistency with historical normal operating conditions, incorporates the logical coherence of switch actions into the fault judgment process. The fault tendency coefficient, obtained by weighted superposition of the two factors, reflects both the degree of abnormality in electrical quantities and whether the switch action behavior is consistent with expectations. This provides a more comprehensive quantitative assessment of the likelihood of a real fault at the node, avoiding the misidentification of normal switch displacements or temporary load fluctuations as faults and reducing the false positive rate.

[0014] By utilizing the time difference in fault characteristics propagating from the fault initiation node to the boundary nodes of each region to determine fault isolation priority, this approach changes the method of performing isolation according to a fixed order or static priority. A similarity comparison method based on each sampling time step is used to calculate the feature propagation delay, quantifying the speed of the fault's impact on the region boundary. Boundary nodes are divided into fast and slow boundary nodes, and the ratio of their numbers is used as a measure of isolation priority, directly linking the isolation order to the urgency of fault propagation. For fault-affected areas with a high proportion of fast boundary nodes, the fault characteristics spread rapidly, triggering the tripping action of the corresponding boundary switches first, cutting off the fault before its impact range expands further. This mechanism, which dynamically determines the isolation sequence based on actual propagation, allows isolation actions to be more closely coordinated with the physical process of fault development, completing critical breakpoint isolation in the early stages of fault expansion, suppressing the spread of the fault range, and improving the timeliness and effectiveness of self-healing control.

[0015] Before executing the recovery path closing, the current phase difference between the first power supply node and the last load node is calculated. Switches are only controlled to close sequentially when the phase difference is less than a preset phase synchronization threshold; otherwise, the power supply voltage output is delayed until synchronization conditions are met. This changes the recovery method that directly closes in sequence or ignores phase angle differences. This process dynamically determines the closing timing based on real-time sampled phase information and uses a proportional-integral control law to dynamically calculate the delay time, ensuring that the closing operation is completed within the window period when the electrical quantities on both sides tend to synchronize. This eliminates the risk of damage to power distribution equipment and sensitive loads caused by asynchronous closing of the inrush circulating current, ensuring a smooth electrical transition during fault recovery. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of the self-healing method for primary and secondary integrated ring network boxes; Figure 2 It is a flowchart of the closing control for the fault recovery path of the distribution network based on phase synchronization; Figure 3 These are curves showing the changes in the steady-state deviation factor, operational reliability factor, and fault tendency coefficient of electrical nodes as a function of a preset sliding window. Figure 4 It is a distribution diagram of the available power capacity and load demand capacity in the non-faulty area after fault isolation. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] See Figure 1This invention provides a self-healing method for a primary and secondary integrated ring main unit based on intelligent decision-making. The overall implementation scheme includes: collecting the electrical operating parameter sequences and switch state sequences of each electrical node within the primary and secondary integrated ring main unit; determining the steady-state deviation factor of each electrical node within a corresponding sliding window based on the statistical differences in the electrical operating parameter sequences of each electrical node within a preset sliding window; determining the operational reliability factor of each electrical node within a corresponding sliding window based on the state transition consistency of the switch state sequences of each electrical node within a continuous sliding window; determining the fault tendency coefficient of each electrical node within a corresponding sliding window based on the steady-state deviation factor and operational reliability factor; determining the fault-affected area and area boundary nodes based on the connection relationships of electrical nodes with fault tendency coefficients greater than a preset initiation threshold in the distribution network topology; determining the fault isolation priority of the fault-affected area based on the propagation delay of the electrical operating parameter sequences of the area boundary nodes within the fault-affected area; completing fault isolation by adjusting the operational timing of corresponding switches within the primary and secondary integrated ring main unit according to the fault isolation priority; and determining the recovery path and closing sequence based on the power capacity and load demand of the remaining non-faulty areas according to the distribution network topology after fault isolation. Based on the recovery path and closing sequence, control the closing action of the corresponding switches in the primary and secondary integrated ring network box to complete the fault recovery.

[0020] Example 1

[0021] In practical implementation, the acquisition process of electrical operating parameter sequences and switch status sequences of each electrical node within the primary and secondary integrated ring network box can be completed through the acquisition steps described in the aforementioned overall scheme. The electrical operating parameter sequence includes instantaneous voltage value sequences and instantaneous current value sequences. The preset sliding window time length is set according to the typical duration of the transient process of a power distribution network fault, for example, it is set to the time length corresponding to 5 power frequency cycles. Each preset sliding window contains multiple sampling moments, and the time interval between adjacent sampling moments is the sampling interval corresponding to the power frequency cycle.

[0022] For each electrical node, obtain the parameter values ​​of the electrical operating parameter sequence for all sampling times within the current preset sliding window. Optionally, when the electrical operating parameter sequence is a voltage instantaneous value sequence, the parameter value is the voltage instantaneous amplitude at each sampling time; when the electrical operating parameter sequence is a current instantaneous value sequence, the parameter value is the current instantaneous amplitude at each sampling time. Calculate the central tendency measure of the parameter values ​​for all sampling times within the current sliding window. The central tendency measure can be the arithmetic mean, calculated by summing the parameter values ​​for all sampling times within the current sliding window and then dividing by the total number of sampling times. Calculate the dispersion measure of the parameter values ​​for all sampling times within the current sliding window relative to the central tendency measure. The dispersion measure can be the standard deviation, calculated by squared the difference between the parameter value and the central tendency measure at each sampling time, summing all squared values, dividing by the total number of sampling times, and then taking the square root.

