A medium-voltage distribution network non-power-off maintenance dynamic balance optimization method and system

Abstract

The application discloses a kind of medium voltage distribution network not maintenance dynamic balance optimization method and system, belong to distribution network not power operation and distributed photovoltaic consumption technology field.The method constructs temporary power supply island model according to maintenance work order dynamically and calculates access point adaptive safety parameter;Establish the state machine containing four stages of maintenance preparation, island power supply, cut back pre-adjustment and power restoration parallel, construct dynamic state vector containing prediction reliability and safety margin, drive objective function penalty weight adaptive adjustment and utilize short-circuit capacity weight differentiation amplification voltage out-of-limit penalty;Rolling optimization is solved to generate car output and photovoltaic utilization, and photovoltaic instruction is issued with set value, and when generating car instruction exceeds climbing ability, it is decomposed into slope sub-instruction and executed step by step;According to the deviation between prediction and execution, enter normal correction, prediction correction or emergency optimization path to realize closed loop, ensure the economy, safety and executability of whole maintenance process.

Description

Technical Field

[0001] This invention belongs to the field of power system distribution network uninterrupted operation and distributed photovoltaic consumption technology, specifically involving a dynamic optimization control method and system for mobile generator vehicles during planned maintenance of medium-voltage distribution networks. Background Technology

[0002] Medium-voltage distribution networks are a crucial link in power supply to end users. Traditional planned maintenance typically requires power outages on the lines under maintenance, impacting power supply reliability and user experience. To mitigate the impact of maintenance outages, medium-voltage generator trucks are often used on-site. These trucks connect to overhead lines, ring main units, switchgear spare bays, or other access points via flexible cables. After synchronous grid connection, the downstream load is seamlessly transferred to the generator truck for power supply. Subsequently, the maintenance section is isolated, and the generator truck supports the downstream load in an islanded manner. After maintenance is completed, the network is then synchronized with the grid side, and the load is switched back to the grid.

[0003] With the large-scale integration of distributed photovoltaic (PV) power into medium- and low-voltage distribution networks, the power balance issue in uninterrupted maintenance scenarios has undergone significant changes. PV output is affected by irradiance, cloud movement, and inverter operating status, and can fluctuate dramatically within minutes; user load, in turn, fluctuates due to changes in industrial production, residential air conditioning, and weekday patterns. Therefore, the net load that the generator unit in the maintenance island actually needs to bear is not a fixed value, but a dynamic quantity resulting from the combined effects of user load, PV output, line losses, and voltage constraints.

[0004] The invention patent with publication number CN121727094A, entitled "Automatic Grid Connection and Disconnection Control Method, System, Equipment and Medium for Mobile Generator Vehicles," relates to an automatic grid connection and disconnection control method for mobile generator vehicles. It controls the opening and closing of the grid connection switch based on real-time phase difference, frequency difference, and voltage difference to achieve generator vehicle connection to the distribution network and disconnection after maintenance. The Chinese invention patent with publication number CN120879716B, entitled "Device, System and Method for Grid Connection and Disconnection of Generator Vehicles," proposes a device and system for uninterrupted power supply operation for grid connection and disconnection of generator vehicles. This system discharges energy to the load through an energy storage unit and synchronizes the current power parameters before synchronizing the target power parameters during power supply switching to achieve uninterrupted operation. The invention patent with publication number CN118801558A, entitled "A Maintenance and Power Supply Protection Method and Integrated Device for Uninterrupted Operation," discloses a maintenance and power supply protection method for uninterrupted operation, enabling quasi-synchronous and non-grid-connected synchronous switching between the low-voltage distribution network and the generator vehicle. However, these solutions mainly focus on how generator trucks connect to the distribution network and how to realize the operational logic of grid connection and disconnection. They are technical solutions at the level of operation process and switching control. Their core is to solve the operational problems of "how to connect" and "how to switch". They do not involve the energy management problem of how to continuously optimize the output of generator trucks based on the dynamic changes of photovoltaic and load after connection, while ensuring safety. Summary of the Invention

[0005] Purpose of the invention: In order to overcome the shortcomings of the prior art, this invention solves the problem of how to generate generator output instructions that are economical, safe and field-executable through a unified optimization framework under the conditions that photovoltaic output and load prediction are uncertain, maintenance access points and isolation ranges change with each operation, generator vehicles have limited climbing speed and need to be smoothly switched back to the grid after maintenance. This enables dynamic balance optimization of the entire process of uninterrupted maintenance of medium-voltage distribution networks with distributed photovoltaics. In addition, this invention also provides a dynamic balance optimization system for uninterrupted maintenance of medium-voltage distribution networks.

[0006] Technical solution: In a first aspect, the present invention provides a dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks, including: Read the uninterrupted maintenance work order of the medium-voltage generator truck and determine the connection point of the medium-voltage generator truck and the status vector of the disconnecting switch during maintenance. Based on the access point and the maintenance disconnect switch state vector, a temporary power supply island model is constructed, and adaptive safety parameters of the access point are calculated based on the temporary power supply island model. Establish a maintenance phase state machine based on the actual progress status of maintenance operations, including the maintenance preparation phase, the islanded power supply phase, the switchback pre-adjustment phase, and the power restoration parallel phase. According to the state machine of the maintenance stage, the objective function and constraints of different stages are activated. The penalty weights of each item in the objective function are dynamically adjusted according to the maintenance dynamic state vector. The short-circuit capacity weight in the adaptive safety parameters of the access point is used to differentiate and amplify the voltage over-limit penalty item. Then, the active power output and photovoltaic utilization of each generator car are solved in the current rolling window. The photovoltaic utilization is sent to the photovoltaic inverter in the form of a power setpoint, and the corresponding power limiting is executed. If the photovoltaic command cannot be executed, the value of the previous cycle is maintained and the deviation information is fed back into the optimization parameter library. For the generator command, if the difference between the current actual output and the command exceeds the ramping capability, it is decomposed into a ramp sub-command that meets the ramping rate constraint and sent out step by step. The photovoltaic command and the generator command are sent out independently. At the end of each execution cycle, the prediction deviation and the execution deviation are respectively entered into one of the three paths: normal rolling correction, prediction deviation correction or emergency optimization, to achieve closed-loop correction.

[0007] Furthermore, including: The objective functions for the different stages are expressed as follows: in, It is the time step number within the scrolling window H. This is expressed as the sum of costs for all time periods within the window. For the first The probability of each scenario occurring. Numbering the risk scenarios. For the firstt Fuel costs at any time For the first t The penalty for the generator car deviating from the economic load rate range at any time. This is a penalty for curtailment of photovoltaic power. This is a penalty item for reverse power risk. This is a penalty term for voltage and branch current margins. For generator command tracking penalty items, For parallel load sharing penalty items, To predict the reserve margin penalty term caused by reduced confidence, Penalty for power-on switching back to the terminal; and All of these correspond to the islanded power supply phase. This corresponds to the switchback to the pre-adjustment stage.

[0008] The constraints include: power balance constraints, node voltage constraints, branch current constraints, reverse power constraints, upper and lower limits of generator output, generator ramp climbing rate constraints, and photovoltaic absorption constraints.

[0009] Furthermore, the node voltage constraint and branch current constraint are constructed based on the sensitivity of the access point adaptive security parameters, including: Sensitivity of generator active power output to node voltage Sensitivity of generator reactive power output to node voltage Construct a linearized estimate, expressed as: ; in, For nodes k The reference voltage, For generator car g Planned active power adjustment relative to the baseline value For generator car g Planned reactive power adjustment relative to the baseline value; The node voltage constraints are determined based on the linearized estimates, and are expressed as follows: ; in, and These are the minimum and maximum values ​​of the node voltage, respectively; Sensitivity of generator car active power output to branch current The linearized estimate of the branch current is determined as follows: ; in, For the first e The reference current of the branch; The branch current constraint is determined based on the linearized estimate of the branch current, and is expressed as follows: ;in, For branch current safety limits.

[0010] Furthermore, this includes: dynamically adjusting the penalty weights in the objective function based on the maintenance dynamic state vector, including: Construct a dynamic state vector for maintenance, including: the real-time net load at the current optimization moment. The standard deviation of the prediction error at the current optimization time. Confidence of prediction at the current optimization time The node at the current optimization moment k Voltage margin The current margin of branch e at the current optimization moment. Inverse power margin of the access point at the current optimization moment The current optimization time is the th i Load rate of generator cars The current optimization time is the th i Tracking margin of a generator car Parallel load sharing deviation at the current optimization moment Combined synchronization deviation with the future cut-back time T ; The prediction confidence Alternative penalty item used to drive prediction confidence The lower the prediction confidence, the greater the spare penalty, prompting the optimizer to reserve more spin-off spare; node voltage margin. Penalty terms for drive voltage and branch current margin In the voltage component, a penalty is applied when the voltage margin falls below a preset safety threshold. The penalty coefficient for weak nodes is amplified by the short-circuit capacity weight. (Branch current margin) Penalty terms for drive voltage and branch current margin The current component in the circuit applies a penalty when the current margin falls below a preset safety threshold; generator tracking margin. Penalties for driving generator car command tracking A squared penalty is applied when the tracking margin is less than zero; synchronization cut-back bias. Penalty for deviation in driving power-back switching terminal The greater the deviation, the greater the penalty.

