Intelligent park emergency resource scheduling method based on cloud collaboration

Abstract

The application relates to the technical field of park resource scheduling, and particularly discloses a smart park emergency resource scheduling method based on cloud cooperation, which comprises the following steps: acquiring a real-time high-frequency electrical waveform sequence and a protection account of a power shortage capacity demand and a candidate path, extracting a static vulnerability index and a reverse transient harmonic distortion rate and an inrush peak value, and mapping the static vulnerability index and the reverse transient harmonic distortion rate and the inrush peak value into a dynamic implicit failure probability. If an available path set is empty, a multi-source power flexible splitting cycle is triggered, the capacity demand is split into sub demands, and a transient stress penalty coefficient is replayed until the failure probability of each sub path is lower than a threshold value, so that a multi-path cooperative scheduling strategy is generated. The method accurately avoids the risk of protection misoperation and provides robust support for emergency mutual aid.

Description

Technical Field

[0001] This invention belongs to the field of park resource scheduling technology, and relates to a smart park emergency resource scheduling method based on cloud collaboration. Background Technology

[0002] Cross-park emergency power sharing is a crucial means of ensuring the continuity of power supply to critical loads in smart parks under extreme conditions. When the emergency power capacity of a single park is insufficient, redundant power generation resources from adjacent parks can be mobilized through distribution network interconnection lines, effectively reducing the scope and duration of power outages. Currently, emergency resource dispatch generally follows the principle of prioritizing available capacity or physical proximity, meaning that, under the premise of meeting the transmission capacity constraints of the interconnection lines, priority is given to selecting adjacent parks with the largest supporting power margin or the shortest electrical distance as power sources.

[0003] However, distribution network lines are not simply power transmission channels; the relay protection devices installed along their routes are the core safeguards for ensuring the safe operation of the system. Emergency power reversal will change the normal power flow direction of the lines, causing some sections to experience reverse current stress. There are significant differences in the configuration level and operating status of protection devices along different reversal paths: some lines still use electromagnetic relays with long service lives, which have limited transient tolerance margins to reverse power flow; while some lines have been upgraded to microprocessor-based protection and support remote switching of setting groups, the latent failure risk under reverse conditions is unevenly distributed due to differences in maintenance cycles, historical operating frequencies, and component aging. If the reversal path selection is based solely on capacity and distance, ignoring the differentiated failure probabilities of protection devices along the route under atypical operating conditions, a malfunction in the protection system due to latent defects during the reversal process will directly lead to the disconnection of the reversal channel. This will not only fail to restore power supply but may also trigger cascading faults, expanding the scope of the power outage.

[0004] It is evident that current emergency resource scheduling methods fail to incorporate the implicit failure risk of protection devices into the quantitative assessment framework during the path decision-making stage, and lack a collaborative optimization mechanism between mutual capacity demand and path protection reliability, which constitutes a key shortcoming of existing technologies in complex distribution network emergency scenarios. Summary of the Invention

[0005] In view of the problems existing in the prior art, the present invention provides a cloud-based collaborative smart park emergency resource scheduling method to solve the above-mentioned technical problems.

[0006] To achieve the above and other objectives, the technical solution adopted by the present invention is as follows: This invention provides a cloud-based collaborative method for emergency resource scheduling in smart parks, the method comprising: Acquire the power shortage capacity demand data of the target park, and receive real-time high-frequency electrical waveform sequences and protection device ledger data uploaded by edge control nodes on the candidate mutual aid path of adjacent parks; The static vulnerability index is calculated by extracting the equipment commissioning time and historical maintenance characteristics from the protection device ledger data. The real-time high-frequency electrical waveform sequence is input into the reverse power flow equivalent circuit model to extract the reverse transient harmonic distortion rate and reverse inrush peak value under the assumed input power shortage capacity demand data. The reverse transient harmonic distortion rate and the reverse inrush peak value are jointly mapped to the transient stress penalty coefficient, and combined with the static vulnerability index, the dynamic implicit failure probability of the corresponding candidate mutual aid path is generated. Candidate mutual aid paths with a dynamic implicit failure probability lower than a preset risk threshold are selected as the initial scheduling set; If the initial scheduling set is empty, a multi-source power flexible splitting loop is triggered. The multi-source power flexible splitting loop includes: splitting the power shortage capacity demand data into at least two sets of sub-capacity demand data according to a predetermined proportional step size matrix. Based on the capacity requirement data of each group, the corresponding re-enactment reverse inrush peak value is re-extracted and the transient stress penalty coefficient is updated to calculate the updated implicit failure probability of the corresponding split path. If the probability of hidden failure after the update of the corresponding split path is lower than the preset risk threshold, the multi-source power flexible split cycle will be exited, and scheduling strategy data containing multi-path collaborative power supply instructions will be generated and sent to the edge control node to execute the closing action.

[0007] As described above, the cloud-based collaborative smart park emergency resource scheduling method provided by this invention has at least the following beneficial effects: This invention first acquires the power shortage capacity demand data of the target park, and simultaneously receives real-time high-frequency electrical waveform sequences and protection device ledger data uploaded by edge control nodes on candidate mutual aid paths in adjacent parks. Based on this, the commissioning time, historical maintenance characteristics, and operation change records of each node device are extracted from the protection device ledger data to form a static vulnerability index that reflects the inherent aging degree and current health status of the equipment. Subsequently, the real-time high-frequency electrical waveform sequence is input into the reverse power flow equivalent circuit model, and the transient electromagnetic response of the mutual aid path is pre-simulated under the assumption of input power shortage capacity demand data. Then, the reverse transient harmonic distortion rate and reverse inrush peak value are extracted, and the two are further mapped together into a transient stress penalty coefficient, which is then coupled with the static vulnerability index to generate the dynamic latent failure probability of the corresponding candidate mutual aid path. The above approach not only uses capacity or path distance as the basis for selecting mutual assistance paths, but also incorporates equipment health aging, transient impact intensity, and operational safety boundaries into the same evaluation link. This elevates the selection of mutual assistance paths from whether they can supply power to whether they can provide stable power supply under controllable risks, thereby significantly reducing the probability of protection malfunctions caused by reverse power flow injection and improving the feasibility and success rate of emergency mutual assistance in complex power grid topologies.

