Method for evaluating and improving toughness of power distribution information physical system under disaster

Abstract

The invention discloses a toughness evaluation and toughness improvement method for a power distribution information physical system under disasters, and belongs to the field of cooperative regulation and optimal operation of power systems. The method comprises the following steps: randomly generating a damage scene of an urban power grid under a disaster through Monte Carlo simulation sampling, establishing a power distribution network information physical system multi-period dynamic joint toughness evaluation model based on a scene basis, and introducing a toughness evaluation index to evaluate the toughness of the power grid; recovery requirements of power network loads and communication system resources are considered, distributed units, energy storage equipment and other resources in a power distribution network are utilized, and strategy optimization is carried out in cooperation with a mobile emergency power supply and a mobile communication vehicle; and the toughness of the power distribution network information physical system under different configurations is evaluated, and an optimal toughness improvement strategy of the power distribution network information physical system is solved. The power distribution network multi-period toughness evaluation model under the extreme disaster scene is established, various resource configurations are combined, power grid toughness improvement is promoted on the basis of considering information network coupling, and the method has high application value.

Description

Technical Field

[0001] The present invention relates to the field of coordinated control and optimized operation of power systems, and in particular to a method for evaluating and improving the resilience of a power distribution information-physical system under disasters. Background Art

[0002] In the context of the intelligent and information-based transformation of the power system, it is a challenging task to promote the assessment and improvement of the resilience of distribution networks under increasingly frequent extreme disasters.

[0003] When extreme disasters occur, power systems are prone to mass failures, which in turn lead to large-scale power outages, seriously threatening social economy and residents' lives. In order to promote the distribution network cyber-physical system to cope with extreme disasters with a higher level of resilience, it is necessary to coordinate various resources and optimize the system operation status. Considering the low probability and high impact characteristics of extreme disasters and the background of rapid development of new energy, the optimization dispatch should give priority to reducing the loss of load loss and coordinate the use of various resources to effectively improve the system resilience. Therefore, based on the complex network theory, a bidirectional dynamic model of the distribution network cyber-physical system is established, and the system damage scenario is established through the Monte Carlo simulation method. With the goal of minimizing the system load shedding, the output of distributed units and energy storage equipment in the system is adjusted, and mobile emergency power supplies and mobile communication vehicles are called to maintain the state of the distribution network cyber-physical system. The resilience evaluation index is introduced to quantitatively analyze the system resilience. Whether it is for power system operators and maintainers or power grid users, it is particularly important to reduce the impact of disasters and promote the distribution network to resume normal operation as soon as possible after encountering extreme disasters. Therefore, the resilience evaluation of the distribution network under extreme disasters and the optimal load shedding model considering information network coupling for urban operation services are of great research significance. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for evaluating and improving the resilience of a power distribution information-physical system under disasters.

[0005] The objective of the present invention is achieved through the following technical solutions:

[0006] A method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters comprises the following steps:

[0007] Step 1: Based on complex network theory, a bidirectional dynamic model of the information-physical system of the distribution network is established, including: a distribution network model, an information network model and a bidirectional connection model. The distribution network model abstracts the substation distribution equipment as nodes and the branch connection lines as edges; the information network model abstracts the sensor communication equipment as nodes and the optical fiber as edges; based on the distribution network model and the information network model, the bidirectional connection model is a bidirectional network architecture and model in which the power node provides energy flow to the information node and the information node provides information flow to the power node;

[0008] Step 2: Generate a set of distribution network damage scenarios under large-scale extreme disasters through Monte Carlo simulation method, establish the scenario basis of distribution network state evolution in the current period, and obtain the data set required for statistical analysis;

[0009] Step 3: Establish an optimal load shedding model that takes into account the system resources of distributed units and energy storage equipment. The model aims to minimize system load loss, takes into account the importance of loads, and considers distribution network constraints, information network constraints, and two-network coupling constraints;

[0010] Step 4: Use the Gurobi solver to solve the bidirectional dynamic model and the optimal load shedding model of the distribution network cyber-physical system, calculate the resilience evaluation index to quantitatively analyze the resilience of the power grid, and use the TOPSIS analysis based on the entropy weight method to obtain the comprehensive coefficient according to the node vulnerability reflected in the evaluation results and the importance of the nodes, and solve the optimal load shedding model, and compare and analyze the system resilience before and after the weighted coefficient is corrected;

[0011] Step 5: To effectively guide the recovery of the cyber-physical system of the distribution network, mobile emergency power supplies and mobile communication vehicles are used as controllable emergency power supplies for the power grid and local emergency wireless communications for the information network, respectively, to carry out post-disaster maintenance of the cyber-physical system of the distribution network, and resilience indicators are used to quantify the resilience improvement effect.

[0012] Furthermore, in step 1, a bidirectional dynamic model of the information-physical system of the distribution network is established based on complex network theory, including a distribution network model, an information network model and a bidirectional connection model, as follows:

[0013] The distribution network model abstracts the substation and generator distribution equipment as nodes, and the branch connection lines and circuit breakers as edges. Characterize the distribution network, the distribution network node set , edge set and the node adjacency matrix As shown below:

[0014] ;

[0015] ;

[0016] A P = [ A P , 1 , 1 A P , 1 , 2 ⋯ A P , 1 , N A P , 2 , 1 ⋱ ⋯ ⋮ ⋮ ⋮ ⋱ ⋮ A P , N , 1 ⋯ ⋯ A P , N , N ] ;

[0017] in, is the number of distribution network nodes; if the node and nodes If there is an edge between , otherwise it is 0; if And node and nodes If there is an edge between , otherwise it is 0;

[0018] The information network model abstracts sensors, data storage devices and communication devices as nodes, and abstracts cables and optical fibers as edges. Representation of information network, information network node collection , edge set and the node adjacency matrix As shown below:

[0019] ;

[0020] ;

[0021] A C = [ A C , 1 , 1 A C , 1 , 2 ⋯ A C , 1 , M A C , 2 , 1 ⋱ ⋯ ⋮ ⋮ ⋮ ⋱ ⋮ A C , M , 1 ⋯ ⋯ A C , M , M ] ;

[0022] in, is the number of nodes in the information network; if the node and nodes If there is an edge between , otherwise it is 0; if And node and nodes If there is an edge between , otherwise it is 0;

[0023] The support matrix for the energy flow provided by the power node to the information node is as follows:

[0024] V PC = [ V PC , 1 , 1 V PC , 1 , 2 ⋯ V PC , 1 , M V PC , 2 , 1 ⋱ ⋯ ⋮ ⋮ ⋮ ⋱ ⋮ V PC , N , 1 ⋯ ⋯ V PC , N , M ] ;

[0025] Among them, if the distribution network node Information Network Node Provide energy flow, then , otherwise it is 0;

[0026] The support matrix of information flow provided by information nodes to power nodes is as follows:

[0027] V CP = [ V CP , 1 , 1 V CP , 1 , 2 ⋯ V CP , 1 , N V CP , 2 , 1 ⋱ ⋯ ⋮ ⋮ ⋮ ⋱ ⋮ V CP , M , 1 ⋯ ⋯ V CP , M , N ] ;

[0028] Among them, if the information network node Can be sent to the distribution network node Provide information flow, then , otherwise it is 0;

[0029] If the distribution network and the information network are bidirectionally coupled, then , if the two networks are not bidirectionally coupled, then and Determine them according to the actual situation; further, this bidirectional connection model is: the topological matrix of the distribution network information physical system:

[0030] V CPS = [ A P V PC V CP A C ] .

