A quick power restoration access method for low-voltage power distribution station emergency power generation vehicle

Abstract

The application relates to the technical field of power emergency power supply and is a quick power restoration access method for a low-voltage power distribution station house emergency power supply vehicle, which comprises the following steps: obtaining historical operation data of each load in a power distribution station house and real-time output parameter data of an emergency power supply vehicle; according to the historical operation data, the starting current characteristic curve and the steady-state power characteristic curve of each load are extracted to construct a load characteristic fingerprint library containing load importance identification; based on the load characteristic fingerprint library, the impact current peak time and the duration of each load are identified, and the transient impact superposition coefficient when adjacent loads are switched is calculated; the application realizes safe and quick power restoration of the loads in the power distribution station house through characteristic fingerprint construction, impact current analysis, switching time sequence optimization, dynamic load access and real-time monitoring and adjustment, has intelligence and reliability, can accurately analyze and risk assess the loads, effectively avoids overload risks caused by impact current, and provides a complete load switching control strategy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power emergency power supply, and in particular to a quick power restoration access method for a low-voltage power distribution station house emergency generator car. BACKGROUND

[0002] In the current low-voltage power distribution station house emergency power supply practice, the load restoration link after the emergency generator car is accessed mainly relies on the manual batch switching of the loads by the operation and maintenance personnel. This method has some practical difficulties. The on-site operation and maintenance personnel often lack accurate data support on the characteristics of the loads, and rely on personal experience to judge the switching order and time interval, resulting in low efficiency and high risk in the power restoration process. In an emergency, such as a power outage in important places such as hospitals and data centers, an operation and maintenance personnel needs to operate dozens of circuit breakers, and must observe the system response after each switching and assess whether the next operation can be performed. The entire power restoration process may take several hours, seriously delaying the critical load recovery time. At the same time, during the manual switching process, due to the inability to accurately calculate the load impact current superposition effect, the operation and maintenance personnel often set the switching interval too conservatively due to concerns about overload risk, or make hasty decisions due to emergency pressure, resulting in unreasonable switching intervals. In addition, the existing manual method lacks real-time monitoring and response mechanism for the dynamic characteristics of the loads, and cannot dynamically adjust the switching plan according to the voltage and frequency fluctuations. It is difficult to respond in a timely manner when the load characteristics change or the generator response is abnormal, further increasing the uncertainty and risk of power restoration. These problems are particularly prominent in large-scale power outages caused by extreme weather, natural disasters, etc., directly affecting the efficiency of emergency response and the business continuity of critical facilities.

[0003] In view of this, the present application provides a quick power restoration access method for a low-voltage power distribution station house emergency generator car to solve the above problems. SUMMARY

[0004] The present application provides a quick power restoration access method for a low-voltage power distribution station house emergency generator car, which overcomes the shortcomings of the prior art and effectively solves the problem of unreasonable switching interval in the existing low-voltage power distribution station house emergency generator car using manual switching to achieve quick power restoration.

[0005] To solve the above problems, the quick power restoration access method for a low-voltage power distribution station house emergency generator car provided by the present application comprises:

[0006] Obtain the historical operation data of each load in the power distribution station house and the real-time output parameter data of the emergency generator car;

[0007] According to the historical operation data, extract the starting current characteristic curve and steady-state power characteristic curve of each load, and construct a load characteristic fingerprint library containing load importance identification;

[0008] Based on the load characteristic fingerprint library, the impact current peak time and duration of each load are identified, and the transient impact superposition coefficient when adjacent loads are switched is calculated;

[0009] According to the real-time output parameter data of the emergency power car, the instantaneous overload capacity threshold of the emergency power car is obtained, and according to the transient impact superposition coefficient and the instantaneous overload capacity threshold, a load switching time sequence constraint matrix is constructed;

[0010] Combined with the load importance identifier in the load characteristic fingerprint library and the load switching time sequence constraint matrix, an initial switching sequence containing switching time and interval duration is generated;

[0011] Based on the load switching time sequence constraint matrix, the initial switching sequence is subjected to impact waveform superposition simulation, and the output current prediction waveform of each switching time of the power car is obtained;

[0012] By analyzing the peak value overrun interval of the output current prediction waveform of the power car, the overload risk switching node in the initial switching sequence is identified;

[0013] For the overload risk switching node, the switching time of the corresponding load is adjusted or the load is split into multiple sub-units for batch switching, and an optimized switching sequence is generated;

[0014] According to the optimized switching sequence, the automatic access operation of the emergency power car and the load of the power distribution station is controlled, and the voltage fluctuation amplitude and frequency deviation in the access process are collected in real time;

[0015] Determine whether the voltage fluctuation amplitude or frequency deviation is outside the preset safety range;

[0016] In response, the subsequent load switching is suspended and the power distribution ratio of the switched load is adjusted until the preset safety range is restored and the optimized switching sequence is continued.

[0017] According to the historical operation data, the starting current characteristic curve and the steady-state power characteristic curve of each load are extracted, and a load characteristic fingerprint library containing load importance identifier is constructed, including:

[0018] The multiple starting processes of each load in the historical operation data are time-aligned, and a unified time axis is established with the starting instruction time as the reference point, and the complete current change curve from zero to steady state in each starting process is extracted;

[0019] The multiple complete current change curves of the same load are subjected to piecewise normalization processing, and the complete current change curve is divided into a sudden change segment, a transition segment and a steady state segment;

[0020] The average current slope and peak current multiple of the starting segment, the transition segment and the steady-state segment are calculated, and the power factor of the steady-state segment is combined to form a multi-dimensional feature vector, denoted as a starting current characteristic curve; wherein the feature vector is calculated as follows:

[0021]

[0022] wherein, is the starting current characteristic vector, is the average current slope of the three stages, is the ratio of the peak current to the steady-state current, is the steady-state power factor;

[0023] The time series data of active power, reactive power and harmonic content of each load in the normal operation state is extracted, and the steady-state power characteristic curve is obtained after low-pass filtering;

[0024] The starting current characteristic curve and the steady-state power characteristic curve are associated with the corresponding load identification, and the importance level label is labeled to construct the load characteristic fingerprint library.

[0025] Based on the above-mentioned load characteristic fingerprint library, the impact current peak time and duration of each load are identified, and the transient impact superposition coefficient when the adjacent loads are switched is calculated; including:

[0026] According to the starting current characteristic curve of each load, the time corresponding to the current peak value is calibrated as the impact current peak time;

[0027] The time experienced by the starting current characteristic curve from the start to the current value decaying to within the preset proportion of the steady-state value is calculated, denoted as the duration;

[0028] A switching time difference matrix of adjacent loads is constructed, and the matrix element value represents the time interval of the switching instructions of two adjacent loads;

[0029] It is judged whether the time interval is less than the sum of the durations of the corresponding two adjacent loads;

[0030] In response to yes, the overlapping area of the impact currents of the two adjacent loads on the time axis is calculated;

[0031] According to the product of the overlapping area and the impact current peak value of each load, the transient impact superposition coefficient is calculated.

[0032] According to the real-time output parameter data of the emergency power generation vehicle, the instantaneous overload capacity threshold of the emergency power generation vehicle is obtained, and according to the transient impact superposition coefficient and the instantaneous overload capacity threshold, a load switching time sequence constraint matrix is constructed, including:

[0033] ​Extracting rated output current and allowed instantaneous overload current multiple from real-time output parameter data of the emergency generator car, calculating the instantaneous overload capacity threshold;

[0034] Traversing the switching time difference matrix, for the load whose transient impact superposition coefficient is not zero, calculating the total current peak value after the impact current superposition;

[0035] Judging whether the total current peak value exceeds the instantaneous overload capacity threshold, and in response, marking the load as prohibited simultaneous switching period, and calculating the minimum switching time interval meeting the instantaneous overload capacity threshold requirement as the constraint time interval;

[0036] According to the constraint time interval, constructing a load switching time sequence constraint matrix, and filling the constraint time interval to the corresponding element position of the load switching time sequence constraint matrix.

[0037] The above combination of load importance identification in the load feature fingerprint library and load switching time sequence constraint matrix generates an initial switching sequence containing switching time and interval duration; including:

[0038] According to the load importance identification in the load feature fingerprint library, all loads are divided into high-priority load group, medium-priority load group and low-priority load group;

[0039] Within each priority load group, sort according to steady-state power from small to large to generate internal switching sequence of each group;

[0040] Serially connecting each priority load group according to priority to form a global load switching sequence chain;

[0041] Taking the time when the emergency generator car starts to stabilize as the starting time, according to the constraint time interval in the load switching time sequence constraint matrix, assigning switching time to each load in turn;

[0042] Calculating the time difference between adjacent switching times as the interval duration;

[0043] Combining the switching time and interval duration of each load to form an initial switching sequence.

