A multi-objective optimal operation method and system for a distribution network

Abstract

The present invention discloses a multi-objective optimal operation method and system for a distribution network, belonging to the technical field of optimal dispatching of distribution networks. The prior art does not comprehensively consider the adjustment rate constraints of power adjustment resources, resulting in the inability to implement the optimization results in the adjustable resource power adjustment in actual projects. A multi-objective optimal operation method for a distribution network considering the adjustment rate of adjustable resources can meet the maximum adjustment rate constraints, thereby making the optimization results more reasonable, conforming to the actual constraints in the industrial community, and having the advantages of making the simulation calculation of the optimal dispatching of the distribution network more accurate and meeting the actual constraint requirements of the equipment participating in the optimization in the project. When dealing with the problem of adjustable resources participating in the optimal operation of the distribution network, the present invention can set different optimal objective function expressions, different decision variables and constraint conditions according to different optimization scenarios, and flexibly adopt numerical optimization algorithms and intelligent algorithms. The scheme is simple, practical and truly feasible.

Description

Technical Field

[0001] The present invention relates to a multi-objective optimal operation method and system for a distribution network, belonging to the technical field of optimal dispatching of distribution networks. Background Art

[0002] In many existing multi-objective optimization problems of distribution networks containing renewable energy, in order to ensure the maximum consumption of renewable energy, it is assumed that renewable energy is fully generated and utilized, while ignoring that in a certain operation scenario, the demand on the load side is not so large and the remaining capacity or regulation rate of the energy storage system is insufficient, resulting in the fact that the generated power of renewable energy cannot be fully consumed and stored in the energy storage system at the optimization moment.

[0003] Furthermore, in the calculation of optimal dispatching of distribution networks, the existing technology lacks the consideration of the regulation rate constraint of power regulation resources, often over-amplifying the regulation ability of adjustable resources participating in the optimal operation of the distribution network, and the optimization result cannot be implemented in the power regulation of adjustable resources in actual projects. For example, when optimizing according to the existing technology, there may be a situation where the increase in the electrical load between adjacent optimization moments is very large. If the planned power generation of a certain power generation device is very small at the previous dispatching moment and its planned power generation is required to be very large at the next dispatching moment to meet the power balance, and the difference in power generation between the front and back moments exceeds its maximum regulation rate, the power supply reliability cannot be met in actual regulation.

[0004] Further, in the control calculation of optimal dispatching of distribution networks, the existing technology rarely considers the countermeasures for the problem that the optimal solution of the lower-layer optimization cannot meet the global optimum of the upper-layer optimization after optimization, thus making the final optimization result not the optimal one. Summary of the Invention

[0005] Aiming at the defects of the existing technology, the first object of the present invention is to provide a multi-objective optimal operation method and system for a distribution network, which can meet the maximum regulation rate constraint, thereby making the optimization result more reasonable, conforming to the actual constraints in the industrial field, having the advantages of making the calculation of optimal dispatching of the distribution network more accurate, and meeting the actual constraint requirements of the equipment participating in the optimization in the project.

[0006] The second object of the present invention is to provide a method for establishing a sub-objective function of an upper-layer optimization objective function as the objective function of a lower-layer optimization, which can avoid to a certain extent the situation that the final optimization result may not be the optimal due to optimization, and has the advantage of making the final optimization result of the adjustable resources participating in the distribution network optimization operation problem more reasonable; and in the lower-layer optimization, a bilinear objective function that restricts each other is constructed, which can avoid to a certain extent the situation that the optimal solution of the lower-layer optimization cannot meet the global optimum of the upper-layer optimization due to optimization, and realizes the effect of "both ensuring the peak shaving and valley filling of the total active load of the system by the orderly charging of electric vehicles and ensuring that no new charging peak is formed" in the lower-layer optimization for the distribution network multi-objective optimization operation method and system.

[0007] To achieve one of the above objects, the first technical solution of the present invention is:

[0008] A distribution network multi-objective optimization operation method considering the adjustable resource regulation rate,

[0009] comprising the following steps:

[0010] Step 1: Obtain several objectives that can assist the distribution network optimization operation;

[0011] The several objectives include renewable energy, energy storage system, temperature-controlled load, and electric vehicle load;

[0012] Step 2: According to the several objectives in Step 1, set the corresponding decision variables and their constraint conditions;

[0013] When renewable energy participates in regulation, select the output power of each grid-connected renewable energy unit in each optimization moment after participating in the optimization as the decision variable, and the regulation of the decision variable does not need to consider the regulation rate constraint, and its constraint condition is that the grid-connected power is greater than zero and less than the predicted output power;

[0014] When the energy storage system participates in regulation, select the output power of each energy storage system in each optimization moment as the decision variable, and both the regulation rate constraint and the energy storage output constraint need to be considered;

[0015] When the temperature-controlled load participates in regulation, select the total output power of all electric refrigeration units supplying cooling to each cold load in each optimization moment as the decision variable, and the regulation rate constraint of the temperature-controlled load needs to be considered while considering the user satisfaction constraint;

[0016] When the electric vehicle load participates in regulation, the main body participating in power regulation is the output power of the charging device at the network topology node in each optimization moment, and the output power of a charging device at a certain place at a certain moment is the opposite of the sum of the charging powers of all electric vehicles at that place at that moment;

[0017] Step 3: Construct system constraints based on the decision variables and their constraints in Step 2;