[0023] For each electrical node, obtain the baseline central tendency and baseline discrete values ​​of the parameter values ​​within the corresponding time window under historical normal operating conditions. In specific implementation, the time length of the corresponding time window under historical normal operating conditions is the same as the time length of the preset sliding window. The historical normal operating conditions are selected through manual confirmation or by selecting a time period with no fault records in historical data. Obtain the parameter values ​​of the electrical operating parameter sequence at all sampling moments within this time window, calculate the arithmetic mean of these parameter values ​​as the baseline central tendency, and calculate the standard deviation of these parameter values ​​as the baseline discrete value.

[0024] Calculate the difference between the central tendency measure and the baseline central tendency measure within the current sliding window, and the difference between the dispersion measure and the baseline dispersion measure within the current sliding window. The difference measures are expressed as absolute values ​​of relative deviations. The steady-state deviation factor is calculated using the following formula:

[0025] in, Indicates electrical node The steady-state deviation factor within the current sliding window; Indicates electrical node The measure of central tendency of parameter values ​​at all sampling times within the current sliding window; Indicates electrical node The benchmark central tendency measure of parameter values ​​within the corresponding time window under historical normal operating conditions; Indicates electrical node The discrete measure of parameter values ​​at all sampling times within the current sliding window; Indicates electrical node The baseline discrete metric value of parameter values ​​within the corresponding time window under historical normal operating conditions; and For preset weighting coefficients, and The range of values ​​is And satisfy In one implementation, Set to 0.6, Set to 0.4, based on the fact that the offset of the central tendency metric is more sensitive to steady-state deviation than the offset of the discrete metric.

[0026] The switch state sequence of each electrical node is acquired by obtaining the state encoding values ​​at multiple sampling moments within each preset sliding window. The switch state sequence is collected through the auxiliary contacts of the switches configured in the primary and secondary fusion ring network box. The state encoding values ​​are binary encoded; for example, the encoding value "0" represents that the switch is in the open state, and the encoding value "1" represents that the switch is in the closed state. Within each preset sliding window, the number of sampling moments is consistent with the number of sampling moments of the electrical operating parameter sequence and is time-aligned.

[0027] For each electrical node, the number of transitions where the state code value changes between adjacent sampling times within the current sliding window is counted. The number of transitions is determined by sequentially comparing the state code values ​​of two adjacent sampling times; if the two codes are different, it is counted as one transition. Simultaneously, the total number of sampling times where the state code value changes within the current sliding window is also counted. A sampling time where the state code value changes is defined as one where the state code value at a given sampling time differs from the state code value at the previous sampling time; the first sampling time within the sliding window is not included in the count. Based on the ratio between the number of transitions within the current sliding window and the total number of sampling times where the state code value changes, the state consistency rate of each electrical node within the current sliding window is determined. The state consistency rate is calculated as follows:

[0028] When the total number of sampling moments in which the state code value changes is zero, the state consistency rate is directly set to 1.

[0029] Obtain the baseline state consistency rate for each electrical node within the corresponding time window under historical normal operating conditions. The corresponding time window under historical normal operating conditions is consistent with the time window used to calculate the baseline central tendency metric, and its duration is the same as the preset sliding window duration. Obtain the state code values ​​for all sampling moments within this time window, and count the number of transitions and the total number of sampling moments where the state code values ​​change in the same way as the current sliding window, and calculate the state consistency rate as the baseline state consistency rate. Use the difference between the state consistency rate in the current sliding window and the baseline state consistency rate as the action reliability factor for each electrical node in the current sliding window, i.e.:

[0030] A positive reliability factor indicates that the switching pattern is more consistent with historical normal operating conditions, while a negative value indicates abnormal fluctuations.

[0031] For each electrical node, the steady-state deviation factor within the current sliding window is amplified according to a preset steady-state influence weight to obtain the steady-state contribution value. The preset steady-state influence weight is denoted as... Steady-state contribution value The reliability factor of the action within the current sliding window is amplified according to a preset action influence weight to obtain the action contribution value. The preset action influence weight is denoted as... Action contribution value Action reliability factor. The values ​​of steady-state influence weight and action influence weight are determined based on the type of electrical node in the distribution network. For power source side nodes, Take 0.7, Take 0.3; for load-side nodes, Take 0.5, The value is set to 0.5. The steady-state contribution value and the action contribution value are superimposed to obtain the fault tendency coefficient of each electrical node within the current sliding window. The calculation method is as follows Operational reliability factor. The fault tendency coefficient comprehensively reflects the probability of failure of electrical nodes in terms of both steady-state operational deviation and abnormal switching operation.

[0032] See Figure 3 In the figure, the horizontal axis represents the sequence number of the preset sliding window, and the vertical axis represents the factor value. The blue solid curve represents the steady-state deviation factor calculated for each preset sliding window of the electrical node. It shows a gradual upward trend, slowly increasing from approximately 0.05 initially to over 0.2 at the end, indicating that the statistical characteristics of the electrical operating parameters gradually deviate from the historical normal operating condition baseline value in the time series. The orange dashed curve represents the action reliability factor, whose value fluctuates significantly and is mainly distributed in the negative range, indicating that the consistency of the state transition of the switch state sequence has many abnormal fluctuations compared to the historical normal operating condition. The green dotted curve represents the fault tendency coefficient, which comprehensively reflects the steady-state deviation factor and the action reliability factor. The curve rises slowly overall, with a trend between the steady-state deviation factor and the action reliability factor. After approximately sequence number 200, it significantly approaches and remains above the preset start-up threshold (gray dashed line) of 0.09, indicating that the fault tendency of the electrical node gradually appears and exceeds the start-up threshold. The preset start-up threshold serves as the trigger threshold for fault monitoring; the gray dashed line in the figure remains basically horizontal, with a constant value of 0.09.