[0044] The above based on the load switching time sequence constraint matrix, the initial switching sequence is subjected to impact waveform superposition simulation, and the generator car output current prediction waveform of each switching time is obtained; including:

[0045] According to the switching time of each load in the initial switching sequence, establishing a switching event sequence on the time axis;

[0046] Extracting load pairs with constraint relationship from the load switching time sequence constraint matrix, and identifying time intervals with impact current superposition risk in the switching event sequence;

[0047] Extracting the starting current characteristic curve of each load from the load feature fingerprint library, aligning the starting time of each curve to the corresponding switching time in the switching event sequence;

[0048] In the time interval where there is a risk of superimposing impact current, the starting current characteristic curve of the related load is sampled at a high time resolution, and in the remaining time interval, it is sampled at a standard time resolution;

[0049] On the unified time axis, the starting current characteristic curve of the load and the steady-state power characteristic curve are vector superimposed according to the corresponding time resolution to obtain the total load current demand curve;

[0050] Establishing an equivalent output impedance model of the emergency generator car, inputting the total load current demand curve into the equivalent output impedance model, calculating the actual output current time sequence data considering the dynamic response of the generator car, and generating the output current prediction wave.

[0051] The above analyzes the peak value overrun interval of the generator car output current prediction waveform, and identifies the overload risk switching node in the initial switching sequence; including:

[0052] Local extremum detection is performed on the generator car output current prediction waveform, and all local peak points and their corresponding time points are extracted;

[0053] Compare the current value of each local peak point with the instantaneous overload capacity threshold, and count the overrun peak points that exceed the threshold;

[0054] For each overrun peak point, trace back to the latest switching event occurrence time to identify the load combination that triggered the overrun peak point;

[0055] Calculate the difference between the current value of the overrun peak point and the instantaneous overload capacity threshold as the overrun current amplitude, and classify the risk level according to the overrun current amplitude;

[0056] The switching event time corresponding to the overrun peak point with a risk level higher than the preset risk threshold is marked as an overload risk switching node.

[0057] The above adjusts the switching time of the corresponding load or splits the load into multiple sub-units for batch switching for the overload risk switching node, and generates an optimized switching sequence; including:

[0058] For each overload risk switching node, extract the corresponding load and its switching time in the initial switching sequence;

[0059] Calculate the waveform peak value change after delaying the switching time of the load by different time lengths;

[0060] If there is a delay duration that makes the waveform peak value below the instantaneous overload capacity threshold, the switching time of the load is delayed for the corresponding duration;

[0061] If any delay duration cannot meet the requirements, the load is split into multiple sub-units according to the list of electrical equipment inside the load, and the switching time is re-allocated;

[0062] The adjusted switching time or the switching time of the split sub-units is updated to the initial switching sequence to form an optimized switching sequence.

[0063] The automatic access operation of the emergency generator car and the load of the power distribution station is controlled according to the optimized switching sequence, and the voltage fluctuation amplitude and frequency deviation during the access process are collected in real time; including:

[0064] According to the optimized switching sequence, a control instruction set containing the switching time of each load and the corresponding circuit breaker number is generated and sent to the automatic control module of the power distribution system of the power distribution station;

[0065] A monitoring window is set before and after the closing of each circuit breaker, and the three-phase voltage instantaneous value and frequency instantaneous value at the output end of the emergency generator car are collected at a high sampling frequency within the monitoring window;

[0066] The effective value of the three-phase voltage instantaneous value is calculated, the maximum voltage and minimum voltage in the monitoring window are extracted, the difference between them and the rated voltage is calculated, and the absolute value of the larger one is taken as the voltage fluctuation amplitude;

[0067] The frequency instantaneous value is statistically analyzed, and the maximum deviation of the frequency in the monitoring window is calculated as the frequency deviation.

[0068] In response, the subsequent load switching is suspended and the power distribution ratio of the switched load is adjusted until the preset safety range is restored and the optimized switching sequence is continued, including:

[0069] Send a pause instruction to the automatic control module of the power distribution system of the power distribution station immediately, freeze the subsequent switching operation in the optimized switching sequence that has not been executed;

[0070] Identify the current switched load set, filter out the adjustable load with power regulation capability, and issue a power regulation instruction to reduce its operating power;

[0071] Continuously monitor the voltage fluctuation amplitude and frequency deviation, and when both are restored to the preset safety range and remain stable for more than a preset stable duration, the pause instruction is lifted;

[0072] Resume the subsequent load switching operation in the optimized switching sequence, and gradually restore the power of the adjustable load to the normal operating level after the subsequent load switching is completed.

[0073] Compared with the prior art, the present application has the following advantages:

[0074] 1、The present application improves the overall efficiency and safety and reliability of the emergency power supply process of the low-voltage power distribution station. Through intelligent decision support, the time period from the connection of the emergency generator truck to the full recovery of the load of the power distribution station is reduced, the emergency response speed and efficiency are improved, especially in critical places with extremely high requirements for power supply continuity, the influence range and duration of power failure can be minimized. At the same time, the present application improves the safety and reliability of the power recovery process, effectively avoids the overload trip phenomenon caused by judgment errors in traditional manual switching, ensures the stable operation of the emergency power supply and the reliable power supply of the load. The present method can maintain more stable system voltage and frequency level, reduce the adverse effects of voltage fluctuation on sensitive equipment, and improve the power experience during emergency power supply. For the operation and maintenance personnel, the present application reduces the decision-making pressure and work intensity in emergency situations, reduces the risk of human error, and enables the operator to focus more on system monitoring and abnormal handling. In addition, the present application also realizes the standardization and intelligentization of the power recovery process, improves the standardization and consistency of emergency operation, and enhances the response ability and recovery flexibility of the power system to various emergencies.

[0075] 2、The present application realizes safe and fast power recovery of the load of the power distribution station through feature fingerprint construction, impact current analysis, switching timing optimization, dynamic load connection and real-time monitoring and adjustment, has intelligence and reliability, can accurately analyze and evaluate the characteristics of each load, effectively avoids the overload risk caused by impact current, and provides complete load switching control strategy. BRIEF DESCRIPTION OF DRAWINGS

[0076] The specific embodiments of the present application will be further described in detail below with reference to the accompanying drawings.

[0077] Figure 1 The present application discloses a fast power recovery connection method for an emergency generator truck of a low-voltage power distribution station. DETAILED DESCRIPTION

[0078] The present application is not limited by the following examples, and the specific implementation can be determined according to the technical scheme of the present application and the actual situation.

[0079] As shown in the accompanying drawings, Figure 1 The present application discloses a fast power recovery connection method for an emergency generator truck of a low-voltage power distribution station, comprising:

[0080] S101, obtain historical operation data of each load in the power distribution station and real-time output parameter data of the emergency generator car; the historical operation data includes key information such as starting current curve, steady-state power data, power factor and harmonic content of each load, which is extracted in real time through the historical database of the station monitoring system. The real-time output parameter data of the emergency generator car includes core parameters such as rated output current, allowed instantaneous overload multiple, output impedance and dynamic response characteristics, which are obtained through the vehicle-mounted monitoring system and communication interface. These data provide comprehensive basic data for subsequent load characteristic analysis and switching sequence optimization, ensuring the accuracy and reliability of the power restoration process.

[0081] S102, according to the historical operation data, extract the starting current characteristic curve and the steady-state power characteristic curve of each load, and construct a load characteristic fingerprint library containing load importance identification; the starting current characteristic curve describes the complete current change process of the load from starting to stable running stage, including the characteristic parameters of the mutation section, the transition section and the steady-state section. The steady-state power characteristic curve reflects the active power, reactive power and harmonic characteristics of the load in the normal running state. These multi-dimensional characteristics together constitute the "electrical DNA" of the load, providing a basis for subsequent impulse current analysis and switching sequence optimization.

[0082] S103, based on the load characteristic fingerprint library, identify the impulse current peak time and duration of each load, and calculate the transient impulse superposition coefficient when adjacent loads are switched; thus, first, the starting current characteristic curve in the characteristic fingerprint library is time-stamped to identify the accurate time point of the current peak value, then the time required for the current to decay from the peak value to the steady state is analyzed as the impulse duration, and finally, by constructing a switching time difference matrix, the overlap of impulse currents of different loads on the time axis is calculated, and the superposition effect is quantitatively evaluated, providing key input for constraint matrix construction.

