A method and system for evaluating power balance risk based on simulation learning

Abstract

The present invention provides a method and system for assessing the risk of power and electricity balance based on simulation learning, and belongs to the technical field of power and electricity balance risk assessment. The technical problem to be solved is: in order to overcome the impact of extreme situations on the power and electricity balance of a power system, resulting in the low efficiency of the current analysis and assessment of the power and electricity balance risk assessment method of the power system, the adopted technical scheme is: inputting the power system's new energy output, energy storage, generator set, and line flow data into a server; the server establishes an operation model of the power system, including: a new energy output model, an energy storage model, a generator set constraint model, and a line flow model; the server establishes a power and electricity balance model for the power system in extreme situations, and calculates the power and electricity imbalance index of the power system in extreme situations; the present invention is applied to the assessment of power and electricity balance risks in a power system.

Description

Technical Field

[0001] The present invention provides a method and system for evaluating power and quantity balance risk based on simulation learning, belonging to the technical field of power and quantity balance risk evaluation. Background Art

[0002] In recent years, with the increasing emphasis on sustainable use of energy, the use of renewable energy for power generation in power systems has developed rapidly and been widely used. However, after renewable energy such as solar energy and wind energy are integrated into the power grid, the unpredictability and volatility brought about by this also affect the power balance of the power system.

[0003] Traditional power system design is mainly based on relatively stable power generation and load patterns. However, with the continuous increase in the proportion of renewable energy, the randomness and uncertainty faced by the system have increased significantly, and the difficulty of balancing supply and demand has increased accordingly. In this context, the risk assessment of power and electricity balance has become increasingly important. The assessment requires quantitative analysis of the potential imbalance between supply and demand in the power system and its risks, so as to help the operators of the power system accurately identify and evaluate possible risk factors, and formulate effective strategies to prevent and deal with power shortages or surpluses. However, the current risk assessment methods for power and electricity balance have defects such as redundant data collection and low efficiency of risk analysis and assessment, which seriously restrict the application and development of renewable energy in the power system and need to be improved. Summary of the invention

[0004] In order to overcome the impact of extreme situations on the power balance of the power system, which leads to the low technical problem of the low efficiency of the current power balance risk assessment method for the power system, the present invention provides a power balance risk assessment method and system based on simulation learning.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a power and quantity balance risk assessment method based on simulation learning, comprising the following assessment steps:

[0006] Step 1: Input the power system’s renewable energy output, energy storage, generator sets, and line flow data to the server;

[0007] Step 2: The server establishes an operation model of the power system, including: new energy output model, energy storage model, generator set constraint model, and line flow model;

[0008] Step 3: The server establishes a power balance model for the power system in extreme situations and calculates the power imbalance index for the power system in extreme situations.

[0009] Step 4: Assess the power balance risk of the power system based on simulation learning.

[0010] The specific method for establishing the operation model of the power system in the second step is:

[0011] The established new energy output models include photovoltaic output model and wind power output model, among which:

[0012] The expression of the photovoltaic output model is:

[0013] ;

[0014] In the formula, P PV is the actual output power of the photovoltaic power station; P s is the rated output power of the photovoltaic power station; I r is the actual solar radiation intensity; T is the actual temperature; I s is the rated solar radiation intensity; T s is the rated temperature; α T is the power temperature coefficient;

[0015] The expression of wind power output model is:

[0016] ;

[0017] In the formula, P W is the output power of the wind power station under the actual wind speed; v r , v c , v e They are cut-in wind speed, cut-out wind speed, and rated wind speed. v is the actual wind speed; A , B , C , D is the wind turbine power fitting parameter; P e is the rated power of the wind power station;

[0018] The expression for establishing the energy storage model is:

[0019] ;

[0020] ;

[0021] ;

[0022] In the formula, SoCi,t For Node i Energy Storage t The state of charge at the moment; S e i,t , P e i,t Node i Energy Storage t The power and power capacity at the moment; β i,t It is a 0-1 variable describing the energy storage state, 1 means charging, and vice versa means discharging; C s is a set of nodes with energy storage devices; P c i,t , P d i,t Node i Energy Storage t Charging and discharging power at all times; or is the charging and discharging efficiency of the energy storage device;

[0023] The expression for establishing the generator set constraint model is:

[0024] ;

[0025] ;

[0026] ;

[0027] ;

[0028] ;

[0029] In the formula, l g min For conventional engines g The minimum output rate; O g,h For conventional engines g Online startup capacity; P g,h For conventional engines g Output power; U g For conventional engines g installed capacity; X g d , X g u Conventional engines gAdjust the climbing rate up and down; O u g,h , O d g,h Represents conventional engine g exist h On / off capacity during time periods; T g on , T g off Represents thermal power units g Minimum on / off time; t Indicates the actual running time;

[0030] The expression for establishing the line power flow model is:

[0031] ;

[0032] In the formula, For Line l The transmission power; , Refers to the line l of built capacity and expansion capacity.