[0008] The present invention further selects candidate mutual assistance paths below a preset risk threshold as the initial scheduling set based on the dynamic latent failure probability. If the initial scheduling set is empty, a multi-source power flexible splitting loop is automatically triggered, which splits the power shortage capacity demand data into at least two sets of sub-capacity demand data according to a predetermined proportional step size matrix. Based on each set of sub-capacity demand data, the corresponding re-enactment reverse inrush peak value is re-extracted, the transient stress penalty coefficient is updated, and the updated latent failure probability of the corresponding splitting path is calculated until each splitting path meets the preset risk threshold requirement. Then, the splitting loop is exited and a scheduling strategy data containing multi-path collaborative power supply instructions is generated and sent to the edge control node to execute the closing action. By introducing a flexible splitting mechanism for insufficient capacity and a dynamic verification mechanism for path risks, the system can automatically switch to a multi-source collaborative bearing mode when a single path cannot meet safety conditions. This disperses the centralized power supply demand, which might otherwise fail due to overload, to multiple manageable mutual aid branches, thereby avoiding dispatch interruptions caused by insufficient capacity at a single point or high risk in a single path. At the same time, by using edge control nodes to execute closing actions in a local closed loop, the command response link can be further shortened, improving the real-time performance of dispatch and the proactive nature of fault handling. This gives the entire mutual aid process the technical advantages of controllable risk, stable execution, and fast emergency recovery. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram showing the connections between the steps of the method of the present invention. Detailed Implementation

[0011] The following description, in conjunction with the implementation of this invention, is merely an example and illustration of the concept of this invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the inventive concept or exceed the scope defined in these claims, all of which should fall within the protection scope of this invention.

[0012] Traditional emergency mutual aid dispatching systems often rely solely on power shortage capacity requirements, line length, or the rated transmission capacity of a single path for path selection and dispatching decisions. They lack the ability to uniformly model the health status of protection devices, transient impact responses, and path-level latent failure risks on candidate mutual aid paths in adjacent industrial parks. When a power shortage occurs in the target industrial park, even if some candidate mutual aid paths meet the power supply requirements in terms of static capacity, if the protection devices along the line have issues such as long service life, missing maintenance records, or degraded operating conditions, they may still malfunction due to excessive transient stress during reverse power flow injection, leading to mutual aid failure, cascaded protection tripping, or even power restoration interruption. Because traditional methods lack dynamic identification of the coupling relationship between equipment aging and electrical transient stress, the selection of dispatching targets often becomes disconnected from actual operational safety boundaries, failing to truly reflect the feasibility of mutual aid paths under complex operating conditions.

[0013] For example, in a scenario where power outage occurs in Park A and Parks B and C are capable of providing support, if Park B is prioritized solely based on available capacity or distance, while ignoring factors such as the long operational history, low frequency of historical maintenance, and insufficient adaptability to reverse power flow of the protection devices along the candidate mutual support path between Park B and Park A, a high peak reverse inrush current and transient harmonic distortion rate may occur at the moment of closing the circuit breaker. This could cause the protection devices to misinterpret the current as a fault and activate prematurely. In this case, although the static capacity may appear sufficient to supply power, the support path may fail rapidly due to protection maloperation during actual operation, forcing the system to reschedule, prolonging the emergency response time, and even leaving loads that could have been restored in a state of power outage. This problem demonstrates that focusing solely on "whether power can be supplied" while ignoring "whether power can be supplied safely" will cause dispatch decisions to remain at the superficial capacity matching level, making it difficult to proactively control the dynamic failure risks during the mutual support process.

[0014] Without addressing the aforementioned issues, existing emergency mutual aid systems will struggle to accurately identify highly sensitive channels that "meet capacity requirements but have high risk" in complex network environments with multiple candidate paths and protection devices. They also will be unable to promptly switch to capacity splitting and path reconstruction strategies when initial scheduling fails. Specifically, traditional systems typically assess the status of protection devices at a static, ledger-style record level, failing to jointly analyze equipment commissioning time, historical maintenance characteristics, and transient impact characteristics reflected in real-time high-frequency electrical waveforms. Therefore, it is difficult to predict the latent failure probability of different paths under reverse power flow before scheduling. Furthermore, without a flexible mechanism for splitting power shortage capacity demands, when a single path cannot meet the safety threshold, the system can only abandon mutual aid or execute a coarse allocation, reducing the power restoration rate and weakening the scheduling flexibility of edge control nodes in local closed-loop execution. Ultimately, this leads to a significant decrease in the robustness and success rate of emergency scheduling.

[0015] When faced with the above problems, this invention links the power shortage capacity demand data of the target park with the real-time high-frequency electrical waveform sequence and protection device ledger data on the candidate mutual aid paths of adjacent parks. First, a static vulnerability index is constructed based on the equipment commissioning time and historical maintenance characteristics. Then, based on the reverse power flow equivalent circuit model, the reverse transient harmonic distortion rate and reverse inrush peak value under the assumed power shortage capacity condition are extracted. After mapping the two to the transient stress penalty coefficient, they are coupled with the static vulnerability index to generate a dynamic hidden failure probability, thereby achieving pre-screening of the safety of candidate mutual aid paths. When the initial scheduling set is empty, a multi-source power flexible splitting loop is triggered to decompose the power shortage capacity demand into at least two sets of sub-capacity demand data according to a predetermined proportional step size matrix. The hidden failure probability of each split path is re-evaluated and updated until a feasible collaborative power supply combination that meets the risk threshold is obtained. Then, scheduling strategy data containing multi-path collaborative power supply instructions is generated and sent to the edge control node to execute the closing action. Through this linkage mechanism, the present invention can not only establish a dynamic balance between capacity satisfaction and operational safety, but also quickly switch to multi-source collaboration and capacity splitting mode when a single path is unavailable. This significantly improves the accuracy of emergency mutual assistance path selection, the stability of protection actions, and the real-time performance of edge-side scheduling, thereby effectively reducing the risk of maloperation caused by reverse power flow and enhancing the continuity and feasibility of power restoration at the park level.

[0016] After introducing the basic concept of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Example

[0017] Please see Figure 1 As shown, the cloud-based collaborative smart park emergency resource scheduling method includes: It acquires power shortage capacity demand data for the target park and receives real-time high-frequency electrical waveform sequences and protection device ledger data uploaded by edge control nodes on candidate mutual aid paths of adjacent parks.

[0018] Preferably, before receiving the real-time high-frequency electrical waveform sequence and protection device ledger data uploaded by the edge control nodes on the candidate mutual assistance path of adjacent parks, the method further includes: The control edge node continuously captures periodic data from the local substation bus at a predetermined high-frequency sampling rate to obtain the original electrical waveform data stream. The variational mode decomposition algorithm is used to perform multi-scale frequency domain stripping on the original electrical waveform data stream, removing power frequency background signals and retaining high-frequency transient feature segments; Absolute time synchronization tags and local topology address identifiers are embedded in high-frequency transient feature segments, and then serialized, recombined, and encoded to form a real-time high-frequency electrical waveform sequence.

[0019] In this embodiment of the invention, the scheduling platform first uses a formula The power shortage capacity demand data of the target park were calculated. ,in The rated power of the i-th type of load within the park. The load importance coefficient is preset, with important loads set to 1, and secondary loads adjusted between 0 and 1 according to the power shortage level. A value of 0.8 is recommended to balance power supply requirements and mutual assistance costs. and These represent the real-time output of photovoltaic and energy storage systems within the park.

