Search

CN-121998178-A - Solving difficulty analysis and difficulty case prediction method for scheduling in power grid mountain fire disaster

CN121998178ACN 121998178 ACN121998178 ACN 121998178ACN-121998178-A

Abstract

The invention discloses a method for analyzing solving difficulty and predicting difficult cases of scheduling in a power grid mountain fire, which comprises the following steps of 1) constructing a mixed integer linear programming model for describing an optimization target of a scheduling process in the power grid fire scene, 2) analyzing the solving difficulty of scheduling in the power grid mountain fire, determining characteristic factors related to power grid faults or scheduling, 3) constructing a difficult scene classification model based on the characteristic factors and the corresponding solving difficulty, and 4) judging the solving difficulty of the optimization target of the scheduling process in the current power grid mountain fire by using the difficult scene classification model. The method can support the power grid in a limited scheduling time window to quickly select the adaptive solving strategy, and remarkably improve the recovery efficiency and the scheduling reliability in a complex mountain fire scene.

Inventors

  • YANG ZHIFANG
  • CAO YISHENG
  • YU JUAN

Assignees

  • 重庆大学

Dates

Publication Date
20260508
Application Date
20260114

Claims (10)

  1. 1. The method for analyzing the solving difficulty and predicting the difficulty case of scheduling in the power grid mountain fire disaster is characterized by comprising the following steps of: Step 1), constructing a mixed integer linear programming model for describing a scheduling process optimization target in a mountain fire scene power grid disaster; step 2) analyzing the solving difficulty of the dispatching in the power grid mountain fire disaster, and determining characteristic factors related to power grid faults or dispatching; Step 3) constructing a difficult scene classification model based on the characteristic factors and the corresponding solving difficulty; and 4) judging the solving difficulty of the optimization target in the scheduling process in the current power grid mountain fire disaster by using the difficult scene classification model.
  2. 2. The method for analyzing solving difficulty and predicting difficulty cases for scheduling in power grid mountain fires according to claim 1, wherein in step 1), the objective function of the mixed integer linear programming model is as follows: ;(1) Wherein, the A penalty coefficient for load loss for adjusting the weight of the security in the objective function; to lose load Weight coefficient of (2); Indicating lost load A value; is a load Load off-time after failure; Is a unit Is used for controlling the output cost coefficient of the (a), Indicating machine set At the moment of time Is a force of the (a); And Respectively the units At the moment of time Is required, and the start-up and shut-down costs of (1) are reduced.
  3. 3. The method for analyzing the solving difficulty and predicting the difficulty case for scheduling in the power grid mountain fire disaster according to claim 1, wherein in the step 1), constraint conditions of the mixed integer linear programming model comprise path constraint of emergency repair personnel, time constraint of the emergency repair personnel reaching fault equipment, coupling constraint of fault equipment repair time and regulation resource availability state and regulation resource constraint; The path constraint of the rush-repair personnel comprises departure constraint of the rush-repair personnel, return constraint of the rush-repair personnel, path uniqueness constraint, capacity constraint of the rush-repair site and destination access constraint; The time constraint of the rush-repair personnel to the fault equipment comprises initial departure time constraint, travel time constraint of the rush-repair station to the fault point, travel time constraint of the repaired fault point to the fault point, waiting time constraint of the rush-repair personnel and the time constraint of the rush-repair personnel from the fault point To the point of failure Is used for repairing incomplete special processing constraint and fault equipment repairing completion time constraint; The coupling constraint of the fault equipment repair time and the regulation resource availability state comprises a unit fault pre-indication variable constraint, a unit recovery post-indication variable constraint, a unit availability joint constraint, a consistency constraint of a system running state and unit availability, a load node fault pre-indication variable constraint, a load node recovery post-indication variable constraint, a load node availability joint constraint, a fault load original load demand, a line fault pre-indication variable constraint, a line recovery post-indication variable constraint and a line availability joint constraint; The regulation and control resource constraint comprises a unit output constraint, a continuous startup and shutdown time constraint of the unit, a unit startup and shutdown and output change constraint, a unit startup and shutdown cost constraint, a total power generation power requirement and total load requirement balance constraint of a system in each period and a tide constraint of each branch in any period.