[0033] Example 2

[0034] In practical implementation, the preset initiation threshold is determined based on the statistical distribution of the fault tendency coefficients of each electrical node before the occurrence of a fault event in the historical operation data of the distribution network. Optionally, the fault tendency coefficients of all electrical nodes within a preset sliding window before each fault occurs in the historical fault records are obtained, and the 95th percentile of these fault tendency coefficients is calculated. The 95th percentile is used as the preset initiation threshold. All electrical nodes with fault tendency coefficients greater than the preset initiation threshold are marked as fault initiation nodes, representing the location where the fault characteristics are first detected.

[0035] Starting from each fault initiation node, a neighborhood search is performed according to the electrical connection paths in the distribution network topology. In practice, the neighborhood search adopts a breadth-first search approach, using the fault initiation node as the starting point. Adjacent electrical nodes with direct electrical connection paths to the fault initiation node are designated as the first-level neighborhood nodes. Then, the adjacent electrical nodes of each of the first-level neighborhood nodes are designated as the second-level neighborhood nodes, and so on, searching outwards layer by layer until all electrical nodes encountered during the search are included in the initial influence area. The search depth of the neighborhood search is set according to the feeder length and the number of branches in the distribution network. The search depth is set to traverse a maximum of 5 levels of adjacent electrical nodes from the fault initiation node. When the number of feeder branches in the distribution network exceeds 20, the search depth is adjusted to traverse a maximum of 3 levels of adjacent electrical nodes.

[0036] Obtain the fault tendency coefficient of each electrical node within the initial influence area. The value of the preset expansion threshold is dynamically determined based on the distribution characteristics of the fault tendency coefficient within the initial influence area. In specific implementation, the arithmetic mean of the fault tendency coefficients of all electrical nodes within the initial influence area is calculated, and the result multiplied by 0.8 is used as the preset expansion threshold. The fault tendency coefficient of each electrical node within the initial influence area is compared one by one with the preset expansion threshold. If the fault tendency coefficient of an electrical node on the boundary of the initial influence area is greater than the preset expansion threshold, then the adjacent electrical nodes of that boundary electrical node that are not included in the initial influence area are included in the fault influence area, thus expanding the boundary. If the fault tendency coefficient of an electrical node on the boundary of the initial influence area is less than or equal to the preset expansion threshold, then the current boundary remains unchanged. If there are multiple consecutive electrical nodes within the initial influence area whose fault tendency coefficients are all less than or equal to the preset expansion threshold and these electrical nodes are located on one side of the edge of the initial influence area, then these electrical nodes are removed from the initial influence area, thus shrinking the boundary. The set of electrical nodes obtained after boundary expansion or shrinkage adjustment is the fault influence area.

[0037] Electrical nodes directly connected to electrical nodes outside the fault-affected area within the fault-affected area are designated as area boundary nodes. In practice, each electrical node within the fault-affected area is traversed, and it is checked whether any of its adjacent electrical nodes are outside the fault-affected area. If such a node exists, it is designated as an area boundary node, and electrical nodes outside the fault-affected area that are directly connected to this area boundary node are not included in the area boundary node count.

[0038] The electrical operating parameter sequence of each regional boundary node within the fault-affected area is obtained. The acquisition time for the electrical operating parameter sequence is the length of a preset sliding window before and after the fault occurrence time. Simultaneously, the electrical operating parameter sequence of the fault initiation node corresponding to each regional boundary node is obtained. The correspondence between the fault initiation node and the regional boundary node is established during the neighborhood search process. That is, the path from a fault initiation node to a regional boundary node through neighborhood search constitutes the correspondence. If a regional boundary node can be reached through the neighborhood search of multiple fault initiation nodes, the fault initiation node with the highest fault tendency coefficient is taken as the fault initiation node corresponding to that regional boundary node.

[0039] The electrical operating parameter sequence of each region boundary node is compared with the electrical operating parameter sequence of the corresponding fault initiation node at each sampling time. Optionally, the similarity comparison uses Pearson correlation coefficient calculation. The two electrical operating parameter sequences are aligned one by one according to the sampling time, and the change curve of the Pearson correlation coefficient between the two sequences with the sliding of the time window is calculated. The specific method for determining the feature propagation delay of fault features from the fault initiation node to each region boundary node is as follows: the difference in sampling time when the Pearson correlation coefficient of the two sequences first decreases to less than 0.5 is taken as the feature propagation delay. If the Pearson correlation coefficient is always greater than or equal to 0.5, then the time from the start to the end of the alignment of the sampling times of the two sequences is taken as the feature propagation delay.

[0040] The preset delay threshold is determined based on the electromagnetic transient propagation speed in the distribution network and the maximum electrical distance in the fault-affected area. In specific implementation, the maximum electrical distance between any two electrical nodes within the fault-affected area is obtained. This electrical distance is expressed as the total length of the electrical connection path between the two electrical nodes, in kilometers. The electrical distance is then divided by the propagation speed of electromagnetic waves in the conductor, where the propagation speed of electromagnetic waves in the conductor is taken as... The propagation time is calculated in kilometers per second, and then multiplied by a safety margin factor of 1.5. The product is used as the preset delay threshold.

[0041] Boundary nodes in regions with feature propagation delays less than a preset delay threshold are marked as fast boundary nodes, while those with feature propagation delays greater than or equal to the preset delay threshold are marked as slow boundary nodes. The fault isolation priority of the affected region is determined by the ratio of the number of fast boundary nodes to the number of slow boundary nodes; a higher ratio indicates a higher fault isolation priority. When the number of slow boundary nodes is zero, the ratio is the number of fast boundary nodes multiplied by 100. The formula for calculating fault isolation priority is:

[0042] in, Indicates the fault isolation priority for the affected area; This indicates the number of fast boundary nodes within the affected area of ​​the fault. Integers greater than or equal to zero; This indicates the number of slow boundary nodes within the affected area of ​​the fault. Integers greater than or equal to zero; This is a preset small positive constant, with a value of 0.001, set to avoid cases where the denominator is zero. When When it is zero, Values Divide by The result.

