A power distribution network adaptive dynamic repair processing method and system considering repair uncertainty

Abstract

The application discloses a kind of power distribution network adaptive dynamic repair processing method and system considering repair uncertainty, the topology of this method after collecting extreme event power distribution network and fault data, constructs four-dimensional fault set, initializes the priori confidence of repair success rate for each fault point.Construct power distribution network dynamic repair collaborative optimization model and add multidimensional constraint, obtain constraint after collaborative optimization model.Set the prediction time domain window, control step and adaptive stop loss trigger threshold of rolling optimization, with the maximum load recovery efficiency as goal, solve model by rolling time domain optimization algorithm, obtain optimal repair path and network reconfiguration action sequence.Execute first step repair instruction, update fault confidence if repair fails, calculate two types of expected value and determine whether to abandon fault point by combining threshold, continuously update power grid state and confidence and iteratively solve, until all repair tasks are completed.The application solves the problems of low load recovery efficiency in the prior art.

Description

Technical Field

[0001] This invention relates to the field of distribution network fault repair and dispatching technology, specifically to an adaptive dynamic repair processing method and system for distribution networks that takes into account repair uncertainties. Background Technology

[0002] In scenarios such as extreme natural disasters and external damage, distribution networks are prone to multi-branch and multi-node faults. Traditional emergency repair methods rely on manual experience to formulate static emergency repair plans, which are difficult to cope with complex operating conditions such as uncertain fault repair success rates, time and space constraints on emergency repair resources, and dynamic changes in network topology. With the development of smart distribution network scheduling technology, rolling optimization methods based on model predictive control have been gradually applied to emergency repair scheduling. By solving the optimization model in different time periods to generate dynamic emergency repair paths and network reconfiguration strategies, some progress has been made in improving load restoration efficiency.

[0003] However, existing technologies still have significant limitations in application: First, they do not fully consider the uncertainties in the fault repair process, usually assuming that a single repair will definitely succeed, and do not establish a confidence update mechanism for the success rate of repair as the number of retries changes, which makes the optimized scheme prone to failure due to repair failure in actual implementation; Second, they lack an adaptive loss-stopping decision mechanism for the repair process, and cannot quantitatively evaluate the difference in benefits between "continuing to retry" and "transferring to other fault points" after repair failure, which easily leads to the ineffective occupation of resources at low-probability fault points; Third, traditional rolling optimization models do not include repair uncertainty and loss-stopping decisions in the constraints, and cannot achieve coordinated optimization of repair paths, network reconstruction and loss-stopping strategies, making it difficult to maximize the overall load recovery efficiency within a limited scheduling time.

[0004] Therefore, there is an urgent need for an adaptive dynamic emergency repair method for distribution networks that takes into account the uncertainty of repair, in order to solve the problems of inefficient use of emergency repair resources and low load restoration efficiency in existing technologies. Summary of the Invention

[0005] To address this, the present invention provides an adaptive dynamic emergency repair method and system for distribution networks that considers repair uncertainties, thereby solving the problems of existing technologies that do not consider repair uncertainties and lack adaptive loss mitigation mechanisms, resulting in inefficient use of repair resources and low load restoration efficiency.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an adaptive dynamic emergency repair method for distribution networks considering repair uncertainties, characterized in that it includes: By collecting distribution network topology and fault data after extreme events, a four-element fault set including disconnection and blocking faults is constructed; based on the four-element fault set, a prior confidence level for the repair success rate is initialized for each fault point. Based on the quaternary fault set and the prior confidence level, a dynamic emergency repair collaborative optimization model for the distribution network is constructed. Linearized power flow equation constraints, radial topology hard constraints, and spatiotemporal transfer constraints of maintenance resources are added to the dynamic emergency repair collaborative optimization model for the distribution network to obtain a constrained collaborative optimization model. Based on the constrained collaborative optimization model, the prediction time-domain window, control step size, and adaptive stop-loss trigger threshold for rolling optimization are set. Based on the set time parameters, the prior confidence level, and the constrained collaborative optimization model, with the objective function of maximizing load recovery efficiency, the constrained collaborative optimization model is solved by the rolling time-domain optimization algorithm to obtain the optimal emergency repair path and network reconstruction action sequence. Based on the optimal repair path and network reconstruction action sequence, the repair instruction of the first control step is intercepted and executed; if the repair fails, the repair success rate confidence of the current fault point is updated through the Bernoulli random trial model, and the expected value of staying and the expected value of transferring are calculated; based on the expected value of staying, the expected value of transferring and the adaptive stop loss trigger threshold, the adaptive stop loss judgment model is used to determine whether to abandon the current fault point. Based on the loss prevention judgment result or after detecting the completion signal of the emergency repair task, the distribution network topology status and fault confidence are updated to obtain the updated distribution network topology status and fault confidence. The constrained collaborative optimization model is then iteratively solved until all repair tasks are completed.

[0007] As a preferred solution for an adaptive dynamic emergency repair method for distribution networks that considers the uncertainty of repair, the four-element fault set includes: a set of disconnected fault branches, a set of disconnected fault nodes, a set of blocked fault branches, and a set of blocked fault nodes. The prior confidence level is the initial probability value of a single successful repair at each fault point.

[0008] As a preferred scheme for an adaptive dynamic emergency repair method for distribution networks that considers repair uncertainties, the objective function expression of the constrained collaborative optimization model is:

[0009] In the formula, For system load recovery efficiency; This is the total recovery time; For the set of load nodes in the power distribution system; t For the current moment; j The node number; For nodes j At any moment t The actual active load; For nodes j Load priority weights; This represents the total load level of the power distribution system under normal operating conditions.

[0010] As a preferred scheme for an adaptive dynamic emergency repair method for distribution networks that considers repair uncertainty, in the process of updating the repair success rate confidence of the current fault point through a Bernoulli random trial model, each repair attempt is regarded as a Bernoulli trial, and the result of the trial is success or failure; if the trial fails, the repair success rate confidence of the fault point is updated through an exponential decay model; the expression of the exponential decay model is:

[0011] In the formula, For the first Expected posterior probability of successful fault repair after a failed repair attempt; This serves as the baseline repair success rate for common faults. This is the preset confidence decay factor.