[0031] Furthermore, in step 2, a scenario basis for the evolution of the distribution network state in the current period is established, wherein the Monte Carlo simulation method generates a distribution network damage scenario set as follows:

[0032] On the basis of known failure probability, the component status of the system is determined by random number sampling of the probability of component status, and then the status of all components is combined to obtain the system status; without considering the mutual influence between component status, we have:

[0033] ;

[0034] in, Indicates The status of each component; For the The failure probability of each component; To obey A uniformly distributed random number; The state vector of a system composed of elements is:

[0035] ;

[0036] Furthermore, in step 3, an optimal load shedding model considering the importance of load is established, and the model objective is to minimize the system load loss. The objective function is as follows:

[0037] ;

[0038] in, is the time index; is the total cost of system operation; is the loss of load cost;

[0039] Lost load cost As shown below:

[0040] ;

[0041] in, is the distribution network node index; A collection of devices that connect to the node; The unit load loss penalty coefficient reflects the importance of the load; is the load loss power.

[0042] Furthermore, in step 3, an optimal load shedding model considering the importance of load is established. The system contains distributed units and energy storage equipment resources. The model considers distribution network constraints, information network constraints and two-network coupling constraints, wherein the distribution network constraints are specifically as follows:

[0043] Node power balance constraints:

[0044] ;

[0045] ;

[0046] ;

[0047] ;

[0048] in, Node is the set of all child nodes of the starting node; is a collection of lines in the distribution network; and They are Time Node is the active power and reactive power of the line at the starting node; and They are Time power lines Active power and reactive power; and They are The active power and reactive power transmitted from the upper power grid to the distribution network at all times; and They are Time Node Predicted values ​​of active and reactive loads at the location; for Always connected to the system node The active power of the load removed at the location is the active load loss; and They are Time Node The active and reactive output of the gas generating units; and They are Time Node Active power output and reactive power output of wind turbines; and They are Time Node The charging and discharging power of the energy storage device; and They are Time Node Active and reactive output of mobile emergency power supply; and Represents the minimum and maximum values ​​of the variable respectively; binary variable Indicates power lines exist The state of the moment, if , it means the line If it is not damaged, it is in a closed state. Otherwise, it means the line is damaged or in an open state.

[0049] AC power flow constraints:

[0050] ;

[0051] ;

[0052] in, and They are Time Node and nodes Voltage; is the reference node voltage; is the line resistance; is the line reactance, is a sufficiently large number;

[0053] Loss of load upper and lower limit constraints:

[0054] ;

[0055] Distributed gas unit constraints:

[0056] ;

[0057] ;

[0058] ;

[0059] ;

[0060] ;

[0061] ;

[0062] ;

[0063] ;

[0064] in, and Node The minimum and maximum output of the gas unit at the location; binary variable Representation Node Gas units at Working status at all times, if , it means that the gas unit is The machine starts at the same time, otherwise it stops. and Gas units The climbing rate and the descending rate; and They are Start-up and shutdown counting time of gas generator sets at all times; and Node Minimum startup and shutdown time of the unit; and They are the heat consumed when the unit is turned on and off once;

[0065] Distributed wind turbine constraints:

[0066] ;

[0067] ;

[0068] in, for Time Node The predicted output of wind turbines; and Node The minimum and maximum values ​​of the reactive output of the wind turbines;

[0069] Energy storage constraints:

[0070] ;

[0071] ;

[0072] ;

[0073] ;

[0074] ;

[0075] Among them, the binary variable and Respectively Time Node The energy storage device is in the charging and discharging state; and They are Time Node The charging and discharging power of the energy storage; and Node The minimum and maximum values ​​of the energy storage device charging; and Node The minimum and maximum values ​​of the energy storage device discharge; for Time Node The capacity of the energy storage; and Node The minimum and maximum values ​​of energy storage; and are the charging and discharging efficiency of the energy storage device respectively;

[0076] Topological constraints:

[0077] ;

[0078] ;

[0079] ;

[0080] in, is the total number of nodes in the distribution network; binary variable Representation Node Is it a root node? , then the node It is the root node, otherwise it is not; and The nodes are Active power of virtual branches acting as receiving and sending ends;

[0081] Furthermore, in step 3, an optimal load shedding model considering the importance of load is established. The system contains distributed units and energy storage equipment resources. The model considers distribution network constraints, information network constraints and two-network coupling constraints, wherein the information network constraints are as follows:

[0082] Node traffic balance constraints:

[0083] ;

[0084] in, It is a collection of nodes of the information network; Index of communication nodes; is the communication source node index; An index for a fiber optic communication link; and Information Link The sending and receiving ends; A node index for configuring a wireless communication vehicle; and Represents wireless communication links The sending and receiving ends; is the information flow on the source node; is the information flow on the information link; Indicates whether the communication node is connected to the control master station. , then it means Time communication node There is a connection with the control master station, otherwise there is no connection; Represents the information flow of wireless communication; the number 1 represents the demand of each node;

[0085] Communication link availability constraints:

[0086] ;

[0087] ;

[0088] ;

[0089] in, is the total number of communication nodes in the information network; Indicates whether the communication link is available. , it means the communication link fails, otherwise it means the communication link is normal.

[0090] Furthermore, in step 3, an optimal load shedding model considering the importance of load is established. The system contains distributed units and energy storage equipment resources. The model considers distribution network constraints, information network constraints and two-network coupling constraints, wherein the two-network coupling constraints are specifically as follows:

[0091] Distributed unit coupling constraints:

[0092] ;

[0093] ;

[0094] ;

[0095] ;

[0096] in, and They are the control states of the corresponding communication nodes on the gas generator sets and wind turbine sets in the distribution network; Indicates the communication node information requirements for controlling gas turbines or wind turbines. , then it means When the communication node is connected to the control master station, the corresponding unit control state variable value is 1, that is, the unit is in a controllable state, otherwise it is uncontrollable;

[0097] Energy storage device coupling constraints:

[0098] ;

[0099] ;

[0100] ;

[0101] in, The control status of the corresponding communication node on the charging and discharging of the distribution network energy storage equipment; Indicates the communication node information requirements for controlling energy storage devices. If , then it means When the communication node is connected to the control master station, the corresponding energy storage control state variable value is 1, which means it is controllable, otherwise it is uncontrollable.

[0102] Furthermore, the toughness evaluation index established in step 4 is as follows:

[0103] Load shedding probability :

[0104] ;

[0105] in, The number of sampling times for Monte Carlo simulation; For Node The number of samples that experienced load shedding during the study period;

[0106] Average node load loss level :

[0107] ;

[0108] in, is the set of all nodes in the distribution network; for Always connected to the system node The active power of the load removed is the active load loss.