[0043] Example 3

[0044] In practice, the affected areas are sorted according to their fault isolation priority from highest to lowest to obtain a region isolation sequence. The fault isolation priority is determined based on the priority set in the aforementioned embodiments. Sort the values ​​in descending order. The higher the fault-affected area's value, the higher its position in the area isolation sequence. When multiple fault-affected areas have the same fault isolation priority value, the number of fault-initiating nodes contained within these areas is compared; the area containing more fault-initiating nodes is ranked higher in the area isolation sequence.

[0045] For the area affected by the current fault in the regional isolation sequence, obtain the switch identifiers corresponding to all regional boundary nodes in the affected area. In specific implementation, each regional boundary node corresponds to a remotely operable switchgear within the primary and secondary integrated ring network box. The switch identifier uses the unique address code of the switchgear within the primary and secondary integrated ring network box. The address code format is a combination of the substation identifier of the primary and secondary integrated ring network box and the port number of the switchgear within that substation. The switch identifier is obtained by querying the distribution network equipment ledger database and retrieving the associated switchgear address code based on the electrical node identifier corresponding to the regional boundary node.

[0046] Based on the real-time trend of the electrical operating parameter sequence of the boundary nodes in the current fault-affected area, the expected operating time of the switch at each boundary node is determined. In practice, the electrical operating parameter sequence of each boundary node in the current fault-affected area within the most recent preset sliding window is obtained; this sequence is selected as the instantaneous current value sequence. A linear fit is performed on the instantaneous current value sequence of the boundary nodes, and the slope of the current value sequence as a function of the sampling time is calculated. The slope represents the rate and direction of current change. A preset trend threshold is set based on the rate of change of the maximum load current during normal operation of the distribution network. The trend threshold is calculated as 10% of the maximum load current of the distribution network divided by the time length corresponding to one power frequency cycle. When the absolute value of the slope of the instantaneous current value sequence at the regional boundary node is consistently greater than the trend threshold and the slope is positive, it indicates that the short-circuit current at the regional boundary node is still in the rising phase. The expected operating time of the switch is determined to be half a power frequency cycle delay after the current slope reaches the preset current peak reference slope. When the absolute value of the slope of the instantaneous current value sequence at the regional boundary node is consistently greater than the trend threshold and the slope is negative, it indicates that the short-circuit current at the regional boundary node has passed its peak and is in the falling phase. The expected operating time of the switch is determined to be a quarter power frequency cycle delay after the current sampling time. When the absolute value of the slope of the instantaneous current value sequence at the regional boundary node is less than or equal to the trend threshold, it indicates that the short-circuit current at the regional boundary node has tended to stabilize. The expected operating time of the switch is determined to be one-eighth power frequency cycle delay after the current sampling time. The preset current peak reference slope is calculated by dividing the maximum short-circuit current peak value at the regional boundary node of the distribution network by the duration from the fault initiation time to the short-circuit current peak value time.

[0047] After determining the expected operating time of the switches at each regional boundary node, all switches at all regional boundary nodes are sorted in ascending order of expected operating time, forming an expected operating time sequence list. Following this ascending order of expected operating time, tripping control commands are sent to the corresponding primary and secondary integrated ring network boxes, controlling the corresponding switches to perform tripping actions. In practice, the tripping control commands are issued through the communication network of the distribution network automation system, using either fiber optic Ethernet or 5G wireless communication. The tripping control commands include switch identifiers and tripping execution command codes. After each switch performs a tripping action, the primary and secondary integrated ring network box collects switch opening / closing indication signals through the switch auxiliary contacts and transmits the tripping completion status information back to the distribution network automation system.

[0048] After all the switches at the boundary nodes of the current fault-affected area have completed their tripping actions, the distribution network automation system confirms that the current fault-affected area has been completely isolated from the distribution network. After the current fault-affected area is isolated, the next fault-affected area immediately following in the sequence is selected from the area isolation sequence. The steps of obtaining the switch identifier corresponding to the area boundary node, determining the expected action time of the switch, and controlling the switch to perform the tripping action are repeated until all fault-affected areas in the area isolation sequence have been isolated.

[0049] After a fault-affected area is isolated, the distribution network automation system continuously monitors the electrical operating parameter sequence of the corresponding fault initiation node and all electrical nodes within that area. If the electrical operating parameter sequence within the isolated fault-affected area recovers to the normal operating parameter range within a preset observation period, and there is no recurrence of a fault tendency coefficient exceeding the preset activation threshold, then the fault isolation operation in the current fault-affected area is marked as active. The preset observation period is set to 3 to 10 seconds depending on the fault type.

[0050] Example 4

[0051] In specific implementation, please refer to Figure 2 This section describes the process of obtaining the available power supply capacity of all power nodes in the non-faulty areas of the distribution network after fault isolation. Power nodes include distributed power source access points that are unaffected by the fault and still operating normally, low-voltage bus outgoing terminals of substations, and grid-connected points of energy storage systems. The available power supply capacity is obtained by querying the current output power and rated capacity of each power node in the distribution network energy management system. The available power supply capacity equals the rated capacity of the power node minus the current output power, and then minus the preset spinning reserve capacity margin, which is 5% of the rated capacity of the power node. The next section describes obtaining the demand capacity of all load nodes in the non-faulty areas. Load nodes include low-voltage load access points of distribution transformers that are unaffected by the fault and not disconnected during fault isolation. The demand capacity is obtained by querying the current active power demand value of each load node in the distribution network energy management system.