[0018] The system constraints include grid power balance constraints and grid power flow constraints;

[0019] The grid power balance constraint is used to control the total power supply to be equal to the total power consumption, which is reflected in the equal input and output power of each node in the network topology. The grid power balance constraint needs to be satisfied through power flow calculation; by inputting the net injection power corresponding to each type of node, the impedance of the distribution network line, and the connectivity information of the distribution network topology, the voltage and power corresponding to each node, as well as the losses on each line, are calculated;

[0020] The grid power flow constraint is used to control the voltage amplitude to fluctuate within the allowable range, and at the same time control the active power transmission and reactive power transmission of the line to fluctuate within the allowable range;

[0021] Step 4: Construct an optimization objective with the lowest daily operating cost and the highest comprehensive energy efficiency based on the system constraints in Step 3;

[0022] The lowest daily operating cost means the lowest operating cost of the distribution network during the whole-day scheduling period, which includes the cost of purchasing electricity from the superior grid by the distribution network, the operating income of the load aggregator, the grid-connected scheduling cost of distributed power sources, and the cost of wind and light abandonment;

[0023] The highest comprehensive energy efficiency means the highest comprehensive energy efficiency of the distribution network during the whole-day scheduling period, which is used to reflect the ratio of the total load power consumption to the total power supply within a day in the distribution network;

[0024] Step 5: Develop an optimization strategy for adjustable resources based on the optimization objective in Step 4;

[0025] The optimization strategy includes the following:

[0026] For the power generation of renewable energy, construct a downward regulation model to control its output power;

[0027] The downward regulation model intentionally reduces the output of renewable energy through power electronic components at each optimization moment, and achieves optimization objectives such as economic optimality and the highest comprehensive energy efficiency on the basis of meeting power balance;

[0028] At the same time, the cost of wind and light abandonment is added to the expression of the objective function of the daily operating cost, as a loss of electricity sales revenue caused by reducing power generation, so that renewable energy and resources on the grid side, load side, and storage side conduct a game according to the differences in the cost of wind and light abandonment and scheduling cost and the optimization objective of the highest comprehensive energy efficiency, and obtain the optimal adjustable resource scheduling plan;

[0029] For the load that can be curtailed, the temperature control units participating in the regulation are uniformly scheduled;

[0030] For the electric vehicle load, in the case of disordered charging before the electric vehicles participate in regulation, the situation of starting charging at the middle moment of a certain scheduling period is equivalent to starting charging from the starting moment of this period, and the situation of stopping charging at the middle moment of a certain scheduling period is equivalent to stopping charging at the end moment of this period. Then, the charging power corresponding to the actual charging duration within the scheduling period where the charging start moment or termination moment is located is converted into the average charging power throughout the period, so as to calculate the load magnitude when the electric vehicles charge disorderly before participating in regulation;

[0031] Obtain the output magnitude of each adjustable resource at each optimization moment; since the adjustable resources are distributed on different nodes in the network topology, therefore, combining the output of the adjustable resources at each optimization moment and the load magnitude of each node, calculate the net injection power of the nodes, and substitute it into the power flow calculation to meet the node power balance constraint and network constraint;

[0032] Step Six: Optimize the operation process according to the optimization strategy in Step Five;

[0033] Adopt an optimization algorithm to find the optimal solution to the problem of the adjustable resources assisting the optimal operation of the distribution network, and execute according to the optimal solution to realize the multi-objective optimal operation of the distribution network.

[0034] Through continuous exploration and experiments, the present invention proposes an optimization method considering the adjustment rate constraint of adjustable resources, which can meet the maximum adjustment rate constraint, and further make the optimization result more reasonable, conform to the actual constraints in the industrial circle, and has the advantages of making the simulation calculation of the optimal dispatching of the distribution network more accurate and meeting the actual constraint requirements of the equipment participating in the optimization in the project.

[0035] The present invention fully considers various operation scenarios and proposes the concept of "downward adjustment" of the output power, that is, by intentionally reducing the output of renewable energy at each optimization moment through power electronic components, to achieve optimization goals such as economic optimality and the highest comprehensive energy efficiency on the basis of meeting power balance. The "downward adjustment" of the renewable energy output power will inevitably cause problems such as wind curtailment and light curtailment. In order to reflect the interactive relationship among the source, network, load, and storage, and to realize meeting power balance not only relying on the "downward adjustment" of the renewable energy output power, the present invention adds the wind curtailment and light curtailment costs to the expression of the daily operation cost of the objective function, so that the renewable energy, network side, load side, and storage side resources conduct a game according to the differences in wind curtailment, light curtailment costs and dispatching costs and the optimization goal of the highest comprehensive energy efficiency, and obtain the optimal adjustable resource dispatching plan.

[0036] Furthermore, when dealing with the problem of adjustable resources participating in the optimal operation of the distribution network, the present invention can set different objective function expressions, different decision variables and constraint conditions according to different optimization scenarios, and flexibly adopt numerical optimization algorithms and intelligent algorithms. The solution is simple, practical and truly feasible.

[0037] Furthermore, after each generation of each group of genes (decision variables) is generated in the present invention, the output power at each optimization moment is obtained according to each adjustable resource model. Since the adjustable resources are distributed on different nodes of the network topology, the net injection power of the nodes can be calculated by combining the output power of the adjustable resources at each optimization moment and the load size of each node, and then substituted into the power flow calculation. Then, under the condition of satisfying the power balance constraint of the power grid and the power flow constraint of the power grid, the fitness function value corresponding to the optimization target is calculated, and more excellent genes are screened accordingly. Finally, according to the screened excellent genes, steps such as crossover and mutation are performed to generate offspring genes, which participate in the optimization again, and this is taken as a cycle.