[0109] Furthermore, in step 4, according to the node vulnerability reflected by the evaluation result, the weighting coefficient is corrected by using the TOPSIS analysis based on the entropy weight method. The TOPSIS analysis steps based on the entropy weight method are specifically:

[0110] Step 4.1: The TOPSIS model is generally calculated on the basis that all indicators are extremely large indicators. Therefore, the type of each indicator is first determined, and the data is positively converted according to the indicator type; load shedding probability and the average load loss level They are all extremely large indicators. The larger the value, the higher the vulnerability of the load node. Prioritizing the protection of nodes with high vulnerability is conducive to reducing the degree of system damage. The load loss penalty coefficient based on load classification , which reflects whether the impact of the node abnormality on social life is significant. It is also a very large indicator. The vulnerability and load loss penalty coefficient comprehensively represent the importance of the node in the system. The matrix obtained after the data is forward-transformed is recorded as the forward-transformed matrix :

[0111] X = [ x 1 , 1 x 2 , 1 ⋮ x 32 , 1 x 1 , 2 x 2 , 2 ⋮ x 32 , 2 x 1 , 3 x 2 , 3 ⋮ x 32 , 3 ] ;

[0112] Step 4.2: Since the indicators have different dimensions, standardization is performed to eliminate the impact of the dimensions so that different indicators can be compared and analyzed on the same scale; a standardized matrix is ​​established , Each element in is:

[0113] ;

[0114] Step 4.3: Calculate the Under the indicator The proportion of samples and regard it as the probability used in the relative entropy calculation , the matrix formed by the above probabilities is recorded as the probability matrix , ;

[0115] Step 4.4: Assumptions Indicates an event Something that might happen, represents the probability of this happening, and defines ,because ,so ; If the event The possible situations are: , define the event The information entropy is , and its calculation formula is H ( x ) = ∑ i = 1 n [ p ( x i ) I ( x i )] =− ∑ i = 1 n [ p ( x i )ln( p ( x i ))] , the essence of information entropy is the expected value of the amount of information; when hour, Take the maximum value, then , so that the information entropy is located at [ 0 , 1 ] In the interval, Divide by ; then for the For each indicator, the calculation formula of information entropy is: , since there may be a value of 0 in the probability matrix, Set to 0;

[0116] Step 4.5: Calculate the information utility value The information utility value reflects the distinguishing ability of the indicator, that is, the larger the information utility value, the stronger the distinguishing ability of the indicator, the more information it carries, the greater its contribution to the comprehensive evaluation, and therefore the greater its weight; according to the formula, we can also know that information entropy The larger the The less information an indicator has;

[0117] Step 4.6: Normalize the information utility value and calculate the entropy weight of each indicator ;

[0118] Step 4.7: Define the maximum value for each indicator and minimum value ,Right now , , , define The distance between each node and the maximum value , and the distance to the minimum ,in, are the elements of the normalized matrix;

[0119] Step 4.8: Calculate the scores before and after normalization; when not normalized, The score of the node ,and , The bigger The smaller it is, that is, the closer it is to the maximum value, the higher the node importance score is; in order to eliminate the influence of different dimensions and numerical ranges, ensure the fairness and comparability of each indicator in decision-making, improve the accuracy of calculation and the robustness of the model, the score is normalized .

[0120] Furthermore, in step 5, the mobile emergency power supply and the mobile communication vehicle are used as the controllable emergency power supply of the power grid and the local area emergency wireless communication of the information network respectively. The constraints satisfied by the two emergency resources are as follows:

[0121] Mobile emergency power restraints:

[0122] ;

[0123] ;

[0124] ;

[0125] in, and They are Time Node Active and reactive output of mobile emergency power supply; and For Node The upper limit of the mobile emergency power supply output; binary variable express Time Node The mobile emergency power supply is configured; The number of mobile emergency power sources in the distribution network;

[0126] Mobile emergency communication vehicle constraints:

[0127] ;

[0128] ;

[0129] in, is the total number of communication nodes in the information network; Representation Node Is there a mobile emergency communication vehicle? If yes, it means it is configured, otherwise it is not configured; represents the information flow of wireless communication; The number of emergency communication vehicles equipped in the information network.

[0130] The beneficial effects of the present invention are:

[0131] (1) This paper uses the Monte Carlo simulation method to randomly generate fault scenarios, avoiding the subjectivity and particularity of artificially setting fault situations, and fully reflects the overall distribution of statistical data by generating a large number of scenarios;

[0132] (2) The proposed optimal load shedding model takes into account the fact that the impact of power outages caused by extreme weather on the public is difficult to fully measure with economic losses. The goal is to minimize system load shedding, and the resilience evaluation index focuses on the load loss of the power system under extreme disasters.

[0133] (3) The information-physical system of the distribution network is equipped with distributed units and energy storage equipment, and can call on mobile emergency power supplies and mobile communication vehicles. It is in line with the current development trend of global energy transformation and the diversification of my country's power system configuration, and provides a solution for building a safe and stable new power system;

[0134] (4) The resilience improvement method takes load priority into consideration and introduces a load loss penalty coefficient as a weight coefficient for the grid load to characterize the importance of the load, giving priority to meeting the power supply needs of important loads and reducing the amount of important load reduction in the power system.

[0135] (5) TOPSIS analysis based on the entropy weight method. The entropy weight method calculates the entropy value of each indicator, assigns weights to the three vulnerability indicators and the original load loss penalty coefficient, and obtains a revised comprehensive importance score, which reduces the bias of subjective weighting and makes the weight allocation more objective. In the face of uncertainty and incomplete information, the TOPSIS model based on the entropy weight method can provide relatively stable and reliable decision support;

[0136] (6) By optimizing the scheduling of controllable resources of the information-physical system of the distribution network under extreme disasters, the problem of severe losses caused by extreme weather to the distribution network is solved, and the resilience of the distribution network to extreme disasters is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0137] Figure 1 It is a flow chart of a method for evaluating the resilience of a power distribution information-physical system under extreme disasters and improving the resilience of multi-source coordination according to the present invention;

[0138] Figure 2 It is a performance change curve diagram of the distribution network under extreme disasters of the present invention;

[0139] Figure 3 The figure is a calculation flow chart of TOPSIS analysis based on the entropy weight method of the present invention;

[0140] Figure 4 The distribution of the load nodes in the distribution network of the present invention before and after weight correction;

[0141] Figure 5 A topological diagram of a cyber-physical system of a distribution network according to an embodiment of the present invention;

[0142] Figure 6 The system active load and reactive load forecast demand curve diagram of Example 1 of the present invention;

[0143] Figure 7 This is a diagram showing the output of the mobile emergency power supply during the dispatching period in Example 4 of the present invention. DETAILED DESCRIPTION

[0144] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0145] Power system resilience has been proposed and applied as a concept to measure the resistance and recovery ability of power grids after severe disturbances. When a line is disconnected or a generator set is lost in extreme weather, if the power generation capacity of the remaining units is redistributed to meet the system load supply and line constraints, then the distribution network does not need to cut load. Otherwise, the system fails and part of the load must be reduced. The output of the remaining units and the power supply of the equipment are adjusted according to the importance of the load to minimize the system loss in the fault state. At this time, the line flow will be redistributed. The coupling of the information network increases the risk of safe operation of the information-physical-physical system of the distribution network. Under the huge destructive power of extreme disasters, unpredictable large-scale failures are prone to cause power supply crises. At the same time, the damage to the communication system will affect the monitoring and control of the power grid, while the maintenance of the communication network will help obtain distribution network information and better control power equipment. Therefore, the deep coupling between the power system and the communication system brings security risks and the lack of a coordinated dispatching mechanism, which needs to be solved urgently.

