Distribution network flexible resource day-ahead robust scheduling method under source load multivariate fluctuation risk

Abstract

The invention discloses a distribution network flexible resource day-ahead robust scheduling method under a source load multivariate fluctuation risk, and belongs to the field of power system optimization operation. The core innovation of the method lies in that an uncertainty set model considering internal and external source load fluctuation risks is constructed, and a two-stage decision-making mechanism and a solving method thereof are designed: resources such as an interconnection switch, an on-load voltage regulation transformer gear, an energy storage charging and discharging mode and the like are pre-allocated in a day-ahead stage; in the second stage, multi-type energy storage and transferable load power scheduling are optimized, and a self-adaptive robust framework coping with the worst scene is formed. On the basis, an optimization model is decomposed into a main problem and a risk stabilization sub-problem, non-convex features brought by endogenous uncertainty are processed by adopting McCormick envelope, and iterative solution is performed through a column constraint generation algorithm. According to the method, through collaborative optimization of source-load double-side flexible resources, wind power, photovoltaic and load multivariate fluctuation risks are effectively stabilized, and the operation economy and flexibility are remarkably improved while the safety robustness of the power distribution network is guaranteed.

Description

A Flexible Resource Day-ahead Robust Scheduling Method for Distribution Networks under Multiple Source-Load Fluctuation Risks Technical Field

[0001] This invention belongs to the field of power system optimization operation, specifically involving an optimized scheduling method for distribution networks, and in particular a flexible resource day-ahead robust scheduling method for distribution networks under the risk of multiple source-load fluctuations. Background Technology

[0002] With the ongoing energy structure transformation, the large-scale integration of distributed power sources (such as photovoltaics and wind turbines) and the emergence of new power loads and consumption patterns have transformed the distribution network from a traditional passive unidirectional radiation network to an active bidirectional interactive system. However, the output of distributed power sources (such as photovoltaics affected by weather) and the actual power demands of new consumption patterns (such as user demand-side response, temperature-controlled loads, post-meter energy storage, rooftop photovoltaics, and electric vehicle charging) exhibit strong randomness, volatility, and uncertainty. This uncertainty on both the source and load sides poses a severe challenge to the safe and stable operation of the distribution system, easily leading to operational risks such as node voltage exceeding limits, line backflow, increased network losses, and transformer overload.

[0003] To address issues such as insufficient carrying capacity and bidirectional transmission capability caused by untimely updates to power distribution system facilities, a large number of flexible and dispatchable resources of various types on both the source and load sides have been incorporated into the power distribution network operation and scheduling to improve the power supply reliability and operational flexibility at the end of the power distribution network and alleviate transmission congestion problems.

[0004] However, the scheduling decisions for some flexible resources in the system (especially demand response flexible resources and mobile flexible resources) inherently possess "endogenous uncertainty." For example, pre-scheduling corresponding resources on the demand side in the first day-ahead phase directly determines the intensity of local load and renewable energy fluctuations it faces in the second day-ahead phase. This superposition of endogenous and exogenous uncertainties leads to a "strong coupling" between random variables and decision variables in the first-stage phase of the scheduling model's sub-problems, exacerbating the difficulty of mitigating renewable resource fluctuation risks and increasing the difficulty of robust day-ahead scheduling decisions for flexible resources. Existing scheduling methods often struggle to handle the non-convex characteristics of strong coupling between random variables and scheduling decisions brought about by the coordinated scheduling of multiple types of flexible resources on both the source and load sides. They lack a unified adaptive robust scheduling framework to simultaneously address the exogenous fluctuation risks of source and load and the endogenous uncertainty risks of flexible resources. Against this backdrop, this invention provides a robust day-ahead scheduling method for flexible resources in distribution networks under multiple source and load fluctuation risks, aiming to overcome the limitations of existing technologies in coordinating multiple types of flexible resources on both the source and load sides to mitigate distribution system operational risks when facing endogenous and exogenous fluctuation risks. Summary of the Invention

[0005] The purpose of this invention is to provide a robust day-ahead scheduling method for flexible resources in distribution networks under the risk of multiple source-load fluctuations. This method can effectively coordinate and allocate multiple types of flexible resources on both the source and load sides, and effectively handle the multiple endogenous and exogenous uncertainties on both the source and load sides of the distribution system. Under the premise of ensuring system safety and robustness, it can improve the economy and flexibility of operation.

[0006] To achieve the above technical objectives, the present invention adopts the following solution:

[0007] A robust method for flexible resource day-ahead scheduling in distribution networks under the risk of multi-source load fluctuations includes the following steps:

[0008] S1. Analyze the source load fluctuation mode and range within the power distribution system, and construct an uncertainty set model to characterize the source load fluctuation risk. The uncertainty set is defined using box inequality constraints.

[0009] S2. In response to the controllable resource response characteristics in the power distribution system, in the first stage (day-ahead), a pre-allocation decision is made on the source-load dual-side multi-type flexible resource operation mode (discrete pre-allocation decision such as tie switch status, on-load tap changer tap position, energy storage charging and discharging mode, substation power transmission mode, etc.). In the second stage, the power scheduling of flexible resources such as multi-type energy storage and multi-type transferable loads is decided, thereby establishing a flexible resource day-ahead adaptive robust scheduling framework to cope with the worst-case scenario under the endogenous and exogenous uncertainty fluctuation risks of renewable resources, load, and ambient temperature.

[0010] S3. Based on the proposed flexible resource day-ahead adaptive robust scheduling framework, construct an adaptive robust optimization model that includes distribution network power flow constraints, source-load dual-side multi-type flexible resource operation scheduling constraints, and operation scheduling objective function.

[0011] S4. For the adaptive robust optimization model that considers multiple endogenous and exogenous uncertainties of source load, analyze its decision mode and model structure, decompose the model into the main problem of pre-scheduling of flexible resource operation mode and the sub-problem of adaptive mitigation of source load fluctuation risk, and use McCormick envelope approximation method to deal with the strong coupling of random variables in the sub-problem caused by endogenous uncertainty.

[0012] S5. Finally, the column constraint generation (C&CG) algorithm is used to iteratively solve the adaptive robust scheduling model established in S1-S4, and obtain a flexible resource day-ahead adaptive robust scheduling scheme that considers the mitigation of multiple fluctuation risks of distribution network sources and loads.

[0013] In step S1 of this invention, a power distribution system with large-scale access to renewable resources such as wind and solar power is considered. First, the deterministic modeling method of each element on the source side and load side of the power distribution system is analyzed. Then, the uncertainty fluctuation mode and range are analyzed, and an uncertain set model representing the source-load fluctuation risk is constructed. The deterministic set is defined by box inequality constraints.

[0014] a) Classification and deterministic modeling methods for loads

[0015] Electrical load classification and characterization methods:

[0016] Electricity load classification and characterization is a fundamental issue in flexible resource day-ahead adaptive robust scheduling methods that consider the multi-source fluctuation risks of distribution network loads. It is mainly divided into load characteristic classification and characterization and adjustable characteristic classification and characterization.

[0017] In terms of load characteristic classification and characterization, the power load data can first be subjected to cluster analysis simulation and feature decomposition to obtain the power load characteristic curve. In this method, it is assumed that the power load type of a certain bus node is the same as its corresponding land use type, which is divided into residential area load, commercial area load and industrial area load. The typical daily load curve obtained by cluster analysis is used as the power load data source for operation scheduling optimization.