[0012] As a preferred solution for an adaptive dynamic emergency repair method for distribution networks that considers repair uncertainties, the formula for calculating the expected value of remaining on standby is as follows:

[0013] In the formula, To achieve a finite step size Internal continuous retry fault The overall expected return; For the overall scheduling time domain; Fault The additional load recovery that can be achieved after being cleared; n For the first n Try again; For nodes j Load priority weights, For nodes j The re-energized state variables, For nodes j The active power load demand; N A set of nodes; The formula for calculating the expected value of the transfer is as follows:

[0014] In the formula, To abandon the current fault and transfer to an alternative fault point Expected returns; The incremental value of the topology after the alternative fault points are repaired; The exact travel time for the team to move from its current location to the alternative point; The base time for performing a single repair.

[0015] As a preferred embodiment of an adaptive dynamic emergency repair method for distribution networks that considers repair uncertainties, the expression for the adaptive loss-stopping determination model is:

[0016] In the formula, k This represents the number of repair attempts made for the current fault point. The minimum number of attempts is the defense line; The remaining available scheduling time, As the threshold for urgent transfer time, To prevent frequent returns and stay-at-home preference penalty coefficients.

[0017] This invention also provides an adaptive dynamic emergency repair system for distribution networks that considers repair uncertainties, employing the above-mentioned adaptive dynamic emergency repair system for distribution networks that considers repair uncertainties, including: The data acquisition and classification module is used to construct a four-element fault set containing disconnection and blocking faults by collecting distribution network topology and fault data after extreme events; based on the four-element fault set, the prior confidence of the repair success rate is initialized for each fault point. The model and constraint construction module is used to construct a dynamic emergency repair collaborative optimization model for the distribution network based on the four-element fault set and the prior confidence level; and to add linearized power flow equation constraints, radial topology hard constraints, and maintenance resource spatiotemporal transfer constraints to the dynamic emergency repair collaborative optimization model for the distribution network to obtain a constrained collaborative optimization model. The time parameter setting module is used to set the prediction time domain window, control step size and adaptive stop loss trigger threshold for rolling optimization based on the constrained collaborative optimization model. The rolling optimization module is used to solve the constrained collaborative optimization model based on the set time parameters, the prior confidence level and the constrained collaborative optimization model, with the objective function of maximizing load recovery efficiency, and to obtain the optimal emergency repair path and network reconstruction action sequence through the rolling time domain optimization algorithm. The emergency repair execution and loss control module is used to intercept and execute the emergency repair instruction of the first control step based on the optimal emergency repair path and network reconstruction action sequence. If the emergency repair fails, the module updates the repair success rate confidence of the current fault point through a Bernoulli random trial model and calculates the expected value of staying and the expected value of transferring. Based on the expected value of staying, the expected value of transferring, and the adaptive loss control trigger threshold, the module determines whether to abandon the current fault point through an adaptive loss control model. The event-driven closed-loop module is used to update the distribution network topology status and fault confidence based on the loss stop judgment result or after detecting the completion signal of the emergency repair task, obtain the updated distribution network topology status and fault confidence, and iteratively solve the constrained collaborative optimization model until all repair tasks are completed.

[0018] As a preferred solution for an adaptive dynamic emergency repair system for distribution networks that considers the uncertainty of repair, the four-element fault set in the data acquisition and classification module includes: a set of disconnection-type fault branches, a set of disconnection-type fault nodes, a set of blocking-type fault branches, and a set of blocking-type fault nodes. The prior confidence level is the initial probability value of a single successful repair at each fault point.

[0019] As a preferred solution for an adaptive dynamic emergency repair system for distribution networks that considers repair uncertainties, the objective function expression of the constrained collaborative optimization model in the model and constraint construction module is as follows:

[0020] In the formula, For system load recovery efficiency; This is the total recovery time; For the set of load nodes in the power distribution system; t For the current moment; j The node number; For nodes j At any moment t The actual active load; For nodes j Load priority weights; This represents the total load level of the power distribution system under normal operating conditions.

[0021] As a preferred solution for an adaptive dynamic emergency repair system for distribution networks that considers repair uncertainty, in the emergency repair execution and loss mitigation determination module, during the process of updating the repair success rate confidence of the current fault point using a Bernoulli random trial model, each repair attempt is considered a Bernoulli trial, with the result being either success or failure. If the trial fails, the repair success rate confidence of the fault point is updated using an exponential decay model; the expression for the exponential decay model is:

[0022] In the formula, For the first Expected posterior probability of successful fault repair after a failed repair attempt; This serves as the baseline repair success rate for common faults. This is the preset confidence decay factor.

[0023] As a preferred solution for an adaptive dynamic emergency repair system for distribution networks that considers the uncertainty of repair, the formula for calculating the expected value of remaining on standby in the emergency repair execution and loss mitigation determination module is as follows:

[0024] In the formula, To achieve a finite step size Internal continuous retry fault The overall expected return; For the overall scheduling time domain; For fault The additional load recovery that can be achieved after being cleared; n For the first n Try again; For nodes j Load priority weights, For nodes j The re-energized state variables, For nodes j The active power load demand; N A set of nodes; The formula for calculating the expected value of the transfer is as follows:

[0025] In the formula, To abandon the current fault and transfer to an alternative fault point Expected returns; The incremental value of the topology after the alternative fault points are repaired; The exact travel time for the team to move from its current location to the alternative point; The base time for performing a single repair.

[0026] As a preferred solution for an adaptive dynamic emergency repair system for distribution networks that considers the uncertainty of repair, the expression of the adaptive loss-stopping determination model in the emergency repair execution and loss-stopping determination module is as follows:

[0027] In the formula, k This represents the number of repair attempts made for the current fault point. The minimum number of attempts is the defense line; The remaining available scheduling time, As the threshold for urgent transfer time, To prevent frequent returns and stay-at-home preference penalty coefficients.