[0038] The adjustment rate is the change magnitude of the output power of the adjustable resource or device per unit time. The adjustment rate can reflect the change amplitude of the output power of the device resource in a short period of time. For a power generation device, it is reflected in the change magnitude of its power generation power per unit time; for an electrical load, it is reflected in the change magnitude of its power consumption power per unit time. Within a unit time, the power adjustment rate of the power generation device cannot exceed its inherent adjustment rate upper limit.

[0039] As a preferred technical measure:

[0040] In the second step, the grid connection power is greater than zero and less than the predicted output power. The specific calculation formula is as follows:

[0041]

[0042] Among them, is the optimized dispatch plan output power of the dth renewable energy grid-connected unit at time t, kW; is the day-ahead predicted output power of the dth renewable energy grid-connected unit at time t, kW;

[0043] The calculation formula for the adjustment rate constraint of the energy storage system is as follows:

[0044]

[0045] Among them, is the output power of the eth energy storage system at time t (negative during charging and positive during discharging), kW; is the output power of the eth energy storage system at time t - 1, kW; is the adjustment rate upper limit of the eth energy storage system, kW / h;

[0046] The calculation formula for the capacity constraint of the energy storage system is as follows:

[0047]

[0048] Among them, is the output electrical energy of the f-th energy storage system at time t, in kWh; is the total capacity of the f-th energy storage system, in kWh; is the initial state of charge of the f-th energy storage system, and T is the number of cumulative charge (discharge) times.

[0049] As a preferred technical measure:

[0050] The calculation formula of the mathematical model of the cooling load is as follows:

[0051]

[0052] Among them, is the indoor temperature of the g-th cooling load (hall) at time t, in °C; is the indoor temperature of the g-th cooling load (hall) at time t + 1, in °C; C g is the equivalent heat capacity of the g-th cooling load (hall), in kW / °C; R g is the equivalent thermal resistance of the g-th cooling load (hall), in °C / kW; is the outdoor ambient temperature of the cooling load (hall) at time t, in °C; is the total output power of all electric refrigeration units supplying cooling to the g-th cooling load (hall) at time t, in kW;

[0053] The user satisfaction is related to the indoor temperature of the cooling load (hall), and its value range is [0,1], which reflects the adjustable ability and adjustable potential of the cooling load; the calculation formula of the user satisfaction constraint of the cooling load (hall) is as follows:

[0054]

[0055] Among them, is the user's expected indoor temperature of the g-th cooling load (hall) at time t, in °C; u is the user satisfaction; if the lowest boundary value of the user satisfaction is specified, the difference between the deduced indoor temperature value and the user's expected temperature is the adjustable range of the corresponding cooling load;

[0056] The set temperature of the electric refrigeration unit should be the same as the user's expected temperature, but considering that there is a temperature control dead zone in the electric refrigeration unit, and δ is defined as the hysteresis width between the upper and lower limits of the operating temperature of the electric refrigeration unit, then the relationship between the upper and lower limits of the actual user's expected temperature fluctuation and the set temperature of the electric refrigeration unit is as follows:

[0057]

[0058] Among them, is the set temperature of all electric refrigeration units supplying cooling to the g-th cooling load (hall) at time t, in °C, and numerically it is the same as is equal; δ is the width of the hysteresis loop between the upper and lower limits of the operating temperature of the electric refrigeration unit, °C; and are the upper and lower limits of the fluctuation of the actual user's desired temperature, °C, respectively, indicating that the indoor temperature within the range of meets the user's expectations and also increases the adjustable potential of the cooling load to a certain extent;

[0059] The mathematical model of the electric refrigeration unit is as follows:

[0060]

[0061] where is the coefficient of performance of all electric refrigeration units for cooling the g-th cooling load (hall); is the total power consumption of all electric refrigeration units for cooling the g-th cooling load (hall) at time t, kW;

[0062] The calculation formula for the regulation rate constraint of all electric refrigeration units for cooling the g-th cooling load (hall) is as follows:

[0063]

[0064] where is the total output power of all electric refrigeration units for cooling the g-th cooling load (hall) at time t-1, kW; is the upper limit of the regulation rate of all electric refrigeration units for cooling the g-th cooling load (hall), kW / h;

[0065] In addition, the constraint on the upper limit of the power consumption value of the electric refrigeration machine also needs to be considered, as shown in the following formula:

[0066]

[0067] where is the upper limit of the total output power of all electric refrigeration units for cooling the g-th cooling load (hall), kW;

[0068] Whether an electric vehicle chooses to charge at a certain charging pile at a certain moment belongs to a 0-1 programming problem. The decision variable is defined as the charging state matrix of the j-th type of electric vehicle at the i-th charging pile at time t The elements in are all 0-1 binary integer variables, taking 1 when choosing to charge and 0 when choosing not to charge;

[0069] The output power of the i-th charging pile at time t The calculation formula of is as shown in the following formula:

[0070]

[0071] Used to reflect the relationship between the output power of a certain charging pile at a certain optimization moment and the charging power of all electric vehicles, and the existing constraints are reflected in the element values of the charging state matrix ;