[0083] S104, obtain the instantaneous overload capacity threshold of the emergency generator car according to the real-time output parameter data of the emergency generator car, and construct a load switching time sequence constraint matrix according to the transient impulse superposition coefficient and the instantaneous overload capacity threshold; this step combines the real-time parameters of the emergency generator car with the load impulse characteristics to accurately calculate the minimum time interval for safe switching between each pair of loads. Each element of the constraint matrix represents the minimum safe interval time between the corresponding two loads, which intuitively shows the time sequence constraint condition of load switching, and provides a constraint framework for initial switching sequence generation.

[0084] S105, generate an initial switching sequence containing switching time and interval length based on the load importance identification in the load characteristic fingerprint library and the load switching time sequence constraint matrix; this step first classifies and groups all loads based on load importance, then sorts them from small to large according to steady-state power within each group, and finally allocates accurate switching time for each load in combination with the time sequence constraints in the constraint matrix to form a complete initial switching sequence, providing a basic scheme for subsequent simulation verification.

[0085] S106, based on the load switching time sequence constraint matrix, perform shock waveform superposition simulation on the initial switching sequence to obtain the output current prediction waveform of each switching time; this step simulates the current waveform and its superposition effect when each load is turned on on a unified time axis by establishing an accurate load current model and a generator car response model, predicts the output current dynamic change of the generator car in the entire switching process, and provides intuitive waveform basis for risk node identification.

[0086] S107, identify the overload risk switching node in the initial switching sequence by analyzing the peak value overrun interval of the generator car output current prediction waveform; this step performs fine analysis on the prediction waveform, extracts all local peak points and compares them with the overload capacity threshold, identifies the time points and their corresponding load switching events that exceed the safe range, and determines the high-risk nodes for priority processing through risk level classification, providing a clear optimization target for switching sequence optimization.

[0087] S108, for the overload risk switching node, adjust the switching time of the corresponding load or split the load into multiple sub-units for batch switching to generate an optimized switching sequence; this step, for each risk node, reduces the impact of current superposition effect through time delay optimization or load splitting strategy, maximizes the reduction of overload risk under the premise of ensuring the priority of important load restoration, and generates the final optimized switching sequence to provide accurate control instructions for actual operation.

[0088] S109, control the automatic connection operation of the emergency generator car and the distribution station load according to the optimized switching sequence, and collect the voltage fluctuation amplitude and frequency deviation in the connection process in real time; this step converts the optimized switching sequence into circuit breaker control instructions, performs load switching operation according to the preset time sequence, and monitors the real-time changes of system voltage and frequency through high-precision acquisition equipment, providing real-time monitoring data for safety control.

[0089] S1010, determine whether the voltage fluctuation amplitude or frequency deviation exceeds the preset safe range;

[0090] S1011, in response to yes, then suspend the subsequent load switching and adjust the power distribution ratio of the switched load until the preset safety range is restored to continue the optimized switching sequence. This step compares the monitoring parameters with the safety threshold in real time, interrupts the switching process immediately when an abnormality is detected, and reduces the power of the adjustable load through the power adjustment instruction. After the system recovers to stable, the subsequent operation is continued to ensure the safety and reliability of the whole power restoration process.

[0091] In the above step S101, the historical operation data includes the starting current curve, steady-state power data, power factor, and harmonic content of each load, and other key information, which is extracted in real time through the historical database of the station monitoring system. The real-time output parameter data of the emergency generator car includes the rated output current, the allowed instantaneous overload multiple, the output impedance, and the dynamic response characteristics, and other core parameters, which are obtained through the vehicle-mounted monitoring system and the communication interface. These data provide comprehensive basic data for subsequent load characteristic analysis and switching sequence optimization, ensuring the accuracy and reliability of the power restoration process.

[0092] In the above step S102, according to the historical operation data, the starting current characteristic curve and the steady-state power characteristic curve of each load are extracted, and a load characteristic fingerprint library containing load importance identification is constructed, including:

[0093] S1021, time alignment is performed on the multiple starting processes of each load in the historical operation data, a unified time axis is established with the starting instruction issuing time as the reference point, and the complete current change curve from zero to steady state in each starting process is extracted; wherein, time alignment is the basic step of feature extraction, which unifies the starting data collected at different times to the same time reference system, facilitating subsequent analysis and comparison. The alignment process uses the starting instruction time as t=0 time to construct a standardized time axis, ensuring the comparability of data collected at different times. The extraction process of the current change curve uses high sampling rate data acquisition, usually 50-100 sampling points per second, to ensure the capture of subtle features of current change, especially the peak value and change rate of instantaneous impact current.

[0094] S1022, segment and normalize the multiple complete current variation curves of the same load, and divide the complete current variation curve into a sudden change segment, a transition segment and a steady state segment; segment and normalization is an effective method for processing the differences in current curves under different starting conditions, and by dividing the starting process into three stages with obvious characteristics, parameters are extracted and standardized respectively. The sudden change segment corresponds to the current surge stage in the initial starting stage, and usually lasts for several to dozens of electrical cycles; the transition segment corresponds to the stage of transition of current from peak to steady state, and the duration is relatively long, usually 3-5 times of the sudden change segment; the steady state segment corresponds to the stage of final stable operation of current. The normalization processing adopts segmented linear transformation to map the current values of each segment to a unified reference range, eliminate the influence of absolute amplitude difference, and highlight the curve form characteristics.

[0095] S1023, calculate the average current slope and peak current multiple of the sudden change segment, the transition segment and the steady state segment, combine the power factor of the steady state segment to form a multi-dimensional feature vector, denoted as a starting current characteristic curve; wherein, the feature vector is calculated as follows:

[0096] ,

[0097] wherein, is the starting current characteristic vector, is the average current slope of the three stages respectively, is the ratio of peak current to steady state current, is the steady state power factor; feature vector extraction is a key step of current curve feature quantization, which converts curve features into numerical representation by calculating key parameters. The average current slope reflects the rate characteristic of current change, which is obtained by difference calculation within each segment; the peak current multiple represents the ratio of maximum current to steady state current, which directly reflects the impact strength; the power factor reflects the energy characteristics of the load.

[0098] S1024, extract the time series data of active power, reactive power and harmonic content of each load in normal running state, and obtain the steady state power characteristic curve after low pass filtering; steady state feature extraction is an important supplement to comprehensive representation of load electrical characteristics, which builds a more complete load feature model by analyzing the power parameters and quality characteristics of the load in normal running state. The extraction process obtains long-time running records through the historical database of the power distribution automation system, and uses a low pass filter to eliminate short-term fluctuations and measurement noise, and retains long-term stable feature patterns. The steady state power characteristic curve contains the power variation mode, daily load characteristics and harmonic characteristics of the load, which provides an important basis for accurate classification and optimal switching of the load.

[0099] S1025, the starting current characteristic curve is associated with the steady-state power characteristic curve to the corresponding load identification, and the importance level label is marked, and the load characteristic fingerprint library is constructed. The feature association is the last step of the fingerprint library construction, the extracted feature data is one-to-one corresponding to the load, and a management label is added. The importance level label is determined based on the functional importance, interruption sensitivity and recovery priority of the load, and is usually divided into high, medium and low three levels. High importance loads such as emergency lighting, fire fighting equipment and key control systems need to be recovered in priority; medium importance loads such as general production equipment and office area power supply can be recovered in secondary priority; low importance loads such as non-essential lighting and leisure facilities can be recovered last after the system is stable. The characteristic fingerprint library is stored in a structured database, supports fast retrieval and dynamic update, and provides comprehensive and reliable data support for subsequent load switching optimization.

[0100] In the above step S103, based on the load characteristic fingerprint library, the impact current peak time and duration of each load are identified, and the transient impact superposition coefficient when adjacent loads are switched is calculated; including:

[0101] S1031, according to the starting current characteristic curve of each load, the time corresponding to the current peak value is marked as the impact current peak time; the peak time marking is the primary step of impact current analysis, which accurately locates the time point of the maximum current value, and provides a time reference for subsequent superposition effect analysis. The marking process first samples the starting current characteristic curve with high precision, then accurately locates the peak point through the extreme value detection algorithm, and finally records the offset of the peak point time relative to the starting time as the impact current peak time. For different types of loads, the peak time distribution is different: motor type loads usually reach the peak value within a few cycles after starting, transformer type loads may reach the peak value at the moment of closing, and electronic device type loads may have multiple peak values. Accurate peak time marking lays the foundation for time analysis of superposition effect.