[0020] At the same time, it receives real-time high-frequency electrical waveform sequences and protection device ledger data uploaded by edge control nodes on candidate mutual aid paths in adjacent parks.

[0021] To achieve accurate identification of reverse current surges, the control edge nodes are controlled at a predetermined high-frequency sampling rate before receiving data. Continuous periodic data capture is performed on the local substation busbar to obtain the raw electrical waveform data stream. .

[0022] Subsequently, the edge control node employs a variational mode decomposition algorithm to perform multi-scale frequency domain stripping on the original electrical waveform data stream. This process aims to remove power frequency background signals and retain high-frequency transient characteristic segments. Specifically, by presetting the number of decomposition layers K, the original electrical waveform data stream is stripped in the frequency domain at multiple scales. It is decomposed into a series of eigenmode components with specific center frequencies. Where K is a positive integer greater than 1, and can be selected from 5 to 10 depending on the waveform complexity to balance computational accuracy and real-time performance. Based on this, the transient energy significance index is calculated using a high-frequency feature extraction formula. The formula is as follows: In the above formula, The transient energy significance index has the following dimensions: This index comprehensively reflects the energy density and frequency distribution intensity of high-frequency components; The k-th modal component signal after VMD decomposition has the dimension A; T is the time length of one power frequency cycle in seconds. is the center frequency of the kth modal component, in Hz; This is the power frequency reference frequency, typically taken as 50Hz.

[0023] The above formula is derived from the traditional definition of signal energy combined with a variation of the frequency-weighted operator. Ordinary energy calculations cannot distinguish between low-frequency fluctuations and high-frequency impacts; therefore, by introducing... The weighting term can significantly amplify the weight of transient components with higher center frequencies in the evaluation index, thereby more sensitively capturing high-frequency abrupt signals that may interfere with older electromagnetic relays.

[0024] After feature extraction, absolute time synchronization tags and local topology address identifiers are embedded in the high-frequency transient feature fragments. The absolute time synchronization tags are generated using BeiDou time synchronization; the local topology address identifiers use a preset hexadecimal encoding format. Finally, The sequence, time synchronization tag, and address identifier are serialized, recombined, and encoded to form a real-time high-frequency electrical waveform sequence. This sequence is uploaded to the cloud dispatch center via a secure encrypted tunnel and registered with the model identifier in the protection device ledger data.

[0025] The static vulnerability index is calculated by extracting the equipment commissioning time and historical maintenance characteristics from the protection device ledger data, including: The equipment commissioning time is compared with the predetermined life cycle benchmark value to obtain the equipment aging weight parameter; Extract the previous maintenance time record and internal analog quantity sampling deviation sequence from the historical maintenance characteristics; The time decay factor is calculated based on the interval between the last maintenance time and the current system time. The reference degradation value is obtained by weighting and summing the mean square error of the internal analog quantity sampling deviation sequence using the time decay factor. The static vulnerability index is obtained by multiplying the baseline degradation value by the equipment aging weight parameter.

[0026] In this embodiment of the invention, the actual operating years of the equipment are first calculated by subtracting the extracted equipment commissioning time from the current time. Then, this actual operating years data is compared with a predetermined lifespan benchmark value to obtain the equipment aging weight parameter. The predetermined lifespan benchmark value is recommended to be set to an adjustable range of 10 to 15 years. To match the physical differences in the hardware lifespan of different microcomputer and electromagnetic protection devices, and to prevent excessive depreciation from leading to the accidental deletion of available mutual aid paths, 12 years is recommended as the preferred lifespan benchmark value. The equipment aging weight parameter is calculated based on an exponential decay model. Specifically, the ratio of the actual operating years data to the predetermined lifespan benchmark value is extracted and multiplied by a preset aging sensitivity coefficient. This coefficient is a dimensionless parameter used to characterize the accelerating effect of the park environment on equipment aging, with a default value of 0.2. Finally, the exponential result is calculated with the natural constant e as the base to generate the equipment aging weight parameter.

[0027] Next, the previous maintenance time record and the internal analog quantity sampling deviation sequence are extracted from the historical maintenance characteristics. The internal analog quantity sampling deviation sequence is the numerical sequence of current channel sampling error in the equipment's previous self-test records, with the dimension A. The interval duration is calculated based on the absolute time difference between the previous maintenance time record and the current system time. And calculate the time decay factor accordingly. The specific formula is expressed as follows:

[0028] In the formula This is the time decay factor; The interval duration is in days; The reference time constant is in days. Its value is set according to the statistical data of the average fault interval time of the same type of protection device. The preferred value is usually 30 days, which represents a standard monthly inspection cycle.

[0029] Based on this, the mean square error of the internal analog quantity sampling deviation sequence is weighted and summed using a time decay factor to obtain the baseline degradation value. The specific calculation formula is as follows: In the formula Here, N represents the baseline degradation value, and N is the total number of data points in the sampling bias sequence. Let be the current deviation value of the i-th sampling.

[0030] Finally, the obtained baseline degradation value is directly multiplied by the equipment aging weight parameter calculated at the front end to generate the final static vulnerability index.

[0031] The real-time high-frequency electrical waveform sequence is input into the reverse power flow equivalent circuit model to extract the reverse transient harmonic distortion rate and reverse inrush current peak value under the assumed input power shortage capacity demand data, including: Extract the fundamental component sequence and high-frequency distortion component sequence from a real-time high-frequency electrical waveform sequence; Using the reverse power flow equivalent circuit model, the transient topology characteristics of closing are deduced based on the assumed input power shortage capacity demand data; The fundamental component sequence is injected into the pre-simulated closing transient topological features to solve the estimated closing current impact envelope, and the maximum amplitude point of the estimated closing current impact envelope is extracted as the reverse inrush peak value. The frequency domain energy spectrum distribution matrix is ​​obtained by performing a discrete Fourier transform on the high-frequency distortion component sequence, and the ratio of the total energy of the predetermined harmonic frequency band to the fundamental frequency energy is calculated as the reverse transient harmonic distortion rate.

[0032] In this embodiment of the invention, the dispatching platform first obtains the emergency power shortage capacity demand data of the target park. This data is calculated by the park's energy management based on the real-time deviation between the current load baseline and the output of distributed power sources. Subsequently, the real-time high-frequency electrical waveform sequence is input into the reverse power flow equivalent circuit model to extract the data under the assumed injection... Reverse transient harmonic distortion rate under operating conditions With reverse surge peak .