[0146] The present invention is based on the resilience assessment method of the information-physical system of the distribution network and the construction of a multi-source coordinated optimization scheduling model. Figure 1 is a specific implementation flow chart of the method, such as Figure 1 As shown, unlike the traditional method of directly controlling power equipment without considering the damage to the communication system, the present invention makes full use of resources such as distributed units, energy storage equipment, mobile emergency power supplies, and mobile communication vehicles to coordinate information maintenance of the power and communication coupling network, and comprehensively improve the system resilience. The construction method includes the following steps:

[0147] Step 1: Based on complex network theory, a bidirectional dynamic model of the information-physical system of the distribution network is established. Specifically, a distribution network model with substations, generator sets and other distribution equipment as nodes, and branch connecting lines, circuit breakers, etc. as edges; an information network model with sensors, data storage devices and other communication equipment as nodes, and cables, optical fibers, etc. as edges; and a bidirectional connection network architecture and model in which power nodes provide energy flow to information nodes and information nodes provide information flow to power nodes.

[0148] The distribution network model is shown below:

[0149] Substations, generators and other power distribution equipment are abstracted as nodes, and branch lines, circuit breakers, etc. are abstracted as edges. Characterize the distribution network, the distribution network node set , edge set and the node adjacency matrix As shown below:

[0150] ;

[0151] ;

[0152] A P = [ A P , 1 , 1 A P , 1 , 2 ⋯ A P , 1 , N A P , 2 , 1 ⋱ ⋯ ⋮ ⋮ ⋮ ⋱ ⋮ A P , N , 1 ⋯ ⋯ A P , N , N ] ;

[0153] in, is the number of distribution network nodes; if the node and nodes If there is an edge between , otherwise it is 0; if And node and nodes If there is an edge between , otherwise it is 0.

[0154] The information network model is as follows:

[0155] Sensors, data storage devices and other communication devices are abstracted as nodes, cables, optical fibers, etc. are abstracted as edges, and Representation of information network, information network node collection , edge set and the node adjacency matrix As shown below:

[0156] ;

[0157] ;

[0158] A C = [ A C , 1 , 1 A C , 1 , 2 ⋯ A C , 1 , M A C , 2 , 1 ⋱ ⋯ ⋮ ⋮ ⋮ ⋱ ⋮ A C , M , 1 ⋯ ⋯ A C , M , M ] ;

[0159] in, is the number of nodes in the information network; if the node and nodes If there is an edge between , otherwise it is 0; if And node and nodes If there is an edge between , otherwise it is 0.

[0160] The bidirectional connection network model is as follows:

[0161] The two-way connection between the distribution network and the information network is reflected in the fact that various monitoring and control devices in the power system rely on the communication system to transmit real-time data, that is, the information node provides information flow to the power node; at the same time, the communication system needs the power provided by the power system to maintain operation, that is, the power node provides energy flow to the information node.

[0162] The support matrix for the energy flow provided by the power node to the information node is as follows:

[0163] V PC = [ V PC , 1 , 1 V PC , 1 , 2 ⋯ V PC , 1 , M V PC , 2 , 1 ⋱ ⋯ ⋮ ⋮ ⋮ ⋱ ⋮ V PC , N , 1 ⋯ ⋯ V PC , N , M ] ;

[0164] Among them, if the distribution network node Information Network Node Provide energy flow, then , otherwise it is 0.

[0165] The support matrix of information flow provided by information nodes to power nodes is as follows:

[0166] V CP = [ V CP , 1 , 1 V CP , 1 , 2 ⋯ V CP , 1 , N V CP , 2 , 1 ⋱ ⋯ ⋮ ⋮ ⋮ ⋱ ⋮ V CP , M , 1 ⋯ ⋯ V CP , M , N ] ;

[0167] Among them, if the information network node Can be sent to the distribution network node Provide information flow, then , otherwise it is 0.

[0168] If the distribution network and the information network are bidirectionally coupled, then , if the two networks are not bidirectionally coupled, then and Determine them according to the actual situation. Furthermore, the topology matrix of the distribution network information physical system is as follows:

[0169] V CPS = [ A P V PC V CP A C ] ;

[0170] Step 2: Generate a set of distribution network damage scenarios under large-scale extreme disasters through the Monte Carlo simulation method, and establish a scenario basis for the evolution of the distribution network state in the current period to obtain the data required for statistical analysis.

[0171] The Monte Carlo simulation method generates a set of distribution network damage scenarios as follows:

[0172] Based on the known failure probability, the probability of component status is randomly sampled to determine the status of system components, and then the status of all components is combined to obtain the system status. Without considering the mutual influence between component status, we have:

[0173] ;

[0174] in, Indicates The status of each component; For the The failure probability of each component; To obey Uniformly distributed random numbers.

[0175] Depend on The state vector of a system composed of elements is:

[0176] .

[0177] Step 3: Establish an optimal load shedding model that takes into account distributed units, energy storage equipment and other system resources. The model aims to minimize system load loss, takes into account the importance of loads, and considers distribution network constraints, information network constraints, and two-network coupling constraints.

[0178] The system load loss cost is determined by setting a load loss penalty coefficient and expressing the load priority with a weight coefficient to prioritize the power supply needs of important loads. Adjust resource allocation to minimize the total cost of system load loss and improve system resilience. The optimization objective function is as follows:

[0179] ;

[0180] in, is the time index; is the total cost of system operation; The loss of load cost.

[0181] Lost load cost As shown below:

[0182] ;

[0183] in, is the distribution network node index; A collection of devices that connect to the node; is the unit load loss penalty coefficient; is the load loss power.

[0184] The constraints are based on the distribution network, information network and the coupling of the two networks. In terms of the distribution network, they include node power balance constraints, AC power flow constraints, upper and lower limit constraints of load loss, distributed unit constraints, energy storage constraints and topology constraints; in terms of the information network, they include node flow balance constraints and communication link availability constraints; in terms of the coupling of the two networks, they include distributed unit coupling constraints and energy storage device coupling constraints.

[0185] Node power balance constraints:

[0186] ;

[0187] ;

[0188] ;

[0189] ;

[0190] in, Node is the set of all child nodes of the starting node; is a collection of lines in the distribution network; and They are Time Node is the active power and reactive power of the line at the starting node; and They are Time power lines Active power and reactive power; and They are The active power and reactive power transmitted from the upper power grid to the distribution network at all times; and They are Time Node Predicted values ​​of active and reactive loads at the location; for Always connected to the system node The active power of the load removed at the location is the active load loss; and They are Time Node The active and reactive output of the gas generating units; and They are Time Node Active power output and reactive power output of wind turbines; and They are Time Node The charging and discharging power of the energy storage device; and They are Time Node Active and reactive output of mobile emergency power supply; and Represents the minimum and maximum values ​​of the variable respectively; binary variable Indicates power lines exist The state of the moment, if , it means the line If it is not damaged, it is in a closed state. Otherwise, it means the line is damaged or in an open state.

[0191] AC power flow constraints:

[0192] ;

[0193] ;

[0194] in, and They are Time Node and nodes Voltage; is the reference node voltage; is the line resistance; is the line reactance, is a sufficiently large number.