[0011] Furthermore, the penalty coefficient for the weak node is amplified by the short-circuit capacity weight, including: The voltage and branch current margin penalty terms are represented by short-circuit capacity weighting, as follows: in, and These are the basic weighting coefficients for voltage margin and current margin penalties, respectively. The preset safety threshold for voltage margin, This is a preset safety threshold for current margin; a penalty is triggered when the actual margin falls below this threshold. The short-circuit capacity weighting ratio is expressed as: ;in, This is the minimum short-circuit capacity weight value within the power supply island. This is the short-circuit capacity weight value, when Characterize weak nodes to obtain a higher voltage over-limit penalty coefficient.

[0012] Furthermore, it includes: the drive power-on switchback terminal deviation penalty term. , is represented as: ; in, The weighting coefficient for the power-on switchback terminal deviation penalty term; the synchronization switchback deviation , is represented as: ; Where T is the predetermined switchback time; These represent the voltage, frequency, and phase angle on the island side at time T, respectively. These are the voltage, frequency, and phase angle at the synchronization point on the grid side, respectively. The system's nominal voltage. The system's rated frequency, To determine the net power exchange between the island and the grid at the time of switchback, This represents the predicted total load within the island at the time of the switchback. The normalized weighting coefficients for each deviation satisfy the following conditions: ; When the state machine is in the switchback pre-adjustment stage during the maintenance phase, Write the objective function and improve The weighting allows the islanded state to approach the synchronous parallel conditions of the grid side in advance.

[0013] Furthermore, this includes: for generator commands, if the difference between the current actual output and the command exceeds the climbing capacity, it is decomposed into ramp sub-commands that satisfy the climbing rate constraint and issued step by step, including: When the instruction is to be issued Compared with the current actual output of the generator car The difference between them does not exceed Direct issuance is permitted, wherein the current instruction to be issued is the active power output of the generator in the previous cycle. Generator car i The maximum ramp rate of active power represents the maximum increase in output power per unit time. For generator car iThe maximum time allowed to complete one instruction tracking, multiplied by the two, represents the maximum power regulation that the generator can complete within the maximum allowed time; When the instruction is to be issued Compared with the current actual output of the generator car The difference between them exceeds At that time, the instruction is broken down into A slope sub-instruction that satisfies the climbing rate constraint, wherein, To optimize the execution cycle, i.e. the time interval between two consecutive instructions, For the floor function, when hour, No sub-instructions need to be issued; After decomposition, the first k The sub-instructions are: .

[0014] Furthermore, including: The rolling correction based on prediction deviation and execution deviation in the next execution cycle includes: In each execution cycle t At the end, the prediction bias was collected. and execution deviation These two types of deviations will serve as the basis for correction in the next execution cycle, and will be categorized and entered into different processing paths. If the prediction is biased and execution deviation If none of the prediction deviations exceed the corresponding thresholds, normal rolling correction begins; The corresponding prediction bias threshold is expressed as: ;in, for t The standard deviation of the net load forecast error at any given time. for t Real-time measured total load; If the prediction is biased Exceeding the prediction bias threshold, but execution bias If the prediction deviation threshold of 3% is not exceeded, the current measured value is used as the latest sample to update the prediction correction coefficient, so that the next round of photovoltaic output and load prediction values ​​are corrected in the direction of the measured values; at the same time, the weight of the prediction credibility reserve penalty term is increased, and more generator cars are reserved for backup in the next round of optimization to compensate for the safety risks caused by prediction uncertainty; finally, the corrected prediction model and the updated state vector are used to re-enter the next round of optimization. When the generator car performs a deviation When the prediction deviation exceeds the 3% threshold and persists for two consecutive sampling periods, regardless of the prediction deviation... If the threshold is exceeded, an emergency optimization will be triggered immediately until the normal rolling correction process is resumed.

[0015] Furthermore, it includes: the prediction bias Represented as: ; in, for t Real-time measured net load for t Forecast net load in real time. for t Real-time measured total load To prevent small positive numbers with a denominator of zero; Execution deviation Represented as: ; in, for t Time-based generator car i Actual output For generator car i Rated power, for t The information is constantly being sent to the generator car. i The target instruction.

[0016] Furthermore, this includes: using the current measured values ​​as the latest sample to update the prediction correction coefficients, so that the next round of photovoltaic output and load predictions are corrected towards the measured values, including: The photovoltaic output and load forecasts are updated using the measured values ​​of the photovoltaic output and load from the previous execution cycle, expressed as: ;

[0017] Furthermore, the normal rolling correction process includes the following steps: The deviation between the measured value and the predicted value in this period is included in the historical error statistics window, and the standard deviation of the prediction error and the prediction confidence are updated. The standard deviation of the prediction error is expressed as: ; Where M is the number of historical sampling points used for statistical error; For the first t-m The measured net load at each sampling time. For the first t-m Predicted net load at each sampling time; The prediction confidence level is expressed as: ; in, This is a power reference value; The margin parameters in the state vector are updated based on the latest measured data; the updated state vector is used as the input for the next round of optimization to resolve the generator output command.

[0018] Secondly, the present invention also provides a dynamic balance optimization system for uninterrupted maintenance of medium-voltage distribution networks, comprising: The maintenance work order parsing module is used to read the uninterrupted maintenance work order of the medium-voltage generator truck and determine the connection point of the medium-voltage generator truck and the status vector of the maintenance disconnect switch. The access point adaptive power supply island modeling module is used to construct a temporary power supply island model based on the access point and the maintenance disconnect switch state vector, and to calculate the access point adaptive safety parameters based on the temporary power supply island model. The maintenance phase state machine module is used to establish and maintain the maintenance phase state machine based on the actual progress status of the maintenance operation, including the maintenance preparation phase, the islanded power supply phase, the switchback pre-adjustment phase, and the power restoration parallel phase. The rolling optimization module is used to activate the objective function and constraints of different stages according to the maintenance stage state machine, dynamically adjust the penalty weights of each item in the objective function according to the maintenance dynamic state vector, and use the short-circuit capacity weight in the access point adaptive safety parameters to differentiate and amplify the voltage over-limit penalty item, thereby solving the active power output and photovoltaic utilization of each generator car within the current rolling window. The instruction execution and feedback module is used to send the photovoltaic utilization amount to the photovoltaic inverter in the form of a power setpoint and execute the corresponding power limiting amount; if the photovoltaic instruction cannot be executed, the value of the previous cycle is maintained and the deviation information is fed back to the optimization parameter library; for the generator vehicle instruction, if the difference between it and the current actual output exceeds the climbing ability, it is decomposed into a ramp sub-instruction that meets the climbing rate constraint and issued step by step, and the photovoltaic instruction and the generator vehicle instruction are issued independently; at the end of each execution cycle, it enters one of the three paths of normal rolling correction, prediction deviation correction or emergency optimization according to the prediction deviation and the execution deviation, to achieve closed-loop correction.

[0019] Thirdly, the present invention also provides a computer device, comprising: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, the steps of the dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks as described in the first aspect are implemented.

[0020] Fourthly, the present invention also provides a computer-readable storage medium, characterized in that it stores a computer program thereon, which, when executed, implements the steps of the dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks as described in the first aspect.

[0021] Beneficial effects: Compared with the prior art, the present invention has the following advantages: (1) This invention uses the real-time changes in net load as input through rolling updates of photovoltaic power output prediction and load assessment. This drives the optimizer to re-solve the generator output command in each execution cycle, so that the generator output is no longer a fixed value or a segmented fixed value, but dynamically adjusted according to the fluctuations in photovoltaic power output and load changes. Example data shows that the strategy of this invention reduces fuel consumption by approximately 26.7% compared to the fixed output strategy, from 176L to 129L, and further reduces it by approximately 12.8% compared to the ordinary rolling optimization strategy, verifying the effect of dynamically tracking net load changes on improving fuel economy.

[0022] (2) By incorporating the photovoltaic curtailment penalty into the objective function, the optimizer tends to maximize photovoltaic utilization under the premise of satisfying the reverse power constraint in the actual solution. When the available photovoltaic power exceeds the load demand, the optimizer does not directly limit the full amount of power generation, but weighs the curtailment penalty and the reverse power penalty, and chooses to reduce curtailment as much as possible within the safety boundary, thereby improving the photovoltaic absorption rate.