[0033] The specific method of establishing the balance model and calculating the imbalance index in the third step is:

[0034] The calculation formula for determining the extreme situation that leads to low output of new energy is:

[0035] ;

[0036] In the formula, r a d The average daily output of renewable energy; r max Rated output for new energy sources; s is the judgment coefficient; d 1. d 2 are the beginning and end of the low output period respectively; d It is a low output time period;

[0037] When it is determined that extreme conditions lead to continuous low output of new energy, a power balance model is established, and the expression is:

[0038] ;

[0039] In the formula, w A collection of different types of generator sets; g For conventional engines, W For wind power, PV For photovoltaics; P w g,h For generator sets g In extreme scenarios h Hours of output; P s,d s,h , P s,c s,h Energy storage s exist h Hours of charging and discharging power; D l n,h For Node n Load power; D u n,h The amount of load shedding that occurs in extreme weather scenarios; Oh w n , Oh s n Connect to the nodes n Generator sets and energy storage collection; Oh T n , Oh F n Respectively represent the nodes n is the set of transmission lines with starting and ending nodes; For Line l The transmission power;

[0040] Calculate the unbalanced power risk in extreme cases, the calculation formula is:

[0041] ;

[0042] By summing up the load shedding power in all extreme cases, the unbalanced power risk index in extreme cases is obtained. D u .

[0043] The specific method for evaluating the power and quantity balance risk in the fourth step is:

[0044] Defined in node N k When performing variable selection on F k Recording Node N k The states of all candidate variables are defined by decision labels L kRecord the decision of which variables should be selected, variable feature matrix F k Includes: Numerical features of each candidate variable itself R k , and the node where the candidate variable is located N k Location characteristics T k , then the variable feature matrix F k Expressed as F k ={ R k , T k};

[0045] Randomly generate multiple mixed integer programming problems. In the process of solving the generated mixed integer programming problems, a set of ( F k , T k ) to complete data collection;

[0046] The collected data is divided into training set, validation set and test set. The training set and validation set are input into the simulation learning model based on deep neural network. The weight vector is learned by minimizing the cross entropy loss function to evaluate the candidate variables. The expression is:

[0047] ;

[0048] In the formula, D is collected ( F k , T k ) number of pairs; is the output of the deep neural network; If the input is F k ,but Probability of occurrence; L is the cross entropy loss function;

[0049] The learned weight vector is used to evaluate the candidate variables to form a probability distribution, and a variable is selected for branching according to the current probability distribution until the mixed integer programming problem is solved.

[0050] The evaluation system used to implement the power and quantity balance risk evaluation method based on simulation learning includes a server, a data acquisition device, a network device and auxiliary equipment, wherein:

[0051] The server includes a processor for building and running models, memory to support data set processing and model concurrency, and storage devices for storing simulation data, model parameters, and learning results;

[0052] Data collection equipment includes sensors and data loggers for real-time collection of renewable energy output, energy storage status, generator set and line flow data, as well as communication equipment for transmitting the collected data to the server for processing;

[0053] Network equipment includes switches and routers for transmitting data;

[0054] Auxiliary devices include displays, input devices, and uninterruptible power supplies.

[0055] The beneficial effects of the present invention compared with the prior art are as follows: the present invention proposes a method for assessing the risk of power and electricity balance in an electric power system based on simulation learning. The method establishes a power system operation model including a new energy output model, an energy storage model, a generator set constraint model and a line flow model, defines extreme weather scenarios, establishes a power and electricity balance model for a new type of power system facing extreme conditions, proposes power and electricity imbalance indicators for a new type of power system facing extreme conditions, and finally assesses the power and electricity balance risk of the power system based on simulation learning. On the one hand, the assessment method can quantitatively define the power and electricity imbalance, and on the other hand, it introduces a simulation learning method, thereby accelerating the efficiency of power and electricity balance risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The present invention will be further described below in conjunction with the accompanying drawings:

[0057] Figure 1 A flowchart of the steps of the power system power and quantity balance risk assessment method of the present invention;