  4. 4. The method for analyzing solving difficulty and predicting difficult cases for scheduling in power grid mountain fires according to claim 3, wherein the departure constraint of the rush-repair personnel is that the rush-repair personnel can only start from the affiliated rush-repair site, and the fault equipment cannot be used as a departure point, namely: ;(2) ;(3) Wherein, the For all the first-aid repair site sets, Collecting all fault points; The repair personnel return constraint means that the repair personnel cannot return to the original site after leaving the repair site, namely: ;(4) Wherein, the A decision variable of 0-1, which indicates whether the rush repair personnel is from the node Moving to a node ; The path uniqueness constraint means that any one path can only be accessed once in the same rush-repair process, and repeated passing is not allowed, namely: ;(5) The capacity constraint of the rush-repair sites means that the number of rush-repair personnel which can be dispatched by each rush-repair site at the same time must not exceed the upper limit of the capacity, namely: ;(6) Wherein, the For the total number of all the rush-repair points, For a station Maximum rush repair personnel capacity; the destination access constraint means that each fault point can only be accessed once at most by one rush-repair person, and the fault point should be left or stay at the point after the access, namely: ;(7) in the formula, A decision variable of 0-1, which indicates whether the rush repair personnel is from the node Moving to a node 。
  5. 5. The method for analyzing and predicting difficulty in solving scheduling in power grid mountain fire according to claim 3, wherein the initial departure time constraint is that the initial time of a rush-repair person from a affiliated rush-repair site is equal to the site The allowable departure time of the middle rush-repair personnel is as follows: ;(8) Wherein, the From the rush repair site for rush repair personnel The time of departure is set to be equal to the time of departure, For a station The allowable departure time of the middle rush repair personnel; Travel time constraint for reaching fault point when starting from rush-repair station means that if rush-repair personnel is led to the rush-repair station Go directly to the fault point The arrival time is equal to the slave station To the point of failure Is to say: ;(9) Wherein, the Is a slave node To the point of Is used for the travel time of the person, Is a sufficiently large constant; travel time constraints to reach a point of failure when starting from the repaired point of failure are as follows: ;(10) Wherein, the Representing a fault The time of occurrence; The rush-repair personnel waiting time constraints are as follows: ;(11) ;(12) Wherein, the Is an auxiliary binary variable; Is a latency variable; from the fault point, the rush repair personnel To the point of failure The scheduling time constraints of (a) are as follows: ;(13) Wherein, the As a point of failure Is time-consuming to repair; the special processing constraint that the repair is not completed means that if the repair personnel fails to complete the repair of the current fault component, the corresponding repair time is assigned as The method comprises the following steps: ;(14) the time constraint of repairing the fault equipment is that the fault equipment Repair completion time of (2) The sum of the time for the emergency repair personnel to reach the equipment and the time required for the repair operation is that: ;(15) in the formula, For faulty devices Is a repair completion time of (a).
  6. 6. The method for analyzing and predicting difficulty in solving scheduling in power grid mountain fire as recited in claim 3, wherein the pre-fault indicating variable constraint is used for guaranteeing that the pre-fault indicating variable constraint is used for ensuring that the pre-fault indicating variable constraint is used for the pre-fault indicating variable constraint Earlier than the occurrence time of unit faults In the time-course of which the first and second contact surfaces, Otherwise, 0; The set pre-fault indicator variable constraints are as follows: ;(16) in the formula, To characterize the unit at the moment 0-1 Variable whether previously nonfaulty; To characterize the unit at the moment 0-1 Variable whether repaired or not later; To characterize the unit at the moment 0-1 Variable if available; is the current moment; Indicating variable constraint after unit recovery is used for ensuring that current moment is later than unit repair time In the time-course of which the first and second contact surfaces, Otherwise, 0; The indicated variable constraints after unit recovery are as follows: ;(17) availability conjunctive constraints refer to the condition that a unit is available for a certain period of time as "not yet failed" or "repaired", i.e.: ;(18) The consistency constraints of the system operating state and the availability of the units are as follows: ;(19) in the formula, For the unit in time period Is used for the power-on state variable of (1), For the level of the force to be exerted, Is rated maximum output; The load node pre-fault indicator variable constraints are as follows: ;(20) Wherein, the To characterize the load at time 0-1 Variable whether previously nonfaulty; To characterize the load at time 0-1 Variable if available; To characterize the load at time 0-1 Variable whether repaired or not later; the load node post-restoration indicated variable constraints are as follows: ;(21) The load node availability conjunctive constraints are as follows: ;(22) in the time period Time fault load The raw load requirements of (2) are as follows: ;(23) Wherein, the Is expressed in time period Time fault load Is added to the original load demand of (1), A state variable that is whether the load is available during the period; the pre-line fault indicator variable constraints are as follows: ;(24) Wherein, the To characterize the line at the moment 0-1 Variable whether previously nonfaulty; to characterize the line at the moment 0-1 Variable whether repaired or not later; to characterize the line at the moment 0-1 Variable if available; The indicated variable constraints after line restoration are as follows: ;(25) The line availability conjunctive constraints are as follows: ;(26) in the formula, To characterize the line at the moment 0-1 Variable if available.