[0072] Among them, is the output power (negative number) of the i-th charging pile at time t, in kW; is the charging state matrix of the j-th type of electric vehicle at the i-th charging pile at time t, and its dimension is 1×N j , where N j is the total number of the j-th type of electric vehicle; is the charging power of the j-th type of electric vehicle at the i-th charging pile at time t, in kW, and its dimension is N j ×1; J is the total number of types of electric vehicles participating in regulation in the system;

[0073] The calculation formula for satisfying the output power limit constraint of each charging pile at each moment is as follows:

[0074]

[0075] Among them, is the upper limit of the charging power of the i-th charging pile at a single optimization moment, in kW;

[0076] The remaining charging duration T of each electric vehicle until it is fully charged ch The calculation formula is as follows:

[0077]

[0078] Among them, for a certain electric vehicle participating in regulation, s is the current driving mileage of the electric vehicle, in km; W 100 is the power consumption of the electric vehicle per 100 km, in kWh / 100 km; P ch is the average charging power, in kW; η ch is the charging efficiency;

[0079] The calculation formula for the charging duration constraint of the electric vehicle is as follows:

[0080]

[0081] Among them, is the maximum charging duration of the w-th electric vehicle starting from time t, in h; is the actual charging duration of the w-th electric vehicle starting from time t, in h.

[0082] As a preferred technical measure:

[0083] In the third step described above, the method of power flow calculation is as follows:

[0084] First, it is necessary to determine the node type. The node type can be divided into three categories: PV nodes, PQ nodes, and Vδ nodes. PQ nodes are nodes with known injected active power and reactive power, but unknown voltage amplitude and phase angle. PV nodes are nodes with known voltage amplitude and active power, but unknown node voltage phase angle and reactive power. Vδ nodes are also called balance nodes. The voltage amplitude and phase angle of this node are known. The essence of power flow calculation is to supply power to other nodes by adjusting the active and reactive power output of the balance node to maintain the power balance of the power grid.

[0085] There are n nodes in the network, among which there are r PV nodes, 1 Vδ node, and n - r - 1 PQ nodes. The power flow equation is obtained, and its calculation formula is as follows:

[0086]

[0087] Among them, P SP and Q SP are the total node net injected active power vector (W) and the total node net injected reactive power vector (Var) respectively; U is the total node voltage vector, V; Y is the node admittance matrix, S; U SP is the given initial node voltage of the total node, V; ΔP is the active power injection deviation of PQ and PV nodes, W; ΔQ is the reactive power injection deviation of PQ nodes, Var; ΔU is the square difference of PV node voltage amplitude, V 2 ;

[0088] The calculation formula of voltage is as follows:

[0089]

[0090] Among them, V g is the voltage of node g, V; and are the upper and lower limits of the voltage amplitude of node g respectively, V;

[0091] The calculation formulas of active power transmission and reactive power transmission of the line are as follows:

[0092]

[0093]

[0094] Among them, P h is the active power on line h, W; and are the upper and lower limits of the active power on line h respectively, W; n b is the total number of branches in the network;

[0095] Q h is the reactive power on line h, Var; and are the upper and lower limits of the reactive power on line h, Var respectively.

[0096] As an optimal technical measure:

[0097] In the fourth step, the calculation formula with the lowest operating cost of the distribution network over the whole-day scheduling period as the optimization objective is as follows:

[0098]

[0099] where is the power purchase cost of the distribution network from the superior power grid at time t, yuan; is the operating income of the load aggregator at time t (the distribution network pays the load aggregator), yuan; is the grid connection scheduling cost of distributed power sources at time t, yuan; is the penalty cost for wind curtailment at time t, yuan; is the cost of photovoltaic curtailment at time t, yuan, and T is the total number of optimization times;

[0100] where the power purchase cost of the distribution network from the superior power grid at time t is shown as the following formula:

[0101]

[0102] where is the unit price of power purchase from the superior power grid at time t, yuan / kWh; P PCC (t) is the power exchange value of the connection line with the superior power grid at time t, kW; Δt is the duration of a single optimization time, h;

[0103] The operating income of the load aggregator at time t The calculation formula is as follows:

[0104]

[0105] where P ESS,a (t) is the output power of the a-th energy storage system at time t (positive during discharging; negative during charging), kW; is the energy storage scheduling price of the a-th energy storage system at time t, yuan / kWh; A is the total number of energy storage systems participating in regulation; P EVA,b (t) is the output power of the b-th electric vehicle at time t (negative during charging, discharging is not considered), kW; is the charging electricity price of the b-th electric vehicle at time t, yuan / kWh; B is the total number of electric vehicles participating in regulation; P FL,c (t) is the response power of the c-th flexible load participating in regulation at time t, kW; The call cost of the c-th flexible load participating in regulation at time t, yuan / kWh; C is the total number of electric vehicles participating in regulation;

[0106] The grid-connected dispatching cost of distributed power sources at time t The calculation formula is as follows:

[0107]

[0108] Among them, P DG,m (t) is the output power of the m-th grid-connected distributed energy source at time t, kW; is the maintenance cost of the m-th grid-connected distributed energy source at time t, yuan / kWh; M is the total number of grid-connected distributed energy sources;

[0109] The curtailment cost at time t and the curtailment cost of light The calculation formula is as follows:

[0110]

[0111]