[0195] Loss of load upper and lower limit constraints:

[0196] ;

[0197] Distributed gas unit constraints:

[0198] ;

[0199] ;

[0200] ;

[0201] ;

[0202] ;

[0203] ;

[0204] ;

[0205] ;

[0206] in, and Node The minimum and maximum output of the gas unit at the location; binary variable Representation Node Gas units at Working status at all times, if , it means that the gas unit is The machine starts at the same time, otherwise it stops. and Gas units The climbing rate and the descending rate; and They are Start-up and shutdown counting time of gas generator sets at all times; and Node Minimum startup and shutdown time of the unit; and The heat consumed when the unit is turned on and off respectively.

[0207] Distributed wind turbine constraints:

[0208] ;

[0209] ;

[0210] in, for Time Node The predicted output of wind turbines; and Node The minimum and maximum reactive output of the wind turbine.

[0211] Energy storage constraints:

[0212] ;

[0213] ;

[0214] ;

[0215] ;

[0216] ;

[0217] Among them, the binary variable and Respectively Time Node The energy storage device is in the charging and discharging state; and They are Time Node The charging and discharging power of the energy storage; and Node The minimum and maximum values ​​of the energy storage device charging; and Node The minimum and maximum values ​​of the energy storage device discharge; for Time Node The capacity of the energy storage; and Node The minimum and maximum values ​​of energy storage; and are the charging and discharging efficiencies of the energy storage device, respectively.

[0218] Topological constraints:

[0219] ;

[0220] ;

[0221] ;

[0222] in, is the total number of nodes in the distribution network; binary variable Representation Node Is it a root node? , then the node It is the root node, otherwise it is not; and The nodes are Active power of the virtual branch acting as the receiving and sending end.

[0223] Node traffic balance constraints:

[0224] ;

[0225] in, It is a collection of nodes of the information network; Index of communication nodes; is the communication source node index; An index for a fiber optic communication link; and Information Link The sending and receiving ends; A node index for configuring a wireless communication vehicle; and Represents wireless communication links The sending and receiving ends; is the information flow on the source node; is the information flow on the information link; Indicates whether the communication node is connected to the control master station. , then it means Time communication node There is a connection with the control master station, otherwise there is no connection; Represents the information flow of wireless communication; the number 1 represents the demand of each node.

[0226] Communication link availability constraints:

[0227] ;

[0228] ;

[0229] ;

[0230] in, is the total number of communication nodes in the information network; Indicates whether the communication link is available. , it means the communication link fails, otherwise it means the communication link is normal.

[0231] Distributed unit coupling constraints:

[0232] ;

[0233] ;

[0234] ;

[0235] ;

[0236] in, and They are the control states of the corresponding communication nodes on the gas generator sets and wind turbine sets in the distribution network; Indicates the communication node information requirements for controlling gas turbines or wind turbines. , then it means When the communication node is connected to the control master station, the corresponding unit control state variable value is 1, that is, the unit is in a controllable state, otherwise it is uncontrollable.

[0237] Energy storage device coupling constraints:

[0238] ;

[0239] ;

[0240] ;

[0241] in, The control status of the corresponding communication node on the charging and discharging of the distribution network energy storage equipment; Indicates the communication node information requirements for controlling energy storage devices. If , then it means When the communication node is connected to the control master station, the corresponding energy storage control state variable value is 1, which means it is controllable, otherwise it is uncontrollable.

[0242] Step 4: Use the Gurobi solver to solve the bidirectional dynamic model and the optimal load shedding model of the distribution network information-physical system, calculate the resilience evaluation index to quantitatively analyze the resilience of the power grid, and use the TOPSIS analysis based on the entropy weight method to obtain the comprehensive coefficient according to the node vulnerability reflected in the evaluation results and the importance of the nodes. Then solve the optimal load shedding model and compare and analyze the system resilience before and after the weighted coefficient is corrected.

[0243] The resilience assessment indicators are as follows:

[0244] After an extreme disaster occurs, the distribution network is very likely to experience mass failures, leading to large-scale power outages and a sharp decline in system performance. During the development of the disaster, the system enters a derated operation state. As the impact of the extreme disaster decreases, the system enters a repair state and gradually returns to its original normal operation state. Figure 2 The “resilience trapezoid” shown in the figure can be used to reflect the performance changes of the power system when it is affected by extreme disasters. Therefore, in addition to the failure of system components due to extreme disasters themselves, they may also be indirectly damaged by group failures. The frequency of load shedding at a node in a Monte Carlo simulation is defined as the load shedding probability :

[0245] ;

[0246] in, The number of sampling times for Monte Carlo simulation; For Node The number of samples that experienced load shedding during the study period.

[0247] Two nodes with the same total load loss may also have different load loss amounts each time due to the large difference in the number of load loss occurrences, that is, once the node is abnormal, the degree of impact on the stable operation of the system is different, and the node resilience level is different. Therefore, for the total load loss amount of the node, it is necessary to find the average value according to the number of load sheddings in all Monte Carlo simulations to calculate the average load loss level of each node. :

[0248] ;

[0249] in, is the set of all nodes in the distribution network; for Always connected to the system node The active power of the load removed is the active load loss.

[0250] Resilience assessment helps identify the vulnerable links and potential risks of the system, and provides scientific basis and direction for decision makers to prioritize the most critical risks and vulnerabilities to improve the overall resilience of the system. In the post-disaster recovery phase, it is necessary to focus on ensuring the power supply of important loads such as hospitals and government agencies. To this end, each node is given a weight according to its importance, which is defined as the unit load loss penalty coefficient , and then combined with the above two resilience indicators reflecting node vulnerability, a new weighting coefficient is obtained.

[0251] In cyber-physical systems, the power network and the information network are interdependent and mutually influential. Failures in information lines and equipment lead to changes in network topology, which may cause information transmission delays and losses, disrupt the normal monitoring of power equipment, and thus fall into a "blind adjustment" state. Abnormalities in the power grid will also interfere with the stable operation of the information network. Therefore, the controllability of the information network, that is, whether the information node is connected to the control center, affects the system resilience to a certain extent.

[0252] The present invention adopts TOPSIS analysis based on entropy weight method to correct the weighted coefficient, and obtains a comprehensive index by node load shedding probability, average load loss level and unit load loss penalty coefficient. Figure 3 The analysis flow shown modifies the weighting coefficients.

[0253] The comprehensive coefficient is obtained by TOPSIS analysis based on the entropy weight method. The specific steps are as follows:

[0254] First, the TOPSIS model is generally calculated on the basis that all indicators are extremely large indicators, so the type of each indicator is first determined, and the data is positively transformed according to the indicator type. and the average load loss level These are all extremely large indicators. The larger the value, the higher the vulnerability of the node. Prioritizing nodes with high vulnerability is conducive to reducing the degree of system damage. The load loss penalty coefficient based on load classification , which reflects whether the impact of the node abnormality on social life is significant. It is also a very large indicator. The vulnerability and load loss penalty coefficient comprehensively represent the importance of the node in the system. The matrix obtained after the data is forward-transformed is recorded as the forward-transformed matrix :

[0255] X = [ x 1 , 1 x 2 , 1 ⋮ x 32 , 1 x 1 , 2 x 2 , 2 ⋮ x 32 , 2 x 1 , 3 x 2 , 3 ⋮ x 32 , 3 ] ;

[0256] Since the indicators have different dimensions, they are standardized to eliminate the impact of the dimensions so that different indicators can be compared and analyzed on the same scale. , Each element in ;

[0257] Secondly, calculate the Under the indicator The proportion of samples and regard it as the probability used in the relative entropy calculation , the matrix formed by the above probabilities is recorded as the probability matrix , ;