[0018] In terms of adjustability classification, loads can be divided into non-adjustable loads (rigid loads) and adjustable electrical loads (flexible load resources) according to their adjustability. These two types can also be distinguished through factory load planning and information technology such as smart meters and smart home appliances, through statistical analysis and modeling prediction. Adjustable loads can be flexibly scheduled on a time scale, as long as the total daily adjustable load demand remains constant; non-adjustable loads, on the other hand, cannot be flexibly adjusted on a time scale.

[0019] Flexible load modeling methods:

[0020] There are various modeling methods for intraday flexible load dispatching models. Essentially, they all use some exogenous or endogenous incentive to adjust the original load demand over time to a certain extent, thereby smoothing load fluctuations at a specific node and optimizing the distribution network operation. The electrical load at a node can be divided into fixed load and flexible load. Flexible load, while maintaining a constant total daily load, can be dispatched and allocated on an intraday time scale. The composition of the electrical load at a specific point is shown in the following formula:

[0021]

[0022]

[0023] In the formula, This represents the active / reactive load at node i within time period t. This represents the magnitude of the fixed electrical load at node i within time period t. This represents the size of the flexible electrical load at node i within time period t.

[0024] Among them, flexible electrical load section It can be used as a flexible resource on the load side to participate in collaborative scheduling, and its model is as follows.

[0025]

[0026]

[0027]

[0028]

[0029] formula This represents the ratio between the original predicted value of flexible electrical load and the fixed electrical load at a certain node during time period t. For ease of calculation, the text will use the ratio... Set to a fixed value. Formula For the energy consumption conservation constraint, where For this node The actual flexible active power load size after time-slot scheduling. Formula This study will present a flexible reactive power equation relationship under the condition of fixed power factor. The power factor angle is set as the electrical load, and it is a fixed value in any typical day. For a certain node The original flexible active load size before time-slot scheduling. For this node Actual flexible reactive load size during the time period. Formula To ensure the non-negativity and prevent overrunning of flexible loads within a single time period, this study sets the upper limit of flexible loads to twice the original flexible load.

[0030] Building cooling / heating load is a significant component of the overall load. This method employs the RC equivalent thermal path method for quasi-steady-state thermodynamic modeling. Here, we take the building's air conditioning load as an example to construct a deterministic model of the cooling / heating load:

[0031]

[0032]

[0033] in, and These are the indoor and outdoor temperatures; R is the equivalent thermal resistance of the building's exterior walls. It is the heat power supplied by the heat exchange station to the heat load; It is the heat capacity of air; The scheduling period is [number].

[0034] when For a duration of 1 hour, assuming that the optimal indoor temperature can be maintained by the predicted cooling / heating supply, according to the formula... The expression for heat load can be obtained as follows:

[0035]

[0036] Assuming that the optimal indoor temperature can be maintained by predictive cooling / heating supply, then we can obtain:

[0037]

[0038] in, It is the predicted heat load required to maintain the target room temperature; This is the target room temperature, the optimal indoor temperature for human comfort. The human comfort temperature is a flexible range as shown below:

[0039]

[0040] To ensure that the indoor temperature remains within the comfortable range for humans in the next period, the thermal load... It also needs to be limited to a reasonable range:

[0041]

[0042]

[0043] in, This is the heating power required to reach the minimum room temperature requirement in the next time period. This is the maximum heating capacity provided when the room temperature does not exceed the maximum temperature limit in the next time period. Given the outdoor temperature for a given time period... This allows us to determine the range of flexible heat load scheduling. The following formula represents the maximum relative value of heat load fluctuations:

[0044]

[0045]

[0046] Clearly, the comfortable temperature for the human body is a flexible range, which allows for a degree of flexibility in the building's thermal load, enabling optimized scheduling within a reasonable range. The constraints for flexible thermal load scheduling are as follows:

[0047]

[0048]

[0049]

[0050] Where, formula It is the total supply and demand balance constraint of heat load in a day; formula Indicates actual thermal load The range of flexible variations; formula The relative range of thermal load variation and Constraints.

[0051] The above formulas can also be used as expressions for cooling load.

[0052] Based on the modeling methods described above, the formula... It is the composition of summer cooling load, formula It constitutes the winter heating load.

[0053]

[0054]

[0055] In the formula, It represents the heat load at a certain point within a time period t during season s. It refers to the cooling load at a certain point within a time period t during season s. They represent spring, summer, autumn, and winter in that order.

[0056] b) Source-load multivariate uncertainty modeling and characterization methods

[0057] 1) Uncertainty Model for Renewable Resource Power Generation

[0058] The uncertainty in the operation of renewable resources such as wind power and photovoltaic power mainly comes from the difference between the predicted and actual values ​​of meteorological data such as wind direction, wind force, and solar intensity. The following uses a budgeted polyhedral uncertainty set to model and describe the uncertainty of renewable resource power generation.

[0059]

[0060] in, This represents the predicted output of the renewable energy power generation unit at node i during time period t. This represents the actual output of the renewable energy power generation device at node i during time period t. This represents the maximum deviation coefficient between the predicted and actual output of the renewable energy power generation unit at node i in time period t. , Let be the uncertain parameter variable representing the upward and downward deviations of the output of the renewable energy power generation unit at node i in time period t. For uncertain budgets, there is a separate daily fluctuation budget. The summation period represents one day within the scheduling cycle.

[0061] 2) Uncertainty Model of Multiple Load Demands

[0062] The uncertainty of electrical load mainly stems from the difference between the predicted and actual values ​​obtained from load forecasting models, the uncertainty of demand response acceptance due to changes in user willingness, and fluctuations caused by random factors such as sudden load surges. It includes both endogenous and exogenous uncertainties, which will affect the flexibility of demand-side response resources to a certain extent. The uncertainty set of electrical load can be characterized by the following model:

[0063]

[0064] in, This represents the predicted value of the active power load at node i during time period t. This represents the actual value of the active power load at node i during time period t. This represents the maximum deviation coefficient between the predicted and actual active power load at node i during time period t. , Let be the uncertain parameter variable of the active power load at node i during time period t. For the uncertain budget of active power load, there is a separate daily fluctuation budget. The summation period represents one day within the scheduling cycle. Similarly, reactive load can be modeled in the same way.

[0065] To account for load-side fluctuations caused by endogenous and exogenous uncertainties in the aforementioned flexible load resource adjustment model, this method incorporates the deterministic predicted values ​​from the flexible load response model within the deterministic model. Using the uncertain set By replacing the existing model, a flexible load dispatching model that takes into account uncertainties can be obtained.

[0066] The uncertainty of building thermal load mainly comes from exogenous environmental uncertainty. Temperature fluctuations caused by relevant endogenous human activities can also be roughly included in the environmental uncertainty for modeling. Therefore, the uncertainty of environmental temperature is modeled below.

[0067]

[0068] in, This represents the predicted outdoor ambient temperature at node i during time period t. This represents the actual outdoor ambient temperature at node i during time period t. This represents the maximum deviation coefficient between the predicted and actual environmental temperature at node i during time period t. , Let i be the uncertain parameter variable of the ambient temperature at node i during time period t, with a separate fluctuation budget for each day. The summation period represents one day within the scheduling cycle. Budgeting for temperature uncertainty.