[0023] (3) The present invention monitors the power flow direction and remaining allowable capacity of the access point in real time by using the reverse power margin index. When the reverse power margin is lower than the preset threshold, the upper limit of the generator output or the photovoltaic utilization is limited in the optimization model by activating the reverse power margin constraint, so as to ensure that the power of the access point is always kept within the allowable set value and the reverse power protection device will not trip, thereby ensuring the continuity of the maintenance operation without power outage. In addition, the safety margin of each node and branch is monitored in real time by using the node voltage margin and branch current margin index. When any margin is lower than the preset threshold, the weight of the corresponding over-limit penalty item is automatically increased and the boundary of the power flow constraint is tightened, so that the optimizer actively pulls the operating point back to the safe area when the safety boundary shrinks, ensuring that the voltage and current are always controlled within the allowable range.

[0024] (4) The present invention uses the tracking margin index of the generator car to determine whether the target instruction exceeds the physical execution capability. When the tracking margin is less than zero, the direct issuance of the instruction is prohibited. Instead, the target instruction is decomposed into multiple ramp sub-instructions that meet the ramp rate constraint and issued step by step. This ensures that the generator car has enough time to complete the tracking in each execution cycle, avoids the optimizer from issuing output jump instructions that exceed the ramp rate capability, and ensures that the optimization result is physically executable. Attached Figure Description

[0025] Figure 1 This is a flowchart of the dynamic balance optimization method for uninterrupted power supply maintenance of medium-voltage distribution networks according to an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] This embodiment provides a dynamic balance optimization method for uninterrupted power supply maintenance of medium-voltage distribution networks, such as... Figure 1 As shown, it includes the following steps: S1 reads the uninterrupted maintenance work order of the medium-voltage generator vehicle and determines the connection point of the medium-voltage generator vehicle and the status vector of the maintenance disconnect switch.

[0028] The work order for uninterrupted maintenance of the medium-voltage generator vehicle specifies the feeder number F and the connection point of the medium-voltage generator vehicle. a , Maintenance disconnector state vector s Scheduled maintenance start time T0, scheduled switchback time T end N sets of generator cars g Distributed photovoltaic node set N pv and load node set N load This forms the work object data packet D0; in this embodiment, the medium-voltage generator vehicle access point... a This is not merely a physical connection location, but also the logical starting point for the construction and optimization of the entire temporary power supply island. It refers not only to the physical connection location between the generator truck and grid equipment, such as the spare bay of a ring main unit, the tower of an overhead line, or the busbar of a substation, via flexible cables (as described below), but also to the access point... a It is also the data input and calculation origin of the algorithm. This embodiment starts from this point, combines the switch status, to search and define the range of the temporary power supply island for this maintenance, and calculate parameters such as electrical distance and sensitivity to each node.

[0029] S2 constructs a temporary power supply island model based on the access point and the maintenance disconnect switch state vector, and calculates the adaptive safety parameters of the access point based on the temporary power supply island model.

[0030] This embodiment is based on the access point a and maintenance disconnect switch state vector s Extract the temporary power supply island model M, which corresponds only to this maintenance operation, from the distribution network topology model G=(N,E,S,A). island ; In a preferred embodiment, the temporary power supply island model M island The set of power supply island nodes N island Power supply island branch collection Load node set Distributed photovoltaic node set The set of generator nodes N g Branch impedance matrix Z, node admittance matrix Y, and equipment limit L im Composition; The state vector of the disconnect switch under maintenance s After writing the distribution network topology model G, from the access point a Search for reachable nodes along closed branches, excluding nodes and branches blocked by disconnecting switches under maintenance.

[0031] Therefore, a power supply island model is generated based on the distribution network topology model G=(N,E,S,A). Where N includes busbars, ring main units, switching stations, distribution transformers, photovoltaic grid connection points, and generator vehicle access points; E includes overhead lines, cables, and transformer branches; S records the open / closed status of each switch; and A records equipment attributes such as line length, cross-section, unit impedance, and rated current. Furthermore, the power supply island nodes are represented as follows: .

[0032] In this embodiment, the access point adaptive security parameters K safe At least including: (1) Access point a to node k The line impedance is expressed as: ;in, Used to calculate the distance from access point a to node k The accumulated impedance is the basis for evaluating electrical distance and voltage sensitivity. In the distribution network topology, from the access point a To the node k The set of all branches along the single power supply path. For a certain branch of the set, branch road Resistance per unit length branch road Reactance per unit length It is the imaginary unit.

[0033] (2) Access point a To node k The electrical distance is expressed as: ; in, This value represents the relative electrical distance between nodes; the greater the distance, the more susceptible the voltage is to influence. It is used to determine the weight of voltage-weak nodes. Representative power supply island set Any node in it.

[0034] (3) Node k The equivalent short-circuit capacity is expressed as: , Reflecting nodes k The smaller the grid strength and short-circuit capacity, the weaker the node, and the easier it is for the voltage to be exceeded. For the system nominal voltage, For nodes k Thevenin equivalent impedance.

[0035] (4) Node k Short-circuit capacity weight: Based on the normalized weights calculated from the short-circuit capacity, nodes with smaller short-circuit capacities have larger weights and will receive higher voltage exceedance penalties. To prevent small positive numbers with a denominator of zero.

[0036] In this embodiment, the aforementioned line impedance Branch impedance data The equipment parameters are derived from the equipment attribute set A of the temporary power supply island model, and the equivalent short-circuit capacity. Thevenin equivalent impedance The inverse of the nodal admittance matrix Y of the temporary power supply island model is obtained, specifically as follows: That is, the corresponding node in the inverse matrix of the node admittance matrix. k The diagonal element. This embodiment uses... and The system can identify nodes that are far from the generator truck's connection point and have low short-circuit capacity. These nodes are the weak points within the power supply island. and This will be used to dynamically adjust the penalty weight for voltage exceedances in the objective function. For identified weak nodes, the optimizer will give them higher priority, ensuring their voltage does not exceed the limit, thereby improving the overall safety of the power supply island. The adaptive safety parameters will change with each maintenance access point. a and switch state vector s The parameters change depending on the situation, which reflects that this solution can adapt to different working conditions for each operation, and is one of the core features that distinguishes it from fixed parameter solutions.

[0037] (5) Power flow sensitivity, which includes the sensitivity of generator active power output to node voltage. Sensitivity of generator reactive power output to node voltage Sensitivity of generator active power output to branch current and access point inverse power sensitivity The aforementioned sensitivity is calculated from the reference power flow disturbance of the power supply island during this maintenance or recursively identified from historical measurements at the same access point, and varies with the access point. a or switch state vector s It changes and updates.

[0038] In this embodiment, To apply a small active power disturbance at the generator node, the power flow is recalculated to obtain the node voltage change. To apply a small reactive power disturbance at the generator node, the power flow is recalculated to obtain the node voltage change. To apply a small active power disturbance at the generator node, the power flow is recalculated to obtain the change in branch current. To obtain the inverse power change at the connection point by applying active or reactive disturbances at the generator node.

[0039] The purpose of constructing power flow sensitivity in this embodiment is to avoid repeatedly calling the complete power flow calculation during the rolling optimization process, to achieve rapid verification of safety constraints, and to enable the optimization solution to converge within the execution cycle.

[0040] Without setting sensitivity, in each iteration of rolling optimization, the optimizer needs to repeatedly try different combinations of generator outputs, i.e., different... The values ​​are selected, and it is determined whether these combinations will lead to node voltage exceeding limits or branch current overload. If a complete Newton-Raphson power flow calculation is performed again for each attempt, i.e., solving a large-scale nonlinear equation system, a single iteration may take several seconds to tens of seconds. Rolling optimization requires repeated verification over hundreds or even thousands of iterations, and the total time may reach several minutes to tens of minutes, far exceeding the execution cycle limit, which is usually 5 minutes. However, with the sensitivity method used in this embodiment, when the optimizer tries a set of output schemes, it does not need to recalculate the complete power flow. It only needs to use the sensitivity to linearly estimate the voltage and current under the current scheme. A single estimation only requires a simple microsecond-level multiply-add operation, instead of solving a nonlinear equation system in seconds, improving the calculation speed by about 1000 times. This may complete hundreds to thousands of scheme evaluations within one execution cycle.

[0041] S3 establishes a maintenance phase state machine based on the actual progress of maintenance operations, including the maintenance preparation phase, islanded power supply phase, switchback pre-adjustment phase, and power restoration parallel phase. .