[0112] Among them, pnt WT (t) is the unit price of curtailment cost of wind at time t, yuan / kWh; is the day-ahead predicted output of the k-th grid-connected wind turbine at time t, kW; is the output of the k-th grid-connected wind turbine participating in the optimized dispatching plan at time t, kW; K is the total number of grid-connected wind turbines;

[0113] pnt PV (t) is the unit price of curtailment cost of light at time t, yuan / kWh; is the day-ahead predicted output of the l-th grid-connected photovoltaic unit at time t, kW; is the output of the l-th grid-connected photovoltaic unit participating in the optimized dispatching plan at time t, kW; L is the total number of grid-connected photovoltaic units;

[0114] The specific expression with the highest comprehensive energy efficiency of the distribution network as the optimization goal for the whole-day dispatching cycle is as follows:

[0115]

[0116] Among them, P LOAD,r (t) is the load size at node r at time t, kW; R is the total number of nodes in the distribution network; ε is the renewable energy penetration rate of the superior grid; η re is the renewable energy generation efficiency of the superior grid; η nre is the non-renewable energy generation efficiency of the superior grid.

[0117] As an optimal technical measure:

[0118] In step 5, for the electric vehicle load, optimize by constructing a lower-layer dual optimization objective model and a total optimization objective model after linear superposition;

[0119] The lower-layer dual optimization objective model includes the following:

[0120] First, perform combined optimization on the electric vehicles charging at each charging pile node in the network topology at each optimization moment, with the lowest total daily charging cost and the lowest total daily net load growth ratio as the optimization objectives, and solve the optimal charging state matrix That is, the optimal combined charging plan for the j-th type of electric vehicle at the i-th charging pile at time t; then, based on the optimal combined charging plan of the electric vehicle obtained from the lower-layer optimization, solve the output power of each charging pile at each optimization moment, and use it as a constant load to participate in the solution of the optimal scheduling plan of other adjustable resources in the distribution network;

[0121] The objective function expression of the total daily charging cost of electric vehicles is as follows:

[0122]

[0123] Where C EVA is the total daily charging cost of electric vehicles, in yuan; P EVA,b (t) is the output power of the b-th electric vehicle at time t (negative during charging, discharging is not considered), in kW; is the charging electricity price of the b-th electric vehicle at time t, in yuan / kWh; B is the total number of electric vehicles participating in regulation; Δt is the duration of a single optimization moment, in h; T is the total number of optimization moments;

[0124] The objective function expression of the total daily net load growth ratio of electric vehicles is as follows:

[0125]

[0126] Where is the daily net load growth ratio of electric vehicles in the network topology; P Load (t) is the magnitude of the total active load in the network topology before the electric vehicle charges at time t, in kW;

[0127] Normalize the values of the two objective functions C EVA and R ΔLoad first, so that the values of C EVA and R ΔLoad obtained from each optimization calculation are in the same order of magnitude, and then specify the weights and Linearly superpose two objective functions into a single objective function to form a total optimization objective model, and its calculation formula is as follows:

[0128]

[0129] Among them, f sublayer is the sum of the normalized objective functions of the lower-layer optimization; and are the weights of the two optimization objectives of the total daily charging cost and the total daily net load growth ratio respectively; and are the normalization coefficients of the two objective function values of the total daily charging cost and the total daily net load growth ratio respectively;

[0130] Since the definition of the total daily net load growth ratio is that for all optimization moments in a day, calculate the ratio of the sum of the charging powers of all electric vehicles at each moment to the total active load in the network at this optimization moment, and then calculate the sum of the net load growth ratios of all optimization moments; assuming that there are T optimization moments in a day, then can take the value of 1 / T, and the value range of the total daily net load growth ratio of the normalized objective function is [0,1]; therefore, can take the value of where is the total charging cost when all electric vehicles charge at the maximum power during peak electricity price moments, and the value range of the total daily charging cost of the normalized objective function is also [0,1]; the weights and of the two optimization objectives should be specified according to the actual optimization objective focus and satisfy

[0131] As an optimal technical measure:

[0132] The total optimization objective model is to minimize the sum f sublayer of the normalized objective functions of the lower-layer optimization, and it is necessary to make C EVA as small as possible while making R ΔLoad as large as possible; during peak electricity price moments, due to the high charging unit price, electric vehicles tend to not charge more. Compared with the values of each parameter in the unordered charging state before scheduling, after scheduling, it will make smaller, and then make C EVA smaller, which is the same as the optimization direction of f sublayer , but at the same time it will also make R ΔLoad smaller, which is the same as f sublayerThe optimization directions are opposite, so there is a contradiction between the two sub-objective functions. The lower-layer optimization aims to find a compromise solution among the optimal solutions. During the low electricity price period, since the charging unit price is low, electric vehicles tend to charge more. Compared with the values of each parameter in the unordered charging state before scheduling, after scheduling, it will make have a larger value, and then make C EVA have a larger value, which is opposite to the optimization direction of f sublayer . However, at the same time, it will also make R ΔLoad have a larger value, which is the same as the optimization direction of f sublayer . Therefore, there is a contradiction between the two sub-objective functions. The lower-layer optimization aims to find a compromise solution among the optimal solutions, which can not only ensure the peak shaving and valley filling of the total active load of the system (including the charging load), but also ensure that no new charging peak will be formed.