[0069] In step S2 of this invention, considering the controllable resource response characteristics in the power distribution system, the pre-allocation of multiple flexible resource operation modes on both the source and load sides is decided in the first stage (day-ahead) (discrete pre-allocation decisions such as tie switch status, on-load tap changer tap position, energy storage charging and discharging mode, and substation power transmission mode). In the second stage, the power scheduling of flexible resources such as multiple types of energy storage and multiple types of transferable loads is decided, thereby establishing a flexible resource day-ahead adaptive robust scheduling framework to cope with the worst-case scenario under the risks of endogenous and exogenous uncertainties in renewable resources, load, and ambient temperature. The method for constructing the flexible resource day-ahead adaptive robust scheduling framework is as follows:

[0070] In a two-stage adaptive robust optimization model, decision-makers can flexibly divide decision variables into ex-ante decision variables and ex-post compensation variables according to their practical significance. For example, the equipment start-up and shutdown plan decision made before the day and the output decision of highly flexible equipment made within the day. This multi-stage dynamic adaptation feature makes adaptive robust optimization more flexible, with lower conservatism, better economy, and good engineering practice value.

[0071] The flexible resource day-ahead adaptive robust scheduling framework is a two-stage, three-layer mathematical optimization model consisting of day-ahead operation mode pre-scheduling and intraday adaptive compensation adjustment. Its compact form is as follows:

[0072]

[0073]

[0074]

[0075]

[0076] Here, x represents the decision variables for the first stage, which need to be decided before the uncertain variables are determined. These include variables with slow adjustment speeds or unsuitable for frequent changes, such as micro gas turbine start-up and shutdown variables, multi-energy storage facility charging and discharging mode variables, on-load tap changer tap position variables, reactive power compensation capacitor switching variables, and demand-side response decisions (mainly discrete variables). u represents the variables in the first layer of the second stage that describe the uncertainty fluctuation risk, including uncertain variables in renewable energy generation. and Uncertainties in electrical load and and outdoor temperature uncertainty and Uncertain parameters that are difficult to predict accurately in advance; y is the adaptive adjustment decision variable in the second layer of the second stage, which represents those compensation decision variables that can be adjusted after the uncertainty is determined, including multiple flexible resource adjustment decisions such as energy storage facility output, distributed power output, and load demand response adjustment, as well as rescheduling decisions such as renewable resource grid connection and emergency load shedding; U is the uncertainty set. c and d are the coefficient matrices of decision variables x and y in the objective function; A, D, G, H, b, and h are constant matrices or column vectors in the corresponding constraints, generally the coupling coefficients between devices, physical characteristic parameters of multi-energy system operation, safety boundaries, and operating characteristic parameters of equipment, etc.

[0077] The first formula above represents finding the minimum value of the sum of the objective function of the ex-ante decision and the objective function of the ex-post compensation decision. To determine the cost of the first-stage decision, the first stage determines the value of x. The cost of the second-stage decision is to find the worst-case scenario and obtain the adaptive adjustment decision y that minimizes the objective function of the second stage under that scenario. The second formula above represents the inequality constraints that are only related to the decision variable x of the first stage. The third formula above uses equality constraints to represent the impact of the pre-scheduling decision of the first-stage operation mode on the relevant operating parameters of the distribution network. The fourth formula above describes the constraints related to x and u in the feasible region of the decision variable y, where x and u have been determined before solving y.

[0078] The aforementioned scheduling framework aims to mitigate the risks of power distribution network operation by addressing the worst-case scenarios under the endogenous and exogenous uncertainties and fluctuations caused by renewable resources, load, and ambient temperature, with the lowest overall operating cost.

[0079] In step S3 of this invention, based on the proposed flexible resource day-ahead adaptive robust scheduling framework, an adaptive robust optimization model is constructed, which includes distribution network power flow constraints, source-load dual-side multi-type flexible resource operation scheduling constraints, and operation scheduling objective function. The relevant model is as follows:

[0080] a) Objective function

[0081] Taking into account factors such as system facility and equipment operation and maintenance costs, energy trading costs, and compensation costs, the objective function of the flexible resource day-ahead adaptive robust scheduling framework is as follows:

[0082]

[0083] This formula corresponds to the objective function of the proposed flexible resource-day adaptive robust scheduling framework. .in, For the operation and maintenance costs of system facilities and equipment, To account for the cost of power exchange with the upper-level power grid, For fuel costs, The cost of curtailment penalties for renewable energy power generation facilities. This refers to the penalty cost for load shedding during demand-side response. The corresponding formula is as follows:

[0084] System operation and maintenance costs:

[0085]

[0086] The average operating and maintenance cost per unit of work done by each type of equipment, of which... Represents a set of device installation nodes; A collection representing the types of installed equipment; The type is The average maintenance cost per unit power generated by device e; Indicates the operating power of the equipment. Indicates the node number. Indicates the time period number; This represents the length of the time interval.

[0087] Power exchange cost:

[0088]

[0089] The above formula differentiates between peak and off-peak electricity prices to calculate the cost of power exchange between the distribution network and the upstream power grid. It is the unit power cost of active power transactions or reactive power auxiliary service transactions within time period t; It refers to the active / reactive power exchanged between the distribution network and the upstream.

[0090] Fuel purchase cost:

[0091]

[0092] The above formula represents the annual cost of fuel purchase. Wherein, A set of nodes equipped with micro gas turbines / diesel generators; Fuel consumption per unit power output of a diesel generator; The gas consumption of a micro gas turbine prime mover to perform work per unit power per hour; and The purchase price per unit of natural gas and diesel.

[0093] Cost of curtailment of renewable energy power generation facilities:

[0094]

[0095] The above formula represents the cost of distributed renewable resources being unable to connect to the grid due to cost or operational safety concerns. Wherein, and Let represent the power curtailment of photovoltaic power sources and the power curtailment of photovoltaic generators at node i during time period t. and The cost of punishment for both.

[0096] Demand-side response load abandonment penalty cost:

[0097]

[0098] When system operation does not meet safety constraints, a portion of non-critical load will be discarded during scheduling. The above formula represents the cost penalty for load shedding: where, A set representing the labels of electrical load nodes; Let be the power of the electrical load that is cut off at node i during time period t. To reduce the cost of load shedding.

[0099] b) Distribution network power flow constraints

[0100]

[0101]

[0102]

[0103]

[0104] The above formula represents the linear power flow equations in a radial distribution network. Its effectiveness in optimal power flow planning for radial distribution networks has been fully demonstrated and widely applied. Where: Represents the set of upstream nodes directly connected to node j; This represents the set of downstream nodes directly connected to node j. This represents the set of nodes in a power distribution network model; Let represent the set of branches in the distribution network model; Rij represents the total line resistance of branch ij; Xij represents the total line reactance of branch ij; and Ui represents the voltage magnitude at node i in the distribution network model. Represents the square of the voltage; Let be the active power flowing into branch ij; The reactive power flowing into branch ij; This represents the size of the unschedulable active load at node j; This represents the size of the unschedulable reactive load at node j; the other relevant active and reactive variables are those in Chapter 2 representing the consumption or generation of active and reactive power related to each node, plus a reactive term with a positive value to represent the consumption of inductive reactive power.

[0105] c) Safety constraints for the operation of the power distribution system

[0106]

[0107]

[0108]

[0109] The first formula is the voltage amplitude constraint of the distribution network to meet the voltage quality requirements of the distribution network load; the latter two formulas constrain the power that the line can accommodate to prevent the current from exceeding the line's carrying capacity and causing a fault.

[0110] d) Heat load supply constraints

[0111]

[0112] The left side of the formula represents the thermal power of all thermal equipment at the node, and the right side represents the thermal load demand of the node. This formula only represents the balance constraint of a single thermal load (such as cooling or heating). When both cooling and heating loads exist, two sets of equations can be written separately according to the formula.