[0028] The present invention has the following advantages: First, a dynamic update mechanism for the confidence level of repair success rate is introduced. Based on Bernoulli random trials and exponential decay models, the uncertainty of fault repair is characterized. It no longer assumes that a single repair will necessarily be successful, making the emergency repair plan more in line with the actual execution scenario and improving the robustness and feasibility of the plan.

[0029] Second, construct a value assessment model for staying on-site versus transferring, and combine it with an adaptive stop-loss judgment criterion to quantitatively compare the difference in benefits between continuing to retry and transferring for emergency repair, so as to avoid the ineffective occupation of emergency repair resources at low-probability fault points and optimize resource allocation efficiency.

[0030] Third, by deeply integrating loss-stopping decisions with rolling time-domain optimization, we can achieve coordinated optimization of repair paths, network reconstruction, and loss-stopping strategies, ensuring that every action is driven by maximizing load restoration efficiency and improving the overall power restoration level within a limited scheduling time.

[0031] Fourth, an event-driven closed-loop iterative update mechanism is adopted. Based on the feedback from emergency repairs, the distribution network topology status and fault confidence are updated in real time, and optimization and re-solution are triggered. This can adapt to dynamic changes in operating conditions and ensure that the solution is always adapted to the current reality. Attached Figure Description

[0032] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0033] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0034] Figure 1 This is a flowchart illustrating an adaptive dynamic emergency repair method for distribution networks that considers repair uncertainties, as provided in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram illustrating the specific implementation process of an adaptive dynamic emergency repair method for distribution networks that considers repair uncertainties, as provided in Embodiment 1 of the present invention. Figure 3 This is a schematic diagram of the architecture of a power distribution network adaptive dynamic emergency repair system that takes into account repair uncertainties, as provided in Embodiment 2 of the present invention. Detailed Implementation

[0035] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0036] Example 1

[0037] See Figure 1 and Figure 2 Embodiment 1 of the present invention provides an adaptive dynamic emergency repair method for distribution networks that considers repair uncertainties, comprising the following steps: S1. By collecting distribution network topology and fault data after extreme events, construct a four-element fault set including disconnection and blocking faults; based on the four-element fault set, initialize the prior confidence of the repair success rate for each fault point; S2. Based on the four-element fault set and the prior confidence level, construct a dynamic emergency repair collaborative optimization model for the distribution network; add linearized power flow equation constraints, radial topology hard constraints, and maintenance resource spatiotemporal transfer constraints to the dynamic emergency repair collaborative optimization model for the distribution network to obtain a constrained collaborative optimization model. S3. Based on the constrained collaborative optimization model, set the prediction time-domain window, control step size, and adaptive stop-loss trigger threshold for rolling optimization; S4. Based on the set time parameters, the prior confidence level and the constrained collaborative optimization model, with the load recovery efficiency as the objective function, the constrained collaborative optimization model is solved by the rolling time-domain optimization algorithm to obtain the optimal emergency repair path and network reconstruction action sequence. S5. Based on the optimal repair path and network reconstruction action sequence, intercept and execute the repair instruction of the first control step; if the repair fails, update the repair success rate confidence of the current fault point through the Bernoulli random trial model, and calculate the expected value of staying and the expected value of transferring; based on the expected value of staying, the expected value of transferring and the adaptive stop loss trigger threshold, determine whether to abandon the current fault point through the adaptive stop loss judgment model. S6. Based on the loss stop judgment result or after monitoring the completion signal of the emergency repair task, update the distribution network topology status and fault confidence to obtain the updated distribution network topology status and fault confidence, and iteratively solve the constrained collaborative optimization model until all repair tasks are completed.

[0038] In this embodiment, in step S1, by collecting distribution network topology and fault data after extreme events, a four-element fault set including disconnection and blocking faults is constructed; based on the four-element fault set, a priori confidence of the repair success rate is initialized for each fault point.

[0039] Specifically, through equipment such as distribution network measurement terminals and fault monitoring devices, comprehensive data is collected on core aspects of the distribution network after extreme events, such as typhoons, rainstorms, and earthquakes, including network topology, line / node operating status, fault location, and fault type. Faults are classified into four categories: branch disconnection, node disconnection, branch blocking, and node blocking, forming a four-element fault set. This accurately distinguishes between conventional explicit faults and hidden blocking faults caused by relay protection malfunctions or switch jamming. For each fault point, based on historical repair experience, fault type severity, and equipment health status, an initial probability value is assigned as the prior confidence level for repair success rate. This breaks the idealistic assumption of "repair will always succeed" in traditional emergency repairs and provides an initial probability benchmark for subsequent uncertain scheduling.

[0040] In this embodiment, in step S2, a dynamic emergency repair collaborative optimization model for the distribution network is constructed based on the quaternary fault set and the prior confidence level; linearized power flow equation constraints, radial topology hard constraints, and spatiotemporal transfer constraints of maintenance resources are added to the dynamic emergency repair collaborative optimization model for the distribution network to obtain a constrained collaborative optimization model.

[0041] Specifically, by using a four-element fault set to identify the repair targets and using prior confidence to anchor the repair uncertainties, a collaborative optimization model that integrates repair path planning and network topology reconstruction is constructed to achieve integrated collaboration between repair and reconstruction.

[0042] After the model is built, three types of core constraints are embedded in sequence: linearized power flow equation constraints, radial topology hard constraints, and spatiotemporal transfer constraints of maintenance resources, so as to obtain a physically feasible and resource-controllable constrained collaborative optimization model.

[0043] Among them, a1, linearized power flow equation constraint: the DistFlow model is adopted to ensure power balance at each node of the distribution network and that the power flow of branches does not exceed the limit, thus meeting the requirements for safe electrical operation.