[0102] S1032, the time experienced by the starting current characteristic curve from starting to the current value decaying to within a preset proportion of the steady-state value is calculated, and is recorded as the duration; the preset proportion is usually 120%. The duration calculation is a key step to evaluate the impact time range, which quantifies the duration effect of the impact current by determining the time required for the current to decay from the peak value to the steady-state value. The calculation process first determines the steady-state current value, then sets the preset proportion (usually 120% of the steady-state value) as the judgment threshold, and finally determines the time when the current curve first falls below the threshold by interpolation or binary search. The time difference between the starting time and the duration reflects the time dimension of the load impact characteristic, directly affecting the arrangement of switching time sequence, and providing time constraint conditions to avoid impact superposition.

[0103] S1033, construct a switching time difference matrix of adjacent loads, the matrix element value represents the time interval of the switching instructions of two adjacent loads; the time difference matrix is a mathematical tool for analyzing impact superposition, and the switching time relationship between all load pairs is directly represented in the form of a two-dimensional matrix. The construction process first creates an n*n zero matrix (n is the number of loads), and then calculates the time interval between each pair of loads according to the switching time of each load in the switching sequence, and fills it into the corresponding position of the matrix. The matrix element represents the time interval from the switching of load i to load j, and the positive value represents that load i is switched before load j, and the negative value represents that load j is switched before load i. The time difference matrix provides a global perspective of the switching time sequence, and provides a structured representation for identifying potential impact superposition risks.

[0104] S1034, judge whether the time interval is less than the sum of the duration of the corresponding two adjacent loads;

[0105] S1035, in response to yes, calculate the overlapping area of the impact currents of the two adjacent loads on the time axis; the overlapping area analysis is the core step of evaluating the impact superposition degree, which determines whether there is a superposition risk by comparing the time interval and the duration, and quantifies the superposition intensity. The judgment process first checks the time interval of each pair of loads in the time difference matrix whether it is less than the sum of the duration , if it is less than, it means that the impact currents of the two exist time overlap. The overlapping area calculation is based on the actual overlapping situation of the two current curves on the time axis, which is realized by integration or discrete sampling point summation. The area calculation formula is:

[0106] ;

[0107] wherein, is the overlapping area, and are the switching times of load and load , and are the corresponding durations, and are the corresponding current curve functions.

[0108] S1036, calculate the transient impact superposition coefficient according to the product of the overlapping area and the peak value of the impact current of each load. The superposition coefficient calculation is the last step of quantifying the impact risk, which obtains a numerical index representing the superposition effect by comprehensively considering the overlapping area and the peak value. The calculation formula is:

[0109] ;

[0110] wherein, is the transient overload capacity threshold of the emergency generator car, and the load is the transient impact superposition coefficient of the load, is the area of the overlap region, and is the corresponding impact current peak value, and is the corresponding duration. The superposition coefficient ranges from 0 to 1, and the larger the value, the stronger the superposition effect and the higher the switching risk. When the coefficient is 0, it means no overlap, and when it is 1, it means complete overlap and peak value occurs at the same time. The superposition coefficient provides a quantitative basis for constraint matrix construction and directly affects the timing arrangement of load switching.

[0111] In the above step S104, the transient overload capacity threshold of the emergency generator car is obtained according to the real-time output parameter data of the emergency generator car, and the load switching timing constraint matrix is constructed according to the transient impact superposition coefficient and the transient overload capacity threshold, including:

[0112] S1041, the rated output current and the allowed transient overload current multiple are extracted from the real-time output parameter data of the emergency generator car, and the transient overload capacity threshold is calculated; the overload capacity threshold calculation is the basis of safety constraints, and the maximum load capacity allowed by the system is determined through the core parameters of the generator car. The calculation process first extracts the rated output current , which represents the maximum current for long-term stable operation; then obtains the allowed transient overload multiple , which represents the current multiple that can be tolerated for a short time, usually 1.5-3 times the rated value, with a duration of several seconds to tens of seconds; finally, the transient overload capacity threshold is calculated as the baseline for judging impact risk. The threshold calculation takes into account the actual performance parameters of the generator car, ensuring that the constraint conditions match the device capacity and providing accurate boundary conditions for switching sequence optimization.

[0113] S1042, traverse the switching time difference matrix, and for the load with a non-zero transient impact superposition coefficient, calculate the total current peak value after superposition; peak value calculation is the core step of risk assessment, and the maximum current value during load combination switching is predicted through the model. The calculation process first selects the load pairs with a superposition coefficient greater than zero, indicating potential superposition risk; then based on the impact current curve of each load and the switching time difference, a superposition model on the time axis is constructed; finally, the total current peak value after superposition is calculated as the key indicator for risk judgment. The total current peak value calculation needs to consider the actual effect of curve superposition, which is not simply the sum of peak values, but based on real-time superposition on the time axis:

[0114] ;

[0115] wherein, is the total current peak value after superposition, is the load At the current value at the moment, is the load switching moment, is the number of loads involved in superposition. The peak value calculation directly reflects the overload risk of switching combination, providing a quantitative basis for constraint setting.

[0116] S1043, whether the total current peak value exceeds the instantaneous overload capacity threshold, in response to yes, mark the load as prohibited simultaneous switching period, and calculate the minimum switching time interval that meets the instantaneous overload capacity threshold requirement as the constraint time interval; the constraint time calculation is the key step to avoid overload risk, and the minimum time interval required for safe switching is determined through comparison and analysis. The comparison process first compares the calculated total current peak value with the overload capacity threshold to determine whether there is an overload risk; if the total current peak value exceeds the threshold, the switching time interval needs to be adjusted, and through binary search or gradient optimization method, the minimum time interval that makes the total current peak value just not exceed the threshold is found, which is recorded as the constraint time interval. The constraint time interval reflects the minimum time requirement for safe switching, and provides an accurate constraint condition for switching sequence optimization.

[0117] S1044, constructing a load switching time sequence constraint matrix according to the constraint time interval, and filling the constraint time interval to the corresponding element position of the load switching time sequence constraint matrix. Matrix filling is the last step of constraint representation, which stores the calculated constraint conditions in a structured form for subsequent use. The filling process first creates an n×n constraint matrix (n is the number of loads), and the initial value is all set to zero, indicating no constraint; then the calculated constraint time interval is filled into the matrix element position of the corresponding load pair, forming a complete constraint relationship expression. The matrix element represents the minimum safe time interval required from the load switching to the load , and a value of zero indicates no constraint and simultaneous switching. The constraint matrix provides a global perspective of switching constraint relationship, and provides a direct decision basis for the switching sequence generation algorithm, ensuring that the generated sequence meets all safety constraint conditions.

[0118] In the above step S105, an initial switching sequence containing switching moments and interval lengths is generated in combination with the load importance identifier in the load feature fingerprint library and the load switching time sequence constraint matrix; including:

[0119] S1051, according to the load importance identification in the load feature fingerprint library, all loads are divided into a high-priority load group, a medium-priority load group and a low-priority load group; priority division is the primary principle of switching sequence design, and the critical load is ensured to be restored in priority through load importance classification. The division process directly classifies the loads into three different priority groups based on the importance labels pre-labeled in the feature fingerprint library. The high-priority load group includes devices that are critical to safe operation and emergency response, such as emergency lighting, fire pump, key monitoring system, etc.; the medium-priority load group includes devices that are important for normal operation but have limited impact on short-term interruption, such as ordinary lighting, ventilation system, office equipment, etc.; the low-priority load group includes non-critical devices that can be delayed to restore, such as landscape lighting, unnecessary air conditioning, etc. Priority division ensures reasonable allocation of limited resources, maximizes system recovery efficiency on the premise of ensuring priority restoration of critical functions.

[0120] S1052, within each priority load group, the internal switching sequence of each group is generated according to the steady-state power from small to large; internal sorting is an important strategy to optimize switching impact, and the impact amplitude of single switching is reduced by power size sorting. The sorting process first extracts the steady-state power data of each load from the feature fingerprint library, and then arranges the power in ascending order within each priority group to form the internal switching sequence. This strategy of small power first and large power later helps the system to gradually adapt to the increase of load, avoids the serious impact caused by the simultaneous switching of high-power loads, and improves the system stability. For loads with similar power, the startup characteristics can be further considered to preferentially switch in loads with low impact multiples and smooth startup, further optimizing the smoothness of the switching process.