[0113] e) Renewable resource power generation constraints

[0114] Photovoltaics:

[0115]

[0116]

[0117] In the formula, Solar irradiance The corresponding photovoltaic power generation potential is a coefficient ranging from 0 to 1; This is the rated solar irradiance of the photovoltaic power source. Once this value is exceeded, the active power output of the photovoltaic power source remains unchanged at the rated value.

[0118] Fan:

[0119]

[0120]

[0121] The output of a wind turbine's prime mover is mainly related to factors such as blade size, air density, and wind speed. Among these, It provides power to the prime mover of the wind turbine; The density of air; The area swept by the blade; The wind speed is the cube of its value. Let be the wind energy utilization coefficient. As shown in the above formula, besides its own performance parameters such as blade size and wind energy utilization efficiency, the wind turbine output is largely related to real-time wind conditions. Assuming that factors such as the wind turbine blade orientation are ideal, its output is a four-stage piecewise function that depends only on wind speed:

[0122]

[0123] In the first stage, when the wind speed is less than the cut-in wind speed... At the first stage, the fan output is 0; in the second stage, when the wind speed is at the cut-in wind speed... With rated wind speed Between, the wind turbine power generation potential coefficient At 0 and rated output Between; the third stage, when the wind speed is within the rated wind speed With cut-off wind speed During this period, the wind turbine power generation potential coefficient remains at its maximum. Unchanged; Fourth stage, when the wind speed is greater than the cutoff wind speed When the wind turbine stops rotating, the protection device adjusts the blade angle, and the wind turbine's power generation potential coefficient becomes 0.

[0124] Micro gas turbines with waste heat recovery capabilities:

[0125] The main operating model includes constraints on the output power of the micro gas turbine prime mover, constraints on power generation and heat production, and constraints on active and reactive power output.

[0126]

[0127]

[0128]

[0129]

[0130]

[0131] in, Minimum prime mover output power considering efficiency factors for a micro gas turbine; This refers to the actual output power of the micro gas turbine prime mover; This refers to the rated output power of the micro gas turbine prime mover; This represents a 0-1 variable indicating the start-stop status of the gas turbine. The apparent power rating of a micro gas turbine; The power generation efficiency of micro gas turbines; This refers to the actual thermal power output of the micro gas turbine. The energy loss rate of a micro gas turbine; The efficiency of the waste heat recovery device for micro gas turbines. This refers to the active power of the actual electrical energy output by the micro gas turbine. This refers to the reactive power of the actual electrical energy output by the micro gas turbine.

[0132] Electrothermal coupling device:

[0133] A typical electrothermal coupling heat pump can be driven with relatively little electricity, converting low-grade heat energy into usable heat energy using a reverse Carnot cycle. It can provide both heating and cooling with a high coefficient of performance (COP). Its operating model is as follows:

[0134]

[0135]

[0136]

[0137]

[0138]

[0139] formula The constraints on the conversion between heating power and electrical power in the electrothermal coupling device, wherein The heating power of the electrothermal coupling device; The heating coefficient of the electrothermal coupling device; This refers to the amount of active power consumed by the electrothermal coupling device for heating. (Formula) The constraints for the conversion of thermal power to electrical power in the electrothermal coupling device, wherein The cooling power of the electrothermal coupling device; The coefficient of performance (COP) of the electrothermal coupling device; This represents the amount of active power consumed by the electrothermal coupling device during cooling operation. (Formula) , , The range of electrical power consumed by the electrothermal coupling device is constrained. This represents the maximum active power consumed by the electrothermal coupling device.

[0140] Energy storage devices:

[0141] For cold, heat, and electricity, the operational constraints of stationary energy storage facilities can be uniformly characterized using the following method:

[0142]

[0143]

[0144]

[0145]

[0146]

[0147]

[0148]

[0149] The first two formulas represent the energy storage constraints of the ES device, where for Energy storage ratio of ES devices during different time periods; for Energy storage capacity of ES devices during a given time period; This refers to the rated electrical capacity of the equipment. This is the self-discharge rate of the energy storage device at each time period. The third formula is the constraint of the equation relationship between the input and output power of the ES device and the stored energy, where... and The charging efficiency and discharging efficiency coefficients of the ES device; and for The first formula represents the charging and discharging power of the ES (Energy Storage) device during a given time period. The second formula represents the energy storage constraints of ES devices of the same type during the first and last time periods of a typical day. The third formula represents the charging and discharging power limit constraints and the logical constraints that prevent a single device from charging and discharging simultaneously. and The maximum charging and discharging power of the ES device; and These are 0-1 variables that characterize the charging and discharging states.

[0150] Voltage regulating device:

[0151]

[0152] in, For voltage adjustment level selection, it belongs to a set of integer candidates. It is the minimum adjustable step size of the voltage regulation ratio.

[0153] Reactive power compensation device:

[0154]

[0155] in, For the selection of reactive power compensation adjustment gears, it belongs to a set of integer candidates; It is the minimum adjustable step size of the reactive power compensation device.

[0156] In step S4 of this invention, the decision-making mode and model structure of the adaptive robust optimization model that considers multiple endogenous and exogenous uncertainties of source and load are analyzed. The model is decomposed into the main problem of pre-scheduling of flexible resource operation mode and the sub-problem of adaptive mitigation of source and load fluctuation risk. The McCormick envelope approximation method is used to deal with the strong coupling of random variables in the sub-problem caused by endogenous uncertainty.

[0157] First, a comprehensive analysis of the robust framework and decision variable types in steps S2 and S3 reveals that the adaptive robust scheduling model presents a two-stage, three-level structure. Discrete variables can be separated from the two-stage model and transferred to the first stage. Therefore, the following method is used to decouple the model stages:

[0158] The model is decomposed into a single-level mixed integer optimization master problem and a two-level optimization sub-problem involving risk adaptive scheduling and mitigation in the worst-case scenario of identifying multivariate uncertain fluctuations in source loads.

[0159]

[0160] This algorithm identifies uncertain scenarios in subproblems and then adds these identified uncertain scenarios back to the main problem as new compensation variables and constraints, exhibiting good convergence. In the main problem, s is the index of the generated uncertain scenario. This indicates the number of scenarios that have been identified in the subproblem, and also the number of iterations. Let be the value of the uncertain variable in the s-th scenario of the subproblem, which has already been determined. It is the set of uncertain variables that have been determined in the subproblem. This is equivalent to the objective function value in the second stage of the first stage. The cutting plane is added based on the uncertain scenario determined in the second stage. This single-stage master problem can be solved directly by a general solver, and the value of the decision variable x obtained is used as a known quantity in the model of the second stage.

[0161] The sub-problems are as follows:

[0162]

[0163] Where y represents the second-stage compensation decision variable, which is a continuous decision variable; The values ​​of the decision variables obtained in the first stage are denoted as u; u is an uncertain variable, and U is the set of all uncertain variables. The dual variables corresponding to the constraints. Subproblems It is a two-level optimization problem that cannot be solved directly by encoding. However, it can be transformed into a maximization problem by equivalently transforming the inner adaptive adjustment problem into the upper-level risk identification problem, which can then be mathematically modeled.

[0164] After transforming the subproblems into duality, we can obtain the following model:

[0165]

[0166] in denoted as the dual variable corresponding to the constraints of the original problem in the subproblem; M is a relatively large constant boundary added to avoid the subproblem from becoming unbounded; u is an uncertain variable; h, G, H, and D are the same coefficient matrices as the original problem.