[0133] The method of the present invention, which establishes the sub-objective function of the upper-layer optimization objective function as the objective function of the lower-layer optimization, can avoid to a certain extent the situation that the final optimization result may not be the optimal due to optimization, and has the advantage of making the final optimization result of the problem of adjustable resources participating in the optimal operation of the distribution network more reasonable.

[0134] Compared with the prior art, the present invention proposes a method of hierarchical optimization. In the lower-layer optimization, the present invention constructs bilinear objective functions that restrict each other. One of them is shown in Equation (28) and belongs to the sub-objective function of the daily operating cost of the upper-layer optimization objective function (shown in Equation (18)), which can avoid to a certain extent the situation that the optimal solution of the lower-layer optimization cannot meet the global optimum of the upper-layer optimization due to hierarchical optimization.

[0135] Furthermore, the present invention normalizes the bilinear objective functions that restrict each other in the lower-layer optimization, assigns weights, and linearly superimposes them into a single linear objective function, achieving the effect of the lower-layer optimization of "ensuring both the peak shaving and valley filling of the total active load of the system (including the charging load) by the orderly charging of electric vehicles and ensuring that no new charging peak will be formed".

[0136] As a preferred technical measure:

[0137] In step six, the optimization algorithm is the NSGA-II optimization algorithm, which specifically includes the following content:

[0138] Before optimization, it is necessary to set the population size N pop and the number of evolution times G pop ; then perform the optimization;

[0139] After the optimization is completed, N pop groups of solutions and the Pareto rank corresponding to each group of solutions will be obtained; after all the solutions with Pareto rank 1 are extracted, groups of solutions will be obtained; subsequently, All corresponding to the group solution The calculated values of all objective functions are normalized. The purpose is to convert all obj calculated values of the objective functions into the same order of magnitude. The normalization method for the calculated value of the m-th

[0140]

[0141] objective function is expressed as follows: where obj is the vector composed of the normalized values of the m-th objective function; is the maximum value among the calculated values of the m-th obj objective function; is the minimum value among the calculated values of the m-th objective function; obj is the vector composed of the calculated values of the m-th objective function; is the vector composed of the calculated values of the m-th obj objective function; After normalizing all M

[0142] objective functions according to Equation (31), the next step is to specify the weights obj of the M obj optimization objectives satisfying Then, through the following formula, calculate the sum of the normalized values of the M obj objective functions corresponding to each group of the group solution. The specific calculation formula is shown as follows:

[0143]

[0144] where is the vector composed of the sum of the normalized values of the M obj objective functions corresponding to each group of solutions in the group solution, is the weight occupied by the m-th obj optimization objective;

[0145] Finally, a group of solutions corresponding to the maximum value found in is the optimal solution to the problem of optimizing the operation of the adjustable resource-assisted distribution network.

[0146] As an optimized technical measure:

[0147] It also includes the initialization of decision variables;

[0148] It is necessary to set the population size participating in the operation, and regard all decision variables participating in the optimization as genes; in the process of initializing the genes, it is necessary to consider the adjustment rate constraint and set the initialization method for different adjustable resources, which includes the unified generation method and the special generation method;

[0149] The unified generation method is divided into the unified generation method of the initial value of the gene at the first optimization moment and the unified generation method of the initial value of the gene at non-first optimization moments;

[0150] The unified generation method of the initial value of the gene at the first optimization moment specifically includes the following content:

[0151] The generation of the initial value of the gene of different adjustable resources at the first optimization moment is not restricted by the adjustment rate, as shown in the following formula:

[0152]

[0153] Among them, is the initial value of the v-th gene in population p; is the lower limit of the value of the v-th gene; is the upper limit of the value of the v-th gene; R is a random real number uniformly distributed within [0,1];

[0154] The unified generation method of the initial value of the gene at non-first optimization moments specifically includes the following content:

[0155] The generation of the initial value of the gene of different adjustable resources at non-first optimization moments is restricted by the value size and adjustment rate in the previous moment, as shown in the following formula:

[0156]

[0157] Among them, the expressions of max(A, B) and min(A, B) respectively represent taking the larger one of elements A and B and taking the smaller one of elements A and B; is the initial value of the (v - 1)-th gene in population p; is the upper limit of the adjustment rate of the adjustable resource represented by the v-th gene in population p at the unit optimization moment; therefore, the essence of the unified generation method of the initial value of the gene at non-first optimization moments is to redefine the upper and lower limits of the gene value;

[0158] The special generation method updates the upper and lower limits of the output power value of the energy storage system at the next optimization moment based on the state of charge (SOC) of the energy storage system at the end of the previous scheduling period and the adjustment rate constraint;

[0159] The generation method of the initial value of the gene at the first optimization moment of the special generation method is the same as the unified generation method of the initial value of the gene at the first optimization moment of the unified generation method;

[0160] In the method for generating the initial value of genes at non-initial optimization moments of a special generation method, further constraints are added to the upper and lower limits of gene values; for the method for generating the initial value of genes at non-initial optimization moments of an energy storage system, as shown in the following formula:

[0161]

[0162] Wherein, is the cumulative value of gene values from the 1st to the (v - 1)th gene in population p, in kW; is the initial value of the u-th gene in population p, in kW; E ess is the capacity of the energy storage system, in kWh; SOC ess is the initial state of charge of the energy storage system.