[0042] In this embodiment, since uninterrupted power supply maintenance is a sequential process, the electrical characteristics and safety requirements are completely different at different stages. Therefore, it is divided into the following four stages: Maintenance preparation stage: At this time, the generator truck has not yet been connected to the grid, the island has not been formed, and the grid is in normal operation. There is no optimization target at this time. It is only necessary to check whether the topology of the access point is correct, whether the switch status is consistent with the work order, and calculate the baseline power flow as the initial value for subsequent optimization; Island power supply stage: At this time, the back-end load has been seamlessly transferred to the generator truck, the maintenance section has been isolated, and the island operates independently. At this time, the optimizer needs to minimize the fuel consumption of the generator truck and maximize the absorption of photovoltaic power while ensuring power balance, voltage not exceeding limits, current not exceeding overload, and reverse power not triggering, while adhering to the hard safety bottom line. Switch back and pre-adjustment stage: At this time, the maintenance work is nearing completion. The system needs to adjust the voltage, frequency, and phase angle of the island side to be basically consistent with the grid side within a few minutes to tens of minutes before the end of the maintenance. At this point, the safety baseline changes from not exceeding the limits within the isolated system to having sufficiently small synchronization deviations. The optimizer needs to sacrifice some economic efficiency, even at the cost of burning more fuel and wasting more light, in order to reduce the synchronization deviations to the allowable range. Power-on parallel phase: Parallel operation is in progress, the isolated system is about to disintegrate, and the optimization task has been completed.

[0043] Based on the segmentation, the activation state of each penalty weight in the objective function can be determined, as well as the activation and tightening degree of the constraints at different stages. The length of the rolling window at different stages is also adaptively adjusted.

[0044] In a preferred embodiment, the scrolling window H is pressed... Sure, This is the maximum allowable comprehensive deviation threshold for simultaneous parallel operation, which is the upper limit of the comprehensive deviation between the island side and the grid side in terms of voltage, frequency, phase angle, and net exchange power.

[0045] when During the maintenance preparation phase or the power restoration phase, ; when During the phase of supplying power to the isolated island, ; when When switching back to the pre-adjustment phase, ; in, For the execution cycle, To determine the reliable prediction duration based on the current prediction error statistics, This represents the remaining time until the planned switchback time. The longest time required to complete the target command adjustment among all online generator cars. Current synchronization deviation The estimated time required to converge to the allowable threshold. The method for determining this is as follows: based on the statistical characteristics of historical prediction errors, extrapolate the standard deviation of prediction errors at future times. ,Pick The maximum time As a reliable prediction duration.

[0046] The method for determining it is: based on the most recent several periods The trend of change, linear extrapolation converges to The required number of cycles, multiplied by To obtain the shortest pre-adjustment time.

[0047] S4 activates the objective functions and constraints of different stages according to the maintenance stage state machine, dynamically adjusts the penalty weights of each item in the objective function according to the maintenance dynamic state vector, and uses the short-circuit capacity weight in the access point adaptive safety parameters to differentiate and amplify the voltage over-limit penalty item, and then solves the active power output and photovoltaic utilization of each generator car within the current rolling window.

[0048] In this embodiment, a maintenance dynamic state vector is first constructed. Includes a global state used to characterize the global power balance and prediction reliability. 、 Network security status used to characterize the safety margins of node voltages and branch currents. The status of the generator car is used to characterize the load rate and command tracking capability of the generator car. and the target state for switching back, used to characterize the synchronization error of the power-on switching back. 。 Each sub-state is used to drive the backup penalty term in the objective function. Safety penalty items Economic and enforceable penalties Switch back to the pre-adjusted penalty item Adaptive weighting.

[0049] Specifically, it is expressed as follows: ;in, The global state, used for power balance and prediction assessment, is represented as: ; in, The real-time net load determines the benchmark for the total output of the generator car; The standard deviation of the prediction error is used to quantify the current prediction risk. To predict reliability, it acts as the master switch for dynamically adjusting and optimizing weights; The network electrical safety state, used for safety constraints, is represented as follows: ; in, For nodesk The voltage margin is used to prevent voltage over-limit. branch road e The current margin is used to prevent equipment overload. This is the reverse power margin at the access point, used to prevent power backflow tripping; The operating status of the generator car, used for economic efficiency and feasibility, is represented as follows: ; in, This refers to the single-machine load rate, used to adjust the economic load rate rewards and penalties. , For single-machine tracing margin, it determines whether the instruction is executable and is used for adjustment. ; To address the load sharing deviation during parallel operation, this is used to adjust load balancing rewards and penalties. , i This is the generator car's number. , A collection of generator cars.

[0050] To switch back to the target state for smooth power-on handover, it is represented as: ; The comprehensive synchronization deviation at the future cutback time T is used for .

[0051] Therefore, in summary, the maintenance dynamic state vector includes the following parameters: the real-time net load at the current optimization moment. The standard deviation of the prediction error at the current optimization time. Confidence of prediction at the current optimization time The node at the current optimization moment k Voltage margin The current margin of branch e at the current optimization moment. Inverse power margin of the access point at the current optimization moment The current optimization time is the th i Load rate of generator cars The current optimization time is the th i Tracking margin of a generator car Parallel load sharing deviation at the current optimization moment Combined synchronization deviation with the future cut-back time T ; In this embodiment, the prediction confidence level Alternative penalty item used to drive prediction confidence The lower the prediction confidence, the greater the spare penalty, prompting the optimizer to reserve more spins for spares; Generator load rate Used to drive economic load rate deviation penalty item A squared penalty is applied when the load factor deviates from the economic range; Node voltage margin Penalty terms for drive voltage and branch current margin In the voltage component, a penalty is applied when the voltage margin is lower than a preset safety threshold, wherein the penalty coefficient of the weak node is amplified by the short-circuit capacity weight; Branch current margin Penalty terms for drive voltage and branch current margin The current component in the circuit is penalized when the current margin falls below a preset safety threshold. Inverse power margin Used to drive reverse power risk penalty item The smaller the reverse power margin, the greater the penalty. Generator car tracking margin Penalties for driving generator car command tracking A squared penalty is applied when the tracking margin is less than zero; Parallel load sharing deviation Penalty for driving parallel load sharing The greater the deviation, the greater the penalty; Synchronous cutback deviation Penalty for deviation in driving power-back switching terminal The greater the deviation, the greater the penalty, causing the islanded state to approach the synchronous parallel conditions of the grid side ahead of time.

[0052] In a preferred embodiment, the state vector is calculated through on-site measurements, which include at least the active and reactive power of the load nodes. Photovoltaics can be used for active power and practical utilization Node voltage ,frequency Phase angle Branch current Actual output of the generator car and grid-side synchronous parameters ;time t Total active power loss within the power supply island The above actual measurements are used to calculate net load, prediction error, safety margin, generator tracking margin, and switchback synchronization deviation, including: Real-time Net Load , is represented as: This determines the benchmark for the total output of the generator, which is then used for power balance and prediction error calculation. Standard deviation of prediction error at the current optimization time , is represented as: , where M is the number of historical sampling points used for statistical error, such as the last 10 periods; For the first t-m The measured net load at each sampling time. For the first t-m The predicted net load at each sampling time.

[0053] The safety margin includes three physical dimensions: voltage, current, and reverse power, and a standardized calculation formula is given: Voltage margin ; Among them, the molecule takes This represents the distance of the node voltage from the nearest operating boundary, with the denominator being the voltage operating bandwidth; when the voltage is between the upper and lower limits, A higher value indicates a greater margin; when the voltage exceeds the limit, The smaller the value, the more severe the restriction.

[0054] Branch current margin ,in, It is a small positive number to prevent the denominator from being 0. It represents the relative margin of the branch current from the rated limit. A positive value indicates that the limit has not been exceeded, while a negative value indicates an overload. The smaller the value, the more dangerous it is.

[0055] Inverse power margin ,in, The maximum allowable reverse power setting for the access point. This represents the actual power currently fed back to the grid from the power supply island via the access point. A positive value indicates power being fed back, while a negative value indicates power being received from the grid. To prevent small positive numbers with a denominator of zero, inverse power margin is used. This indicates that the current reverse power is within the allowable range. This indicates that the limit has been exceeded, and the output of the generator truck needs to be reduced or the photovoltaic system needs to be restricted.

[0056] Generator car tracking margin , is represented as: ;in, For generator car i Maximum active power ramp rate, unit: kW / min For generator car i The maximum time allowed to complete one instruction trace, in minutes; the product of the two represents the time the generator can complete the current execution cycle from... The maximum power adjustment that can be achieved upon departure. The target instruction value calculated by the optimizer. To make practical contributions at present; when When this occurs, it indicates that the target instruction exceeds the dynamic tracking capability of the generator car and should not be issued directly.

[0057] The timing starts at the current optimization time and ends at the instruction activation time of the next execution cycle; its value is typically the execution cycle. The values ​​must be integer multiples of the specified values ​​to ensure that the tracking process is synchronized with the optimization cycle.

[0058] Parallel operation to share deviation , is represented as: , This is the difference between the maximum and minimum load rates among all online generators. When Exceeding the preset threshold ,like In the objective function A penalty is imposed to balance the output of each generator.