[0167] It can be noted that the objective function contains This strongly coupled nonlinear term is approximated by a McCormick envelope method:

[0168]

[0169] Define the auxiliary variables as described in the above formula. And the following constraint group is given:

[0170]

[0171] in, , , , They are random variables With dual variables The upper and lower bound parameters.

[0172] The dual problem after approximate linear substitution is as follows:

[0173]

[0174] Where m and n are respectively and The number of elements.

[0175] In step S5 of the present invention, a column constraint generation algorithm is used to iteratively solve the adaptive robust scheduling model established in steps S1-S4 to obtain a flexible resource day-ahead adaptive robust scheduling scheme that considers the mitigation of multiple fluctuation risks of distribution network sources and loads.

[0176] The algorithm for solving the flexible resource day-ahead adaptive robust scheduling framework is as follows:

[0177]

[0178] The beneficial effects of this invention are as follows:

[0179] This invention provides a robust day-ahead scheduling method for flexible resources in distribution networks under multi-source and load fluctuation risks. By constructing a day-ahead adaptive robust scheduling framework for flexible resources, it fully considers the exogenous fluctuation risks from renewable energy and loads in the distribution network, as well as the endogenous uncertainty risks caused by demand-side flexible resource scheduling decisions. It also considers the multi-source and load fluctuation risks arising from the endogenous and exogenous uncertainties of multiple types of flexible resources on both the source and load sides. To address the strong coupling between decision variables and random variables in sub-problems caused by endogenous uncertainty, this invention employs the McCormick envelope approximation method for linear transformation, ensuring the solvability of the model. Finally, it collaboratively optimizes multiple types of flexible resources on both the source and load sides during the day-ahead stage to mitigate distribution network operation risks, maximizing the economic efficiency and flexibility of distribution network operation while ensuring its operational safety and robustness. The day-ahead adaptive robust scheduling method for flexible resources considering multi-source and load fluctuation risks provided by this invention can overcome the limitations of existing technologies in coordinating multiple types of flexible resources on both the source and load sides to mitigate distribution system operation risks when facing endogenous and exogenous multi-source and load fluctuation risks. Attached Figure Description

[0180] Figure 1 is a schematic diagram of the method flow of the present invention;

[0181] Figure 2 is a schematic diagram of the IEEE 33-node distribution network system topology used in a specific embodiment of the present invention;

[0182] Figure 3 is an outdoor temperature curve of the system facilities and equipment during operation in a specific embodiment of the present invention;

[0183] Figure 4 is a schematic diagram of the endogenous uncertainty fluctuation of the adjustable flexible load in a specific embodiment of the present invention;

[0184] Figure 5 is a schematic diagram of load rate (left axis) and load abandonment (right axis) under risk fluctuation conditions in a specific embodiment of the present invention. Detailed Implementation

[0185] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0186] Figure 1 is a schematic flowchart of the method of the present invention, illustrating the basic steps of the method. Figure 2 is a topology diagram of the IEEE 33-node distribution network system used in the embodiments of the present invention, which includes system topology information and device location information of distributed flexible resources.

[0187] As shown in Figure 1, a flexible resource-daytime robust scheduling method for distribution networks under the risk of multiple source-load fluctuations includes the following steps:

[0188] S1. Considering a power distribution system with large-scale integration of renewable resources such as wind and solar power, as shown in Figure 2, and the risk of source-load fluctuations, analyze the deterministic modeling methods for various elements on the source and load sides of the power distribution system, as well as the corresponding sources of randomness. Then, determine the fluctuation pattern and range, and construct an uncertain set model characterizing the source-load fluctuation risk. The deterministic set is modeled in the following manner:

[0189] a) Flexible load scheduling methods

[0190] Flexible load model:

[0191] The flexible load model corresponding to the node in the system that can act as a flexible electrical load for demand-side response is as follows.

[0192]

[0193]

[0194] in, This is the flexible and adjustable part, and its scheduling constraints are as follows:

[0195]

[0196]

[0197]

[0198]

[0199] This invention employs the RC equivalent thermal circuit method to model the quasi-steady-state model of building thermal load. Here, we take the building's air conditioning load as an example to construct a deterministic model of cooling / heating load:

[0200]

[0201]

[0202] The scheduling constraints for flexible thermal loads are as follows:

[0203]

[0204]

[0205]

[0206] In this embodiment, the relative range of heat load variation and It is 0.2.

[0207] b) Source-load multivariate uncertainty modeling and characterization methods

[0208] 1) Uncertainty Model for Renewable Resource Power Generation

[0209] The uncertainty in the operation of renewable resources such as wind power and photovoltaic power mainly comes from the difference between the predicted and actual values ​​of meteorological data such as wind direction, wind force, and solar intensity. The following uses a budgeted polyhedral uncertainty set to model and describe the uncertainty of renewable resource power generation.

[0210]

[0211] 2) Uncertainty Model of Multiple Load Demands

[0212] The uncertainty of electrical load mainly stems from the difference between the predicted and actual values ​​obtained from load forecasting models, the uncertainty of demand response acceptance due to changes in user willingness, and fluctuations caused by random factors such as sudden load surges. It includes both endogenous and exogenous uncertainties, which will affect the flexibility of demand-side response resources to a certain extent. The uncertainty set of electrical load can be characterized by the following model:

[0213]

[0214] Similarly, reactive loads can be modeled in the same way.

[0215] To account for load-side fluctuations caused by endogenous and exogenous uncertainties in the aforementioned flexible load resource adjustment model, this method incorporates the deterministic predicted values ​​from the flexible load response model within the deterministic model. Using the uncertain set By replacing the existing model, a flexible load dispatching model that takes into account uncertainties can be obtained.

[0216] The uncertainty of building thermal load mainly comes from exogenous environmental uncertainty. The endogenous uncertainty of temperature fluctuation caused by relevant human activities can also be roughly included in the environmental uncertainty for modeling. Therefore, the uncertainty of environmental temperature is modeled below.

[0217]

[0218] In this embodiment, the relevant uncertain fluctuation range is set to 20%, and the uncertain budget is set to fluctuate in 6 time periods per day.

[0219] S2. Based on the controllable resource response characteristics of the power distribution system, the system makes decisions on the pre-allocation of multiple flexible resource operation modes on both the source and load sides in the first stage (day-ahead) (discrete pre-allocation decisions such as tie switch status, on-load tap changer tap position, energy storage charging and discharging mode, and substation power transmission mode). In the second stage, it makes decisions on the power scheduling of flexible resources such as multiple types of energy storage and multiple types of transferable loads. This establishes a flexible resource day-ahead adaptive robust scheduling framework to cope with the worst-case scenario under the risks of endogenous and exogenous uncertainties in renewable resources, load, and ambient temperature. The construction method of the flexible resource day-ahead adaptive robust scheduling framework is as follows:

[0220] The flexible resource day-ahead adaptive robust scheduling framework is a two-stage, three-layer mathematical optimization model consisting of day-ahead operation mode pre-scheduling and intraday adaptive compensation adjustment. Its compact form is as follows:

[0221]

[0222]

[0223]

[0224]

[0225] Here, x represents the decision variables for the first stage, which need to be decided before the uncertain variables are determined. These include variables with slow adjustment speeds or unsuitable for frequent changes, such as micro gas turbine start-up and shutdown variables, multi-energy storage facility charging and discharging mode variables, on-load tap changer tap position variables, reactive power compensation capacitor switching variables, and demand-side response decisions (mainly discrete variables). u represents the variables in the first layer of the second stage that describe the uncertainty fluctuation risk, including uncertain variables in renewable energy generation. and Uncertainties in electrical load and and outdoor temperature uncertainty and Uncertain parameters that are difficult to predict accurately in advance; y is the adaptive adjustment decision variable in the second layer of the second stage, which represents those compensation decision variables that can be adjusted after the uncertainty is determined, including multiple flexible resource adjustment decisions such as energy storage facility output, distributed power output, and load demand response adjustment, as well as rescheduling decisions such as renewable resource grid connection and emergency load shedding; U is the uncertainty set. c and d are the coefficient matrices of decision variables x and y in the objective function; A, D, G, H, b, and h are constant matrices or column vectors in the corresponding constraints, generally the coupling coefficients between devices, physical characteristic parameters of multi-energy system operation, safety boundaries, and operating characteristic parameters of equipment, etc.