[0033] In this embodiment of the invention, the construction of the reverse power flow equivalent circuit model does not directly use the geographical connections of the candidate mutual assistance paths as the object. Instead, it first performs electrical equivalent conversion on the cable parameters, transformer parameters, and load-side power shortage capacity demand data in the candidate mutual assistance paths, and then maps the conversion results uniformly to series equivalent resistance. With equivalent inductance This is to form a lumped parameter model for reverse power flow transient analysis. Specifically, the cable material resistivity of candidate mutual aid paths is first extracted from protection device ledger data and line asset ledger data. Cable cross-sectional area S, actual laying length and inductance per unit length ,in The dimensions are , The dimension is H / m; then the relationship is calculated based on the conductor resistance. The cable foundation resistance is determined, and combined with the transient dominant frequency extracted in the previous steps. and power frequency reference frequency Through frequency correction term The cable resistance is corrected for high frequency operation to obtain the cable equivalent resistance under high frequency conditions. ,in The skin effect amplification factor is a dimensionless parameter, calibrated from historical transient test data, used to characterize the additional losses caused by uneven current distribution within the conductor cross-section under high-frequency conditions; simultaneously, based on... Calculate the equivalent inductance of the cable to characterize its resistance to changes in current rate.

[0034] Next, the rated line voltage of the interconnecting transformer is extracted from the transformer nameplate and ledger data. Rated apparent capacity and short-circuit impedance percentage And based on the transformer short-circuit impedance conversion relationship Calculate the comprehensive short-circuit impedance of the transformer ; Furthermore, extract the preset transformer. ratio Preferably, it is set based on the historical short-circuit test average value of similar interconnecting transformers, used to describe the proportional relationship between the transformer's leakage reactance and leakage resistance, and based on... as well as The equivalent resistance and equivalent reactance of the transformer are calculated separately, and then based on... The equivalent reactance is converted into leakage inductance, thus obtaining the equivalent inductance of the transformer. .

[0035] Finally, the high-frequency equivalent resistance of the cable is... Equivalent resistance of transformer By performing series addition, the lumped parameter equivalent resistance of the candidate mutual assistance path is obtained. Equivalent inductance of the cable Equivalent inductance of a transformer By adding them in series, we can obtain the lumped-parameter equivalent inductance. Based on this, a series connection is constructed. A reverse power flow equivalent circuit model is established. Furthermore, based on this equivalent circuit model, a reverse power flow transient response equation is constructed. ,in For the mutual aid branch current, To support the equivalent power supply voltage on the support side, This is the equivalent grid voltage on the receiving side.

[0036] After the model is ready, a digital twin mapping is performed on the real-time high-frequency electrical waveform sequence, and a pre-set 50Hz digital low-pass filter is used to perform lossless phase-frequency separation to extract the fundamental component sequence. The remaining residual signal with frequencies higher than the power frequency is then stripped into a high-frequency distortion component sequence. Subsequently, using the aforementioned reverse power flow equivalent circuit model, the simulated closing transient topology characteristics are derived based on the assumed input power shortage capacity demand data. Specifically, the fundamental component sequence is injected into the simulated closing transient topology characteristics to solve for the estimated closing current impact envelope. To accurately capture the extreme value of the transient current impact under the worst mutual assistance injection condition, this embodiment uses a nonlinear impact envelope function superimposed with a natural logarithmic decay factor for calculation. The specific variant formula is as follows: In the formula The estimated closing current impulse envelope varies with time and is in amperes. Assume input power shortage capacity demand data; The rated line voltage of the mutual aid line; This is the preset comprehensive load power factor, a dimensionless constant, with a default value of 0.9; and These are the combined resistance and combined inductance output from the reverse power flow equivalent circuit model, respectively. The preset closing phase angle penalty factor is preferably set to 1.86. The maximum amplitude point of the estimated closing current impulse envelope at the t=0 limiting state is directly extracted and used as the peak value of the reverse inrush current. Output.

[0037] On the other hand, for the extracted high-frequency distortion component sequence, a fast discrete Fourier transform is performed on it to obtain the frequency domain energy spectrum distribution matrix, and the ratio of the total energy of the predetermined harmonic frequency band to the fundamental frequency energy is calculated as the reverse transient harmonic distortion rate. The calculation is performed using a dynamic weighted distortion evaluation formula, which is: In the formula This represents the reverse transient harmonic distortion rate; Let be the effective value of the current of the h-th harmonic component in the frequency domain energy spectrum matrix. is the effective value of the fundamental component; H is the preset maximum harmonic analysis order, preferably 50th to fully cover the high-frequency noise range of the inverter. The historical baseline steady-state transmission power of this candidate mutual assistance path is obtained based on the average transmission power of the line over the past 30 days under non-emergency conditions. is the harmonic penetration amplification factor, and is a dimensionless coefficient. Based on historical power quality testing experience of microgrids, a preferred value of 0.15 is recommended.

[0038] The reverse transient harmonic distortion rate and the reverse inrush peak value are jointly mapped to the transient stress penalty coefficient, and combined with the static vulnerability index, the dynamic implicit failure probability of the corresponding candidate mutual aid path is generated.

[0039] Preferably, the reverse transient harmonic distortion rate and the reverse inrush peak value are jointly mapped to a transient stress penalty coefficient, including: Obtain the hardware architecture type identifier from the protection device ledger data; When the hardware architecture type is identified as electromagnetic characterization data, the corresponding exponential amplification base is matched based on the interval where the reverse inrush peak is located, and the first transient stress penalty coefficient is generated by combining the exponential amplification base. When the hardware architecture type is identified as microcomputer-type characterization data, the reverse transient harmonic distortion rate is substituted into the preset linear mapping function for calculation to generate the second transient stress penalty coefficient. The first and second transient stress penalty coefficients are input into a normalization converter to be converted into transient stress penalty coefficients with uniform dimensions.

[0040] Preferably, by combining the static vulnerability index, the dynamic implicit failure probability of the corresponding candidate mutual aid path is generated, including: Obtain the operating characteristic curve data of adjacent upstream and downstream protection devices on the candidate mutual assistance path; Extract the time difference margin of the upper and lower level protection devices under the condition of reverse inrush peak crossing; The time difference margin of the action is compared with the predetermined safety time margin and converted into a cascaded coordination risk factor; The initial node failure base is generated by multiplying the static vulnerability index of all nodes along the path with the transient stress penalty coefficient. The initial node failure base is multiplied by the cascaded risk factors to construct the single-node corrected failure probability; Based on the series reliability aggregation formula, the single-node corrected failure probability of all nodes along the route is multiplied together to obtain the dynamic implicit failure probability of the entire candidate mutual aid path.