  7. 7. The method for analyzing and predicting difficulty in solving scheduling in power grid mountain fire according to claim 3, wherein the unit is arranged at any time period The force constraints of (2) are as follows: ;(27) Wherein, the Is a unit In the time period Is started up; 、 the upper and lower limits of the output; the continuous power-on and power-off time constraint of the unit is as follows: ;(28) Wherein, the 、 Respectively the units Is started up at minimum the minimum shutdown time required for the system to be shut down, 、 Respectively accumulating continuous startup and shutdown time; The unit start-stop and output change constraint is as follows: ;(29) Wherein, the 、 The maximum climbing and falling rates of the unit are respectively; The start-stop cost constraint of the unit is as follows: ;(30) Wherein, the 、 Respectively the units Is a starting-up and stopping cost coefficient, 、 The starting-up and stopping costs are respectively; The balance constraint of the total power generation power requirement and the total load requirement of the system in each period is as follows: ;(31) the flow constraints of each branch in any period are as follows: ;(32) Wherein, the 、 The minimum and maximum allowable power flows of the branches; 、 coefficients are assigned to the node-to-branch power flow, Is a node Is a voltage level of (a) in the battery.
  8. 8. The method for analyzing and predicting difficulty in solving scheduling in power grid mountain fire according to claim 1, wherein the characteristic factors related to power grid faults or scheduling include fault equipment category, fault number, fault occurrence time and degree of attention to safety and economy in the process of recovering power grid.
  9. 9. The method for analyzing solving difficulty and predicting difficult cases for scheduling in power grid mountain fires according to claim 1, wherein in step 3), the step of constructing a difficult scene classification model comprises: Step 3.1) constructing a characteristic factor vector related to grid faults or scheduling, namely: ;(33) Wherein, the Representing characteristic factors related to grid faults or scheduling; Step 3.2) describing the relation between the characteristic factors and the solving difficulty, namely: ;(34) Wherein, the Representing the time required for the solver to solve at a convergence accuracy Gap of 0.01%; step 3.3) generating an initial sample of the feature vector by adopting a random sampling method, namely: ;(35) Wherein, the Representing a predetermined joint profile of the set-up, Is the total number of samples; Step 3.4) carrying out feasibility test on the samples, removing infeasible scenes and forming an effective sample set, wherein the step of carrying out feasibility test is to take a fault scene sample generated based on random sampling as input, solve a mixed integer linear programming model by utilizing Gurobi solver, and if no solution exists, the current fault scene sample is infeasible; Step 3.5) inputting the effective sample set into a mixed integer linear programming model to obtain the solving time corresponding to each group of samples, and constructing a classification data set, namely: ;(36) Wherein, the In order for the feature vector of the difficult scene, The corresponding solving time is obtained; For a simple scene feature vector, The corresponding solving time is obtained; step 3.6) constructing a classification model based on a support vector machine; The discriminant function of the classification model is as follows: ;(37) Wherein, the In order to be a lagrange multiplier, As a function of the kernel, As a result of the bias term, The number of training samples; step 3.7) introducing a priori empirical features Prior, namely: ;(38) Wherein, the Representing characteristics Importance weights of (2); And 3.8) writing the Prior experience feature Prior into the input of the classification data set, and training the classification model by utilizing the updated classification data set to obtain a difficult scene classification model.
  10. 10. The method for analyzing the difficulty of solving the scheduling problem and predicting the difficulty case in the mountain fire disaster of the power grid, which is disclosed by claim 9, is characterized in that the step of generating an initial sample of the feature vector by adopting a random sampling method comprises the steps of setting a value interval of each factor according to the structure of the power system and the mountain fire spread characteristic; sampling the continuous variable with uniform distribution or normal distribution, sampling the discrete variable with discrete uniform distribution, and generating initial sample.