[0258] Assumptions Indicates an event Something that might happen, represents the probability of this happening, and defines ,because ,so If the event The possible situations are: , define the event The information entropy is , and its calculation formula is H ( x ) = ∑ i = 1 n [ p ( x i ) I ( x i )] =− ∑ i = 1 n [ p ( x i )ln( p ( x i ))] , the essence of information entropy is the expected value of the amount of information. hour, Take the maximum value, then , so that the information entropy is located at [ 0 , 1 ] In the interval, Divide by . Then for the For each indicator, the calculation formula of information entropy is: , since there may be a value of 0 in the probability matrix, Set to 0;

[0259] Next, calculate the information utility value The information utility value reflects the distinguishing ability of the indicator, that is, the larger the information utility value, the stronger the distinguishing ability of the indicator, the more information it carries, the greater its contribution to the comprehensive evaluation, and therefore the greater its weight. According to the formula, information entropy The larger the The less information an indicator has;

[0260] Then, the information utility value is normalized and the entropy weight of each indicator is calculated. ;

[0261] Define the maximum value corresponding to each indicator and minimum value ,Right now , , , define The distance between each node and the maximum value , and the distance to the minimum ,in, are the elements of the normalized matrix;

[0262] Finally, calculate the scores before and after normalization. The score of the node ,and , The bigger The smaller it is, that is, the closer it is to the maximum value, the higher the node importance score is. In order to eliminate the influence of different dimensions and numerical ranges, ensure the fairness and comparability of each indicator in decision-making, and improve the accuracy of calculation and the robustness of the model, the scores are normalized: .

[0263] Step 5: To effectively guide the recovery of the cyber-physical system of the distribution network, mobile emergency power supplies and mobile communication vehicles are used as controllable emergency power supplies for the power grid and local emergency wireless communications for the information network, respectively, to carry out post-disaster maintenance of the cyber-physical system of the distribution network, and resilience indicators are used to quantify the resilience improvement effect.

[0264] Mobile emergency power supply is a reliable and adjustable power supply for the distribution network. When the system is affected by extreme disasters, it can be used as a backup power supply to reduce load shedding. The output constraints and configuration state constraints to be met are as follows:

[0265]

[0266]

[0267]

[0268] in, and They are Time Node Active and reactive output of mobile emergency power supply; and For Node The upper limit of the mobile emergency power supply output; binary variable express Time Node The mobile emergency power supply is configured; is the number of mobile emergency power supplies in the distribution network.

[0269] Mobile emergency communication vehicles can help restore communications, enabling adjacent nodes to communicate wirelessly, thereby helping the power network to mobilize distributed units and energy storage equipment earlier to speed up the recovery of load demand and reduce overall losses. The information restrictions and configuration state constraints to be met are as follows:

[0270]

[0271]

[0272] in, is the total number of communication nodes in the information network; Representation Node Is there a mobile emergency communication vehicle? If yes, it means it is configured, otherwise it is not configured; represents the information flow of wireless communication; The number of emergency communication vehicles equipped in the information network.

[0273] From the perspective of the system as a whole, Figure 2 The load curve in the integral is the system resilience index It is defined as the ratio of the area between the actual load curve and the time axis to the area between the expected load curve and the time axis, reflecting the level of overall load loss in the system:

[0274]

[0275] in, is the mathematical expectation; For the research cycle; This is the system operation curve under normal conditions; It is the system operation curve under extreme disasters.

[0276] The same set of scenarios is set up, and the load shedding loss of the distribution network is calculated under different configurations of the distribution network information-physical system, that is, whether it contains a mobile emergency power supply and a mobile emergency communication vehicle, so as to measure the effectiveness of the resilience improvement method.

[0277] Examples:

[0278] Based on the IEEE 33-node network, by investigating historical data, it is assumed that the line failure probability under extreme disasters is , the faulty line is likely to cause insufficient load supply to its neighboring nodes. Due to the close connection within the system, cascading faults are easy to propagate in the network and expand the scope of the fault. Therefore, the abnormality of the front-end node will also threaten the normal operation of the back-end node. The sampling number is set to 1000 times, and a series of scenarios are generated by the Monte Carlo simulation method based on random sampling to simulate the damage of the distribution network under extreme disasters. With the goal of minimizing the sum of the weighted loss of the system load, the optimal load shedding model is solved, and the resilience of the distribution network is evaluated according to the statistical data results. The evaluation results are shown in Table 1. It is noted that the distribution network nodes 6 and 12 equipped with gas units have a lower load shedding probability than other nodes.

[0279] Table 1. Toughness evaluation results under original load weight coefficient

[0280]

[0281] The results of resilience assessment reflect the vulnerability of the nodes. It is believed that when different load nodes are affected under the same external conditions, the greater the possibility and degree of the affected nodes, the higher the vulnerability, and vice versa. Combined with the results of vulnerability analysis, the original load weight coefficient is revised to obtain a new comprehensive load index. The TOPSIS analysis based on the entropy weight method is used to process the indicator matrix and solve and determine the weight of each indicator, as shown in Table 2. In order to eliminate the influence of different dimensions and numerical ranges, ensure the fairness and comparability of each indicator in decision-making, improve the accuracy of calculation and the robustness of the model, the scores are normalized, and the scores of each node are shown in Table 3. Before and after the weighting coefficient is corrected, that is, the comparison between the unit load loss penalty coefficient that only reflects the importance and the comprehensive coefficient is shown in Table 3. Figure 4 As shown in the figure, in order to intuitively show the difference between the two coefficients, they have been processed into [ 0 . 1 , 1 ] For load nodes with higher comprehensive indicators, such as nodes 24, 25, and 32, it means that these nodes play a greater role in ensuring social stability and reducing system load shedding, and corresponding reinforcement and maintenance measures should be taken when necessary.

[0282] Table 2. Evaluation index weights

[0283]

[0284] Table 3. TOPSIS evaluation results

[0285]

[0286] After the weighted coefficient of load importance is corrected, the fault scenario set generated by Monte Carlo simulation is used as the initial scenario again to solve the optimal load shedding model. The original weight coefficient of load classification is multiplied by the load shedding of the corresponding node to obtain the system weighted load loss that can reflect the level of important load protection. The comparison of the total load shedding of the system before and after correction is shown in Table 4. Without using any emergency resources, the average system load loss is reduced by , the load shedding of important loads has been reduced, indicating that taking active preventive measures such as reinforcement of nodes with high scores before extreme events can improve the overall resilience of the system.

[0287] Table 4. Comparison of system average load shedding and total load shedding weighted by important coefficients before and after coefficient correction

[0288]

[0289] Using the improved IEEE 33-node distribution network model, the coupling relationship between various power supply parameters and their locations and the distribution network and information network corresponding to the equipment is shown in Table 5, and the coupling relationship with the 21-node information network is shown in Table 5. Figure 5 As shown, the dotted line in the figure represents the coupling relationship between the power system and the communication system.

[0290] Table 5. Parameters of various power sources and their corresponding coupling relationships between distribution network and information network

[0291]

[0292] Extreme disasters attack both the distribution network and the information network. Assume that the fault line of the distribution network is , , , , and The fault line of the information network is , , , , and The specific distribution is as follows: Figure 5 As shown in the figure, the system's active load and reactive load demand curves are as follows: Figure 6 shown.