[0041] In this embodiment of the invention, the protection device ledger data is parsed to obtain its hardware architecture type identifier; when the parsing reveals that the hardware architecture type identifier is electromagnetic type characterization data, since the electromagnetic relay core is extremely prone to magnetic saturation under high amplitude current, leading to malfunction, based on the reverse inrush current peak value... The ampere interval it falls into matches the corresponding exponential amplification base. The cardinality is dimensionless, and the default rule is: if If it is less than the rated tolerance value It starts at 1.0, and jumps to 1.5 if the limit is exceeded; The first transient stress penalty coefficient is generated by combining this base number, and its mapping formula is as follows: ,in This is the first transient stress penalty coefficient. The preset critical current for magnetic saturation of the electromagnet core; Conversely, when the hardware architecture type is identified as microprocessor-based, considering that microprocessor protection has strong inrush current resistance but A / D sampling is susceptible to high-frequency noise interference, the reverse transient harmonic distortion rate is used. Substituting the preset linear mapping function into the calculation, the second transient stress penalty coefficient is generated. The calculation formula is as follows: ,in This is the second transient stress penalty coefficient. To determine the predetermined harmonic amplification slope, Preset the tolerance distortion threshold for microcomputer-based A / D converters.

[0042] Subsequently, the obtained first or second transient stress penalty coefficient is input into the normalization converter. The input logic here follows a "mutually exclusive two-choice" routing mechanism based on the hardware architecture type identifier. That is, for any single protection device node, its physical hardware attributes determine that it can only belong to either the electromagnetic or microprocessor type; therefore, two penalty coefficients will never be generated simultaneously for a single node. Specifically, when the hardware architecture type identifier is determined to be electromagnetic, only the first transient stress penalty coefficient is extracted and transmitted to the normalization converter; conversely, when the hardware architecture type identifier is determined to be microprocessor type, only the second transient stress penalty coefficient is extracted and transmitted to the converter. Although the first and second transient stress penalty coefficients are calculated from different sources, their mathematical essence has been ensured to be dimensionless penalty multipliers in the preceding operators. Therefore, the normalization converter does not perform scaling, addition, or weighting changes on the absolute magnitude of the values. Specifically, based on the uniqueness of the data channel source, the normalization converter directly maps the received first or second transient stress penalty coefficients to equivalent values ​​and uniformly defines the output as the transient stress penalty coefficient. .

[0043] Based on this, in order to generate the dynamic latent failure probability, the action characteristic curve data of adjacent upper and lower level protection devices on the candidate mutual assistance path are obtained, and the action time difference margin of the upper and lower level protection devices under the condition of facing the reverse inrush peak crossing is extracted. (Calculated by subtracting the action time of the primary protection from the action time of the secondary protection, in seconds), and the action time difference margin is compared with the predetermined safety time margin. Comparison transformed into cascaded coordination risk factors The conversion process uses a formula. calculate; When the action time difference margin When decreasing, The value will increase significantly, thereby amplifying the risk level of the corresponding node in subsequent failure probability calculations.

[0044] Following this, the dispatch center analyzed the static vulnerability index of all nodes along the path. With the corresponding transient stress penalty coefficient Perform direct multiplication to generate the initial node failure base. Its dimensions are Furthermore, the initial node failure base is multiplied by the cascaded coordination risk factor and divided by the preset limit failure tolerance area base. The formula for constructing the single-node correction failure probability is as follows: ,in The single-node correction failure probability of the i-th node, with a value range of [0,1]; This is the preset limit failure tolerance area base, with dimensions A². The physical meaning is: the theoretical value calculated when this type of protection device withstands the ultimate reverse inrush flow specified in the standard type test under the factory rated operating conditions. The reference value is retrieved from a preset equipment parameter library based on the hardware model of the protection device, thereby realizing the normalized calculation of the failure probability of different models of equipment.

[0045] Finally, based on the series reliability aggregation formula The dynamic implicit failure probability data of the entire candidate mutual assistance path is obtained by multiplying the single-node corrected failure probabilities of all N nodes along the route. .

[0046] Candidate mutual aid paths with a dynamic implicit failure probability lower than a preset risk threshold are selected as the initial scheduling set.

[0047] Preferably, after selecting candidate mutual aid paths with a dynamic implicit failure probability lower than a preset risk threshold as the initial scheduling set, the process further includes: Obtain the physical cable laying distance parameters for each candidate mutual assistance path in the initial scheduling set; A capacity adequacy score is constructed by dividing the available mutual assistance capacity limit by the power shortage capacity demand data. The safety gain score is constructed by subtracting the absolute difference between the preset risk threshold and the probability of dynamic latent failure. The comprehensive utility score for the corresponding path is generated by performing a weighted normalized summation operation on the reciprocal of the capacity adequacy score, the security gain score, and the physical cable laying distance parameter. The path priority allocation sequence within the initial scheduling set is reconstructed according to the sorting rule of comprehensive utility score values ​​from largest to smallest.

[0048] In this embodiment of the invention, the scheduling center first filters the probability of dynamic hidden failures. Candidate mutual assistance paths below a preset risk threshold are used as the initial scheduling set. The preset risk threshold takes into account the extremely low tolerance of high-voltage distribution networks for unplanned power outages; to ensure the deterministic operation of protection devices under reverse power flow impacts, a preferred upper limit of 0.1% is recommended. Those skilled in the art can adjust this threshold according to the actual power supply reliability requirements of the industrial park.

[0049] After constructing the initial scheduling set, in order to achieve the optimal ranking of resource scheduling schemes, the physical cable laying distance parameters of each candidate mutual assistance path in the initial scheduling set are first obtained through the GIS asset management database. Subsequently, the maximum available mutual aid capacity was utilized. (Calculated by superimposing the real-time surplus output of energy storage and distributed power sources in adjacent parks) divided by the power shortage capacity demand data Construct a capacity adequacy score. Next, the dynamic latent failure probability is subtracted from the preset risk threshold. Obtain the absolute difference and construct the security gain score. The larger this value, the higher the safety margin of the path against protection malfunctions during power mutual assistance.

[0050] To comprehensively evaluate the economy, safety, and capacity matching of the route, a multi-index fusion formula is used to perform a weighted normalized summation of the capacity adequacy score, the safety gain score, and the reciprocal of the physical cable laying distance parameter to generate a comprehensive utility score. The specific formula for the overall utility score is as follows: In the above formula, This is a dimensionless numerical score representing the overall utility level. This is the maximum value of the capacity sufficiency in the set, used for normalization; This is the preset baseline path reference length; These are the capacity weighting coefficient, the security weighting coefficient, and the distance weighting coefficient, respectively, whose algebraic sum equals 1. In this embodiment, the security weighting coefficient... The preset logic is set to dynamically increase the value when the path contains an electromagnetic protection device. The weight is reduced to 0.6, while the remaining weights are proportionally reduced to strongly mitigate the hidden risks of outdated equipment; under normal operating conditions, the preferred option is recommended. .