[0044] The expression for the power balance constraint is:

[0045] In the formula, For nodes i At time step t Active power output of distributed power sources / generators; For nodesi At time step t Reactive power output of distributed power sources / generators; For nodes The load at time Is it powered? For nodes i Active load; For nodes i reactive load; For the line l At time step t The active power; For the line l At time step t reactive power; and They are nodes Great merit and no merit in effort; and These are collections of substation nodes and distributed power generation nodes, respectively.

[0046] The expression for the linearized power flow equation constraints is:

[0047] In the formula, , They are nodes i and i respectively. Voltage amplitude squared; and branch road Resistance and reactance; For the line l At time step t The topological state variables; M is a maximal positive number in the Big M method; and These are the minimum and maximum limits of the voltage amplitude; branch road At the rated power; L represents the set of branches.

[0048] a2. Radial topology hard constraint: Through connectivity determination and spanning tree constraint, ensure that the distribution network always operates in a loop-free radial structure, which is in line with the inherent operating characteristics of the distribution network.

[0049] For disconnection-type faults, the system is forcibly locked to the disconnected state; for interlocking-type faults, the system is forcibly locked to the conducting state, ensuring the physical rationality of the fault state.

[0050] The expression for the hard constraints on network topology state is: For branch circuit faults:

[0051] In the formula, branch road At any moment Topological state variables, 1 for closed, 0 for open; For node failures:

[0052] In the formula, For nodes The load at time Is it powered? For nodes j At time step t The normal operating condition energized state variables.

[0053] a3. Constraints on the spatial and temporal transfer of maintenance resources: It is clear that the emergency repair team can only be in a single location at any given time, the transfer between fault points requires a fixed amount of travel time, and the maintenance tasks must be executed in a logically coherent manner to prevent the plan from being out of sync with the actual emergency repair capabilities.

[0054] The expression for the spatiotemporal transition constraint of the resource is:

[0055] In the formula, For maintenance personnel exist Is the time at the point of failure? If maintenance is performed, the value is 1; otherwise, it is 0. The maximum time for maintenance status; To move to the fault point and the total time required to complete its repair; It is a sufficiently large positive number; F This is the set of all possible movement paths for the repair team. m This represents the maximum number of moves a repair team can make during the entire repair cycle. R The assembly of the emergency repair team.

[0056] The expression for the logical constraint of the maintenance task is:

[0057] In the formula, Fault At any moment Whether it has been repaired: 1 for yes, 0 for no; This serves as a start marker for maintenance; if at time... Start repairing the fault The value is 1 if it is 1, otherwise it is 0. To fix the fault The time.

[0058] In this embodiment, the objective function expression of the constrained collaborative optimization model is:

[0059] In the formula, For system load recovery efficiency; This is the total recovery time; For the set of load nodes in the power distribution system; t For the current moment; j The node number; For nodes j At any moment t The actual active load; For nodes j Load priority weights; This represents the total load level of the power distribution system under normal operating conditions.

[0060] In this embodiment, in step S3, based on the constrained collaborative optimization model, the prediction time-domain window, control step size, and adaptive stop-loss trigger threshold for rolling optimization are set.

[0061] Specifically, the prediction time window is the future projection duration for rolling optimization. A reasonable setting balances the overall optimization perspective with online computing power, avoiding computational redundancy due to an excessively long time window and short-sighted decision-making due to an excessively short time window. The control step size is the actual execution duration of the instruction after a single optimization, preventing scheduling chaos caused by frequent adjustments to the repair plan and ensuring the smooth implementation of repair instructions. The adaptive stop-loss trigger threshold includes a minimum number of attempts and a time threshold for the urgency of the task to be transferred. The minimum number of attempts avoids blindly turning back due to a single accidental failure, while the urgency time threshold ensures that after issuing the abandonment instruction, the repair team still has sufficient time to transfer and complete alternative tasks, defining a clear trigger boundary for subsequent stop-loss determination.

[0062] In this embodiment, in step S4, based on the set time parameters, the prior confidence level, and the constrained collaborative optimization model, with the load recovery efficiency as the objective function, the constrained collaborative optimization model is solved by the rolling time-domain optimization algorithm to obtain the optimal emergency repair path and network reconstruction action sequence.

[0063] Specifically, the prediction time-domain window, control step size, and stop-loss threshold set in step S3 are used as time parameters. Combined with the prior confidence of repair uncertainty in step S1 and the constrained collaborative optimization model in step S2, an objective function is established with maximizing load restoration efficiency as the core. Essentially, this maximizes the weighted cumulative load restoration amount within the prediction time domain, prioritizing power supply to high-priority loads. A rolling time-domain optimization algorithm is adopted, and through iterative deduction using mixed-integer linear programming within a preset time domain, the time cost of the repair team's movement, the time spent on fault repair, and the benefits of restoring power to nodes are dynamically weighed. Finally, the optimal action sequence consisting of the repair team's movement path, fault repair order, and distribution network switch operations within the current time domain is output, achieving globally optimal scheduling decision-making under uncertain conditions.

[0064] In this embodiment, a mixed integer linear programming (MILP) solution engine based on branch and bound method, such as Gurobi or Cplex, is preferably used. For ultra-large-scale distribution networks with more than 1,000 nodes, in order to pursue the ultimate computing speed, the solution engine can be replaced with heuristic optimization algorithms such as genetic algorithms and particle swarm optimization algorithms, or deep reinforcement learning (DRL) algorithms. Although such algorithms will slightly sacrifice the optimality of the solution, the solution process still follows the rolling time-domain optimization framework of this invention and is adapted to the scheduling needs of distribution networks of different sizes.

[0065] In this embodiment, in step S5, based on the optimal repair path and network reconstruction action sequence, the repair instruction of the first control step is intercepted and executed; if the repair fails, the confidence level of the repair success rate of the current fault point is updated through the Bernoulli random trial model, and the expected value of staying and the expected value of transferring are calculated; based on the expected value of staying, the expected value of transferring and the adaptive stop-loss trigger threshold, the adaptive stop-loss judgment model is used to determine whether to abandon the current fault point.