[0058] Figure 2 This is a schematic diagram of the average solution time of an improved 5-node system in an embodiment of the present invention;

[0059] Figure 3 This is a solution time distribution diagram of the improved 5-node system at T=720h in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] like Figure 1As shown, the present invention provides an assessment method for power balance risk in an electric power system based on simulation learning. The method first establishes an operation model of a new electric power system, which includes a new energy output model, an energy storage model, a generator set constraint model and a line flow model. Then, extreme situations mainly based on extreme weather are considered, extreme weather scenarios are defined, and a power balance model for a new electric power system facing extreme situations is established. Power imbalance indicators for a new electric power system facing extreme situations are proposed, and finally, the power balance risk of the electric power system is assessed based on simulation learning. The power balance risk assessment method proposed in the present invention, on the one hand, quantifies the power imbalance, and on the other hand, introduces a simulation learning method, which speeds up the efficiency of power balance risk assessment and can solve the impact of extreme situations on the power balance of the electric power system.

[0061] Specifically, the power and quantity balance risk assessment method based on simulation learning proposed in the present invention includes the following assessment steps:

[0062] Step 1: Initialization, input data such as the power system's renewable energy output, energy storage, generator sets, and line flows.

[0063] Step 2: Establish an operation model of the power system including a new energy output model, an energy storage model, a generator set constraint model and a line flow model.

[0064] Among them, the new energy output model mainly includes photovoltaic and wind power output models. Among them, the photovoltaic output model is established based on the physical characteristics and environmental conditions of the photovoltaic power generation system. The output of the photovoltaic power generation system is mainly affected by factors such as solar radiation intensity and photovoltaic panel temperature. The photovoltaic output model is:

[0065] ;

[0066] In the formula, P PV is the actual output power of the photovoltaic power station; P s is the rated output power of the photovoltaic power station; I r is the actual solar radiation intensity; T is the actual temperature; I s is the rated solar radiation intensity; T s is the rated temperature; α T is the power temperature coefficient;

[0067] The actual solar radiation intensity data is collected in real time through sensors on photovoltaic panels or equipment such as meteorological stations; the actual temperature data is collected in real time through temperature sensors on the surface temperature of photovoltaic panels; the rated solar radiation intensity, actual output power, and rated output power are all technical parameters of photovoltaic panels, which can be directly obtained according to the photovoltaic panel model.

[0068] The wind power output model is established based on the physical characteristics of the wind power generation system and wind conditions. The output of the wind power generation system is mainly affected by factors such as wind speed and wind direction. The wind power output model is:

[0069] ;

[0070] In the formula, P W is the output power of the wind power station under the actual wind speed; v r , v c , v e They are cut-in wind speed, cut-out wind speed, and rated wind speed. v is the actual wind speed; A , B , C , D is the wind turbine power fitting parameter; P e is the rated power of the wind power station;

[0071] Cut-in wind speed, cut-out wind speed, rated wind speed, rated power, and power fitting parameters are technical parameters of the fan and can be directly obtained according to the fan model; the actual wind speed is obtained in real time through the wind speed sensor.

[0072] The energy storage model is established based on the physical characteristics and operating data of the energy storage device. The output of the energy storage device is mainly affected by factors such as charging and discharging efficiency, capacity limitation, and charging and discharging status. The energy storage model is:

[0073] ;

[0074] ;

[0075] ;

[0076] In the formula, SoC i,t For Node i Energy Storage t The state of charge at the moment; S e i,t , P e i,tNode i Energy Storage t The power and power capacity at the moment; β i,t It is a 0-1 variable describing the energy storage state, 1 means charging, and vice versa means discharging; C s is a set of nodes with energy storage devices; P c i,t , P d i,t Node i Energy Storage t Charging and discharging power at all times; or It is the charging and discharging efficiency of the energy storage device; the power, power capacity, charging and discharging power and charging and discharging efficiency are the technical parameters of the energy storage equipment, which can be directly obtained according to the model of the energy storage equipment; the charging and discharging status is obtained by collecting the voltage, current, temperature and other operating data of the energy storage equipment in real time through sensors to obtain its charging and discharging status.