[0163] To achieve one of the above purposes, the second technical solution of the present invention is:

[0164] A multi-objective hierarchical optimization operation system for a distribution network considering the adjustable resource regulation rate, which includes:

[0165] One or more processors;

[0166] A storage device for storing one or more programs;

[0167] When the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned multi-objective hierarchical optimization operation method for a distribution network considering the adjustable resource regulation rate.

[0168] Compared with the prior art, the present invention has the following beneficial effects:

[0169] Through continuous exploration and experiments, the present invention proposes an optimization method considering the adjustable resource regulation rate constraint, which can meet the maximum regulation rate constraint, and further make the optimization result more reasonable, meet the actual constraints in the industrial field, and has the advantages of making the simulation calculation of the distribution network optimization dispatch more accurate and meeting the actual constraint requirements of the equipment participating in the optimization in the project.

[0170] The present invention fully considers various operating scenarios and proposes the concept of "downward adjustment" of output power, that is, by means of power electronic components, the output of renewable energy is deliberately reduced at each optimization moment, and on the basis of meeting power balance, optimization goals such as economic optimization and the highest comprehensive energy efficiency are achieved; the "downward adjustment" of the output power of renewable energy will inevitably cause problems such as wind curtailment and light curtailment. In order to reflect the interactive relationship between the power source, grid, load, and storage, and to achieve power balance not only by relying on the "downward adjustment" of the output power of renewable energy, the present invention adds the costs of wind curtailment and light curtailment to the expression of the daily operating cost of the objective function, enabling the resources of renewable energy, the grid side, the load side, and the storage side to play a game according to the differences in the costs of wind curtailment, light curtailment, and scheduling costs and the optimization goal of the highest comprehensive energy efficiency, and obtaining the optimal adjustable resource scheduling plan.

[0171] Furthermore, the present invention uses the method of establishing the sub-objective function of the upper-layer optimization objective function as the objective function of the lower-layer optimization, which to a certain extent avoids the situation that the final optimization result may not be optimal caused by hierarchical optimization, and has the advantage of making the final optimization result of the problem of adjustable resources participating in the optimal operation of the distribution network more reasonable.

[0172] Compared with the prior art, the present invention proposes a method of hierarchical optimization. In the lower-layer optimization, the present invention constructs bilinear objective functions that restrict each other and superimposes them into a linear objective function, avoiding the unsatisfactory extreme optimization results that may be caused by single-objective optimization.

[0173] The present invention normalizes the bilinear objective functions in the lower-layer optimization, assigns weights, and linearly superimposes them into a single linear objective function, achieving the effect of the lower-layer optimization of "both ensuring the peak shaving and valley filling of the total active load (including the charging load) of the system by the orderly charging of electric vehicles and ensuring that no new charging peaks are formed". BRIEF DESCRIPTION OF THE DRAWINGS

[0174] Figure 1 The execution block diagram of the branch and bound method of the present invention;

[0175] Figure 2 The execution block diagram of the NSGA-II algorithm of the present invention;

[0176] Figure 3 The network topology structure diagram of the IEEE 33-node of the present invention

[0177] Figure 4 The diagram showing the unordered charging load vs. the ordered charging load of electric vehicles of the present invention;

[0178] Figure 5 The effect diagram of peak shaving and valley filling of the total active load of the grid by the orderly charging of the present invention;

[0179] Figure 6 The distribution diagram of all feasible solutions in the feasible region of the bi-objective optimization of the present invention;

[0180] Figure 7 Diagram showing the optimal value of the objective function corresponding to the Pareto front of the present invention;

[0181] Figure 8 Diagram showing the daily optimal allocation plan of adjustable resources of the present invention. Detailed implementation manners

[0182] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0183] On the contrary, the present invention covers any alternatives, modifications, equivalent methods and solutions made within the spirit and scope of the present invention defined by the claims. Further, in order to enable the public to better understand the present invention, some specific details are described in detail in the following detailed description of the present invention. Those skilled in the art can fully understand the present invention without the description of these details.

[0184] A multi-objective optimal operation method for a distribution network considering the adjustable resource regulation rate includes the following steps:

[0185] Step 1: Define decision variables and their constraints.

[0186] 1. The regulation of renewable energy such as wind and light does not need to consider the regulation rate constraint. The power generation of wind, light, etc. has the characteristics of intermittency and volatility, and it is impossible to truly control its over-generation and under-generation. Therefore, an energy storage system is usually set up to cooperate with it in the optimal operation of the distribution network, and the excess power generation is stored when the power generation is excessive. However, considering the limitation of the energy storage system capacity or the fact that the state of charge (SOC) of the energy storage system at the previous moment is already at a relatively high level, that is, it is impossible to store too much power at this moment, and only wind and light curtailment can be selected to limit the grid-connected power of renewable energy under the condition of meeting the power balance of the distribution network, resulting in the concept of "power down-regulation". Therefore, for renewable energy such as wind and light, the output of each grid-connected renewable energy unit participating in the optimized dispatching plan at each optimization moment can be selected as the decision variable, and the regulation of the decision variable does not need to consider the regulation rate constraint, but only needs to consider that its grid-connected power is greater than zero and less than the predicted output size, which is expressed by formula (1):

[0187]

[0188] In formula (1), is the output of the dth renewable energy grid-connected unit participating in the optimized dispatching plan at time t, in kW; The predicted output of the d-th renewable energy grid-connected unit at time t, in kW.