[0226] The aforementioned scheduling framework aims to mitigate the risks of power distribution network operation by addressing the worst-case scenarios under the endogenous and exogenous uncertainties and fluctuations caused by renewable resources, load, and ambient temperature, with the lowest overall operating cost.

[0227] S3. Based on the proposed flexible resource day-ahead adaptive robust scheduling framework, an adaptive robust optimization model is constructed, including distribution network power flow constraints, source-load dual-side multi-type flexible resource operation scheduling constraints, and operation scheduling objective function. The relevant model is as follows:

[0228] a) Objective function

[0229] Taking into account factors such as system facility and equipment operation and maintenance costs, energy trading costs, and compensation costs, the objective function of the flexible resource day-ahead adaptive robust scheduling framework is as follows:

[0230]

[0231] This formula corresponds to the objective function of the proposed flexible resource-day adaptive robust scheduling framework. .in, For the operation and maintenance costs of system facilities and equipment, To account for the cost of power exchange with the upper-level power grid, For fuel costs, The cost of curtailment penalties for renewable energy power generation facilities. This refers to the penalty cost for load shedding during demand-side response. The corresponding formula is as follows:

[0232] System operation and maintenance costs:

[0233]

[0234] Power exchange cost:

[0235]

[0236] Fuel purchase cost:

[0237]

[0238] Cost of curtailment of renewable energy power generation facilities:

[0239]

[0240] Demand-side response load abandonment penalty cost:

[0241]

[0242] b) Distribution network power flow constraints

[0243]

[0244]

[0245]

[0246]

[0247] c) Safety constraints for the operation of the power distribution system

[0248]

[0249]

[0250]

[0251] d) Heat load supply constraints

[0252]

[0253] The left side of the formula represents the thermal power of all thermal equipment at the node, and the right side represents the thermal load demand of the node. This formula only represents the balance constraint of a single thermal load (such as cooling or heating). When both cooling and heating loads exist, two sets of equations can be written separately according to the formula.

[0254] e) Renewable resource power generation constraints

[0255] Photovoltaics:

[0256]

[0257]

[0258] Fan:

[0259]

[0260]

[0261]

[0262] Micro gas turbines with waste heat recovery capabilities:

[0263] The main operating model includes constraints on the output power of the micro gas turbine prime mover, constraints on power generation and heat production, and constraints on active and reactive power output.

[0264]

[0265]

[0266]

[0267]

[0268]

[0269] Electrothermal coupling device:

[0270] A typical electrothermal coupling heat pump can be driven with relatively little electricity, converting low-grade heat energy into usable heat energy using a reverse Carnot cycle. It can provide both heating and cooling with a high coefficient of performance (COP). Its operating model is as follows:

[0271]

[0272]

[0273]

[0274]

[0275]

[0276] Energy storage devices:

[0277] For cold, heat, and electricity, the operational constraints of stationary energy storage facilities can be uniformly characterized using the following method:

[0278]

[0279]

[0280]

[0281]

[0282]

[0283]

[0284]

[0285] Voltage regulating device:

[0286]

[0287] in, For voltage adjustment level selection, it belongs to a set of integer candidates. It is the minimum adjustable step size of the voltage regulation ratio.

[0288] Reactive power compensation device:

[0289]

[0290] in, For the selection of reactive power compensation adjustment gears, it belongs to a set of integer candidates; It is the minimum adjustable step size of the reactive power compensation device.

[0291] The relevant parameters for the example are as follows:

[0292] Equipment Main Parameter Table

[0293]

[0294] S4. For the adaptive robust optimization model that considers multiple endogenous and exogenous uncertainties of source and load, analyze its decision mode and model structure. Decompose the model into the main problem of pre-scheduling of flexible resource operation mode and the sub-problem of adaptive mitigation of source and load fluctuation risk. The McCormick envelope approximation method is used to deal with the strong coupling of random variables in the sub-problem caused by endogenous uncertainty.

[0295] First, a comprehensive analysis of the robust framework and decision variable types in steps S2 and S3 reveals that the adaptive robust scheduling model presents a two-stage, three-level structure. Discrete variables can be separated from the two-stage model and transferred to the first stage. Therefore, the following method is used to decouple the model stages:

[0296] The model is decomposed into a single-level mixed integer optimization master problem and a two-level optimization sub-problem involving risk adaptive scheduling and mitigation in the worst-case scenario of identifying multivariate uncertain fluctuations in source loads.

[0297] The main question is as follows:

[0298]

[0299] The sub-problems are as follows:

[0300]

[0301] Where y represents the second-stage compensation decision variable, which is a continuous decision variable; The values ​​of the decision variables obtained in the first stage are denoted as u; u is an uncertain variable, and U is the set of all uncertain variables. The dual variables corresponding to the constraints. Subproblems It is a two-level optimization problem that cannot be solved directly by encoding. However, it can be transformed into a maximization problem by equivalently transforming the inner adaptive adjustment problem into the upper-level risk identification problem, which can then be mathematically modeled.

[0302] After transforming the subproblems into duality, we can obtain the following model:

[0303]

[0304] in denoted as the dual variable corresponding to the constraints of the original problem in the subproblem; M is a relatively large constant boundary added to avoid the subproblem from becoming unbounded; u is an uncertain variable; h, G, H, and D are the same coefficient matrices as the original problem.

[0305] It can be noted that the objective function contains This strongly coupled nonlinear term is approximated by a McCormick envelope method:

[0306]

[0307] Define the auxiliary variables as described in the above formula. And the following constraint group is given:

[0308]

[0309] in, , , , They are random variables With dual variables The upper and lower bound parameters.

[0310] The dual problem after approximate linear substitution is as follows:

[0311]

[0312] Where m and n are respectively and The number of elements.

[0313] S5. Using the constraint generation algorithm as follows, and using the open-source toolbox Yalmip and commercial / academic solvers, iteratively solve the adaptive robust scheduling model established in steps S1-S4. The optimal solution obtained is the flexible resource-ahead adaptive robust scheduling scheme that considers the mitigation of multiple fluctuation risks of distribution network sources and loads.

[0314] In this embodiment, the algorithm basically converges after 3-4 iterations when solving the day-ahead adaptive robust scheduling scheme, demonstrating good stability.