[0077] The generator set constraint model is established based on the technical characteristics and operating limitations of the generator set. The output of the generator set is mainly affected by factors such as the capacity, ramp rate, start-stop time, etc. The following formulas are the upper and lower limit constraints of the generator output, the generator ramp constraint, the generator startup capacity change process, and the minimum on / off time constraint of the generator. The upper and lower limit constraints of the generator output ensure that the output of the generator set is within a safe range to ensure the stable operation of the system; the generator ramp constraint limits the rate of change of the generator set output to avoid mechanical stress and system instability caused by rapid changes; the generator startup capacity change process can accurately describe the output changes of the thermal power unit in adjacent time periods and reflect the dynamic response characteristics of the thermal power unit; the minimum on / off time constraint ensures that the generator set has enough time to complete preheating and cooling when starting and stopping to extend the equipment life and improve operating efficiency. The generator set constraint model is:

[0078] ;

[0079] ;

[0080] ;

[0081] ;

[0082] ;

[0083] In the formula, l g min For conventional engines g The minimum output rate; Og,h For conventional engines g Online startup capacity; P g,h For conventional engines g Output power; U g For conventional engines g installed capacity; X g d , X g u Conventional engines g Adjust the climbing rate up and down; O u g,h , O d g,h Represents conventional engine g exist h On / off capacity during time periods; T g on , T g off Represents thermal power units g Minimum on / off time; t Indicates the actual running time; the above parameters are the technical parameters of the generator and can be directly obtained from the generator model.

[0084] The network flow model assumes that the power flow of the transmission line can be freely dispatched within the capacity limit. The expansion plan of the transmission line is only reflected in the investment and expansion of the transmission capacity. Therefore, the line flow only considers the capacity factor of the transmission line. The line flow model is:

[0085] ;

[0086] In the formula, For Line l The transmission power; , Refers to the line l The built capacity and expanded capacity are both technical parameters of the transmission line.

[0087] The third step: establish an electric power balance model for the power system facing extreme situations, and calculate the electric power imbalance index for the power system facing extreme situations, wherein the extreme situations involved mainly include environmental conditions such as no wind and no sun for many consecutive days, extremely hot and no wind, and extremely cold and no sun. The present invention specifically judges extreme situations through the continuous low output of new energy.

[0088] In extreme cases, the output of new energy is extremely low. The judgment formula for extreme cases leading to low output of new energy is:

[0089] ;

[0090] In the formula, r a d The average daily output of renewable energy; r max Rated output for new energy sources; s is the judgment coefficient; d 1. d 2 are the beginning and end of the low output period respectively; d This is the low output time period.

[0091] When extreme conditions cause the renewable energy to continue to have low output, an electric power balance model is established, and the node power balance is:

[0092] ;

[0093] In the formula, w A collection of different types of generator sets; g For conventional engines, W For wind power, PV for photovoltaics; P w g,h For generator sets g In extreme scenarios h Hours of output; P s,d s,h , P s,c s,h Energy storage s exist h Hours of charging and discharging power; D l n,h For Node n Load power; D u n,h The amount of load shedding that occurs in extreme weather scenarios;

[0094] Oh w n , Oh s n Connect to the nodes n Generator sets and energy storage collection; Oh T n , OhF n Respectively represent the nodes n is the set of transmission lines with starting and ending nodes; For Line l of transmission power.

[0095] Load shedding is a quantitative indicator of power balance risk. The unbalanced power risk in extreme cases is calculated as follows:

[0096] ;

[0097] By summing up the load shedding power in all extreme cases, the unbalanced power risk index in extreme cases is obtained. D u ;

[0098] The objective function for solving the load shedding power is: minD u The constraints are the operation model of the power system and the node power balance formula. The risk assessment model constructed based on this is a mixed integer linear programming problem.

[0099] Step 4: Assess the power balance risk of the power system based on simulation learning.

[0100] The risk assessment model constructed in the third step is a mixed integer linear programming problem, which requires the use of the B&B algorithm to iteratively partition the solution space to find the optimal solution. The main idea of ​​the B&B algorithm is to divide the solution space of the problem into multiple subspaces, and gradually prune by estimating the upper bound (or lower bound) of the solution of each subspace, excluding subspaces that are unlikely to contain the optimal solution, thereby gradually narrowing the search range. Variable selection is the key to the success of the B&B algorithm, and the selection of variables depends on the quality of candidate variables. Simulation learning selects variables based on sample imitation relative penalty cost branching rules, collects data by randomly generating power balance risk assessment problems, divides the data set to train a simulation learning model based on a deep neural network, learns weight vectors to evaluate candidate variables, and selects variable branches to solve the problem.

[0101] The process of solving the power balance risk problem using simulation learning is as follows. N k For example, the variable feature matrix F k Record N k The states of all candidate variables, decision labels L k Document decisions about which variables should be selected. F k Not only does it include the numerical characteristics of each candidate variable itself Rk , including the node where the candidate variable is located N k Location characteristics T k . F k It can be expressed as F k ={ R k , T k}.