[0052] Based on the available power supply capacity of power supply nodes and the demand capacity of load nodes, load nodes whose available power supply capacity exceeds their demand capacity are marked as recoverable nodes. In practice, the demand capacity of each load node is compared one by one with the available power supply capacity of all power supply nodes. If the available power supply capacity of at least one power supply node exceeds the demand capacity of a load node, that load node is marked as a recoverable node. Load nodes whose demand capacity exceeds the available power supply capacity of all power supply nodes are marked as temporarily unrecoverable nodes, and these temporarily unrecoverable nodes are not included in the subsequent recovery path determination.

[0053] Starting from each recoverable node, a power supply path from any power source node to the recoverable node is searched within the fault-isolated distribution network topology. In practice, the search employs an undirected graph traversal algorithm based on the fault-isolated distribution network topology. The distribution network topology is abstracted as an undirected graph, where vertices represent electrical nodes and edges represent electrical connection paths. For each recoverable node, a depth-first search is performed, starting from the recoverable node. During the search, the search proceeds only along electrical connection paths that are not fault-isolated, recording the vertices and edges traversed. When any power source node is encountered, the recorded vertex and edge sequences are considered as a power supply path. All searched power supply paths are obtained, and the path with the fewest switches is selected as the candidate recovery path for the recoverable node. The number of switches refers to the total number of switching devices within the primary and secondary integrated ring network boxes along the power supply path, including circuit breakers and load switches.

[0054] All candidate recovery paths for recoverable nodes are sorted in ascending order of path length. Path length is represented by the number of switches traversed along the candidate recovery path. When two or more candidate recovery paths have the same number of switches, the ratio of the available power supply capacity of the power nodes on these candidate recovery paths to the required capacity of the recoverable nodes is compared. The candidate recovery path with the larger ratio is ranked higher in the order. The sorted candidate recovery paths are then sequentially designated as recovery paths, with the first candidate recovery path becoming the first recovery path, the second candidate recovery path becoming the second recovery path, and so on. The closing sequence of switches within the recovery path is determined to be consistent with the order of the recovery paths; that is, the switches on the first recovery path are closed first, followed by the switches on the second recovery path.

[0055] The current phase value of the electrical operating parameter sequence of the first power node in the current recovery path is obtained. The electrical operating parameter sequence is selected as the voltage instantaneous value sequence. In specific implementation, the voltage instantaneous value sequence within the most recent power frequency cycle is collected through the voltage transformer configured at the first power node. The current phase value is extracted from the voltage instantaneous value sequence using a phase-locked loop (PLL) algorithm. The input of the PLL algorithm is the voltage instantaneous value sequence, and the output is the current phase value, which ranges from 0 to 360 degrees. The current phase value of the electrical operating parameter sequence of the last load node in the current recovery path is obtained in the same way as at the first power node, by collecting the voltage instantaneous value sequence through the voltage transformer configured at the last load node and extracting the current phase value.

[0056] Calculate the phase difference between the current phase value of the first power node and the current phase value of the last load node. The phase difference is the absolute value of the current phase value of the first power node minus the current phase value of the last load node. The phase difference ranges from 0 to 180 degrees. When the absolute value is greater than 180 degrees, use 360 ​​degrees minus the absolute value as the phase difference.

[0057] The preset phase synchronization threshold is determined based on the allowable phase difference of closing the switchgear in the primary and secondary integrated ring network box. In specific implementation, the technical parameter manual of the switchgear is consulted to obtain the nominal maximum allowable phase difference of closing the switchgear. The typical value of the maximum allowable phase difference of closing is 30 degrees. The maximum allowable phase difference of closing is multiplied by a safety factor of 0.8, and the product is used as the preset phase synchronization threshold, which is set to 24 degrees.

[0058] If the phase difference is less than the preset phase synchronization threshold, the switches on the current recovery path will be controlled to perform closing actions sequentially according to the closing sequence of the current recovery path. The closing action involves sending a closing control command to the corresponding switchgear within the primary and secondary integrated ring network box. The closing control command includes a switch identifier and a closing execution command code. The switch identifier uses the unique address code of the switchgear. The execution sequence of the closing action is as follows: starting from the first switch closest to the first-end power node on the current recovery path, closing is performed sequentially towards the last-end load node. After each switch performs the closing action, a preset switch action confirmation time is waited for. After confirming the closing completion status information returned by the switch's auxiliary contacts, the next switch is controlled to perform the closing action. The preset switch action confirmation time is determined based on the inherent closing time of the switchgear, and is set to 1.5 times the inherent closing time.

[0059] If the phase difference is greater than or equal to the preset phase synchronization threshold, the voltage output of the first power node in the current recovery path is delayed. The delay operation is executed as follows: a voltage regulation command is sent to the grid-connected interface device of the first power node. This command includes a set value for the voltage output delay duration. The voltage output delay duration of the first power node is dynamically calculated based on the difference between the phase difference and the preset phase synchronization threshold, using a proportional-integral control law. The dynamic calculation formula is:

[0060] in, This indicates the voltage output delay time, in milliseconds. This represents the phase difference between the current phase value of the first-end power node and the current phase value of the last-end load node, expressed in degrees. This indicates the preset phase synchronization threshold, in degrees. This represents the proportional control coefficient. The value is set to 0.5 milliseconds per degree, and the setting is based on the time conversion value corresponding to each phase change of 1 degree within one power frequency cycle multiplied by the proportional gain factor. Represents the integral control coefficient. The value is set to 0.05 milliseconds per second, based on one-tenth of the proportional control coefficient value. This indicates the cumulative time, in seconds, during which the phase difference value remains greater than or equal to the preset phase synchronization threshold. During the delay, the phase difference value is continuously recalculated. Once the phase difference value falls below the preset phase synchronization threshold, the delay stops, and the switches are controlled to perform closing actions sequentially according to the closing sequence of the current recovery path.