[0051] If the initial scheduling set is empty, a multi-source power flexible splitting loop is triggered. The multi-source power flexible splitting loop includes: splitting the power shortage capacity demand data into at least two sets of sub-capacity demand data according to a predetermined proportional step size matrix.

[0052] Preferably, the power shortage capacity demand data is decomposed into at least two sets of sub-capacity demand data according to a predetermined proportional step size matrix, including: Extract the available mutual support capacity dataset for each adjacent park participating in the split; Calculate the capacity ratio matrix of the available mutual aid capacity dataset; The greatest common divisor of the capacity ratio matrix is ​​used as the initial splitting base, and the step exploration parameter space is constructed in combination with the predetermined traversal step size. In the step-by-step exploration parameter space, the combined proportional feature vector is extracted in descending order of spatial distance. By using the combined proportional feature vector to perform a vector dot product operation on the power shortage capacity demand data, at least two sets of sub-capacity demand data are obtained.

[0053] In this embodiment of the invention, if the initial scheduling set selected by the front-end logic is empty, it indicates that no single mutual aid path can safely bear the overall reverse power flow impact within the preset implicit failure risk threshold. To avoid emergency rescue failure due to deadlock, a multi-source power flexible splitting loop is immediately triggered. This loop first sends a query command to the other normally powered microgrids adjacent to it, extracts the available mutual aid capacity dataset of each adjacent park participating in the splitting, and divides the available mutual aid capacity of each park by the sum of the available capacity of all participating parks to generate a dimensionless capacity natural proportion sequence. Subsequently, the proportion sequence is transposed and multiplied by itself to construct a capacity ratio matrix that can globally represent the correlation between the relative sufficiency of multi-source capacity. The greatest common divisor of each non-repeating element outside the main diagonal of the capacity ratio matrix is ​​extracted and used as the initial splitting base. Based on this, combined with the predetermined traversal step size Constructing the step exploration parameter space The predetermined traversal step size. Set to an adjustable range between 0.01 and 0.10; based on the initial split base. As the starting point, according to Discretization and incremental recursion are performed to generate a series of candidate scale vectors whose sum of all components is constant at 1, thereby constructing the step exploration parameter space. .

[0054] After completing the spatial construction, all candidate scale vectors are traversed within the step-by-step exploration parameter space, and combined scale feature vectors are extracted in descending order of spatial distance. The core of this extraction process lies in calculating the capacity margin Euclidean spatial distance corresponding to each candidate scale vector. Its calculation uses a variant distance evaluation formula: N is the total number of adjacent parks participating in the split; This represents the available mutual support capacity of the i-th park, in kilowatts. This refers to the original power shortage capacity demand data for the target industrial park. This represents the splitting ratio value assigned to the i-th park from the candidate ratio vector currently being traversed; This represents the statistical average of the available mutual aid capacity of each adjacent park, which is used as the denominator here to eliminate the influence of the absolute magnitude of capacity on the distance measurement scale. The smaller this spatial distance modulus, the higher the balance of the remaining available capacity of each adjacent park after allocation, the more balanced the spatial margin of the entire mutual aid community from the origin of capacity depletion danger, and the stronger the redundancy capability against secondary load fluctuations.

[0055] Traditional microgrid load allocation objective functions often use the minimization of variance to pursue absolute equalization. However, in the scenario of mutual assistance of heterogeneous capacity, rigid absolute averaging can easily lead to instantaneous over-limit collapse of small-capacity areas. This embodiment transforms the allocation logic into calculating the vector magnitude of the remaining available breathing capacity of each area after splitting and stripping in a multi-dimensional space. The larger this spatial distance magnitude, the higher the spatial margin of the entire mutual assistance community from the origin of capacity depletion danger after allocation, and the stronger the redundancy capability against secondary load fluctuations.

[0056] The calculated capacity margin Euclidean space distance The candidate proportion vector with the smallest value is locked as the optimal combined proportion feature vector. Based on the capacity allocation ratio represented by each component in the vector, the power shortage capacity demand data is analyzed. Perform proportional decoupling operations to obtain at least two sets of sub-capacity demand data. Among them, the sub-capacity demand data respectively represent the active power compensation share that adjacent parks need to undertake in the process of mutual power supply, and meet the following requirements. This breaks down a single shortfall capacity into multiple sub-task capacities that can be collaboratively undertaken by different power supply points, thereby reducing the risk of single-node overload and improving the feasibility of mutual assistance scheduling.

[0057] Based on the capacity requirement data of each group, the corresponding re-enactment reverse inrush peak value is re-extracted and the transient stress penalty coefficient is updated to calculate the updated implicit failure probability of the corresponding split path.

[0058] If the probability of hidden failure after the update of the corresponding split path is lower than the preset risk threshold, the multi-source power flexible split cycle will be exited, and scheduling strategy data containing multi-path collaborative power supply instructions will be generated and sent to the edge control node to execute the closing action.

[0059] Preferably, the generation of scheduling strategy data containing multi-channel coordinated power supply commands is sent to the edge control node to execute the closing action, including: Obtain the switchable setting group characteristic data of the microprocessor-based protection device on the corresponding split path; Extract the reverse mutual assistance setpoint parameter from the switchable setpoint group feature data and replace the current working setpoint memory block; The coordination delay between upper and lower level protection is shifted and corrected based on the reverse mutual assistance setting parameters in order to re-determine the coordination relationship of time level difference. The protection setting switching instruction containing the reverse mutual assistance setting parameter is packaged and encapsulated with the physical closing action instruction to form the scheduling strategy data; The scheduling strategy data is sent to the edge control node through an encrypted communication channel and executed sequentially according to the built-in time tags of the instructions.

[0060] Preferably, after performing the closing action, the following steps are also included: Activate the high-frequency transient waveform monitoring task built into the edge control node; Collect transient voltage waveform data during the implementation phase after the closing action is completed, and compare the waveform deviation value between the transient voltage waveform data during the implementation phase and the nominal voltage waveform of the system. When the waveform deviation exceeds the predetermined waveform safety boundary threshold, a truncation trigger signal is directly generated locally at the edge control node. The cutoff trigger signal is converted into a tripping electrical pulse to directly drive the local circuit breaker to trip, thus completing the active degradation cutoff protection.

[0061] In this embodiment of the invention, after the flexible splitting of multi-source power is completed, the capacity demand data of each subgroup obtained from the decomposition are re-substituted into the equivalent circuit model of the reverse power flow to extract the corresponding replayed reverse inrush peak value. The exponential amplification base or harmonic calculation function is then re-matched based on the updated peak data to update the transient stress penalty coefficient. Subsequently, the updated latent failure probability of the corresponding splitting path is calculated in conjunction with the static vulnerability index. If the updated latent failure probability of the corresponding splitting path is lower than a preset risk threshold, it is determined that the current capacity allocation scheme has diluted the transient impact within the equipment safety envelope. The multi-source power flexible splitting loop is then exited, and scheduling strategy data containing multi-path coordinated power supply commands is generated and sent to the edge control node to execute the closing action.