Description

Solving difficulty analysis and difficulty case prediction method for scheduling in power grid mountain fire disaster Technical Field The invention relates to the field of power systems and automation thereof, in particular to a method for solving difficulty analysis and difficult case prediction of scheduling in power grid mountain fires. Background In recent years, extreme climate events are increasingly frequent, continuous extreme drought occurs in the global area, the risk of forest fires is obviously increased, and the safe and stable operation of a power grid is seriously threatened. Actual events show that mountain fires often cause damage to large-scale power grid equipment, and the characteristics of various fault types, complex fault time sequence evolution and the like are presented. Under the background, the scheduling in the power grid disaster not only needs to respond to equipment faults quickly, but also needs to be balanced between safety and economy, and the recovery efficiency directly influences the power supply reliability and disaster coping capacity. Therefore, how to construct a disaster recovery scheduling system with rapid recovery capability becomes a common requirement for research and engineering practice. The existing research has made a certain progress in the aspect of power grid dispatching under the mountain fire scene, but mainly focuses on two types of contents, namely, dispatching optimization of power grid regulation and control resources oriented to power generation, energy storage, load side response and the like, and path planning and dispatching strategies oriented to emergency resources such as emergency repair teams, fire extinguishing vehicles, emergency power supplies and the like. However, most of these studies only consider single type resources, or focus on repair recovery at post-disaster stage, lacking space-time collaborative modeling of emergency resources and regulatory resources at the post-disaster stage. Because the state coupling and scheduling cooperative relation between two types of resources cannot be established, the existing recovery strategy is difficult to consider the safety and economy of the system, and the scheduling efficiency and the overall recovery capability are insufficient. From the model structure, the scheduling problem in mountain fire is related to a large number of discrete decision variables, time sequence state constraints and strong coupling relations among resources, and the complexity is obviously higher than that of the conventional power grid scheduling problem. In addition, the mountain fire has the characteristics of high spreading speed, wide influence range, complex fault chain and the like, so that the scheduling problem presents challenges of large solving scale, high calculation pressure, strong timeliness requirement and the like. In a limited scheduling time window, a default strategy depending on a general solver is difficult to ensure rapid convergence, and particularly, under the conditions of serious equipment failure, resource shortage, high load requirement and the like, the solving time and the solving quality cannot meet the actual rapid recovery requirement. In emergency practice, different forest fire fault scenarios often lead to distinct solution difficulties. However, at the present stage, system research on solving difficulty of scheduling problems in power grid mountain fires is not available, key factors influencing solving efficiency are not known enough, and a possibly-occurring difficult scene cannot be predicted in advance. The scheduling center cannot select a proper acceleration strategy according to different scenes, so that the practicability and the response speed of scheduling in the disaster are restricted. In summary, the current power grid mountain fire disaster scheduling mainly has two core problems, namely, a first system model for collaborative scheduling of emergency resources and regulation resources is lacking, a recovery scheme with both safety and economy is difficult to form, and a second system analysis and difficult scene prediction method for solving different mountain fire fault scenes is lacking, so that a targeted solving strategy selection basis cannot be provided for rapid recovery in the disaster. Therefore, a new modeling framework and a data driving technology are necessary to be provided, a solution difficulty evolution rule is systematically described, and accurate prediction of a difficult scene is realized, so that theoretical and technical support is provided for rapid solution and recovery of scheduling in power grid mountain fire. Disclosure of Invention The invention aims to provide a method for solving difficulty analysis and difficult case prediction of scheduling in power grid mountain fire, which comprises the following steps: Step 1), constructing a mixed integer linear programming model for describing a scheduling process optimization target in a