[0061] See Figure 4 In the figure, the horizontal axis represents the load node number, sorted from largest to smallest load demand capacity, ranging from 1 to 500; the vertical axis represents capacity, in kilovolt-amperes (kVA). The gray bar chart shows the distribution of demand capacity for each load node, indicating that the demand capacity of each load node gradually decreases from approximately 100 kVA to near zero, exhibiting a long-tail distribution characteristic, with most load nodes having relatively small demand capacity. The blue solid curve and orange dashed curve represent the available power supply capacity of power node A and power node B in the non-faulty area after fault isolation, respectively. The available power supply capacity of power node A is stable between approximately 430 kVA and 480 kVA, with relatively small fluctuations; the available power supply capacity of power node B is stable between approximately 370 kVA and 400 kVA, with even smaller fluctuations. The available power supply capacity of the two power nodes is significantly higher than the demand capacity of the vast majority of load nodes, satisfying the criteria for recoverable nodes in Example 4.

[0062] Example 5

[0063] In practice, the voltage instantaneous value sequence of each electrical node is collected by the voltage transformer configured in the primary and secondary integrated ring network box. The primary winding of the voltage transformer is connected in parallel with the primary conductive circuit of the electrical node. The output of the secondary winding of the voltage transformer is connected to the analog quantity acquisition unit in the primary and secondary integrated ring network box through the signal conditioning circuit. The analog quantity acquisition unit performs analog-to-digital conversion on the voltage instantaneous value at a preset sampling frequency. The time interval between adjacent sampling moments in the voltage instantaneous value sequence is equal to the power frequency period divided by the number of sampling points in each power frequency period. The number of sampling points in each power frequency period is set to 128 points, and the corresponding sampling frequency is 6400 Hz. The instantaneous current value sequence of each electrical node is collected by current transformers configured within the primary and secondary integrated ring main unit. The primary winding of the current transformer is connected in series to the primary conductive circuit of the electrical node. The output of the secondary winding of the current transformer is connected to the analog signal acquisition unit within the primary and secondary integrated ring main unit via a signal conditioning circuit. The analog signal acquisition unit performs analog-to-digital conversion on the instantaneous current value at the same sampling frequency as the instantaneous voltage value sequence. The time interval between adjacent sampling moments in the instantaneous current value sequence is consistent with the time interval between adjacent sampling moments in the instantaneous voltage value sequence. The instantaneous voltage value sequence and the instantaneous current value sequence together serve as the electrical operating parameter sequence. The unit of each sampling moment in the instantaneous voltage value sequence is volts, and the unit of each sampling moment in the instantaneous current value sequence is amperes.

[0064] The switch opening and closing indication signal sequence of each electrical node is collected through the auxiliary contacts configured in the primary and secondary integrated ring network box. The auxiliary contacts are mechanically linked to the operating mechanism of the switchgear in the primary and secondary integrated ring network box. When the switchgear is in the closed position, the auxiliary contacts output a closed status signal; when the switchgear is in the open position, the auxiliary contacts output an open status signal. The status signals of the auxiliary contacts are hard-wired into the digital quantity acquisition unit in the primary and secondary integrated ring network box. The digital quantity acquisition unit collects the status of the auxiliary contacts at the same sampling frequency as the analog quantity acquisition unit. At each sampling moment, the digital quantity acquisition unit reads the on / off status of the auxiliary contacts, encodes the closed status signal as a value 1, and the open status signal as a value 0, forming a switch opening and closing indication signal sequence. The value at each sampling moment in the switch opening and closing indication signal sequence is 1 or 0, and the switch opening and closing indication signal sequence serves as the switch status sequence.

[0065] The electrical operating parameter sequence and switch status sequence are aligned according to the sampling time and then stored in a local buffer queue. The local buffer queue is located in the random access memory of the local control unit within the primary and secondary integrated ring network box. The local buffer queue adopts a ring buffer data structure, and the total storage capacity of the ring buffer is determined based on the preset sliding window duration and sampling frequency. In specific implementation, the preset sliding window duration is set to the length corresponding to 5 power frequency cycles, i.e., 100 milliseconds, and the sampling frequency is 6400 Hz. Each electrical operating parameter sequence contains 640 sampling times within each preset sliding window. Each sampling time simultaneously contains an instantaneous voltage value, an instantaneous current value, and a switch opening / closing indication signal value. Each data element in the local buffer queue is a triplet structure, with the three fields of the triplet structure storing the instantaneous voltage value, instantaneous current value, and switch opening / closing indication signal value, respectively. The total storage capacity of the ring buffer is set to accommodate the data volume corresponding to 120 preset sliding windows, i.e., the ring buffer can store 76,800 data elements.

[0066] The electrical operating parameter sequence and switch status sequence are aligned according to the sampling time as follows: the analog quantity acquisition unit and the switch quantity acquisition unit share the same sampling clock signal. The sampling clock signal is provided by the GPS timing module in the primary and secondary fusion ring network box. The GPS timing module outputs a second pulse signal and absolute time information. The rising edge of the second pulse signal serves as the synchronous trigger signal for the analog quantity acquisition unit and the switch quantity acquisition unit to start sampling. At each sampling time, the analog quantity acquisition unit and the switch quantity acquisition unit simultaneously latch the sampling data. After the sampling data is tagged with an absolute time tag, it is transmitted to the local control unit. The local control unit combines the instantaneous voltage value, instantaneous current value, and switch on / off indicator signal value at the same sampling time into a triplet data element according to the absolute time tag. The triplet data element is written into the ring buffer in the order of the absolute time tag.