[0121] S1053, the priority load groups are sequentially connected in order of priority to form a global load switching sequence chain; sequence concatenation is the basic step to build a complete switching sequence, and the overall switching sequence is organized through the priority principle. The concatenation process first takes the internal sequence of the high-priority load group as the starting segment, and then connects the internal sequences of the medium-priority group and the low-priority group in order to form a complete sequence chain that runs through all the loads. This priority-based concatenation method ensures that the needs of important loads are met to the greatest extent under any power supply capacity, while ensuring the systematicness and controllability of the switching process. The sequence chain construction takes into account the importance and power characteristics of the loads, providing a reasonable basic sequence for time allocation.

[0122] S1054, starting from the moment the emergency generator truck stabilizes after startup, assigns switching times to each load sequentially according to the constraint time intervals in the load switching timing constraint matrix. Time allocation is a crucial step in transforming the sequence into an actual execution plan, ensuring that the switching process meets all constraints through time scheduling. The allocation process first determines the starting time t0, typically the time point after the generator truck starts and outputs steadily; then, it sequentially allocates the switching time for each load according to the global sequence chain. For the first load in the sequence... Each load, its switching time Must meet:

[0123] ;

[0124] in, For the constraint matrix from the load to load The constraint time interval, For the constraint matrix from the load to load The constraint time interval, For the constraint matrix from load 1 to load The constraint time interval, and so on, shows that... For the constraint matrix from the load to load The constraint time interval. This dynamic programming method ensures that the switching time of each load satisfies the constraint relationship with all previous loads, minimizing the risk of cumulative impact.

[0125] S1055, calculate the time difference between adjacent switching moments as the interval duration;

[0126] S1056, the initial switching sequence is formed by combining the switching times and intervals of each load allocation.

[0127] Sequence combination is the final step in the deployment plan, representing the complete execution scheme through the pairing of times and intervals. The combination process first calculates the time difference between adjacent deployment times. The interval is set as the operation interval. Then, the identifier of each load, the switching time, and the interval of the previous operation are combined into a triplet to form a complete switching instruction sequence. The initial switching sequence adopts a time-driven execution mode. The system automatically executes the corresponding switching operations according to the preset time, ensuring the accuracy and automation of the process. The sequence combination completes the transformation from theoretical constraints to actual operation plans, providing a specific execution scheme for subsequent simulation verification.

[0128] In step S106, based on the load switching time sequence constraint matrix, the initial switching sequence is subjected to shock waveform superposition simulation to obtain the predicted output current waveform of each switching time of the power generation vehicle; including:

[0129] S1061, according to the switching time of each load in the initial switching sequence, a switching event sequence is established on the time axis; the event sequence establishment is a preparation step of the simulation, and the framework structure of the simulation is constructed through the time axis marker. The establishment process first marks all the switching events of the loads on the unified time axis with the switching time as the coordinate point to form a discrete event sequence; then the identification and characteristic data of the load are associated with each event point to establish the mapping relationship between the event and the characteristic. The event sequence provides the time skeleton of the simulation, defines the distribution of the key calculation points, indicates the key attention interval for subsequent waveform calculation, and improves the simulation efficiency.

[0130] S1062, the load pairs with constraint relationship are extracted from the load switching time sequence constraint matrix, and the time interval with the risk of superimposed shock current in the switching event sequence is identified; the risk interval identification is an important step of optimizing the simulation resource allocation, and the time period with possible problems is predicted through the constraint relationship. The identification process first screens out the load pairs with constraint values greater than zero from the constraint matrix, indicating that there is a potential superposition risk; then according to the switching time of the corresponding load in the switching event sequence, the time interval in which the shock superposition may exist is determined. The risk interval is usually composed of two switching event times and the shock duration period thereafter, and is the key calculation area of the waveform simulation. The risk interval identification realizes the reasonable allocation of simulation resources, and the key areas are calculated in detail and the non-key areas are simplified, thereby improving the simulation efficiency.

[0131] S1063, the starting current characteristic curves of each load are extracted from the load characteristic fingerprint library, and the starting time of each curve is aligned to the switching time of the corresponding switching event sequence; the waveform alignment is a basic operation for building a global model, and the time sequence relationship of the current of each load is uniformly established through the time reference. The alignment process first extracts the standard starting current characteristic curve of each load from the characteristic fingerprint library; then the time axis origin of the curve is translated to the switching time of the corresponding switching event sequence to realize time alignment; finally, according to the difference between the actual power of the load and the standard condition, the current amplitude is proportionally adjusted to ensure the accuracy of the model. The waveform alignment integrates the dispersed load characteristics into a unified time framework, provides basic data for waveform superposition calculation, and ensures the time sequence consistency of the model.

[0132] S1064, in the time interval where the risk of superimposed impact current exists, the starting current characteristic curve of the relevant load is sampled at a high time resolution, and in the remaining time interval, it is sampled at a standard time resolution; resolution optimization is a balanced strategy to improve calculation accuracy and efficiency, and the accuracy of key results is improved by focusing on area refinement. The sampling process first determines the standard time resolution, usually 10-20ms, which meets the general waveform calculation demand; then the time interval with risk is identified, and the resolution is increased to 1-2ms in these intervals to ensure the capture of the rapid change characteristics of the impact current; finally, the current curve of each load is resampled according to the resolution requirement to generate discrete data points for superimposed calculation. The resolution optimization strategy ensures the calculation accuracy of the key area while avoiding the resource waste caused by global high-precision calculation, significantly improving the simulation efficiency.

[0133] S1065, on the unified time axis, the starting current characteristic curve of the load and the steady-state power characteristic curve are vector superimposed according to the corresponding time resolution to obtain the total load current demand curve; waveform superposition is the core step of simulation calculation, and the total demand of the system is obtained by time sequence accumulation of current vector. The superposition process first constructs discrete calculation points on the unified time axis according to the sampling resolution; then for each calculation point, the current contribution of all the loads that have been put into at that time is accumulated to form the total current demand; finally, the sampling error is eliminated by smoothing processing to obtain the continuous total load current demand curve. The superposition calculation considers the phase relationship and power factor influence of the current, not only the amplitude superposition, but also the vector synthesis effect, to ensure the physical accuracy of the model. Waveform superposition intuitively shows the total current dynamic change of the system in the switching process, providing input conditions for the response analysis of the generator car.

[0134] S1066, an equivalent output impedance model of the emergency generator car is established, the total load current demand curve is input into the equivalent output impedance model, the actual output current time series data considering the dynamic response of the generator car is calculated, and the output current prediction wave is generated. Response simulation is the last link of the simulation, and the actual running state is predicted through the device characteristic model. The modeling process first establishes the equivalent circuit model of the generator car, including internal impedance, regulation time constant and overload response characteristics; then the total load current demand curve is input as an input, and the actual output current after considering voltage drop and regulation response is calculated through the model; finally, a complete output current prediction waveform is generated as the basis for risk assessment. The model calculation considers the nonlinear response characteristics of the generator car, especially the dynamic response process when the load suddenly changes, which improves the accuracy of the prediction. The output current prediction waveform fully reflects the dynamic performance of the system in the switching process, providing an intuitive waveform basis for risk node identification.

[0135] In step S107 above, by analyzing the peak over-limit range of the predicted waveform of the generator output current, the overload risk switching nodes in the initial switching sequence are identified; including:

[0136] S1071, perform local extremum detection on the predicted waveform of the generator output current, extracting all local peak points and their corresponding times. Extremum detection is a fundamental step in identifying risk points; numerical analysis is used to find all peak points in the current waveform. The detection process employs a sliding window method, setting an appropriately wide time window on the predicted waveform. When the current value at the center point is greater than all other points within the window, it is marked as a local peak point. The detection algorithm must consider the influence of waveform noise and subtle fluctuations, ensuring the identified peak points have practical significance through smoothing preprocessing and minimum amplitude threshold filtering. For each identified peak point, its time is recorded. and current value This forms a set of peak feature points, providing key data points for risk assessment.

[0137] S1072 compares the current values ​​at each local peak point with the instantaneous overload capacity threshold, and counts the peak points exceeding the threshold. Threshold comparison is a direct method for risk assessment, using preset standards to screen out time points where actual risks exist. The comparison process first determines the instantaneous overload capacity threshold. This serves as the upper limit standard for safe operation; then the current value at each peak point is... Compared with a threshold, when When a point is reached, it is marked as an overload peak, indicating an overload risk at that moment. Statistical analysis of overload peaks reflects potential safety hazards in the switching sequence and is a key objective of sequence optimization. The statistical results include the number, distribution, and severity of overload points, providing a quantitative basis for risk rating.