[0059] The current optimization time is the th i Load rate of generator cars , is represented as: ,in, Rated power, Indicates generator car i exist t The active power output at any given moment is also a decision variable in this method, due to the load factor. Its function is to drive the economic load rate to deviate from the penalty term in the objective function. It is necessary to evaluate in real time whether each candidate value falls within the economic range; therefore, the load factor is calculated based on the decision variables currently being solved. The time series solved by the optimizer is... ,in, This is the first value in the entire time series, and it is also used as the instruction to be issued. Then, compared with the current actual output. Compare the results to determine if the climb is within the candidate's climbing ability.

[0060] Prediction reliability at the current optimization time , is represented as: , The lower the value, the higher the weighting of backup capacity, risk scenarios, and security margin. This is the power reference value. For example: assume the standard deviation of the prediction error... benchmark value .therefore, This indicates that the prediction error is only 5% of the rated capacity, and the reliability is high. If the error is as high as 500kW, the ratio is 0.5. As can be seen from the two results above, the credibility has decreased significantly.

[0061] Secondly, based on the above, we construct objective functions for different stages, expressed as follows: in, It is the time step number within the scrolling window H. This is expressed as the sum of costs for all time periods within the window. For the first The probability of each scenario occurring. Numbering the risk scenarios. For the first t Fuel costs at any time For the first t The penalty for the generator car deviating from the economic load rate range at any time. This is a penalty for curtailment of photovoltaic power. This is a penalty item for reverse power risk. This is a penalty term for voltage and branch current margins. For generator command tracking penalty items, For parallel load sharing penalty items, To predict the reserve margin penalty term caused by reduced confidence, Penalty for power-on switching back to the terminal; and All of these correspond to the islanded power supply phase. This corresponds to the switchback to the pre-adjustment stage.

[0062] Fuel cost in this embodiment , is represented as: , This is the fuel consumption characteristic coefficient of the generator. To contribute in a way that yields results; therefore, this embodiment continuously tries different methods. By taking combinations of values ​​and substituting them into the above formula and other constraints, we can determine which combination is the most economical and safest.

[0063] It should be noted that in this embodiment, the weighting coefficient λ itself may have dimensions, which converts the dimensions of the multiplicand into a unified cost dimension, for example... The dimension is not 1, but cost / power; The dimension of is not 1, but rather cost / dimensionless quantity; therefore, each The specific dimensions are determined by the corresponding penalty term, but the common result is to make the final dimensions of all terms consistent. In this embodiment, the penalty weights of each term in the objective function are dynamically adjusted according to the maintenance dynamic state vector, including: (1) The prediction confidence Alternative penalty item used to drive prediction confidence The lower the prediction confidence, the greater the spare penalty, prompting the optimizer to reserve more spins.

[0064] In one preferred approach, a backup penalty is applied to the prediction confidence level. Represented as: ;in, The weighting coefficients for the backup penalty term of prediction credibility are used to adjust the degree of impact of the decrease in prediction credibility on the backup penalty.

[0065] (2) Node voltage margin Penalty terms for drive voltage and branch current margin In the voltage component, a penalty is applied when the voltage margin is lower than a preset safety threshold, wherein the penalty coefficient of the weak node is amplified by the short-circuit capacity weight; Branch current margin Penalty terms for drive voltage and branch current margin The current component in the circuit is penalized when the current margin falls below a preset safety threshold. In this embodiment, the short-circuit capacity weight is used to represent the voltage and branch current margin penalty terms, as follows: ; in, and These are the basic weighting coefficients for voltage margin and current margin penalties, respectively. The preset safety threshold for voltage margin, This is a preset safety threshold for current margin; a penalty is triggered when the actual margin falls below this threshold. The short-circuit capacity weighting ratio is expressed as: ; in, This is the minimum short-circuit capacity weight value within the power supply island. This is the short-circuit capacity weight value, when Characterize weak nodes to obtain a higher voltage over-limit penalty coefficient.

[0066] In this embodiment, the short-circuit capacity weight ratio is introduced into the voltage and branch current margin penalty term to achieve differentiated safety protection for nodes with different grid strengths. That is, weak nodes with long electrical distance and small short-circuit capacity are prioritized for protection by the optimizer, while strong nodes with short electrical distance and large short-circuit capacity give up some adjustment margin under the premise of ensuring safety.

[0067] Short circuit capacity The smaller the value, the more likely it is to be a node. k The weaker the power grid, the more susceptible the voltage is to disturbances. The short-circuit capacity weight is calculated accordingly. The larger the magnification factor, the higher the magnification factor. The larger it is, the greater it becomes. Ultimately, in the voltage over-limit penalty item... In the context, the penalty coefficient obtained by this node The larger the value, the more the optimizer automatically allocates more adjustment resources to the weakest nodes at the far end during the solution process, rather than distributing them evenly across the entire power supply island. Essentially, this gives more attention to the truly problematic locations, improving overall power supply security while avoiding overly conservative constraints on near-end nodes.

[0068] (3) Generator car tracking margin Penalties for driving generator car command tracking A squared penalty is applied when the tracking margin is less than zero.

[0069] In this embodiment, the generator command tracking penalty... Represented as: (4) Synchronous cutback deviation Penalty for deviation in driving power-back switching terminal The greater the deviation, the greater the penalty.

[0070] In this embodiment, the driving power-on switching back terminal deviation penalty term , is represented as: ; in, The weighting coefficient for the power-on switchback terminal deviation penalty term; the synchronization switchback deviation , is represented as: ; Where T is the predetermined switchback time; These represent the voltage, frequency, and phase angle on the island side at time T, respectively. These are the voltage, frequency, and phase angle at the synchronization point on the grid side, respectively. The system's nominal voltage. The system's rated frequency, To determine the net power exchange between the island and the grid at the time of switchback, This represents the predicted total load within the island at the time of the switchback. The normalized weighting coefficients for each deviation satisfy the following conditions: ; When the state machine is in the switchback pre-adjustment stage during the maintenance phase, Write the objective function and improve The weighting allows the islanded state to approach the synchronous parallel conditions of the grid side in advance.

[0071] (5) Penalty item for deviation of economic load ratio Represented as: ; Only when the load rate ρ i,t Below or higher A squared penalty is applied only when necessary to encourage generators to operate in the high-efficiency zone. Among these, , where is the weighting coefficient for the economic load rate deviation penalty, and is a preset positive real number used to adjust the impact of the generator car operating outside the economic range on the overall objective function. This represents the lower limit of the economic load rate range for generator cars, typically between 0.3 and 0.4. When the load rate of a generator car falls below this value, fuel efficiency decreases significantly. This represents the upper limit of the economic load factor range for generator cars, typically between 0.7 and 0.85. When the load factor of a generator car exceeds this value, it approaches its rated output. Although fuel efficiency remains acceptable, reserve capacity is insufficient, making it unsuitable for handling sudden load increases. ρ i,t No. i Taiwan generator car t The load rate at any given time is a dimensionless parameter. This is the squared penalty function.

[0072] (6) Penalties for curtailment of photovoltaic power Represented as: ; in, This is a weighting coefficient for the photovoltaic curtailment penalty, a preset positive real number used to balance the trade-off between limiting photovoltaic power generation and increasing the output of mobile generators. For the first j A photovoltaic power station at time t Available active power, that is, the maximum active power that a photovoltaic inverter can generate under the current illumination conditions without considering power curtailment. For the first j A photovoltaic power station at time t The actual utilization of active power is used as a decision variable for optimization.

[0073] (7) Reverse power risk penalty Represented as: ;in, This is the base weighting coefficient for the reverse power risk penalty, a preset positive real number used to adjust the priority of preventing reverse power as a safety objective in the entire optimization process. This is the sensitivity coefficient for inverse power penalty.

[0074] (8) Parallel load sharing penalty , represented as ;in, The weighting coefficient for the load sharing penalty in parallel operation is a preset positive real number used to adjust the priority of the load balancing objective among the generators in the overall optimization process. For a moment t The parallel load sharing deviation is the difference between the maximum and minimum load rates among all online generator cars, calculated as follows: In all generator car pairs In the process, identify the two generator cars with the largest difference in load rate and take the absolute value of this difference. The larger this value, the more unbalanced the load among the generator cars. and For the first i Taiwan and the j Taiwan generator car t Load rate at any given time.

[0075] Then, the constraints include: power balance constraints, node voltage constraints, branch current constraints, reverse power constraints, upper and lower limits of generator output constraints, generator ramp climbing rate constraints, and photovoltaic absorption constraints.