[0066] Specifically, following the rolling optimization logic of "step-by-step execution and real-time feedback," the emergency repair command corresponding to the first control step is extracted from the optimal action sequence and issued for execution, ensuring that the scheduling commands are implemented step by step. When the on-site feedback indicates that the emergency repair has failed, a single repair attempt is treated as a Bernoulli random trial, with the result being either success or failure. The confidence level of the repair success rate at the current fault point is dynamically lowered through exponential decay logic, accurately reflecting the resistance to repairing hidden faults.

[0067] The expression for the exponential decay model is as follows:

[0068] In the formula, For the first Expected posterior probability of successful fault repair after a failed repair attempt; This represents the baseline repair success rate for common faults. This is the preset confidence decay factor.

[0069] In practical applications, the standard Bayesian posterior probability update equation can also be used to replace the exponential confidence decay model. Although this method is highly dependent on the initial prior distribution, in scenarios with extremely rich historical fault data, it can also perceive the resistance to repairing hidden faults through posterior feedback and achieve the same technical effect.

[0070] In this embodiment, two types of core values ​​are quantified: the expected value of staying is the comprehensive expected benefit of continuously retrying the current fault within a finite number of times; the expected value of transferring is the global topology recovery benefit of abandoning the current fault and transferring to the optimal alternative fault point.

[0071] The formula for calculating the expected value of staying behind is as follows:

[0072] In the formula, To achieve a finite step size Internal continuous retry fault The overall expected return; For the overall scheduling time domain; For fault The additional load recovery that can be achieved after being cleared; n For the first n Try again; For nodes j Load priority weights, For nodes j The re-energized state variables, For nodes j The active power load demand; N A set of nodes; The formula for calculating the expected value of the transfer is as follows:

[0073] In the formula, To abandon the current fault and transfer to an alternative fault point Expected returns; The incremental value of the topology after the alternative fault points are repaired; The exact travel time for the team to move from its current location to the alternative point; The base time for performing a single repair.

[0074] In this embodiment, a single-step heuristic evaluation method based on alternative optimal paths is preferred to calculate the expected value of the transfer. It can also be extended to multi-step prediction or a value evaluation model based on Markov decision process (MDP). Although such methods will increase the online computing load, the underlying logic is still a game between global opportunity cost and local retry value, which does not change the core concept of the stop-loss decision of this invention.

[0075] After calculating the two types of core values, the adaptive stop-loss judgment model, combined with conditions such as minimum number of attempts, remaining scheduling time, and urgency threshold, compares the two types of values ​​to determine whether to abandon the current stubborn fault, thereby avoiding the ineffective consumption of repair resources at low-probability fault points from the root.

[0076] The expression for the adaptive stop-loss determination model is as follows:

[0077] In the formula, k This represents the number of repair attempts made for the current fault point. The minimum number of attempts is the defense line; The remaining available scheduling time, As the threshold for urgent transfer time, To prevent frequent returns and stay-at-home preference penalty coefficients.

[0078] In this embodiment, in step S6, based on the loss prevention judgment result or after monitoring the completion signal of the emergency repair task, the distribution network topology state and fault confidence are updated to obtain the updated distribution network topology state and fault confidence. The constrained collaborative optimization model is iteratively solved until all repair tasks are completed.

[0079] Specifically, two types of discrete events serve as trigger signals: first, the stop-loss judgment result output in step S5 (abandoning the current fault); and second, the detection of a repair task completion signal (fault successfully repaired). Upon triggering, the distribution network topology is immediately updated, correcting for network connectivity changes caused by fault repair or switching operations. Simultaneously, the repair success rate confidence of each fault point is updated, and the latest uncertainty assessment results are synchronized. Based on the updated topology and confidence, the constrained collaborative optimization model is re-introduced, and a new round of rolling optimization is performed in step S4, continuously generating repair and reconfiguration schemes adapted to the current operating conditions. This iterative process continues until all fault points are repaired and no repair tasks remain, ultimately achieving efficient, adaptive, and dynamic repair of the distribution network under extreme events.

[0080] In this embodiment, the emergency repair resources mainly refer to manual emergency repair teams. In practical applications, this method is also applicable to collaborative scheduling scenarios that include mobile energy storage vehicles, mobile power supply vehicles, or remotely automatic operation switches. Meanwhile, the control logic can be deployed on the centralized server of the distribution network dispatch center, or it can be deployed in the regional controller or microgrid gateway based on edge computing, adapting to the needs of different deployment scenarios.

[0081] In one possible implementation, a verification example is provided as follows: This embodiment uses the standard IEEE 33-node distribution network system as the test object to simulate the scenario of multiple faults damaging the distribution network after extreme weather disasters. A typical working condition with multiple hidden blocking faults is selected. The present invention is compared with the traditional post-disaster repair and reconstruction baseline scheme (DMRS-NR) to verify the practical application effect of the present invention from three aspects: power restoration efficiency, dispatch robustness, and engineering feasibility.

[0082] I. Power restoration efficiency test: The test selected a typical scenario of the IEEE 33-node system (Scenario #46). This scenario has four hidden blocking faults caused by relay protection maloperation and switch mechanical jamming on the critical branches (such as the fault on branch 21). It is a high-difficulty working condition that is prone to causing ineffective occupation of emergency repair resources.

[0083] Test results show that the traditional baseline solution, lacking an adaptive stop-loss mechanism, blindly retried for hidden faults, resulting in 13 retries at 4 fault points over 26 hours, directly causing a severe deadlock of repair resources and a final network power restoration rate of only 45.69%. In contrast, this invention, by assessing the expected value of staying on-call versus transferring resources in real time, decisively implemented a stop-loss and transfer strategy after 13 hours of repair work, promptly releasing ineffectively occupied repair resources and prioritizing high-value fault repair tasks. Ultimately, the network power restoration rate increased to 88.11%, nearly doubling the power restoration performance of the baseline solution, effectively solving the repair deadlock problem caused by stubborn faults and significantly improving overall power restoration efficiency.