[0102] First, multiple mixed integer programming problems are randomly generated. In the process of solving the generated mixed integer programming problems, a set of ( F k , T k ) to complete the data collection.

[0103] Secondly, the collected data is divided into training set, validation set and test set. The training set and validation set are input into the deep neural network based simulation learning model to learn the weight vector to evaluate the candidate variables by minimizing the cross entropy loss function, as shown below:

[0104] ;

[0105] In the formula, D is collected ( F k , T k ) number of pairs; is the output of the deep neural network; If the input is F k ,but Probability of occurrence; L is the cross entropy loss function.

[0106] Finally, the learned weight vector is used to evaluate the candidate variables to form a probability distribution. According to the current probability distribution, a variable is selected for branching until the mixed integer programming problem is solved. Specifically, the learned weight vector is used instead of the relative penalty cost branching rule to evaluate the candidate variables and form a probability distribution. According to the current probability distribution, the variable to be branched is selected, and then the branch and bound process is continued until the mixed integer programming problem is solved, the minimum load shedding amount is obtained, and the corresponding unbalanced power risk index is calculated. D u .

[0107] The present invention also provides a power and electricity balance risk assessment system based on simulation learning, which mainly includes a server with high-performance computing performance, data acquisition equipment, network equipment and other auxiliary equipment.

[0108] The server needs to have a processor capable of handling complex simulation models and learning algorithms, sufficient memory to support the processing of large data sets and concurrent running of models, and storage devices for storing simulation data, model parameters, and learning results.

[0109] The data acquisition equipment includes sensors and data loggers for real-time collection of data related to renewable energy output, energy storage status, generator sets and line flows, and communication equipment for transmitting the collected data to high-performance computing servers for processing.

[0110] Network equipment includes switches and routers used to ensure network connections for data transmission, and firewalls and network security devices used to protect the security of data transmission.

[0111] Other auxiliary equipment includes displays and input devices for operators to monitor and operate the system, and uninterruptible power supplies to ensure that the system continues to operate during power fluctuations or interruptions.

[0112] like Figure 2 The figure shows the average solution time diagram of the improved 5-node system provided in the embodiment of the present invention. In order to verify the effectiveness of the method of the present invention, it is verified in the improved 5-node system, using the data in the time domain T (68h, 338h, 720h, 1440h) as input, and using the simulation learning method and the relative penalty cost branch rule method to perform the power imbalance risk assessment calculation; the efficiency of solving the power balance risk assessment problem of the proposed method and the existing relative penalty cost branch rule method is compared. The improved 5 nodes include five generators, a wind farm and a solar farm. The total power generation of the system is 3155MW. When T=1440h, the average solution time of the simulation learning method can be reduced by about 40% compared with the relative penalty cost branch rule method, which verifies that the method proposed in the present invention has a higher evaluation efficiency;

[0113] like Figure 3 The figure shows the solution time distribution diagram of the improved 5-node system provided in the embodiment of the present invention at T=720h. It can be seen from the figure that the solution time of the simulation learning method is generally lower than the solution time of the relative penalty cost branch rule method, and the solution time distribution is more concentrated. Compared with the relative penalty cost branch rule method, the proposed method can provide the optimal guarantee or a smaller gap faster, and has a more stable solution time for the problem.

[0114] This is an average solution time diagram of the improved 5-node system provided in an embodiment of the present invention. It is verified in the improved 5-node system, and data in the time domain T (68h, 338h, 720h, 1440h) is used as input. The simulation learning method and the relative penalty cost branch rule method are used to perform power imbalance risk assessment calculation respectively; when T=1440h, the average solution time of the simulation learning method can be reduced by about 40% compared with the relative penalty cost branch rule method, which verifies that the method proposed in the present invention has higher evaluation efficiency.