[0315] The corresponding robust scheduling decision results and associated costs are as follows:

[0316] Scheduling scheme cost

[0317] Deterministic solutions with no source load uncertainty: Deterministic solutions face multiple source load uncertainties; Robust solutions face multiple source load uncertainties; Overall cost (RMB): 9561.31, 6651.61, 0849.3; Fuel purchase cost (RMB): 2094.75, 2233.42, 2386.23; Operation and maintenance cost (RMB): 718.13, 700.91, 694.015; Electricity purchase and sales cost (RMB): 6748.28, 7676.97, 751.1; Wind curtailment penalty cost (RMB): 018.58, 7117.4734; Solar curtailment penalty cost (RMB): 000; Load curtailment penalty (RMB): 06021.610 surface

[0318] Comparative examples demonstrate that, when facing uncertain fluctuations in multiple power sources and loads, the flexible, day-ahead adaptive robust dispatch scheme proposed in this invention, which considers mitigating the risk of fluctuations in multiple power sources and loads in the distribution network, provides extremely strong resilience against fluctuations at a relatively low robust cost. Under a 20% fluctuation in multiple power sources and loads, this robust scheme only curtails a very small amount of wind power, without curtailing solar power or other loads; while the deterministic dispatch scheme, although possessing a slight cost advantage under ideal conditions, suffers from severe load curtailment when facing the risk of fluctuations in multiple power sources and loads in the distribution network.

[0319] This embodiment illustrates that the flexible resource day-ahead adaptive robust scheduling method provided by the present invention, which considers the multi-factor fluctuation risk of the distribution network source and load, can solve the limitations of the existing technology in coordinating multiple types of flexible resources on both the source and load sides to mitigate the operational risks of the distribution system when facing the multi-factor fluctuation risk of the source and load.

[0320] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0321] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A robust method for flexible resource day-ahead scheduling in distribution networks under the risk of multi-source load fluctuations, characterized in that, The method includes the following steps: S1. Analyze the source load fluctuation mode and range within the power distribution system, and construct an uncertainty set model to characterize the source load fluctuation risk. The uncertainty set is defined using box inequality constraints. S2. Based on the controllable resource response characteristics of the distribution system, in the first stage (day-ahead), pre-allocation decisions are made for multiple flexible resource operation modes on both the source and load sides (discrete pre-allocation decisions such as tie switch status, on-load tap changer tap position, energy storage charging and discharging mode, and substation power transmission mode). In the second stage, decisions are made regarding the power scheduling of multiple types of energy storage and multiple types of transferable loads, thereby establishing a day-ahead adaptive robust scheduling framework for flexible resources to cope with the worst-case scenario under the risks of endogenous and exogenous uncertainties in renewable resources, load, and ambient temperature. S3. Based on the proposed day-ahead adaptive robust scheduling framework for flexible resources, a system is constructed that includes distribution network power flow constraints and multiple types of flexible resources on both the source and load sides. An adaptive robust optimization model including scheduling constraints and scheduling objective function is established. S4. For this adaptive robust optimization model considering multiple endogenous and exogenous uncertainties of source and load, its decision-making mode and model structure are analyzed. The model is decomposed into a pre-scheduling master problem of flexible resource operation mode and an adaptive mitigation sub-problem of source and load fluctuation risk. The McCormick envelope approximation method is used to handle the strong coupling of random variables in the sub-problem caused by endogenous uncertainties. S5. Finally, the column constraint generation (C&CG) algorithm is used to iteratively solve the adaptive robust scheduling model established in S1-S4, obtaining a flexible resource day-ahead adaptive robust scheduling scheme considering the mitigation of multiple source and load fluctuation risks in the distribution network.

2. The flexible day-ahead adaptive robust scheduling method for resources considering the multi-source load fluctuation risk in the distribution network as described in claim 1, characterized in that, In step S1, a power distribution system with large-scale access to renewable resources such as wind and solar power and source-load fluctuation risk is considered. First, the deterministic modeling method of each element on the source side and load side of the power distribution system is analyzed. Then, the uncertainty fluctuation mode and range are analyzed, and an uncertainty set model representing the source-load fluctuation risk is constructed. The deterministic set is defined by box inequality constraints. a) Load Classification and Deterministic Modeling Methods: Flexible Load Modeling Methods: There are various modeling methods for intraday flexible load scheduling models. Essentially, they all use some exogenous or endogenous incentive to adjust the original load demand over time to a certain extent, thereby smoothing load fluctuations at a certain node and optimizing the distribution network operation. The electrical load at a certain node can be divided into fixed electrical load and flexible electrical load. Flexible electrical load, under the premise of maintaining a constant total intraday load, can be scheduled and allocated on an intraday time scale. The composition of flexible electrical load at a certain point is shown in the following formula: Among them, flexible electrical load section It can be used as a flexible resource on the load side to participate in collaborative scheduling, and its model is as follows: The cooling / heating load of a building is an important component of the overall load. This method uses the RC equivalent thermal circuit method to model a quasi-steady-state thermodynamic model. Here, we take the building's air conditioning load as an example to construct a deterministic model of the cooling / heating load: The scheduling constraints for flexible thermal loads are as follows: b) Modeling and characterization method of source and load multivariate uncertainty 1) Uncertainty model of renewable resource power generation The uncertainty of renewable resource operation such as wind power and photovoltaic mainly comes from the difference between the predicted and actual values ​​of meteorological data such as wind direction, wind force and light intensity. The following uses a polyhedral uncertainty set with budget to model and describe the uncertainty of renewable resource power generation. 2) Multivariate Load Demand Uncertainty Model: The uncertainty of electrical load mainly stems from the difference between the predicted and actual values ​​obtained from the load forecasting model, the uncertainty of demand response acceptance due to changes in user willingness, and fluctuations caused by random factors such as sudden load surges. It includes both endogenous and exogenous uncertainties, which will affect the flexibility of demand-side response resources to a certain extent. The uncertainty set of electrical load can be characterized by the following model: Similarly, reactive load can be modeled in the same way. To account for load-side fluctuations caused by endogenous and exogenous uncertainties in the aforementioned flexible load resource adjustment model, this method uses the deterministic predicted values ​​from the flexible load response model in the deterministic model. Using the uncertain set By substitution, a flexible load scheduling model that takes into account uncertainties can be obtained. The uncertainty of building thermal load mainly comes from exogenous environmental uncertainties. Temperature fluctuations caused by relevant endogenous human activities can also be roughly included in the environmental uncertainty for modeling. Therefore, the uncertainty of environmental temperature is modeled below.