[0138] S1073, for each exceeding-limit peak point, trace back to the most recent switching event to identify the load combination that triggered the exceeding-limit peak point; tracing analysis is a key step in clarifying responsibility, identifying the specific operation that caused the problem through time-series correlation. The tracing process first determines the time of the exceeding-limit peak point. Then search for events less than in the switching event sequence. The analysis records the most recent switching events and the corresponding loads. Simultaneously, considering the duration of the inrush current, all switching events occurring within a period preceding the peak (typically the longest inrush duration) are included in the analysis to identify potentially contributing load combinations. This time-series-based retrospective analysis clarifies risk responsibility, provides precise targets for targeted optimization, and avoids efficiency losses caused by blind adjustments.

[0139] S1074, calculate the difference between the current value of the out-of-limit peak point and the instantaneous overload capacity threshold as the out-of-limit current amplitude, and classify the risk level according to the out-of-limit current amplitude; the risk classification is the decision basis for optimizing the priority, and the risk severity is quantified by the out-of-limit degree. The calculation formula is:

[0140] ,

[0141] wherein, is the out-of-limit current amplitude, is the peak point current value, is the overload capacity threshold. According to the relative value of the out-of-limit amplitude classify the risk level, usually divided into three levels: mild risk (0-10%), moderate risk (10%-30%), and severe risk (>30%). The risk classification provides a scientific basis for the priority of optimization work, ensuring that limited adjustment resources are prioritized to address the most serious risk points, improving optimization efficiency.

[0142] S1075, mark the switching event time corresponding to the out-of-limit peak point with a risk level higher than the preset risk threshold as an overload risk switching node. Node marking is the final output of risk identification, which clearly indicates the specific operation point that needs to be adjusted. The marking process first sets a preset risk threshold (usually moderate risk and above), and then marks the switching event time corresponding to the out-of-limit peak point with a risk level exceeding the threshold as an overload risk switching node. Each risk node records the load identification, switching time and risk level, forming a risk-operation mapping relationship. The identification result of the overload risk switching node directly guides the subsequent optimization adjustment, providing a clear optimization object and target for sequence optimization, ensuring the accuracy and effectiveness of the adjustment work.

[0143] In the above step S108, for the overload risk switching node, adjust the switching time of the corresponding load or split the load into multiple sub-units for batch switching, and generate an optimized switching sequence; including:

[0144] S1081, for each overload risk switching node, extract the corresponding load and its switching time in the initial switching sequence; node extraction is the starting point of optimization, which determines the specific operation that needs to be adjusted through risk identification results. The extraction process first reads the node information from the risk node list, including the load identification and original switching time; then obtains the detailed characteristic data of the load from the feature fingerprint library to provide a reference for the adjustment strategy; at the same time, analyze the position and context relationship of the load in the switching sequence, and evaluate the feasibility and influence range of the adjustment. Node extraction converts abstract risk analysis into specific optimization objects, laying the foundation for targeted adjustment and ensuring accurate positioning and effective implementation of optimization work.

[0145] S1082, calculate the waveform peak value change situation after delaying the switching time of the load for different lengths of time; delay optimization is the simplest and most effective adjustment strategy, which reduces the impact superposition by time staggering. The calculation process first sets a series of tentative delay lengths, usually starting from 10% of the impact duration, increasing by 10% to 100%; then, for each delay value, the adjusted switching sequence corresponding to the current prediction waveform is recalculated, focusing on the change in the peak value point; finally, a delay-peak relationship curve is generated, which intuitively shows the effects of different delay strategies. Delay optimization is based on the concept of "time for safety", which avoids the high-risk period of current peak value superposition by appropriately delaying the input time of non-urgent loads, and is the most economical and simple optimization method, especially suitable for handling light risk nodes.

[0146] S1083, if there is a delay length that reduces the waveform peak value below the instantaneous overload capacity threshold, delay the switching time of the load by the corresponding length of time; delay decision is an optimization implementation based on simulation results, which selects the optimal delay parameter through quantitative analysis. The decision process first analyzes the delay-peak relationship curve to identify the minimum delay length that reduces the peak value below the threshold ; then consider the system recovery time requirement and operation smoothness, appropriately round the adjustment (usually take 110% of the minimum effective delay), to ensure sufficient safety margin; finally, update the switching time of the load in the switching sequence from the original time to . This minimum effective delay strategy maximizes the total time of power supply recovery while ensuring safety, balancing the dual demands of safety and efficiency.

[0147] S1084, if any delay length cannot meet the requirements, according to the list of electrical equipment inside the load, split the load into multiple sub-units and reassign the switching time; the splitting strategy is an effective method to handle severe risk nodes, which reduces the impact amplitude of single switching by load subdivision. The splitting process first analyzes the composition of the electrical equipment inside the load, identifies the subsystems or equipment groups that can be independently controlled; then, according to the importance and starting characteristics of the equipment, the original load is split into 2-4 sub-units; finally, assign a new switching time to each sub-unit, usually using an even distribution strategy to ensure that there is enough time interval between sub-units to avoid impact superposition. The splitting strategy solves the severe risk problem that cannot be solved by simple delay through the "divide and conquer" method, which is an ideal solution for handling large power complex loads, and can ensure the rapid recovery of core functions while effectively controlling the impact risk.

[0148] S1085, update the adjusted switching time or the split sub-unit switching time to the initial switching sequence to form an optimized switching sequence. Sequence updating is the last step of optimization, which integrates local adjustments into the global plan. The updating process first unifies the results of delay adjustment and split adjustment into a standard switching instruction format; then replaces the original instructions in the initial switching sequence with these adjusted instructions; finally, reorders them in time sequence and updates the interval time between adjacent instructions to form a new complete sequence. The updated switching sequence needs to be verified globally, and all risk points have been effectively handled through complete waveform simulation. The optimized switching sequence not only retains the priority framework of the original sequence, but also solves specific risk points, and is a safe and efficient load restoration execution scheme.

[0149] In the above step S109, the automatic connection operation of the emergency power generation vehicle and the distribution station load is controlled according to the optimized switching sequence, and the voltage fluctuation amplitude and frequency deviation in the connection process are collected in real time; including:

[0150] S1091, according to the optimized switching sequence, generate a control instruction set containing the switching time of each load and the corresponding circuit breaker number, and issue it to the automatic control module of the distribution system of the distribution station; instruction generation is a preparation step for execution control, which converts logical sequence into specific device control command. The generation process first converts each record in the switching sequence into a standard format control instruction, including instruction type, target device, operation parameter and execution time, etc. key information; then according to the system communication protocol requirements, encode the instruction into a transmissible data packet; finally, through a safe communication channel, the instruction set is issued to the automatic control unit. The control instruction adopts a time-triggered execution mode, each instruction has an accurate time stamp, and the system automatically executes the corresponding operation at the specified time according to the internal clock. This time-based centralized control method ensures the accurate execution of complex switching sequence, avoiding the uncertainty and delay risk of manual operation.

[0151] S1092, set a monitoring window before and after each circuit breaker closing, collect the three-phase voltage instantaneous value and frequency instantaneous value at the output end of the emergency generator car at a high sampling frequency within the monitoring window; parameter monitoring is the information basis of safety control, and the system operation state is evaluated through real-time data acquisition. The monitoring setting first determines the monitoring window range, which is usually 1 second before the circuit breaker closing to 5 seconds after the closing, covering the complete process of the impact current; then a high sampling frequency is set, which is usually 64-128 points per cycle, to ensure the capture of the detailed characteristics of transient changes; finally, the acquisition channel is configured to record the instantaneous values of three-phase voltage and frequency simultaneously, forming a complete state monitoring. High-precision power parameter analyzers or special acquisition modules are used for data acquisition, and an isolation transformer is used to ensure the safety and accuracy of the measurement. Real-time monitoring provides first-hand information on the dynamic response of the system, providing data support for safety judgment and intervention decision-making, and is a key link in closed-loop control.

[0152] S1093, calculate the effective value of the three-phase voltage instantaneous value, extract the maximum voltage and minimum voltage within the monitoring window, calculate the difference between the two and the rated voltage, and take the larger absolute value as the voltage fluctuation amplitude; fluctuation calculation is a numerical processing of state evaluation, and the system stability is quantified by statistical characteristics. The calculation process first converts the collected three-phase voltage instantaneous value into an effective value to eliminate the influence of waveform details and focus on amplitude changes; then searches for the maximum value and minimum value of the effective value sequence within the monitoring window, reflecting the voltage fluctuation range; finally, calculate the deviation of the two extreme values from the rated voltage , and take the larger absolute deviation as the voltage fluctuation amplitude :

[0153] ;

[0154] The voltage fluctuation amplitude directly reflects the voltage stability of the system during load input, and is an important indicator for evaluating the voltage regulation capability of the generator car and the impact of load impact, providing a key reference value for safety monitoring.