[0293] In order to study the damage and resilience level of the information-physical system of the distribution network under extreme disasters and verify the feasibility of the resilience improvement plan, four case scenarios are established, as shown in Table 6. The factors considered in the examples are represented by √, and vice versa. Specifically, Example 1 is a basic scenario without taking any measures; Example 2 considers the node vulnerability combined with the load importance as a comprehensive indicator to provide a reference for taking active preventive measures such as strengthening grid construction before the disaster; Example 3 uses a mobile communication vehicle as a local emergency wireless communication; Example 4 adds a mobile power supply vehicle as an emergency power supply. Examples 3 and 4 are both emergency recovery after the fault occurs. Examples 2-4 divide the resilience improvement into two situations: pre-disaster prevention and post-disaster recovery from a time perspective, and the factors considered in post-disaster recovery are from few to many, so as to study the effect of various emergency resources on improving the system operation status.

[0294] Table 6. Considerations for each scenario

[0295]

[0296] Considering different emergency resource elements, in the disaster scenario where both the power lines and information lines are damaged, the optimal load shedding scheme is solved. Table 7 shows the comparison of the total system load shedding, the average total load shedding weighted by importance coefficients, and the system resilience of Examples 1-4.

[0297] Table 7. System load loss and toughness comparison of examples 1-4

[0298]

[0299] From the analysis of Example 1, we can see that if we do not pay attention to nodes with high vulnerability before the disaster and do not take any measures after the failure, the damaged communication nodes will not be able to control the corresponding distributed units and energy storage equipment in the distribution network. The system recovery degree is limited and it is difficult to support the operation needs of the system after extreme disasters. The average load loss of the system during the dispatch period is as high as 100%. If nodes with higher comprehensive indicators can be assessed before a disaster and corresponding measures can be taken, the system load loss will be reduced and the resilience level will be improved.

[0300] Example 3 considers the use of mobile communication vehicles on the basis of Example 1. The closer the information network node is to the control center, the wider the communication range it affects, and the more important it is to the recovery of the faulty system. Two mobile communication vehicles are configured at information network nodes 3 and 14. After the fault occurs, the mobile communication vehicle located at communication node 3 helps to restore local communication, and wireless communication can be achieved between nodes 2, 3, 4 and 19; the wireless communication vehicle located at communication node 14 helps to restore communication between nodes 1 and 14. Since the communication network is a ring network structure, all nodes in the network quickly restore communication needs, so that the distributed units and energy storage equipment in the power grid are in a controllable state, and can adjust the working state according to the actual situation to respond to load demand and reduce load losses.

[0301] Mobile emergency power supply is a flexible emergency power supply. It is often used as a stable and reliable backup power supply when extreme disasters occur to reduce system operation losses. Considering that the load node at the end of the line is the farthest from the power supply, the line between the two is the longest, and the possibility of failure is the highest, at the same time, it can be seen from the resilience evaluation results that the probability of load shedding at nodes 30-33 is the highest. Therefore, in Example 4, mobile power supply vehicles are added at distribution network nodes 22 and 33, which can be flexibly adjusted according to actual load demand. The active output change curve during the dispatching period is as follows: Figure 7 As shown, the system resilience is significantly improved, further improving the system operation.

[0302] The present invention aims at the challenges of distribution network resisting disaster disturbances under extreme weather conditions. Based on the information-physical system of the distribution network, the present invention establishes a coordinated optimization scheduling model containing multiple power sources, minimizes system load shedding under the consideration of information network coupling, and meets the power demand of important loads in combination with load priority, thereby reducing the removal of important loads in the system. Through case analysis based on relevant data of power networks and communication networks, it is found that considering information network coupling and utilizing resources such as mobile emergency power supplies and communication vehicles can effectively improve system resilience in scenarios where lines are damaged by extreme disasters, reduce the huge losses caused by disasters, and protect social stability and residents' lives, thus verifying the feasibility of the present invention.

[0303] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art shall not deviate from the spirit and scope of the present invention, and shall be within the scope of protection of the claims attached to the present invention.

Claims

1. A method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters, characterized in that: The following steps are involved: Step 1: Based on complex network theory, a bidirectional dynamic model of the information-physical system of the distribution network is established, including: a distribution network model, an information network model and a bidirectional connection model. The distribution network model abstracts the substation distribution equipment as nodes and the branch connection lines as edges; the information network model abstracts the sensor communication equipment as nodes and the optical fiber as edges; based on the distribution network model and the information network model, the bidirectional connection model is a bidirectional network architecture and model in which the power node provides energy flow to the information node and the information node provides information flow to the power node; Step 2: Generate a set of distribution network damage scenarios under large-scale extreme disasters through Monte Carlo simulation method, establish the scenario basis of distribution network state evolution in the current period, and obtain the data set required for statistical analysis; Step 3: Establish an optimal load shedding model that takes into account the system resources of distributed units and energy storage equipment. The model aims to minimize system load loss, takes into account the importance of loads, and considers distribution network constraints, information network constraints, and two-network coupling constraints; Step 4: Use the Gurobi solver to solve the bidirectional dynamic model and the optimal load shedding model of the distribution network cyber-physical system, calculate the resilience evaluation index to quantitatively analyze the resilience of the power grid, and use the TOPSIS analysis based on the entropy weight method to obtain the comprehensive coefficient according to the node vulnerability reflected in the evaluation results and the importance of the nodes, and solve the optimal load shedding model, and compare and analyze the system resilience before and after the weighted coefficient is corrected; Step 5: Use the mobile emergency power supply and mobile communication vehicle as the controllable emergency power supply for the power grid and the local emergency wireless communication for the information network respectively, carry out post-disaster maintenance on the information-physical system of the distribution network, and use resilience indicators to quantify the resilience improvement effect.

2. The method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters according to claim 1, characterized in that: In step 1, a bidirectional dynamic model of the information-physical system of the distribution network is established based on complex network theory, including a distribution network model, an information network model and a bidirectional connection model, as follows: The distribution network model abstracts the substation and generator distribution equipment as nodes, and the branch connection lines and circuit breakers as edges. Characterize the distribution network, the distribution network node set , edge set and the node adjacency matrix As shown below: ； ； ； in, is the number of distribution network nodes; if the node and nodes If there is an edge between , otherwise it is 0; if And node and nodes If there is an edge between , otherwise it is 0; The information network model abstracts sensors, data storage devices and communication devices as nodes, and abstracts cables and optical fibers as edges. Representation of information network, information network node collection , edge set and the node adjacency matrix As shown below: ； ； ； in, is the number of nodes in the information network; if the node and nodes If there is an edge between , otherwise it is 0; if And node and nodes If there is an edge between , otherwise it is 0; The support matrix for the energy flow provided by the power node to the information node is as follows: ； Among them, if the distribution network node Information Network Node Provide energy flow, then , otherwise it is 0; The support matrix of information flow provided by information nodes to power nodes is as follows: ； Among them, if the information network node Can be sent to the distribution network node Provide information flow, then , otherwise it is 0; If the distribution network and the information network are bidirectionally coupled, then , if the two networks are not bidirectionally coupled, then and Determine them according to the actual situation; further, this bidirectional connection model is: the topological matrix of the distribution network information physical system: 。 3. The method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters according to claim 1, characterized in that: In step 2, a scenario basis for the evolution of the distribution network state in the current period is established, wherein the Monte Carlo simulation method generates a distribution network damage scenario set as follows: With known failure probability, we randomly sample the probability of component states to determine the state of system components, and then combine the states of all components to get the system state; without considering the mutual influence between component states, we have: ； in, Indicates The status of each component; For the The probability of failure of a component; To obey A uniformly distributed random number; The state vector of a system composed of elements is: 。 4. The method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters according to claim 1, characterized in that: In step 3, an optimal load shedding model considering the importance of load is established. The model objective is to minimize the system load loss. The objective function is as follows: ； in, is the time index; is the total cost of system operation; is the loss of load cost; Lost load cost As shown below: ； in, is the distribution network node index; A collection of devices that connect to the node; The unit load loss penalty coefficient reflects the importance of the load; is the load loss power.