[0084] II. Scheduling Robustness Test: To further verify the scheduling stability of this invention under unknown and complex operating conditions, three test scenarios with different fault scales (number of faults F=10, F=20, and F=30) were set up. Each scenario underwent 50 independent Monte Carlo simulation tests, with the cumulative resilience area (AUC) as the core evaluation metric. The test results are shown in Table 1.

[0085] Table 1. Performance Statistical Comparison of Different Strategies under Different Fault Scales

[0086] As shown in Table 1, compared with the traditional benchmark scheme, the present invention achieves a robust improvement of 5.46%, 7.61%, and 9.04% in cumulative resilience area under three fault scale scenarios, respectively. At the same time, through statistical significance testing, the p-values ​​corresponding to the improvement effect in all scenarios are less than 0.05, and the p-value in the high fault scale scenario is less than 0.001, which fully demonstrates that the present invention has excellent scheduling robustness in unknown and uncertain post-disaster emergency repair environments and can stably adapt to complex working conditions with different fault scales.

[0087] III. Project Feasibility: To assess the feasibility and scalability of practical engineering applications, a collaborative scheduling test analysis of multiple emergency repair teams was conducted, focusing on verifying the physical execution reliability and online scheduling efficiency of the solution.

[0088] Test results show that in a scenario with two emergency repair teams working together, the adaptive scheduling advantage of this invention is fully realized, with the cumulative resilience area (AUC) increasing by a peak of 7.07%. Simultaneously, the average time for a single rolling optimization solution is only 11.15 seconds, far lower than the minute-level online rolling scheduling time requirement in actual engineering projects. During the test, this invention, by embedding radial topology hard constraints and locking inoperable fault constraints, ensures that the generated repair paths and network reconfiguration instructions fully comply with the underlying physical operation rules of the distribution network, with no instruction execution conflicts, guaranteeing 100% executable reconfiguration schemes. It can be directly applied to actual post-disaster emergency repair scheduling scenarios in distribution networks.

[0089] The application scenarios of this invention are as follows: In scenarios where extreme weather disasters such as typhoons, rainstorms, and earthquakes cause large-scale power distribution network failures, this invention can characterize the uncertainty of fault repair, dynamically plan emergency repair paths, and coordinate network reconstruction to improve overall power restoration efficiency.

[0090] In emergency repair scenarios where there are hidden interlocking faults such as relay protection malfunctions and switch mechanical jamming in the distribution network, this invention can sense the repair resistance through confidence updates and adopt an adaptive loss-stopping strategy to avoid the ineffective occupation of repair resources.

[0091] In the scenario of joint emergency repair of power distribution networks involving multiple emergency repair teams, mobile energy storage vehicles, and mobile power supply vehicles, this invention can be adapted to multiple types of emergency repair resources, dynamically optimize resource allocation and scheduling, and ensure the orderly conduct of collaborative emergency repairs.

[0092] In large-scale distribution network post-disaster scheduling scenarios with large node scale and numerous faults, this invention can be adapted to different network scales, balance computational efficiency and scheduling effect, and meet the engineering requirements of online rolling scheduling.

[0093] In the distributed emergency repair scheduling scenario under the edge computing architecture of the distribution network, this invention can be flexibly deployed on the regional controller or microgrid gateway to respond to local fault feedback in real time and adapt to the needs of distributed emergency repair management.

[0094] It should be noted that the method of this disclosure embodiment can be executed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this disclosure embodiment, and the multiple devices will interact with each other to complete the method described.

[0095] It should be noted that the above description describes some embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0096] Example 2

[0097] See Figure 3 Embodiment 2 of the present invention also provides an adaptive dynamic emergency repair system for distribution networks that considers repair uncertainties, comprising: The data acquisition and classification module 001 is used to construct a four-element fault set containing disconnection and blocking faults by collecting distribution network topology and fault data after extreme events; and to initialize the prior confidence of the repair success rate for each fault point based on the four-element fault set. The model and constraint construction module 002 is used to construct a dynamic emergency repair collaborative optimization model for the distribution network based on the four-element fault set and the prior confidence level; and to add linearized power flow equation constraints, radial topology hard constraints, and maintenance resource spatiotemporal transfer constraints to the dynamic emergency repair collaborative optimization model for the distribution network to obtain a constrained collaborative optimization model. The time parameter setting module 003 is used to set the prediction time domain window, control step size and adaptive stop loss trigger threshold for rolling optimization based on the constrained collaborative optimization model. The rolling optimization module 004 is used to solve the constrained collaborative optimization model based on the set time parameters, the prior confidence level and the constrained collaborative optimization model, with the objective function of maximizing load recovery efficiency, and to obtain the optimal emergency repair path and network reconstruction action sequence through the rolling time domain optimization algorithm. The emergency repair execution and loss control determination module 005 is used to intercept and execute the emergency repair instruction of the first control step based on the optimal emergency repair path and network reconstruction action sequence; if the emergency repair fails, the repair success rate confidence of the current fault point is updated through the Bernoulli random trial model, and the expected value of staying and the expected value of transferring are calculated; based on the expected value of staying, the expected value of transferring and the adaptive loss control trigger threshold, the adaptive loss control determination model determines whether to abandon the current fault point; The event-driven closed-loop module 006 is used to update the distribution network topology state and fault confidence based on the loss stop judgment result or after detecting the completion signal of the emergency repair task, to obtain the updated distribution network topology state and fault confidence, and to iteratively solve the constrained collaborative optimization model until all repair tasks are completed.

[0098] In this embodiment, the data acquisition and classification module 001 includes the following four-element fault set: a set of disconnection-type fault branches, a set of disconnection-type fault nodes, a set of interlocking-type fault branches, and a set of interlocking-type fault nodes. The prior confidence level is the initial probability value of a single successful repair at each fault point.