[0155] S1094, perform statistical analysis on the frequency instantaneous value, and calculate the maximum deviation of the frequency within the monitoring window as the frequency deviation. Frequency analysis is an important means of evaluating the dynamic characteristics of the system, and the power balance state is reflected through the frequency change. The analysis process first smoothes the collected frequency instantaneous value to eliminate measurement noise and high-frequency fluctuations; then searches for the maximum and minimum value of the processed sequence within the monitoring window, reflecting the frequency fluctuation range; finally, calculate the maximum deviation of the two extreme values from the nominal frequency as the frequency deviation :

[0156] ;

[0157] The frequency offset intuitively reflects the power balance state and adjustment ability of the system during the load change process, is an important indicator for evaluating the performance of the power generation car speed regulation system and the dynamic characteristics of the load, and provides a reliable basis for stability judgment. The voltage fluctuation amplitude and the frequency offset jointly constitute the comprehensive evaluation index of the system state, and provide comprehensive data support for safety monitoring and adjustment decision.

[0158] In the above step S1011, in response to yes, the subsequent load switching is suspended and the power distribution ratio of the switched load is adjusted until the preset safety range is restored and the optimized switching sequence is continued to be executed, including:

[0159] S201, immediately send a suspension instruction to the automation control module of the power distribution station building power distribution system, freeze the subsequent switching operation in the optimized switching sequence which has not been executed; emergency suspension is the primary measure for risk response, which avoids further deterioration of the state by stopping the operation. The suspension process first detects whether the voltage fluctuation amplitude or the frequency offset is out of the preset safety range, and the safety range is usually ±7% for voltage and ±0.5Hz for frequency; when the out-of-limit condition is detected, the system immediately generates a suspension instruction with the highest priority, which is sent to the automation control unit through a fast communication channel; the control unit immediately interrupts the execution of the current switching sequence after receiving the instruction, and keeps the current state unchanged. The suspension mechanism adopts a hard real-time response design, which ensures that the detection-decision-execution process is completed in a very short time (usually <100ms), effectively preventing further deterioration of the system. This fast suspension strategy provides self-protection capability for the system, actively interrupts the operation when abnormal conditions occur, and prevents chain failures, which is the key defense line of safety control.

[0160] S202, identify the current set of switched loads, select the adjustable loads with power adjustment capability, and issue power adjustment instructions to reduce their operating power; power adjustment is an active measure for steady-state recovery, which improves system balance through load-side adjustment. The identification process first extracts all the loads currently in operation from the execution record; then according to the load characteristic database, it selects a subset of loads with power adjustable capability, such as frequency conversion devices, adjustable lighting, temperature control devices, etc.; finally, based on the deviation of the system state, the total power adjustment amount required is calculated. The adjustment strategy adopts a proportional distribution principle, which distributes the total adjustment amount to each adjustable load according to its maximum adjustment capacity, ensuring the fairness and effectiveness of the adjustment. The power adjustment instruction is issued for execution through the load controller or the energy management system, usually using a gentle gradual adjustment method to avoid sudden changes causing new shocks. This active adjustment strategy quickly improves the system balance without interrupting the operation of critical loads, creates conditions for stable recovery by power concessions of non-critical loads.

[0161] S203, continuously monitor the voltage fluctuation amplitude and frequency deviation, and when both are restored to the preset safe range and remain stable for more than the preset stable duration, the suspension command is lifted; state monitoring is the information basis for recovery decision, and the system stability is confirmed by continuous observation. The monitoring process uses continuous sampling strategy to continuously collect system voltage and frequency parameters at a high frequency (usually 5-10 times per second); at the same time, the fluctuation amplitude and deviation in the sliding window are calculated to evaluate the stability trend of the system; when the parameters remain in the safe range for more than the preset stable duration (usually 10-30 seconds), the system confirms that the state has been stably restored. The preset safe range is usually set to voltage ±5% and frequency ±0.3Hz, which is more stringent than the threshold for triggering suspension, ensuring sufficient safety margin; the preset stable duration is determined according to the system size and generating car characteristics, ensuring that the transient process is completely over. State monitoring provides a reliable criterion for system recovery, avoiding the risk of repeated oscillation caused by premature recovery, and ensuring that the system continues to operate after it is truly stable.

[0162] S204, restore the subsequent load switching operations in the optimized switching sequence, and gradually restore the power of adjustable loads to the normal operating level after the subsequent load switching is completed. Sequence recovery is the continuation after interruption, ensuring the continuity of the whole process through smooth transition. The recovery process first lifts the suspension state, allowing the automatic control unit to continue executing the subsequent switching instructions; then, according to the difference between the current time and the original sequence time, adjust the execution time of the subsequent instructions appropriately to ensure that the instruction interval meets the safety requirements; finally, after all the loads are put in, gradually restore the adjustable loads whose power was reduced before, usually using a step-by-step increase strategy, increasing the power by 10-20% every 2-3 minutes until it returns to normal. This gradual recovery strategy not only ensures the safe completion of the power restoration process, but also realizes the normal operation of all loads, reflecting the balance between safety and efficiency, and drawing a perfect full stop to the entire power restoration process.

[0163] To sum up, the application improves the overall efficiency and safety and reliability of the emergency power supply process of the low-voltage power distribution station. Through intelligent decision support, the time period from the connection of the emergency generator truck to the full recovery of the power distribution station load is reduced, the emergency response speed and efficiency are improved, especially in critical places with extremely high requirements for power supply continuity, the influence range and duration of power failure can be minimized. At the same time, the safety and reliability of the power recovery process are improved, the overload trip phenomenon caused by judgment errors in traditional manual switching is effectively avoided, and the stable operation of the emergency power supply and the reliable power supply of the load are ensured. The method can maintain more stable system voltage and frequency level, reduce the adverse effects of voltage fluctuation on sensitive equipment, and improve the power supply experience during emergency power supply. For the operation and maintenance personnel, the application reduces the decision-making pressure and work intensity in emergency situations, reduces the risk of human error, and enables the operator to focus more on system monitoring and abnormal handling. In addition, the application also realizes the standardization and intelligentization of the power recovery process, improves the standardization and consistency of emergency operation, and enhances the response ability and recovery flexibility of the power system to various emergencies.

[0164] The application realizes safe and rapid power recovery of the power distribution station load through feature fingerprint construction, impact current analysis, switching timing optimization, dynamic load access and real-time monitoring adjustment, has intelligence and reliability, can accurately analyze and evaluate the risk of each load, effectively avoid the overload risk caused by impact current, and provide complete load switching control strategy.