[0062] In the specific execution phase of generating the scheduling strategy data, the switchable setting group characteristic data of the microprocessor-based protection devices on the corresponding split path is first obtained through the asset ledger. Then, the reverse mutual assistance setting parameters specifically pre-set for bidirectional microgrids are automatically extracted from this characteristic data. These parameters are used to directly overwrite and replace the currently active conventional working setting memory block in the protection device's memory. Since the initiation of the reverse setting will disrupt the original radial protection timing coordination sequence of the distribution network, the coordination delay between upper and lower level protections must be shifted and corrected based on the reverse mutual assistance setting parameters to re-determine the time-level coordination relationship. This embodiment uses a logarithmic time compensation variant formula based on the setting drift rate for calculation. The formula is: In the formula The time delay for the new protection system after translation correction is in seconds; To protect the original positive reference of this microcomputer model, a time delay is required. The operating current threshold is the newly extracted reverse mutual assistance setting parameter. The original forward operating current threshold that was being replaced; The preset time compensation coefficient has a constant dimension of seconds. This coefficient is used to characterize the inertial time drift constant of the inverse time-limited action element inside the protection device in response to current changes, and is preferably 0.05s.

[0063] The reverse mutual assistance setting parameter switching instruction containing the recalculated time difference is packaged and encapsulated with the closing action instruction of the physical switch mechanism to form scheduling strategy data. This scheduling strategy data is then sent to the edge control node through a 5G or fiber optic encrypted communication channel. The underlying logic controller on the edge side parses the data and executes the "first switch setting, then close switch" action in sequence according to the microsecond-level absolute time sequence tag embedded in the instruction.

[0064] After the physical circuit breaker is fully closed and the closing action is performed, the high-frequency transient waveform monitoring task built into the edge control node is immediately activated. The edge node continuously collects transient voltage waveform data of the implementation phase after the closing action is completed at a preset high-frequency sampling rate, and calculates the waveform deviation between the transient voltage waveform data of this implementation phase and the system nominal voltage waveform (ideal sine wave) pre-stored in the controller in real time. To accurately quantify the threat level of high-frequency, high-amplitude transient spikes, a secondary distortion penalty model that integrates frequency domain features is used for calculation. The specific formula is as follows: in The value represents the overall waveform deviation; M represents the total number of discrete sampling points within the first power frequency cycle after the circuit breaker is closed. This represents the actual transient voltage waveform data at the j-th sampling point; This refers to the nominal voltage waveform data corresponding to the j-th sampling point; System rated peak voltage; The transient distortion frequency within this cycle is extracted in real time by the local phase-locked loop, and the unit is Hertz; This is the power frequency reference frequency, with a value of 50Hz. The preset transient sensitivity weight is a dimensionless coefficient used to calibrate the system's tolerance to high-frequency oscillations. The recommended default value is 0.20.

[0065] When the calculation results When the predetermined waveform safety boundary threshold is exceeded, it is recommended that the predetermined waveform safety boundary threshold be set to an adjustable range between 0.05 and 0.15. The purpose of this setting is to avoid false disconnection caused by minor disturbances and to ensure that the inverter equipment of the adjacent microgrid is not damaged by large-scale peak overvoltage. The edge control node does not need to wait for the lengthy confirmation from the cloud scheduling platform. Instead, it generates a cutoff trigger signal directly in the interrupt program deployed in the local high-speed DSP chip. The digital level cutoff trigger signal is converted into a high-power trip pulse through an optocoupler isolation amplification circuit. This pulse directly drives the trip coil of the local circuit breaker to complete the mechanical tripping action, thereby proactively degrading the protection before the accident spreads.

[0066] It should be noted that if, within the preset maximum number of splitting rounds, the probability of hidden failure after the update of all splitting schemes still cannot be lower than the preset risk threshold, the multi-source power flexible splitting cycle will terminate and return an alarm signal that mutual assistance is not feasible to the upper-level scheduling platform. At the same time, it is recommended to perform graded offloading of non-critical loads to reduce the demand for power shortage capacity.

[0067] It should be noted that the interval and threshold sizes are set for ease of comparison. The size of the threshold depends on the amount of sample data and the base number set by those skilled in the art for each set of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value. Furthermore, the above formulas are all dimensionless calculations, and the formulas are derived from software simulations using a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0068] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0069] The above are merely specific embodiments 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.

[0070] In conclusion, the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A cloud-based collaborative method for emergency resource scheduling in smart industrial parks, characterized in that: include: Acquire the power shortage capacity demand data of the target park, and receive real-time high-frequency electrical waveform sequences and protection device ledger data uploaded by edge control nodes on the candidate mutual aid path of adjacent parks; The static vulnerability index is calculated by extracting the equipment commissioning time and historical maintenance characteristics from the protection device ledger data. The real-time high-frequency electrical waveform sequence is input into the reverse power flow equivalent circuit model to extract the reverse transient harmonic distortion rate and reverse inrush peak value under the assumed input power shortage capacity demand data. The reverse transient harmonic distortion rate and the reverse inrush peak value are jointly mapped to the transient stress penalty coefficient, and combined with the static vulnerability index, the dynamic implicit failure probability of the corresponding candidate mutual aid path is generated. Candidate mutual aid paths with a dynamic implicit failure probability lower than a preset risk threshold are selected as the initial scheduling set; If the initial scheduling set is empty, a multi-source power flexible splitting loop is triggered. The multi-source power flexible splitting loop includes: splitting the power shortage capacity demand data into at least two sets of sub-capacity demand data according to a predetermined proportional step size matrix. Based on the capacity requirement data of each group, the corresponding re-enactment reverse inrush peak value is re-extracted and the transient stress penalty coefficient is updated to calculate the updated implicit failure probability of the corresponding split path. If the probability of hidden failure after the update of the corresponding split path is lower than the preset risk threshold, the multi-source power flexible split cycle will be exited, and scheduling strategy data containing multi-path collaborative power supply instructions will be generated and sent to the edge control node to execute the closing action.

2. The cloud-based collaborative smart park emergency resource scheduling method according to claim 1, characterized in that, The real-time high-frequency electrical waveform sequence is input into the reverse power flow equivalent circuit model to extract the reverse transient harmonic distortion rate and reverse inrush current peak value under the assumed input power shortage capacity demand data, including: Extract the fundamental component sequence and high-frequency distortion component sequence from a real-time high-frequency electrical waveform sequence; Using the reverse power flow equivalent circuit model, the transient topology characteristics of closing are deduced based on the assumed input power shortage capacity demand data; The fundamental component sequence is injected into the pre-simulated closing transient topological features to solve the estimated closing current impact envelope, and the maximum amplitude point of the estimated closing current impact envelope is extracted as the reverse inrush peak value. The frequency domain energy spectrum distribution matrix is ​​obtained by performing a discrete Fourier transform on the high-frequency distortion component sequence, and the ratio of the total energy of the predetermined harmonic frequency band to the fundamental frequency energy is calculated as the reverse transient harmonic distortion rate.