[0099] In this embodiment, in the model and constraint construction module 002, the objective function expression of the constrained collaborative optimization model is:

[0100] In the formula, For system load recovery efficiency; This refers to the total recovery time. For the set of load nodes in the power distribution system; t For the current moment; j The node number; For nodes j At any moment t The actual active load; For nodes j Load priority weights; This represents the total load level of the power distribution system under normal operating conditions.

[0101] In this embodiment, in the emergency repair execution and loss mitigation determination module 005, during the process of updating the repair success rate confidence of the current fault point using the Bernoulli random trial model, each repair attempt is considered as a Bernoulli trial, with the result being either success or failure. If the trial fails, the repair success rate confidence of the fault point is updated using an exponential decay model; the expression for the exponential decay model is:

[0102] In the formula, For the first Expected posterior probability of successful fault repair after a failed repair attempt; This represents the baseline repair success rate for common faults. This is the preset confidence decay factor.

[0103] In this embodiment, the formula for calculating the expected value of remaining personnel in the emergency repair execution and loss mitigation determination module 005 is as follows:

[0104] In the formula, To achieve a finite step size Internal continuous retry fault The overall expected return; For the overall scheduling time domain; For fault The additional load recovery that can be achieved after being cleared; n For the first n Try again; For nodes j Load priority weights, For nodes j The re-energized state variables, For nodes j The active power load demand; N A set of nodes; The formula for calculating the expected value of the transfer is as follows:

[0105] In the formula, To abandon the current fault and transfer to an alternative fault point Expected returns; The incremental value of the topology after the alternative fault points are repaired; The exact travel time for the team to move from its current location to the alternative point; The base time for performing a single repair.

[0106] In this embodiment, in the emergency repair execution and loss control determination module 005, the expression of the adaptive loss control determination model is:

[0107] In the formula, k This represents the number of repair attempts made for the current fault point. The minimum number of attempts is the defense line; The remaining available scheduling time, As the threshold for urgent transfer time, To prevent frequent returns and stay-at-home preference penalty coefficients.

[0108] It should be noted that the information interaction and execution process between the modules of the above system are based on the same concept as the method embodiment in Embodiment 1 of this application, and the resulting technical effects are the same as those in the method embodiment of this application. For details, please refer to the description in the method embodiment shown above in this application, and it will not be repeated here.

[0109] Example 3

[0110] Embodiment 3 of the present invention provides a non-transitory computer-readable storage medium storing program code for a distribution network adaptive dynamic emergency repair processing method that considers repair uncertainty. The program code includes instructions for executing the distribution network adaptive dynamic emergency repair processing method that considers repair uncertainty in Embodiment 1 or any possible implementation thereof.

[0111] Computer-readable storage media can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

[0112] Example 4

[0113] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor; The processor and the memory communicate with each other via a bus; the memory stores program instructions that can be executed by the processor, and the processor can call the program instructions to execute an adaptive dynamic emergency repair method for distribution networks that takes into account the repair uncertainty, as described in Embodiment 1 or any possible implementation thereof.

[0114] Specifically, a processor can be implemented in hardware or software. When implemented in hardware, the processor can be a logic circuit, an integrated circuit, etc. When implemented in software, the processor can be a general-purpose processor that reads software code stored in memory. This memory can be integrated into the processor or located outside the processor and exist independently.

[0115] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable system. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0116] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing systems. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Optionally, they can be implemented using program code executable by a computing system, thereby storing them in a storage system for execution by the computing system. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0117] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for adaptive dynamic emergency repair of distribution networks considering repair uncertainties, characterized in that, include: By collecting distribution network topology and fault data after extreme events, a four-element fault set including disconnection and blocking faults is constructed. Based on the aforementioned quaternary fault set, a prior confidence level for the repair success rate is initialized for each fault point; Based on the quaternary fault set and the prior confidence level, a dynamic emergency repair collaborative optimization model for the distribution network is constructed. Linearized power flow equation constraints, radial topology hard constraints, and spatiotemporal transfer constraints of maintenance resources are added to the dynamic emergency repair collaborative optimization model for the distribution network to obtain a constrained collaborative optimization model. Based on the constrained collaborative optimization model, the prediction time-domain window, control step size, and adaptive stop-loss trigger threshold for rolling optimization are set. Based on the set time parameters, the prior confidence level, and the constrained collaborative optimization model, with the objective function of maximizing load recovery efficiency, the constrained collaborative optimization model is solved by the rolling time-domain optimization algorithm to obtain the optimal emergency repair path and network reconstruction action sequence. Based on the optimal repair path and network reconstruction action sequence, the repair instruction of the first control step is intercepted and executed; if the repair fails, the repair success rate confidence of the current fault point is updated through the Bernoulli random trial model, and the expected value of staying and the expected value of transferring are calculated; based on the expected value of staying, the expected value of transferring and the adaptive stop loss trigger threshold, the adaptive stop loss judgment model is used to determine whether to abandon the current fault point. Based on the loss prevention judgment result or after detecting the completion signal of the emergency repair task, the distribution network topology status and fault confidence are updated to obtain the updated distribution network topology status and fault confidence. The constrained collaborative optimization model is then iteratively solved until all repair tasks are completed.

2. The adaptive dynamic emergency repair method for distribution networks considering repair uncertainties according to claim 1, characterized in that, The four-element fault set includes: a set of disconnection-type fault branches, a set of disconnection-type fault nodes, a set of interlocking-type fault branches, and a set of interlocking-type fault nodes; The prior confidence level is the initial probability value of a single successful repair at each fault point.

3. The adaptive dynamic emergency repair method for distribution networks considering repair uncertainties according to claim 2, characterized in that, The objective function expression of the constrained collaborative optimization model is: In the formula, For system load recovery efficiency; This refers to the total recovery time. For the set of load nodes in the power distribution system; t For the current moment; j The node number; For nodes j At any moment t The actual active load; For nodes j Load priority weights; This represents the total load level of the power distribution system under normal operating conditions.