Claims

1. A method for quick power restoration access of an emergency generator vehicle of a low-voltage power distribution station, characterized in that, The method comprises the following steps: acquiring historical operation data of each load in a power distribution house and real-time output parameter data of an emergency generator car; extracting starting current characteristic curves and steady-state power characteristic curves of each load according to the historical operation data, and constructing a load characteristic fingerprint library containing load importance identifiers; based on the load characteristic fingerprint library, identifying the impact current peak time and duration of each load, and calculating the transient impact superposition coefficient when adjacent loads are switched; according to the real-time output parameter data of the emergency generator car, obtaining the instantaneous overload capacity threshold of the emergency generator car, and according to the transient impact superposition coefficient and the instantaneous overload capacity threshold, constructing a load switching time sequence constraint matrix; combining the load importance identifiers in the load characteristic fingerprint library and the load switching time sequence constraint matrix to generate an initial switching sequence containing switching time and interval duration; based on the load switching time sequence constraint matrix, performing impact waveform superposition simulation on the initial switching sequence to obtain the predicted waveform of the generator car output current at each switching time; by analyzing the peak value overrun interval of the predicted waveform of the generator car output current, identifying the overload risk switching node in the initial switching sequence; for the overload risk switching node, adjusting the switching time of the corresponding load or splitting the load into multiple sub-units for batch switching to generate an optimized switching sequence; controlling the automatic access operation of the emergency generator car and the loads in the power distribution house according to the optimized switching sequence, and collecting the voltage fluctuation amplitude and frequency deviation in the access process in real time; determining whether the voltage fluctuation amplitude or frequency deviation exceeds the preset safety range; in response to yes, pausing the subsequent load switching and adjusting the power distribution ratio of the switched load until the preset safety range is restored and the optimized switching sequence is continued to be executed; the load characteristic fingerprint library is used to identify the impact current peak time and duration of each load, and to calculate the transient impact superposition coefficient when adjacent loads are switched; The method comprises the following steps: According to the starting current characteristic curve of each load, the time corresponding to the current peak value is calibrated as the impact current peak time; calculate the time experienced by the starting current characteristic curve from starting to the current value decaying to the steady-state value within a preset proportion, and record it as the duration; construct a switching time difference matrix of adjacent loads, and the matrix element value represents the time interval of the switching instructions of two adjacent loads; determine whether the time interval is less than the sum of the durations of the two adjacent loads; in response to yes, calculate the overlapping area of the impact currents of the two adjacent loads on the time axis; according to the product of the overlapping area and the impact current peak value of each load, the transient impact superposition coefficient is calculated; according to the real-time output parameter data of the emergency generator car, obtaining the instantaneous overload capacity threshold of the emergency generator car, and according to the transient impact superposition coefficient and the instantaneous overload capacity threshold, constructing a load switching time sequence constraint matrix, comprising: extracting the rated output current and the allowed instantaneous overload current multiple from the real-time output parameter data of the emergency generator car to calculate the instantaneous overload capacity threshold; traverse the switching time difference matrix, and for the load with a non-zero transient impact superposition coefficient, calculate the total current peak value after the impact current superposition; determining whether the total current peak value exceeds the transient overload capability threshold, and in response, marking the load as prohibited from being switched simultaneously, and calculating a minimum switching time interval satisfying the transient overload capability threshold requirement as a constraint time interval; constructing a load switching time sequence constraint matrix according to the constraint time interval, and filling the constraint time interval into a corresponding element position of the load switching time sequence constraint matrix.

2. The method for quick restoration of the low-voltage distribution station emergency power generation vehicle according to claim 1, characterized in that, According to the historical operation data, the starting current characteristic curve and the steady-state power characteristic curve of each load are extracted, and a load characteristic fingerprint library containing load importance identification is constructed, including: aligning the multiple starting processes of each load in the historical operation data in time, establishing a unified time axis with the starting instruction issuing time as the reference point, and extracting the complete current change curve from zero to steady state in each starting process; segmenting and normalizing the multiple current complete change curves of the same load, dividing the current complete change curve into a mutation segment, a transition segment and a steady-state segment; calculate the average current slope and peak current multiple of the mutation segment, the transition segment and the steady-state segment, and combine the power factor of the steady-state segment to form a multi-dimensional feature vector, denoted as the starting current characteristic curve; wherein the feature vector is calculated as follows: ， wherein, is the start current characteristic vector, is the average current slope of the three phases, respectively, is the ratio of peak current to steady state current, is the steady state power factor; extracting the time sequence data of active power, reactive power and harmonic content of each load in the normal operation state, and obtaining the steady-state power characteristic curve after low-pass filtering; associate the starting current characteristic curve and the steady-state power characteristic curve to the corresponding load identification, and label the importance level tag, to construct the load characteristic fingerprint library.

3. The method of claim 1, wherein the method further comprises: The load importance identification in the load characteristic fingerprint library and the load switching time sequence constraint matrix are combined to generate an initial switching sequence containing switching time and interval length; including: According to the load importance identification in the load characteristic fingerprint library, all loads are divided into high-priority load group, medium-priority load group and low-priority load group; sort the internal switching sequence of each group according to the steady-state power from small to large within each priority load group; link the priority load groups in turn according to the priority to form a global load switching sequence chain; take the time when the emergency generator car starts to stabilize as the starting time, and according to the constraint time interval in the load switching time sequence constraint matrix, assign switching times to each load in turn; calculate the time difference between adjacent switching times as the interval length; combine the switching time and interval length of each load to form the initial switching sequence.

4. The method for quick restoration of the low-voltage distribution station emergency power generation vehicle according to claim 3, characterized in that, Based on the load switching time sequence constraint matrix, the initial switching sequence is subjected to shock waveform superposition simulation to obtain the predicted waveform of the generator car output current at each switching time; including: According to the switching time of each load in the initial switching sequence, a switching event sequence is established on the time axis; extract the load pairs with constraint relationship from the load switching time sequence constraint matrix, and identify the time interval with the risk of superimposed shock current in the switching event sequence; extract the starting current characteristic curve of each load from the load characteristic fingerprint library, and align the starting time of each curve to the corresponding switching time in the switching event sequence; In the time interval where the risk of superimposed inrush current exists, the inrush current characteristic curve of the relevant load is sampled at a high time resolution, and in the remaining time interval, it is sampled at a standard time resolution; On the unified time axis, the inrush current characteristic curve of the load and the steady-state power characteristic curve are vector superimposed according to the corresponding time resolution to obtain a total load current demand curve; An equivalent output impedance model of the emergency generator car is established, the total load current demand curve is input into the equivalent output impedance model, the actual output current time sequence data considering the dynamic response of the generator car is calculated, and an output current prediction wave is generated.

5. The method for quick restoration of the low-voltage distribution station emergency power generation vehicle according to claim 4, characterized in that, The peak exceeding interval of the output current prediction waveform of the generator car is analyzed to identify an overload risk switching node in the initial switching sequence; comprising: Local extreme value detection is performed on the output current prediction waveform of the generator car, and all local peak points and their corresponding time points are extracted; The current values of the local peak points are compared with the instantaneous overload capacity threshold, and the exceeding peak points exceeding the threshold are counted; For each exceeding peak point, the nearest switching event time point is traced back to identify the load combination that causes the exceeding peak point; The difference between the current value of the exceeding peak point and the instantaneous overload capacity threshold is calculated as the exceeding current amplitude, and the risk level is classified according to the exceeding current amplitude; The switching event time point corresponding to the exceeding peak point with a risk level higher than a preset risk threshold is marked as an overload risk switching node.

6. The method for quick restoration of the low-voltage distribution station emergency power generation vehicle of claim 5, characterized in that, For the overload risk switching node, the switching time of the corresponding load is adjusted or the load is split into multiple sub-units for batch switching to generate an optimized switching sequence; comprising: For each overload risk switching node, the corresponding load and its switching time in the initial switching sequence are extracted; The waveform peak value change after delaying the switching time of the load by different time lengths is calculated; If there is a delay time that makes the waveform peak value below the instantaneous overload capacity threshold, the switching time of the load is delayed by the corresponding time length; If any delay time cannot meet the requirements, the load is split into multiple sub-units according to the list of electrical equipment inside the load, and the switching time is re-allocated; The adjusted switching time or the switching time of the split sub-units is updated to the initial switching sequence to form an optimized switching sequence.

7. The method for quick restoration of the low-voltage distribution station emergency power generation vehicle according to claim 6, characterized in that, According to the optimized switching sequence, the automatic connection operation of the emergency generator car and the load of the power distribution station is controlled, and the voltage fluctuation amplitude and frequency deviation during the connection process are collected in real time; comprising: According to the optimized switching sequence, a control instruction set containing the switching time of each load and the corresponding circuit breaker number is generated and sent to the automatic control module of the power distribution system of the power distribution station; A monitoring window is set before and after the closing of each circuit breaker, and the three-phase voltage instantaneous value and frequency instantaneous value of the output end of the emergency generator car are collected at a high sampling frequency in the monitoring window; The effective value of the three-phase voltage instantaneous value is calculated, the voltage maximum and voltage minimum in the monitoring window are extracted, the difference between them and the rated voltage is calculated, and the absolute value of the larger one is taken as the voltage fluctuation amplitude; The frequency instantaneous value is statistically analyzed, and the maximum deviation of the frequency in the monitoring window is calculated as the frequency deviation.

8. The method for quick restoration of the low-voltage distribution station emergency power generation vehicle of claim 7, characterized in that, The response is then, suspend the subsequent load switching and adjust the power distribution ratio of the switched load until the preset safety range is restored to continue the optimized switching sequence, including: Immediately send a pause instruction to the automation control module of the power distribution station building power distribution system, freeze the subsequent switching operation in the optimized switching sequence that has not been executed; Identify the current switched load set, filter out the adjustable load with power regulation capability, and issue a power regulation instruction to reduce its operating power; Continuously monitor the voltage fluctuation amplitude and frequency deviation, and when both are restored to the preset safety range and remain stable for more than the preset stable time, the pause instruction is removed; Resuming the subsequent load switching operation in the optimized switching sequence, and gradually restoring the power of the adjustable load to the normal operating level after the subsequent load switching is completed.

Images