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for assessing power balance risk based on simulation learning, characterized in that: The evaluation steps include: Step 1: Input the power system’s renewable energy output, energy storage, generator sets, and line flow data to the server; Step 2: The server establishes an operation model of the power system, including: new energy output model, energy storage model, generator set constraint model, and line flow model. The specific method is as follows: The established new energy output models include photovoltaic output model and wind power output model, among which: The expression of the photovoltaic output model is: Where P PV is the actual output power of the photovoltaic power station; P s is the rated output power of the photovoltaic power station; I r is the actual solar radiation intensity; T is the actual temperature; I s is the rated solar radiation intensity; T s is the rated temperature; α T is the power temperature coefficient; The expression of wind power output model is: Where P W is the output power of the wind power station under the actual wind speed; v r 、v c 、v e are cut-in wind speed, cut-out wind speed, and rated wind speed, respectively; v is the actual wind speed; A, B, C, and D are wind turbine power fitting parameters; P e is the rated power of the wind power station; The expression for establishing the energy storage model is: Where, SoC i,t is the state of charge of the energy storage at node i at time t; S e i,t , P e i,t are the energy storage capacity and power capacity of node i at time t; β i,t is a 0-1 variable describing the energy storage state, 1 for charging, and vice versa for discharging; C s is the set of nodes with energy storage devices; P c i,t , P d i,t are the charging and discharging power of the energy storage at node i at time t; η is the charging and discharging efficiency of the energy storage device; the expression for establishing the constraint model of the generator set is: In the formula, λ g min is the minimum output rate of the conventional engine g; g,h is the online startup capacity of the conventional engine g; P g,h is the output power of the conventional engine g; U g is the installed capacity of conventional engine g; X g d , X g u are the upward and downward climbing rates of the conventional engine g respectively; u g,h , O d g,h They represent the on and off capacity of conventional engine g in period h; T g on , T g of They represent the minimum start and stop time of thermal power unit g respectively; τ represents the actual operation time; The expression for establishing the line power flow model is: In the formula, F l，h is the transmission power of line l; U l They refer to the built capacity and expansion capacity of line l respectively; Step 3: The server establishes a power balance model for the power system in extreme situations and calculates the power imbalance index for the power system in extreme situations. The specific method is as follows: The calculation formula for determining the extreme situation that leads to low output of new energy is: In the formula, ρ a d is the average daily output of renewable energy; max is the rated output of new energy; σ is the judgment coefficient; d1 and d2 are the beginning and end of the low output period respectively; δ is the low output time period; When it is determined that extreme conditions lead to continuous low output of new energy, a power balance model is established, and the expression is: In the formula, w is a collection of different types of generator sets; g is a conventional engine, W is wind power, and PV is photovoltaic; P w g,h is the output of generator set g at the hth hour in the extreme scenario; P s,d s,h , P s,c s,h are the charging and discharging power of energy storage s in h hours; D l n,h is the load power of node n; D u n,h is the load shedding power in extreme weather scenarios; Ω w n ,Ω s n are the generator set and energy storage set connected to node n respectively; Ω T n ,Ω F n They represent the set of transmission lines with node n as the starting and ending nodes respectively; is the transmission power of line l; Calculate the unbalanced power risk in extreme cases, the calculation formula is: By summing up the load shedding in all extreme cases, we can get the unbalanced power risk index D in extreme cases. u ; Step 4: Evaluate the power balance risk of the power system based on simulation learning. The specific method is as follows: Defined at node N k When performing variable selection on k Record Node N k The states of all candidate variables are defined by the decision labels L k Record the decision of which variables should be selected, the variable feature matrix F k Includes: the numerical features R of each candidate variable itself k , and the node N where the candidate variable is located k The position feature T k , then the variable feature matrix F k Indicated as F k = {R k ,T k }; Randomly generate multiple mixed integer programming problems. In the process of solving the generated mixed integer programming problems, a set of (F k ,T k ) to complete data collection; The collected data is divided into training set, validation set and test set. The training set and validation set are input into the simulation learning model based on deep neural network. The weight vector is learned by minimizing the cross entropy loss function to evaluate the candidate variables. The expression is: Where D is the collected (F k ,T k ) number of pairs; is the output of the deep neural network; If the input is F k ,but The probability of occurrence; L is the cross entropy loss function; The learned weight vector is used to evaluate the candidate variables to form a probability distribution, and a variable is selected for branching according to the current probability distribution until the mixed integer programming problem is solved.

2. An evaluation system for implementing the power and quantity balance risk evaluation method based on simulation learning as described in claim 1, characterized in that: Including servers, data acquisition equipment, network equipment and auxiliary equipment, including: The server includes a processor for building and running models, memory to support data set processing and model concurrency, and storage devices for storing simulation data, model parameters, and learning results; Data collection equipment includes sensors and data loggers for real-time collection of renewable energy output, energy storage status, generator set and line flow data, as well as communication equipment for transmitting the collected data to the server for processing; Network equipment includes switches and routers for transmitting data; Auxiliary devices include displays, input devices, and uninterruptible power supplies.

Images