[0067] The local cache queue manages data storage according to the first-in, first-out (FIFO) principle. In its implementation, the circular buffer has write and read pointers. The write pointer points to the storage location of the next data element to be written, and the read pointer points to the storage location of the current data element to be read. Each time new sampled data is generated, the triplet data element is written to the storage location pointed to by the write pointer, which then moves forward one position. When the write pointer reaches the end of the circular buffer, it wraps back to the beginning of the circular buffer. The read pointer is used in subsequent self-healing decision-making processes to retrieve data from the local cache queue. The read pointer reads the triplet data element from the circular buffer, moves forward one position after reading, and wraps back to the beginning of the circular buffer when it reaches the end. When the positions of the write and read pointers meet preset conditions, the state of the circular buffer is checked. If the write pointer approaches the read pointer from behind and is about to overlap, writing stops and a buffer overflow alarm signal is generated.

[0068] The electrical operating parameter sequence and switch state sequence within a preset time length are extracted from the local cache queue according to the first-in-first-out principle. The preset time length is determined based on the time length of the preset sliding window required for calculating the steady-state deviation factor and action reliability factor in the subsequent self-healing decision. The preset time length is the sum of the time lengths of 5 preset sliding windows, i.e., 500 milliseconds. The specific method for extracting data is as follows: The local control unit calculates the starting absolute time corresponding to the preset time length based on the absolute time tag of the current sampling time. The starting absolute time is equal to the absolute time of the current sampling time minus the preset time length. Starting from the current position of the pointer in the circular buffer, the control unit searches backward for triplet data elements. All triplet data elements whose absolute time tags are greater than or equal to the starting absolute time and less than or equal to the absolute time of the current sampling time are extracted and arranged in ascending order of absolute time tags. From the extracted triplet data elements, the voltage instantaneous value sequence, the current instantaneous value sequence, and the switch opening and closing indication signal sequence are separated. The voltage instantaneous value sequence and the current instantaneous value sequence are used together as the extracted electrical operating parameter sequence, and the switch opening and closing indication signal sequence is used as the extracted switch state sequence. These sequences are used for the calculation of steady-state deviation factor, action reliability factor, fault tendency coefficient, and fault isolation priority in the subsequent self-healing decision-making process.

[0069] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A self-healing method for primary and secondary fusion ring network boxes based on intelligent decision-making, characterized in that, include: Collect the electrical operating parameter sequence and switch status sequence of each electrical node in the primary and secondary integrated ring network box; Based on the statistical differences in the electrical operating parameter sequences of each electrical node within a preset sliding window, the steady-state deviation factor of each electrical node within the corresponding sliding window is determined. Based on the consistency of the state transitions of the switching state sequence of each electrical node within a continuous sliding window, the reliability factor of the operation of each electrical node within the corresponding sliding window is determined. Based on the steady-state deviation factor and operational reliability factor of each electrical node, determine the fault tendency coefficient of each electrical node within the corresponding sliding window; Based on the connection relationship of electrical nodes with fault tendency coefficients greater than the preset start threshold in the distribution network topology, the fault-affected area and the boundary nodes of the area are determined; Based on the propagation delay of the electrical operating parameter sequence of the regional boundary node within the fault-affected area, the fault isolation priority of the fault-affected area is determined. Based on the fault isolation priority, fault isolation is achieved by adjusting the action timing of the corresponding switches in the primary and secondary fusion ring network box; Based on the distribution network topology after fault isolation, the recovery path and closing sequence are determined by utilizing the power capacity and load demand of the remaining non-faulty areas. Based on the recovery path and closing sequence, control the closing action of the corresponding switches in the primary and secondary integrated ring network box to complete the fault recovery.

2. The self-healing method for a primary and secondary fusion ring network box based on intelligent decision-making according to claim 1, characterized in that, The step of determining the steady-state deviation factor of each electrical node within a corresponding sliding window based on the statistical characteristic differences of the electrical operating parameter sequences of each electrical node within a preset sliding window includes: Obtain the electrical operating parameter sequence of each electrical node and the parameter values ​​at multiple sampling times within each preset sliding window; For each electrical node, calculate the central tendency measure of the parameter values ​​at all sampling times within the current sliding window, and calculate the discrete measure of the parameter values ​​at all sampling times within the current sliding window relative to the central tendency measure. Obtain the baseline central tendency metric and baseline discrete metric of the parameter values ​​of each electrical node within the corresponding time window under historical normal operating conditions; The steady-state deviation factor of each electrical node in the current sliding window is obtained by weighted summing the difference between the central tendency measure and the benchmark central tendency measure, and the difference between the discrete measure and the benchmark discrete measure.

3. The self-healing method for a primary and secondary fusion ring network box based on intelligent decision-making according to claim 2, characterized in that, The step of determining the operational reliability factor of each electrical node within a corresponding sliding window based on the consistency of state transitions in the switching state sequence of each electrical node within a continuous sliding window includes: Obtain the state encoding values ​​of the switching state sequence of each electrical node at multiple sampling times within each preset sliding window; For each electrical node, count the number of transitions where the state code value changes between adjacent sampling times within the current sliding window, and count the total number of sampling times where the state code value changes within the current sliding window; The state consistency rate of each electrical node within the current sliding window is determined based on the ratio between the number of transitions within the current sliding window and the total number of changes in the state code value. Obtain the baseline state consistency rate of each electrical node within the corresponding time window under historical normal operating conditions; The difference between the state consistency rate within the current sliding window and the baseline state consistency rate is used as the reliability factor for the operation of each electrical node within the current sliding window.