3. The cloud-based collaborative smart park emergency resource scheduling method according to claim 1, characterized in that, The reverse transient harmonic distortion rate and the reverse inrush peak value are jointly mapped to a transient stress penalty coefficient, including: Obtain the hardware architecture type identifier from the protection device ledger data; When the hardware architecture type is identified as electromagnetic characterization data, the corresponding exponential amplification base is matched based on the interval where the reverse inrush peak is located, and the first transient stress penalty coefficient is generated by combining the exponential amplification base. When the hardware architecture type is identified as microcomputer-type characterization data, the reverse transient harmonic distortion rate is substituted into the preset linear mapping function for calculation to generate the second transient stress penalty coefficient. The first transient stress penalty coefficient and the second transient stress penalty coefficient are input into the normalization converter to be converted into transient stress penalty coefficients.

4. The cloud-based collaborative smart park emergency resource scheduling method according to claim 1, characterized in that, The static vulnerability index is calculated by extracting the equipment commissioning time and historical maintenance characteristics from the protection device ledger data, including: The equipment commissioning time is compared with the predetermined life cycle benchmark value to obtain the equipment aging weight parameter; Extract the previous maintenance time record and internal analog quantity sampling deviation sequence from the historical maintenance characteristics; The time decay factor is calculated based on the interval between the last maintenance time and the current system time. The reference degradation value is obtained by weighting and summing the mean square error of the internal analog quantity sampling deviation sequence using the time decay factor. The static vulnerability index is obtained by multiplying the baseline degradation value by the equipment aging weight parameter.

5. The cloud-based collaborative smart park emergency resource scheduling method according to claim 1, characterized in that, By combining the static vulnerability index, the dynamic implicit failure probability of the corresponding candidate mutual aid path is generated, including: Obtain the operating characteristic curve data of adjacent upstream and downstream protection devices on the candidate mutual assistance path; Extract the time difference margin of the upper and lower level protection devices under the condition of reverse inrush peak crossing; The time difference margin of the action is compared with the predetermined safety time margin and converted into a cascaded coordination risk factor; The initial node failure base is generated by multiplying the static vulnerability index of all nodes along the path with the transient stress penalty coefficient. The initial node failure base is multiplied by the cascaded risk factors to construct the single-node corrected failure probability; Based on the series reliability aggregation formula, the single-node corrected failure probability of all nodes along the route is multiplied together to obtain the dynamic implicit failure probability of the entire candidate mutual aid path.

6. The cloud-based collaborative smart park emergency resource scheduling method according to claim 1, characterized in that, The power shortage capacity demand data is broken down into at least two sets of sub-capacity demand data according to a predetermined proportional step size matrix, including: Extract the available mutual support capacity dataset for each adjacent park participating in the split; Calculate the capacity ratio matrix of the available mutual aid capacity dataset; The greatest common divisor of the capacity ratio matrix is ​​used as the initial splitting base, and the step exploration parameter space is constructed in combination with the predetermined traversal step size. In the step-by-step exploration parameter space, the combined proportional feature vector is extracted in descending order of spatial distance. By using the combined proportional feature vector to perform a vector dot product operation on the power shortage capacity demand data, at least two sets of sub-capacity demand data are obtained.

7. The cloud-based collaborative smart park emergency resource scheduling method according to claim 1, characterized in that, The scheduling strategy data containing multi-channel coordinated power supply commands is generated and sent to the edge control node to execute the closing action, including: Obtain the switchable setting group characteristic data of the microprocessor-based protection device on the corresponding split path; Extract the reverse mutual assistance setpoint parameter from the switchable setpoint group feature data and replace the current working setpoint memory block; Based on the reverse mutual assistance setting parameters, the coordination delay of the upper and lower level protection is shifted and corrected, and the coordination relationship of the time level difference is re-deduced. The protection setting switching instruction containing the reverse mutual assistance setting parameter is packaged and encapsulated with the physical closing action instruction to form the scheduling strategy data; The scheduling strategy data is sent to the edge control node through an encrypted communication channel and executed sequentially according to the built-in time tags of the instructions.

8. The cloud-based collaborative smart park emergency resource scheduling method according to claim 7, characterized in that, After performing the closing action, the following is also included: Activate the high-frequency transient waveform monitoring task built into the edge control node; Collect transient voltage waveform data during the implementation phase after the closing action is completed, and compare the waveform deviation value between the transient voltage waveform data during the implementation phase and the nominal voltage waveform of the system. When the waveform deviation exceeds the predetermined waveform safety boundary threshold, a truncation trigger signal is directly generated locally at the edge control node. The cutoff trigger signal is converted into a tripping electrical pulse to directly drive the local circuit breaker to trip.

9. The cloud-based collaborative smart park emergency resource scheduling method according to claim 1, characterized in that, Before receiving the real-time high-frequency electrical waveform sequence and protection device ledger data uploaded by the edge control nodes on the candidate mutual assistance path of adjacent parks, the process also includes: The control edge node continuously captures periodic data from the local substation bus at a predetermined high-frequency sampling rate to obtain the original electrical waveform data stream. The variational mode decomposition algorithm is used to perform multi-scale frequency domain stripping on the original electrical waveform data stream, removing power frequency background signals and retaining high-frequency transient feature segments; Absolute time synchronization tags and local topology address identifiers are embedded in high-frequency transient feature segments, and then serialized, recombined, and encoded to form a real-time high-frequency electrical waveform sequence.

10. The cloud-based collaborative smart park emergency resource scheduling method according to claim 1, characterized in that, After selecting candidate mutual aid paths with a dynamic implicit failure probability lower than a preset risk threshold as the initial scheduling set, the following steps are also included: Obtain the physical cable laying distance parameters for each candidate mutual assistance path in the initial scheduling set; A capacity adequacy score is constructed by dividing the available mutual assistance capacity limit by the power shortage capacity demand data. The safety gain score is constructed by subtracting the absolute difference between the preset risk threshold and the probability of dynamic latent failure. The comprehensive utility score for the corresponding path is generated by performing a weighted normalized summation operation on the reciprocal of the capacity adequacy score, the security gain score, and the physical cable laying distance parameter. The path priority allocation sequence within the initial scheduling set is reconstructed according to the sorting rule of comprehensive utility score values ​​from largest to smallest.

Images