3. The flexible resource-day-ahead adaptive robust scheduling method considering the multi-source load fluctuation risk of the distribution network as described in claim 1, characterized in that, In step S2, considering the controllable resource response characteristics of the power distribution system, the first stage (day-ahead) determines the pre-allocation of multiple flexible resource operation modes on both the source and load sides (discrete pre-allocation decisions such as tie switch status, on-load tap changer tap position, energy storage charging and discharging mode, and substation power transmission mode). The second stage determines the power scheduling of multiple types of energy storage and multiple types of transferable loads, thereby establishing a flexible resource day-ahead adaptive robust scheduling framework to cope with the worst-case scenario under the risks of endogenous and exogenous uncertainties in renewable resources, load, and ambient temperature. The construction method of the flexible resource day-ahead adaptive robust scheduling framework is as follows: The flexible resource day-ahead adaptive robust scheduling framework is a two-stage, three-layer mathematical optimization model consisting of day-ahead operation mode pre-scheduling and intraday adaptive compensation adjustment, with the following compact form: Here, x represents the decision variables for the first stage, which need to be decided before the uncertain variables are determined. These include variables with slow adjustment speeds or unsuitable for frequent changes, such as micro gas turbine start-up and shutdown variables, multi-energy storage facility charging and discharging mode variables, on-load tap changer tap position variables, reactive power compensation capacitor switching variables, and demand-side response decisions (mainly discrete variables). u represents the variables in the first layer of the second stage that describe the uncertainty fluctuation risk, including uncertain variables in renewable energy generation. and Uncertainties in electrical load and and outdoor temperature uncertainty and Uncertain parameters that are difficult to predict accurately in advance; y is the adaptive adjustment decision variable in the second layer of the second stage, which represents those compensation decision variables that can be adjusted after the uncertainty is determined, including multiple flexible resource adjustment decisions such as energy storage facility output, distributed power output, and load demand response adjustment, as well as rescheduling decisions such as renewable resource grid connection and emergency load shedding; U is the uncertainty set. c and d are the coefficient matrices of decision variables x and y in the objective function; A, D, G, H, b, and h are constant matrices or column vectors in the corresponding constraints, generally the coupling coefficients between devices, physical characteristic parameters of multi-energy system operation, safety boundaries, and operating characteristic parameters of devices; the scheduling framework aims to cope with the worst-case scenario under the endogenous and exogenous uncertainty fluctuation risks brought about by renewable resources, load, and ambient temperature, and to mitigate the operation risks of the distribution network with the lowest comprehensive operating cost.

4. The flexible day-ahead adaptive robust scheduling method for resources considering the multi-source load fluctuation risk in the distribution network according to claim 1, characterized in that, In step S3, based on the proposed flexible resource day-ahead adaptive robust scheduling framework, an adaptive robust optimization model is constructed, including distribution network power flow constraints, source-load dual-side multi-type flexible resource operation scheduling constraints, and operation scheduling objective function. The relevant model is as follows: a) The objective function comprehensively considers factors such as system facility and equipment operation and maintenance costs, energy trading costs, and compensation costs. The objective function of the flexible resource day-ahead adaptive robust scheduling framework is as follows: This formula corresponds to the objective function of the proposed flexible resource-day adaptive robust scheduling framework. .in, For the operation and maintenance costs of system facilities and equipment, To account for the cost of power exchange with the upper-level power grid, For fuel costs, The cost of curtailment penalties for renewable energy power generation facilities. This is the cost of demand-side response load abandonment penalty; the corresponding formula is as follows: System operation and maintenance cost: Power exchange cost: Fuel purchase cost: Cost of curtailment of renewable energy power generation facilities: Demand-side response load abandonment penalty cost: b) Distribution network power flow constraints c) Safety constraints for the operation of the power distribution system d) Heat load supply constraints The left side of the formula represents the thermal power of all thermal equipment at the node, and the right side represents the node's thermal load demand. This formula only characterizes the balance constraint of a single thermal load (such as cooling or heating). When both cooling and heating loads exist, two sets of equations can be written separately according to the formula described above; e) Renewable resource power generation constraint photovoltaic: Fan: Micro gas turbine with waste heat recovery function: The main operating model includes constraints on the output power of the micro gas turbine prime mover, constraints on power generation and heat production, and constraints on active and reactive power output. Electrothermal coupling devices: Typical electrothermal coupling devices, such as heat pumps, can be driven with relatively little electricity to convert low-grade heat energy into usable heat energy using a reverse Carnot cycle. They can provide both heating and cooling with a high coefficient of performance (COP). Their operating model is as follows: Energy storage devices: For cold, heat, and electricity, the operational constraints of stationary energy storage facilities can be uniformly characterized using the following methods: Voltage regulating device: in, For voltage adjustment level selection, it belongs to a set of integer candidates. It is the minimum adjustable step size of the voltage regulation ratio; reactive power compensation device: in, For the selection of reactive power compensation adjustment gears, it belongs to a set of integer candidates; It is the minimum adjustable step size of the reactive power compensation device.

5. The flexible day-ahead adaptive robust scheduling method for resources considering the multi-source load fluctuation risk in the distribution network according to claim 1, characterized in that, In step S4, for the adaptive robust optimization model considering the multivariate endogenous and exogenous uncertainties of source loads, its decision-making mode and model structure are analyzed. The model is decomposed into a pre-scheduling master problem of flexible resource operation mode and an adaptive mitigation sub-problem of source load fluctuation risk. The McCormick envelope approximation method is used to handle the strong coupling of random variables in the sub-problem caused by endogenous uncertainties. First, the robust framework and decision variable types in steps S2 and S3 are analyzed comprehensively. The adaptive robust scheduling model presents a two-stage, three-level style. Discrete variables can be separated from the two-stage model to the first stage. Therefore, the model stages are decoupled in the following way: The model is decomposed into a single-level mixed integer optimization master problem and a two-level optimization sub-problem of risk adaptive scheduling mitigation in the worst case of multivariate uncertain fluctuations of source loads. The master problem is as follows: The sub-problems are as follows: Where y represents the second-stage compensation decision variable, which is a continuous decision variable; The values ​​of the decision variables obtained in the first stage are denoted as u; u is an uncertain variable, and U is the set of all uncertain variables. The dual variables corresponding to the constraints; subproblems This is a two-level optimization problem that cannot be directly solved by encoding. However, it can be equivalently transformed from the inner adaptive adjustment problem into a maximization problem integrated into the upper-level risk identification problem, thus allowing for mathematical modeling. After the dual transformation of the subproblems, the following model can be obtained: in Let be the dual variables corresponding to the constraints of the original problem in the subproblem; M is a relatively large constant boundary added to avoid the subproblem becoming unbounded; u is an uncertain variable; h, G, H, and D are the same coefficient matrices as in the original problem; it can be noted that the objective function contains This strongly coupled nonlinear term is approximated by a McCormick envelope method: Define the auxiliary variables as described in the above formula. And the following constraint group is given: in, 、 、 、 They are random variables With dual variables The upper and lower bound parameters; the dual subproblem after approximate linear substitution is as follows: Where m and n are respectively and The number of elements.

6. The flexible day-ahead adaptive robust scheduling method for resources considering the multi-source load fluctuation risk in the distribution network according to claim 1, characterized in that, In step S5, the following constraint generation algorithm is used to iteratively solve the adaptive robust scheduling model established in steps S1-S4; the solution algorithm for the flexible resource day-ahead adaptive robust scheduling framework is as follows: Step 1: Initialization: Let the upper bound The lower realm Iteration counter With the ideal value of the random variable Perform iterative initialization and select an appropriate convergence gap. Step 2: Analyze the fluctuations of parameters such as wind, solar, and load in the risk scenario obtained from the l-th generation sub-problem. Substitute into the main scheduling problem In the process, the optimal flexible resource operation mode pre-scheduling decision for the current iteration is obtained by solving the problem. Coupled with cost variables The value of is determined and the lower bound is updated. Step 3: Take the ingredients from Step 2... Received Substituting the subproblem, we can obtain the values ​​of random variables in the most risky scenario when the distribution network is affected by wind, solar, and load fluctuations during the current pre-scheduling decision. and update the upper bound. Step 4: Add the following constraints to the main problem to generate column constraints: Step 5: If Then proceed to Step 2 and let ； Otherwise, the problem converges, and the algorithm terminates; The optimal solution obtained is a flexible resource-ahead adaptive robust scheduling scheme that takes into account the mitigation of multiple fluctuation risks of power generation and load in the distribution network.

Images