4. The self-healing method for a primary and secondary fusion ring network box based on intelligent decision-making according to claim 3, characterized in that, The step of determining the fault tendency coefficient of each electrical node within the corresponding sliding window based on the steady-state deviation factor and operational reliability factor of each electrical node includes: For each electrical node, the steady-state deviation factor within the current sliding window is amplified according to a preset steady-state influence weight to obtain the steady-state contribution value; The reliability factor of the action within the current sliding window is amplified according to the preset action influence weight to obtain the action contribution value; The steady-state contribution value and the action contribution value are superimposed to obtain the fault tendency coefficient of each electrical node within the current sliding window.

5. The self-healing method for a primary and secondary fusion ring network box based on intelligent decision-making according to claim 4, characterized in that, The step of determining the fault-affected area and the boundary nodes of the area based on the connection relationship of electrical nodes with fault tendency coefficients greater than a preset initiation threshold in the distribution network topology includes: Mark all electrical nodes with a fault tendency coefficient greater than the preset start threshold as fault initiation nodes; Starting from each fault initiation node, a neighborhood search is performed according to the electrical connection path in the distribution network topology, and all electrical nodes found are included in the initial influence area. Based on the comparison between the fault tendency coefficient of each electrical node in the initial influence area and the preset expansion threshold, the boundary of the initial influence area is expanded or contracted to obtain the fault influence area. Electrical nodes that are directly connected to electrical nodes outside the fault-affected area are designated as area boundary nodes.

6. The self-healing method for a primary and secondary fusion ring network box based on intelligent decision-making according to claim 5, characterized in that, The step of determining the fault isolation priority of the affected area based on the propagation delay of the electrical operating parameter sequence of the regional boundary node within the affected area includes: Obtain the electrical operating parameter sequence of each regional boundary node within the fault-affected area, and obtain the electrical operating parameter sequence of the fault initiation node corresponding to each regional boundary node; The electrical operating parameter sequence of each region boundary node is compared with the electrical operating parameter sequence of the corresponding fault initiation node at each sampling time to determine the feature propagation delay of the fault feature from the fault initiation node to each region boundary node. Region boundary nodes with feature propagation delay less than a preset delay threshold are marked as fast boundary nodes, and region boundary nodes with feature propagation delay greater than or equal to the preset delay threshold are marked as slow boundary nodes. The fault isolation priority of the affected area is determined by the ratio of the number of fast boundary nodes to the number of slow boundary nodes within the affected area. The larger the ratio, the higher the fault isolation priority.

7. The self-healing method for a primary and secondary fusion ring network box based on intelligent decision-making according to claim 6, characterized in that, The process of achieving fault isolation by adjusting the timing of corresponding switches within the primary and secondary fusion ring network box according to fault isolation priority includes: According to the fault isolation priority from high to low, the affected areas of each fault are sorted to obtain the area isolation sequence; For the current fault-affected area in the regional isolation sequence, obtain the switch identifiers corresponding to all regional boundary nodes of the fault-affected area; Based on the real-time changing trend of the electrical operating parameter sequence of the area boundary nodes in the current fault-affected area, determine the expected operating time of the switch of each area boundary node; Following the expected action times from smallest to largest, control the corresponding switches to perform tripping actions. After all switches at the boundary nodes of the current fault-affected area have completed tripping actions, continue processing the next fault-affected area in the area isolation sequence.

8. The self-healing method for a primary and secondary fusion ring network box based on intelligent decision-making according to claim 7, characterized in that, The process of determining the recovery path and closing sequence based on the distribution network topology after fault isolation, utilizing the power capacity and load demand of the remaining non-faulty areas, includes: Obtain the available power supply capacity of power nodes in all non-faulty areas of the distribution network after fault isolation, and obtain the required capacity of load nodes in all non-faulty areas. Based on the available power supply capacity of the power supply node and the demand capacity of the load node, the load node with the available power supply capacity greater than the demand capacity is marked as a recoverable node; Starting from each recoverable node, search for the power supply path from any power source node to the recoverable node in the distribution network topology after fault isolation, and take the path with the fewest switches among all power supply paths as the candidate recovery path for the recoverable node. Sort all candidate recovery paths for recoverable nodes in ascending order of path length, determine the sorted candidate recovery paths as recovery paths in sequence, and determine the closing order of switches in the recovery paths to be consistent with the sorting of the recovery paths.

9. The self-healing method for a primary and secondary fusion ring network box based on intelligent decision-making according to claim 8, characterized in that, The process of controlling the closing action of the corresponding switches in the primary and secondary integrated ring network box according to the recovery path and closing sequence to complete the fault recovery includes: Obtain the current phase value of the electrical operating parameter sequence of the first power node in the current recovery path, and obtain the current phase value of the electrical operating parameter sequence of the last load node in the current recovery path; Calculate the phase difference between the current phase value of the first power node and the current phase value of the last load node; If the phase difference is less than the preset phase synchronization threshold, the switches on the current recovery path will be controlled to perform closing actions in sequence according to the closing order of the current recovery path. If the phase difference is greater than or equal to the preset phase synchronization threshold, the voltage output of the first power node of the current recovery path is delayed until the phase difference is less than the preset phase synchronization threshold. Then, the control switches are used to perform closing actions in sequence according to the closing sequence of the current recovery path.

10. The self-healing method for a primary and secondary fusion ring network box based on intelligent decision-making according to claim 9, characterized in that, The voltage output delay time of the first-end power node is dynamically calculated and determined according to the difference between the phase difference and the preset phase synchronization threshold, using a proportional-integral control law.

Images