4. The adaptive dynamic emergency repair method for distribution networks considering repair uncertainties according to claim 3, characterized in that, In updating the repair success rate confidence of the current fault point using the Bernoulli random trial model, each repair attempt is considered a Bernoulli trial, with a result of success or failure. If the trial fails, the repair success rate confidence of the fault point is updated using an exponential decay model; the expression for the exponential decay model is: In the formula, For the first Expected posterior probability of successful fault repair after a failed repair attempt; This represents the baseline repair success rate for common faults. This is the preset confidence decay factor.

5. The adaptive dynamic emergency repair method for distribution networks considering repair uncertainties according to claim 4, characterized in that, The formula for calculating the expected value of staying behind is as follows: In the formula, To achieve a finite step size Internal continuous retry fault The overall expected return; For the overall scheduling time domain; For fault The additional load recovery that can be achieved after being cleared; n For the first n Try again; For nodes j Load priority weights, For nodes j The re-energized state variables, For nodes j The active power load demand; N A set of nodes; The formula for calculating the expected value of the transfer is as follows: In the formula, To abandon the current fault and transfer to an alternative fault point Expected returns; The incremental value of the topology after the alternative fault points are repaired; The exact travel time for the team to move from its current location to the alternative point; The base time for performing a single repair.

6. The adaptive dynamic emergency repair method for distribution networks considering repair uncertainties according to claim 5, characterized in that, The expression for the adaptive stop-loss determination model is: In the formula, k This represents the number of repair attempts made for the current fault point. The minimum number of attempts is the defense line; The remaining available scheduling time, As the threshold for urgent transfer time, To prevent frequent returns and stay-at-home preference penalty coefficients.

7. A distribution network adaptive dynamic emergency repair system considering repair uncertainty, employing the distribution network adaptive dynamic emergency repair method considering repair uncertainty as described in any one of claims 1-6, characterized in that, include: The data acquisition and classification module is used to construct a four-element fault set containing disconnection and blocking faults by collecting distribution network topology and fault data after extreme events; based on the four-element fault set, the prior confidence of the repair success rate is initialized for each fault point. The model and constraint construction module is used to construct a dynamic emergency repair collaborative optimization model for the distribution network based on the four-element fault set and the prior confidence level; and to add linearized power flow equation constraints, radial topology hard constraints, and maintenance resource spatiotemporal transfer constraints to the dynamic emergency repair collaborative optimization model for the distribution network to obtain a constrained collaborative optimization model. The time parameter setting module is used to set the prediction time domain window, control step size and adaptive stop loss trigger threshold for rolling optimization based on the constrained collaborative optimization model. The rolling optimization module is used to solve the constrained collaborative optimization model based on the set time parameters, the prior confidence level and the constrained collaborative optimization model, with the objective function of maximizing load recovery efficiency, and to obtain the optimal emergency repair path and network reconstruction action sequence through the rolling time domain optimization algorithm. The emergency repair execution and loss control module is used to intercept and execute the emergency repair instruction of the first control step based on the optimal emergency repair path and network reconstruction action sequence. If the emergency repair fails, the module updates the repair success rate confidence of the current fault point through a Bernoulli random trial model and calculates the expected value of staying and the expected value of transferring. Based on the expected value of staying, the expected value of transferring, and the adaptive loss control trigger threshold, the module determines whether to abandon the current fault point through an adaptive loss control model. The event-driven closed-loop module is used to update the distribution network topology status and fault confidence based on the loss stop judgment result or after detecting the completion signal of the emergency repair task, obtain the updated distribution network topology status and fault confidence, and iteratively solve the constrained collaborative optimization model until all repair tasks are completed.

8. The adaptive dynamic emergency repair system for distribution networks considering repair uncertainty according to claim 7, characterized in that, In the data acquisition and classification module, the four-element fault set includes: a set of disconnection-type fault branches, a set of disconnection-type fault nodes, a set of interlocking-type fault branches, and a set of interlocking-type fault nodes; The prior confidence level is the initial probability value of a single successful repair at each fault point.

9. A distribution network adaptive dynamic emergency repair system considering repair uncertainty as described in claim 8, characterized in that, In the model and constraint construction module, the objective function expression of the constrained collaborative optimization model is: In the formula, For system load recovery efficiency; This refers to the total recovery time. For the set of load nodes in the power distribution system; t For the current moment; j The node number; For nodes j At any moment t The actual active load; For nodes j Load priority weights; This represents the total load level of the power distribution system under normal operating conditions.

10. A distribution network adaptive dynamic emergency repair system considering repair uncertainty according to claim 9, characterized in that, In the emergency repair execution and loss mitigation determination module, during the process of updating the repair success rate confidence of the current fault point using the Bernoulli random trial model, each repair attempt is considered a Bernoulli trial, with a result of success or failure. If the trial fails, the repair success rate confidence of the fault point is updated using an exponential decay model; the expression for the exponential decay model is: In the formula, For the first Expected posterior probability of successful fault repair after a failed repair attempt; This represents the baseline repair success rate for common faults. This is a preset confidence decay factor; In the emergency repair execution and loss control determination module, the formula for calculating the expected value of remaining on the job is: In the formula, To achieve a finite step size Internal continuous retry fault The overall expected return; For the overall scheduling time domain; For fault The additional load recovery that can be achieved after being cleared; n For the first n Try again; For nodes j Load priority weights, For nodes j The re-energized state variables, For nodes j The active power load demand; N A set of nodes; The formula for calculating the expected value of the transfer is as follows: In the formula, To abandon the current fault and transfer to an alternative fault point Expected returns; The incremental value of the topology after the alternative fault points are repaired; The exact travel time for the team to move from its current location to the alternative point; The base time for performing a single repair; In the emergency repair execution and loss control determination module, the expression of the adaptive loss control determination model is: In the formula, k This represents the number of repair attempts made for the current fault point. The minimum number of attempts is the defense line; The remaining available scheduling time, As the threshold for urgent transfer time, To prevent frequent returns and stay-at-home preference penalty coefficients.

Images