[0076] In this embodiment, the node voltage constraint and branch current constraint are constructed based on the sensitivity of the access point adaptive security parameters, including: Sensitivity of generator active power output to node voltage Sensitivity of generator reactive power output to node voltage Construct a linearized estimate, expressed as: ; in, For nodes k The reference voltage, For generator car g Planned active power adjustment relative to the baseline value For generator car g Planned reactive power adjustment relative to the baseline value; The node voltage constraints are determined based on the linearized estimates, and are expressed as follows: ;in, and These are the minimum and maximum values ​​of the node voltage, respectively; Sensitivity of generator car active power output to branch current The linearized estimate of the branch current is determined as follows: ;in, For the first e The reference current of the branch; The branch current constraint is determined based on the linearized estimate of the branch current, and is expressed as follows: ;in, For branch current safety limits.

[0077] In this embodiment, the rolling optimization solves a constrained multi-objective optimization problem. The standard mathematical forms corresponding to the power balance constraint, node voltage constraint, branch current constraint, inverse power constraint, generator output upper and lower limit constraint, generator ramping rate constraint, and photovoltaic absorption constraint are as follows: ; The above constraints result in a complete feasible domain boundary when solving for the active power output and photovoltaic utilization of each generator car within the current rolling window.

[0078] In this embodiment, the decision variables of the rolling optimization model include the active power output Pi,t of each generator and the actual utilization power of each photovoltaic power station. Among these constraints, the output of the generator truck is directly limited by power balance constraints and upper and lower limits of generator truck output; the upper limit of photovoltaic utilization is the available photovoltaic power. The lower limit is zero; the photovoltaic power generation limit is... The optimization is derived from the decision results of photovoltaic utilization. The optimizer determines the optimal values ​​of the above decision variables by minimizing the objective function J and satisfying all constraints, thereby achieving joint optimization of generator output, photovoltaic utilization, and photovoltaic curtailment.

[0079] S5 sends the photovoltaic utilization amount to the photovoltaic inverter in the form of a power setpoint and executes the corresponding power limiting amount; if the photovoltaic command cannot be executed, the value of the previous cycle is maintained and the deviation information is fed back to the optimization parameter library; for the generator command, if the difference between the current actual output and the command exceeds the climbing ability, it is decomposed into a ramp sub-command that meets the climbing rate constraint and sent out step by step, and the photovoltaic command and the generator command are sent out independently; at the end of each execution cycle, according to the prediction deviation and the execution deviation, it enters one of the three paths of normal rolling correction, prediction deviation correction or emergency optimization to achieve closed-loop correction.

[0080] In this embodiment, the photovoltaic utilization obtained from the optimization solution is sent to the photovoltaic inverter in the form of a power setpoint. Under normal circumstances, the photovoltaic inverter directly tracks to the target set value within one execution cycle, without needing to execute in steps; If a photovoltaic command cannot be executed due to communication anomalies or equipment failure, the photovoltaic utilization rate in the current execution cycle will remain at the value of the previous cycle, and this deviation information will be fed back into the optimization parameter library for rolling correction in the next round. For generator commands, when the difference between the command to be issued and the generator's current actual output exceeds the product of its maximum ramp rate and the execution cycle, the command will be decomposed into ramp sub-commands that meet the ramp rate constraint and issued step by step to ensure that the generator output smoothly transitions to the target value.

[0081] The photovoltaic (PV) commands and generator vehicle commands are issued independently to the PV inverter and generator vehicle controller, respectively. These commands are decoupled at the issuing level and do not have a direct correlation. In this embodiment, the power limiting is implemented through a power setpoint, preventing the PV inverter from generating excessive power and avoiding waste.

[0082] Therefore, in a preferred embodiment, the scroll optimization solution obtains a solution that covers the entire scroll window. HThe system generates a scheduling plan sequence, but only extracts the instruction value for the current execution cycle for distribution. The distribution channel consists of two independent paths: Photovoltaic command channel: The photovoltaic utilization obtained from the solution is sent to the photovoltaic inverter in the form of power setpoint; Generator car command channel: The obtained generator car output is sent to the generator car controller.

[0083] Under normal circumstances, photovoltaic inverters have excellent dynamic response capabilities, far exceeding the hill-climbing response speed of generator vehicles. Therefore, photovoltaic commands do not need to be executed in steps; the inverter directly tracks the current actual power to the target setpoint within one execution cycle and executes the corresponding power limiting. It will also display a success message.

[0084] When a photovoltaic command cannot be executed due to communication errors, inverter failure, or execution feedback timeout, the system performs the following actions: The system maintains the photovoltaic (PV) utilization rate in the current execution cycle at the actual value of the previous cycle. Simultaneously, it calculates the PV command execution deviation and feeds this deviation information back into the optimization parameter library for rolling correction in the next round. In the next optimization cycle, the system adjusts the actual calculated value of available PV power based on the feedback information, corrects the prediction model, or triggers emergency optimization.

[0085] The generator command must meet the ramp rate constraint and cannot be directly abruptly changed. The system determines executability by tracking margin. .

[0086] When the instruction is to be issued Compared with the current actual output of the generator car The difference between them does not exceed Direct issuance is permitted, wherein the current instruction to be issued is the active power output of the generator in the previous cycle. For generator car i The maximum ramp rate of active power represents the maximum increase in output power per unit time. For generator car i The maximum time allowed to complete one instruction tracking, multiplied by the two, represents the maximum power regulation that the generator can complete within the maximum allowed time; When the difference exceeds the climbing capacity range, the system decomposes the target instruction into... n A slope sub-instruction that satisfies the climbing rate constraint is decomposed into... A slope sub-instruction that satisfies the climbing rate constraint, wherein, To optimize the execution cycle, i.e. the time interval between two consecutive instructions, For the floor function, when hour, No sub-instructions need to be issued; After decomposition, the first k The sub-instructions are: ; in, Or, the step size between two adjacent sub-instructions is: ; Ensure that the slope of each sub-instruction satisfies: That is, the generator car only needs to be adjusted within each execution cycle. The power change is limited to the level of its climbing ability, ensuring that the generator output smoothly transitions to the target value.

[0087] In each execution cycle t At the end, the prediction bias was collected. and execution deviation These two types of deviations will serve as the basis for correction in the next execution cycle, and will be categorized and entered into different processing paths. (1) If the prediction is biased and execution deviation If none of the prediction deviations exceed the corresponding thresholds, normal rolling correction begins; The corresponding threshold is expressed as: ;in, for The standard deviation of the net load forecast error at any given time. for Real-time measured total load; In this step, the normal rolling correction process includes the following steps: The deviation between the measured value and the predicted value in this period is included in the historical error statistics window, and the standard deviation of the prediction error and the prediction confidence are updated. The standard deviation of the prediction error is expressed as: Where M is the number of historical sampling points used for statistical error; For the first t-m The measured net load at each sampling time. For the first t-m Predicted net load at each sampling time; Update the margin parameters in the state vector based on the latest measured data.

[0088] (2) If the prediction is biased Exceeding the prediction bias threshold, but execution bias If the value does not exceed 3%, the current measured value is used as the latest sample to update the prediction correction coefficient, so that the next round of photovoltaic output and load prediction values ​​are corrected towards the measured values; at the same time, the weight of the prediction credibility reserve penalty term is increased, and more generator cars are reserved for backup in the next round of optimization to compensate for the safety risks caused by prediction uncertainty; finally, the corrected prediction model and the updated state vector are used to re-enter the next round of optimization. The correction methods include: The current measured values ​​are used as the latest sample to update the prediction correction coefficients, so that the next round of photovoltaic power output and load predictions are corrected towards the measured values, including: The photovoltaic output and load forecasts are updated using the measured values ​​of the photovoltaic output and load from the previous execution cycle, expressed as: ;in, For the current optimization time, For the previous execution cycle, These are the raw forecast values ​​for photovoltaic output and load, respectively, from day-ahead or hourly forecast models, or ultra-short-term forecasts provided by photovoltaic inverter / new energy management systems; and These are the measured values ​​of photovoltaic output and load from the previous execution cycle, respectively. and These are the predicted values ​​for photovoltaic power output and load for the previous execution cycle, respectively. and These are the photovoltaic output and load forecast values ​​used for the current optimization time after recursive correction; The recursive correction coefficient, ranging from [0,1], represents the degree of confidence in the measured deviation. When α is close to 1, the measured deviation is fully used to correct the predicted value; when α is close to 0, the predicted value is hardly corrected. The correction coefficient can be a fixed empirical value, such as 0.7–0.9, or it can be adaptively adjusted based on the statistical characteristics of historical prediction errors. That is, when the recent prediction error is large, α is increased to increase the correction weight of the measured deviation; when the prediction error is small, α is decreased to maintain prediction smoothness.

[0089] Simultaneously increase the weight of the backup penalty term for prediction reliability. ,make In the next round of optimization, more generator cars will be reserved for standby operation. Finally, the next round of optimization will be re-entered with the revised prediction model and updated state vector. If the prediction deviation exceeds the threshold for several consecutive cycles but execution remains normal, the reserve capacity ratio will be further increased based on the above corrections, such as... The deviation increases periodically until it decreases.