5. The method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters according to claim 1, characterized in that: In step 3, an optimal load shedding model considering the importance of load is established. The system contains distributed units and energy storage equipment resources. The model considers distribution network constraints, information network constraints and two-network coupling constraints. The distribution network constraints are as follows: Node power balance constraints: ； ； ； ； in, Node is the set of all child nodes of the starting node; is a collection of lines in the distribution network; and They are Time Node is the active power and reactive power of the line at the starting node; and They are Time power lines Active power and reactive power; and They are The active power and reactive power transmitted from the upper power grid to the distribution network at all times; and They are Time Node Predicted values ​​of active and reactive loads at the location; for Always connected to the system node The active power of the load removed at the location is the active load loss; and They are Time Node The active and reactive output of the gas generating units; and They are Time Node Active power output and reactive power output of wind turbines; and They are Time Node The charging and discharging power of the energy storage device; and They are Time Node Active and reactive output of mobile emergency power supply; and Represents the minimum and maximum values ​​of the variable respectively; binary variable Indicates power lines exist The state of the moment, if , it means the line If it is not damaged, it is in a closed state. Otherwise, it means the line is damaged or in an open state.

6. The method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters according to claim 1, characterized in that: In step 3, an optimal load shedding model considering the importance of load is established. The system contains distributed units and energy storage equipment resources. The model considers distribution network constraints, information network constraints and two-network coupling constraints, wherein the information network constraints are as follows: Node traffic balance constraints: ； in, It is a collection of nodes of the information network; Index of communication nodes; is the communication source node index; An index for a fiber optic communication link; and Information Link The sending and receiving ends; A node index for configuring a wireless communication vehicle; and Represents wireless communication links The sending and receiving ends; is the information flow on the source node; is the information flow on the information link; Indicates whether the communication node is connected to the control master station. , then it means Time communication node There is a connection with the control master station, otherwise there is no connection; Represents the information flow of wireless communication; the number 1 represents the demand of each node; Communication link availability constraints: ； ； ； in, is the total number of communication nodes in the information network; Indicates whether the communication link is available. , it means the communication link fails, otherwise it means the communication link is normal.

7. The method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters according to claim 1, characterized in that: In step 3, an optimal load shedding model considering the importance of load is established. The system contains distributed units and energy storage equipment resources. The model considers distribution network constraints, information network constraints and two-network coupling constraints, wherein the two-network coupling constraints are specifically as follows: Distributed unit coupling constraints: ； ； ； ； in, and They are the control states of the corresponding communication nodes on the gas generator sets and wind turbine sets in the distribution network; Indicates the communication node information requirements for controlling gas turbines or wind turbines. , then it means When the communication node is connected to the control master station, the corresponding unit control state variable value is 1, that is, the unit is in a controllable state, otherwise it is uncontrollable.

8. The method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters according to claim 1, characterized in that: The resilience evaluation index established in step 4 is as follows: Load shedding probability : ； in, The number of sampling times for Monte Carlo simulation; For Node The number of samples that experienced load shedding during the study period.

9. The method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters according to claim 1, characterized in that: In step 4, according to the node vulnerability reflected by the evaluation result, the weighting coefficient is corrected by using the TOPSIS analysis based on the entropy weight method. The TOPSIS analysis steps based on the entropy weight method are specifically: Step 4.1: The TOPSIS model is generally calculated on the basis that all indicators are extremely large indicators. Therefore, the type of each indicator is first determined, and the data is positively converted according to the indicator type; load shedding probability and the average load loss level They are all extremely large indicators. The larger the value, the higher the vulnerability of the load node. Prioritizing the protection of nodes with high vulnerability is conducive to reducing the degree of system damage. The load loss penalty coefficient based on load classification , which reflects whether the impact of the node abnormality on social life is significant. It is also a very large indicator. The vulnerability and load loss penalty coefficient comprehensively represent the importance of the node in the system. The matrix obtained after the data is forward-transformed is recorded as the forward-transformed matrix : ； Step 4.2: Since the indicators have different dimensions, standardization is performed to eliminate the impact of the dimensions so that different indicators can be compared and analyzed on the same scale; a standardized matrix is ​​established , Each element in is: ； Step 4.3: Calculate the Under the indicator The proportion of samples and regard it as the probability used in the relative entropy calculation , the matrix formed by the above probabilities is recorded as the probability matrix , ; Step 4.4: Assumptions Indicates an event Something that might happen, represents the probability of this happening, and defines ,because ,so ; If the event The possible situations are: , define the event The information entropy is , and its calculation formula is , the essence of information entropy is the expected value of the amount of information; when hour, Take the maximum value, then , so that the information entropy is located at In the interval, Divide by ; then for the For each indicator, the calculation formula of information entropy is: , since there may be a value of 0 in the probability matrix, Set to 0; Step 4.5: Calculate the information utility value The information utility value reflects the distinguishing ability of the indicator, that is, the larger the information utility value, the stronger the distinguishing ability of the indicator, the more information it carries, the greater its contribution to the comprehensive evaluation, and therefore the greater its weight; according to the formula, we can also know that information entropy The larger the The less information an indicator has; Step 4.6: Normalize the information utility value and calculate the entropy weight of each indicator ; Step 4.7: Define the maximum value for each indicator and minimum value ,Right now , , , define The distance between each node and the maximum value , and the distance to the minimum ,in, are the elements of the normalized matrix; Step 4.8: Calculate the scores before and after normalization; when not normalized, The score of the node ,and , The bigger The smaller it is, that is, the closer it is to the maximum value, the higher the node importance score is; in order to eliminate the influence of different dimensions and numerical ranges, ensure the fairness and comparability of each indicator in decision-making, improve the accuracy of calculation and the robustness of the model, the score is normalized .

10. The method for evaluating and improving the resilience of a power distribution cyber-physical system under disasters according to claim 1, characterized in that: In step 5, the mobile emergency power supply and the mobile communication vehicle are used as the controllable emergency power supply of the power grid and the local emergency wireless communication of the information network respectively. The constraints satisfied by the two emergency resources are as follows: Mobile emergency power restraints: ； ； ； in, and They are Time Node Active and reactive output of mobile emergency power supply; and For Node The upper limit of the mobile emergency power supply output; binary variable express Time Node The mobile emergency power supply is configured; The number of mobile emergency power sources in the distribution network; Mobile emergency communication vehicle constraints: ； ； in, is the total number of communication nodes in the information network; Representation Node Is there a mobile emergency communication vehicle? If yes, it means it is configured, otherwise it is not configured; represents the information flow of wireless communication; The number of emergency communication vehicles equipped in the information network.

Images