[0090] (3) When the generator car performs a deviation When the percentage exceeds 3% and persists for two sampling periods, regardless of the prediction bias... If the threshold is exceeded, emergency optimization will be triggered immediately. Specifically: First, freeze all currently executed switch states and generator car start / stop states to prevent the risk from escalating due to the execution of new switch operations or start / stop actions under abnormal conditions. Then, the following measures are executed in order of priority: adjust the output command of the online generator with the largest deviation to near the current actual value, start the standby generator and connect it to the grid to increase power generation to make up for the power shortage, implement photovoltaic power limiting to eliminate reverse power risk, and disconnect interruptible loads as a last resort.

[0091] After each level of measures is implemented, the implementation deviation is reassessed to see if it has decreased. Once the implementation deviation has decreased to within 3%, the freeze is lifted until the normal rolling correction process is resumed.

[0092] Prediction bias in this embodiment Represented as: ;in, for Real-time measured net load for Forecast net load in real time. for Real-time measured total load To prevent small positive numbers with a denominator of zero; Execution deviation Represented as: ;in, for Time-based generator car i Actual output For generator car i Rated power, for The information is constantly being sent to the generator car. i The target instruction, that is It is the result of the previous cycle. By comparing actual output With target instructions The difference determines whether emergency optimization or parameter adjustment for the next round of optimization is needed.

[0093] The effectiveness of this invention was verified using a planned maintenance of a switching station as an example. The maintenance power supply island includes one medium-voltage generator truck access point, three main load nodes, and two distributed photovoltaic grid-connected points. The peak load is approximately 820kW, and the maximum available photovoltaic output is approximately 360kW. The execution cycle is 5 minutes, and the rolling window is adaptively set to 45 to 60 minutes. Comparison strategies include: A) a fixed output strategy, setting the generator truck output based on the historical maximum net load; B) a conventional rolling optimization strategy, focusing only on fuel cost and power balance, without considering access point sensitivity, tracking completion time, and switchback terminal constraints; and C) the strategy of this invention. The verification results are shown in the table below:

[0094] The above results show that, compared with the fixed output strategy, the present invention reduces fuel consumption by approximately 26.7%, increases photovoltaic absorption rate to 98.1%, reduces reverse power risk to zero, achieves a voltage qualification rate of 100%, controls command tracking deviation within 2.6%, and reduces switchback impact power to approximately 18kW. This verifies the comprehensive technical effectiveness of the present invention in dynamically tracking net load changes, improving photovoltaic absorption, eliminating reverse power risk, ensuring voltage safety, ensuring command execution, and reducing power restoration impact.

[0095] Secondly, the present invention also provides a dynamic balance optimization system for uninterrupted maintenance of medium-voltage distribution networks, comprising: The maintenance work order parsing module is used to read the uninterrupted maintenance work order of the medium-voltage generator truck and determine the connection point of the medium-voltage generator truck and the status vector of the maintenance disconnect switch. The access point adaptive power supply island modeling module is used to construct a temporary power supply island model based on the access point and the maintenance disconnect switch state vector, and to calculate the access point adaptive safety parameters based on the temporary power supply island model. The maintenance phase state machine module is used to establish and maintain the maintenance phase state machine based on the actual progress status of the maintenance operation, including the maintenance preparation phase, the islanded power supply phase, the switchback pre-adjustment phase, and the power restoration parallel phase. The rolling optimization module is used to activate the objective function and constraints of different stages according to the maintenance stage state machine, dynamically adjust the penalty weights of each item in the objective function according to the maintenance dynamic state vector, and use the short-circuit capacity weight in the access point adaptive safety parameters to differentiate and amplify the voltage over-limit penalty item, thereby solving the active power output and photovoltaic utilization of each generator car within the current rolling window. The instruction execution and feedback module is used to send the photovoltaic utilization amount to the photovoltaic inverter in the form of a power setpoint and execute the corresponding power limiting amount; if the photovoltaic instruction cannot be executed, the value of the previous cycle is maintained and the deviation information is fed back to the optimization parameter library; for the generator vehicle instruction, if the difference between it and the current actual output exceeds the climbing ability, it is decomposed into a ramp sub-instruction that meets the climbing rate constraint and issued step by step, and the photovoltaic instruction and the generator vehicle instruction are issued independently; at the end of each execution cycle, it enters one of the three paths of normal rolling correction, prediction deviation correction or emergency optimization according to the prediction deviation and the execution deviation, to achieve closed-loop correction.

[0096] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. "A plurality of" means two or more, unless otherwise explicitly specified.

[0097] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0098] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0099] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0100] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0101] Therefore, this embodiment discloses a computer device.

[0102] The computer device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the above-described embodiments of the dynamic balancing optimization method for uninterrupted maintenance of medium-voltage distribution networks. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described system embodiments.

[0103] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks.

[0104] The computer device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device may include, but is not limited to, a processor and memory.

[0105] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0106] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0107] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0108] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0109] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A dynamic balance optimization method for uninterrupted power supply maintenance of medium-voltage distribution networks, characterized in that, include: Read the uninterrupted maintenance work order of the medium-voltage generator truck and determine the connection point of the medium-voltage generator truck and the status vector of the disconnecting switch during maintenance. Based on the access point and the maintenance disconnect switch state vector, a temporary power supply island model is constructed, and adaptive safety parameters of the access point are calculated based on the temporary power supply island model. Establish a maintenance phase state machine based on the actual progress status of maintenance operations, including the maintenance preparation phase, the islanded power supply phase, the switchback pre-adjustment phase, and the power restoration parallel phase. According to the state machine of the maintenance stage, the objective function and constraints of different stages are activated. The penalty weights of each item in the objective function are dynamically adjusted according to the maintenance dynamic state vector. The short-circuit capacity weight in the adaptive safety parameters of the access point is used to differentiate and amplify the voltage over-limit penalty item. Then, the active power output and photovoltaic utilization of each generator car are solved in the current rolling window. The photovoltaic utilization is sent to the photovoltaic inverter in the form of a power setpoint, and the corresponding power limiting is executed. If the photovoltaic command cannot be executed, the value of the previous cycle is maintained and the deviation information is saved. For the generator command, if the difference between the current actual output and the generator command exceeds the ramping capability, it is decomposed into a ramp sub-command that meets the ramping rate constraint and sent out step by step. The photovoltaic command and the generator command are sent out independently. At the end of each execution cycle, the generator enters one of the three paths of normal rolling correction, prediction deviation correction or emergency optimization according to the prediction deviation and the execution deviation, so as to achieve closed-loop correction.

2. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 1, characterized in that, The objective functions for the different stages are expressed as follows: in, It is the time step number within the scrolling window H. This is expressed as the sum of costs for all time periods within the window. For the first The probability of each scenario occurring. Numbering the risk scenarios. For the first t Fuel costs at any time For the first t The penalty for the generator car deviating from the economic load rate range at any time. This is a penalty for curtailment of photovoltaic power. This is a penalty item for reverse power risk. This is a penalty term for voltage and branch current margins. For generator command tracking penalty items, For parallel load sharing penalty items, To predict the reserve margin penalty term caused by reduced confidence, Penalty for power-on switching back to the terminal; and All of these correspond to the islanded power supply phase. The corresponding pre-adjustment phase; the constraints include: power balance constraints, node voltage constraints, branch current constraints, reverse power constraints, upper and lower limits of generator output constraints, generator ramp climbing rate constraints, and photovoltaic absorption constraints.

3. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 2, characterized in that, The node voltage constraints and branch current constraints are constructed based on the sensitivity of the access point adaptive security parameters, including: Sensitivity of generator active power output to node voltage Sensitivity of generator reactive power output to node voltage Construct a linearized estimate, expressed as: ; in, For nodes k The reference voltage, For generator car g Planned active power adjustment relative to the baseline value For generator car g Planned reactive power adjustment relative to the baseline value; The node voltage constraints are determined based on the linearized estimates, and are expressed as follows: ; in, and These are the minimum and maximum values ​​of the node voltage, respectively; Sensitivity of generator car active power output to branch current The linearized estimate of the branch current is determined as follows: ; in, For the first e The reference current of the branch; The branch current constraint is determined based on the linearized estimate of the branch current, and is expressed as follows: ;in, For branch current safety limits.

4. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 3, characterized in that, The step of dynamically adjusting the penalty weights in the objective function based on the maintenance dynamic state vector includes: Construct a dynamic state vector for maintenance, including: the real-time net load at the current optimization moment. The standard deviation of the prediction error at the current optimization time. Confidence of prediction at the current optimization time The node at the current optimization moment k Voltage margin The current margin of branch e at the current optimization moment. Inverse power margin of the access point at the current optimization moment The current optimization time is the th i Load rate of generator cars The current optimization time is the th i Tracking margin of a generator car Parallel load sharing deviation at the current optimization moment Combined synchronization deviation with the future cut-back time T ; The prediction confidence Alternative penalty item used to drive prediction confidence The lower the prediction confidence, the greater the spare penalty, prompting the optimizer to reserve more spin-off spare; node voltage margin. Penalty terms for drive voltage and branch current margin In the voltage component, a penalty is applied when the voltage margin falls below a preset safety threshold. The penalty coefficient for weak nodes is amplified by the short-circuit capacity weight. (Branch current margin) Penalty terms for drive voltage and branch current margin The current component in the circuit applies a penalty when the current margin falls below a preset safety threshold; generator tracking margin. Penalties for driving generator car command tracking A squared penalty is applied when the tracking margin is less than zero; synchronization cut-back bias. Penalty for deviation in driving power-back switching terminal The greater the deviation, the greater the penalty.

5. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 4, characterized in that, The penalty coefficient for the weak node is amplified by the short-circuit capacity weight, including: The voltage and branch current margin penalty terms are represented by short-circuit capacity weighting, as follows: in, and These are the basic weighting coefficients for voltage margin and current margin penalties, respectively. The preset safety threshold for voltage margin, This is a preset safety threshold for current margin; a penalty is triggered when the actual margin falls below this threshold. The short-circuit capacity weighting ratio is expressed as: ;in, This is the minimum short-circuit capacity weight value within the power supply island. This is the short-circuit capacity weight value, when Characterize weak nodes to obtain a higher voltage over-limit penalty coefficient.

6. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 4, characterized in that, The drive power-on switchback terminal deviation penalty item , is represented as: ; in, The weighting coefficient for the power-on switchback terminal deviation penalty term; the synchronization switchback deviation , is represented as: ; Where T is the predetermined switchback time; These represent the voltage, frequency, and phase angle on the island side at time T, respectively. These are the voltage, frequency, and phase angle at the synchronization point on the grid side, respectively. The system's nominal voltage. The system's rated frequency, To determine the net power exchange between the island and the grid at the time of switchback, This represents the predicted total load within the island at the time of the switchback. The normalized weighting coefficients for each deviation satisfy the following conditions: ; When the state machine is in the switchback pre-adjustment stage during the maintenance phase, Write the objective function and improve The weighting allows the islanded state to approach the synchronous parallel conditions of the grid side in advance.

7. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 6, characterized in that, If the difference between the command to the generator car and the current actual output exceeds the climbing capacity, it is decomposed into a step-by-step issuance of slope sub-commands that satisfy the climbing rate constraint, including: When the instruction is to be issued Compared with the current actual output of the generator car The difference between them does not exceed Direct issuance is permitted, wherein the current instruction to be issued is the active power output of the generator in the previous cycle. Generator car i The maximum ramp rate of active power represents the maximum increase in output power per unit time. For generator car i The maximum time allowed to complete one instruction tracking, multiplied by the two, represents the maximum power regulation that the generator can complete within the maximum allowed time; When the instruction is to be issued Compared with the current actual output of the generator car The difference between them exceeds At that time, the instruction is broken down into A slope sub-instruction that satisfies the climbing rate constraint, wherein, To optimize the execution cycle, i.e. the time interval between two consecutive instructions, For the floor function, when hour, No sub-instructions need to be issued; After decomposition, the first k The sub-instructions are: .

8. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 7, characterized in that, The process involves entering one of three paths—normal rolling correction, prediction deviation correction, or emergency optimization—based on both prediction and execution deviations to achieve closed-loop correction, including: In each execution cycle t At the end, the prediction bias was collected. and execution deviation These two types of deviations will serve as the basis for correction in the next execution cycle, and will be categorized and entered into different processing paths. If the prediction is biased and execution deviation If none of the prediction deviations exceed the corresponding thresholds, normal rolling correction begins; The corresponding prediction bias threshold is expressed as: ;in, for t The standard deviation of the net load forecast error at any given time. for t Real-time measured total load; If the prediction is biased Exceeding the prediction bias threshold, but execution bias If the prediction deviation threshold of 3% is not exceeded, the current measured value is used as the latest sample to update the prediction correction coefficient, so that the next round of photovoltaic output and load prediction values ​​are corrected in the direction of the measured values; at the same time, the weight of the prediction credibility reserve penalty term is increased, and more generator cars are reserved for backup in the next round of optimization to compensate for the safety risks caused by prediction uncertainty; finally, the corrected prediction model and the updated state vector are used to re-enter the next round of optimization. When the generator car performs a deviation When the prediction deviation exceeds the 3% threshold and persists for two consecutive sampling periods, regardless of the prediction deviation... If the threshold is exceeded, an emergency optimization will be triggered immediately until the normal rolling correction process is resumed.

9. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 8, characterized in that, The prediction bias Represented as: ; in, for t Real-time measured net load for t Forecast net load in real time. for t Real-time measured total load To prevent small positive numbers with a denominator of zero; Execution deviation Represented as: ; in, for t Time-based generator car i Actual output For generator car i Rated power, for t The information is constantly being sent to the generator car. i The target instruction.

10. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 9, characterized in that, The current measured values ​​are used as the latest sample to update the prediction correction coefficients, so that the next round of photovoltaic output and load predictions are corrected towards the measured values, including: The photovoltaic output and load forecasts are updated using the measured values ​​of the photovoltaic output and load from the previous execution cycle, expressed as: ; in, For the current optimization time, For the previous execution cycle, These are the raw forecast values ​​for photovoltaic output and load, respectively, from day-ahead or hourly forecast models, or ultra-short-term forecasts provided by photovoltaic inverter / new energy management systems; and These are the measured values ​​of photovoltaic output and load from the previous execution cycle, respectively. and These are the predicted values ​​for photovoltaic power output and load for the previous execution cycle, respectively. and These are the photovoltaic output and load forecast values ​​used for the current optimization time after recursive correction; These are the recursive correction coefficients.

11. The dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks according to claim 9, characterized in that, The normal rolling correction process includes the following steps: The deviation between the measured value and the predicted value in this period is included in the historical error statistics window, and the standard deviation of the prediction error and the prediction confidence are updated. The standard deviation of the prediction error is expressed as: ; Where M is the number of historical sampling points used for statistical error; For the first tm The measured net load at each sampling time. For the first tm Predicted net load at each sampling time; The prediction confidence level is expressed as: ; in, This is a power reference value; The margin parameters in the state vector are updated based on the latest measured data; the updated state vector is used as the input for the next round of optimization to resolve the generator output command.

12. A dynamic balance optimization system for uninterrupted maintenance of medium-voltage distribution networks implemented according to any one of claims 1 to 11, characterized in that, include: The maintenance work order parsing module is used to read the uninterrupted maintenance work order of the medium-voltage generator truck and determine the connection point of the medium-voltage generator truck and the status vector of the maintenance disconnect switch. The access point adaptive power supply island modeling module is used to construct a temporary power supply island model based on the access point and the maintenance disconnect switch state vector, and to calculate the access point adaptive safety parameters based on the temporary power supply island model. The maintenance phase state machine module is used to establish and maintain the maintenance phase state machine based on the actual progress status of the maintenance operation, including the maintenance preparation phase, the islanded power supply phase, the switchback pre-adjustment phase, and the power restoration parallel phase. The rolling optimization module is used to activate the objective function and constraints of different stages according to the maintenance stage state machine, dynamically adjust the penalty weights of each item in the objective function according to the maintenance dynamic state vector, and use the short-circuit capacity weight in the access point adaptive safety parameters to differentiate and amplify the voltage over-limit penalty item, thereby solving the active power output and photovoltaic utilization of each generator car within the current rolling window. The instruction execution and feedback module is used to send the photovoltaic utilization amount to the photovoltaic inverter in the form of a power setpoint and execute the corresponding power limiting amount; if the photovoltaic instruction cannot be executed, the value of the previous cycle is maintained and the deviation information is fed back to the optimization parameter library; for the generator vehicle instruction, if the difference between it and the current actual output exceeds the climbing ability, it is decomposed into a ramp sub-instruction that meets the climbing rate constraint and issued step by step, and the photovoltaic instruction and the generator vehicle instruction are issued independently; at the end of each execution cycle, it enters one of the three paths of normal rolling correction, prediction deviation correction or emergency optimization according to the prediction deviation and the execution deviation, to achieve closed-loop correction.

13. A computer device, characterized in that, include: One or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, the steps of the dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks as described in any one of claims 1 to 11 are implemented.

14. A computer-readable storage medium, characterized in that, It contains a computer program, which, when executed, implements the steps of the dynamic balance optimization method for uninterrupted maintenance of medium-voltage distribution networks as described in any one of claims 1 to 11.

Images