[0189] 2. For the energy storage system to participate in regulation, the output power of each energy storage system at each optimization time can be selected as the decision variable. Both the regulation rate constraint and the energy storage output constraint need to be considered. The regulation rate constraint limits the upper and lower limits of the discharge power change of the energy storage system at adjacent times; since the total capacity of the energy storage system is fixed, and the state of charge (SOC) of the energy storage system at the end of each time represents the remaining electrical energy at the end of this time, the upper and lower limits of the charging power or the upper and lower limits of the discharge power of the energy storage system in the next optimization time are restricted, forming the energy storage output constraint.

[0190] The regulation rate constraint of the energy storage system can be expressed by Equation (2):

[0191]

[0192] In Equation (2), is the output power of the e-th energy storage system at time t (negative for charging and positive for discharging), in kW; is the output power of the e-th energy storage system at time t - 1, in kW; is the upper limit of the regulation rate of the e-th energy storage system, in kW / h.

[0193] The capacity constraint of the energy storage system can be expressed by Equation (3):

[0194]

[0195] In Equation (3), is the output electrical energy of the f-th energy storage system at time t, in kWh; is the total capacity of the f-th energy storage system, in kWh; is the initial state of charge of the f-th energy storage system, and T is the number of cumulative charge (discharge) times.

[0196] 3. For the load that can be curtailed, taking the temperature control load (electric chiller) of a cinema and a shopping mall as an example to participate in regulation, the total output power of all electric chillers supplying cooling to each cooling load (such as a certain hall of a cinema or a shopping mall) at each optimization time can be selected as the decision variable (assuming that all electric chillers supplying cooling to the same cooling load have the same model and working conditions), and the regulation rate constraint of the temperature control load needs to be considered while considering the user satisfaction constraint.

[0197] The mathematical model of the cooling load (a certain hall of the building for cooling) can be expressed by Equation (4):

[0198]

[0199] In formula (4), is the indoor temperature of the g-th cooling load (hall) at time t, in °C; is the indoor temperature of the g-th cooling load (hall) at time t + 1, in °C; C g is the equivalent heat capacity of the g-th cooling load (hall), in kW / °C; R g is the equivalent thermal resistance of the g-th cooling load (hall), in °C / kW; is the outdoor ambient temperature of the cooling load (hall) at time t, in °C; is the total output power of all electric refrigeration units cooling the g-th cooling load (hall) at time t, in kW.

[0200] The user satisfaction is related to the indoor temperature of the cooling load (hall), and its value range is [0, 1]. It reflects the adjustable ability and potential of the cooling load. The user satisfaction constraint of the cooling load (hall) can be expressed by formula (5):

[0201]

[0202] In formula (5), is the desired indoor temperature of the g-th cooling load (hall) at time t, in °C; u is the user satisfaction. If the minimum boundary value of the user satisfaction is specified, the difference between the deduced indoor temperature value and the desired temperature corresponds to the adjustable range of the cooling load.

[0203] Ideally, the set temperature of the electric refrigeration unit should be the same as the user's desired temperature. However, considering the temperature control dead zone of the electric refrigeration unit, if δ is defined as the hysteresis width between the upper and lower limits of the operating temperature of the electric refrigeration unit, then the relationship between the upper and lower limits of the fluctuation of the actual user's desired temperature and the set temperature of the electric refrigeration unit can be expressed by formula (6).

[0204]

[0205] In formula (6), is the set temperature of all electric refrigeration units cooling the g-th cooling load (hall) at time t, in °C, and numerically equal to ; δ is the hysteresis width between the upper and lower limits of the operating temperature of the electric refrigeration unit, in °C; and are the upper and lower limits of the fluctuation of the actual user's desired temperature, in °C, indicating that the indoor temperature within the range of meets the user's expectations and also increases the adjustable potential of the cooling load to a certain extent.

[0206] The mathematical model of the electric refrigeration unit can be expressed by formula (7).

[0207]

[0208] In formula (7), is the coefficient of performance of all electric refrigeration units for cooling the g-th cooling load (hall); is the total power consumption of all electric refrigeration units for cooling the g-th cooling load (hall) at time t, in kW.

[0209] The regulation rate constraint for all electric refrigeration units for cooling the g-th cooling load (hall) can be expressed by formula (8):

[0210]

[0211] In formula (8), is the total output power of all electric refrigeration units for cooling the g-th cooling load (hall) at time t-1, in kW; is the upper limit of the regulation rate of all electric refrigeration units for cooling the g-th cooling load (hall), in kW / h.

[0212] In addition, for the electric refrigeration machine, the constraint on the upper limit of the power consumption value also needs to be considered, which can be expressed by formula (9).

[0213]

[0214] In formula (9), is the upper limit of the total output power of all electric refrigeration units for cooling the g-th cooling load (hall), in kW.

[0215] 4. For the translatable load, taking electric vehicles as an example, the main body participating in power regulation is the output power of the charging piles at the network topology nodes at each optimization moment, and the output power of a charging pile at a certain place at a certain moment is the opposite number (negative number, not considering V2G) of the sum of the charging powers of all electric vehicles at that place at that moment. Since whether an electric vehicle chooses to charge at a certain charging pile at a certain moment belongs to a 0-1 programming problem, the decision variable can be defined as the charging state matrix of the j-th type of electric vehicle at the i-th charging pile at time t The elements in are all 0-1 binary discrete integer variables, taking 1 when choosing to charge and 0 when choosing not to charge. Assuming that all electric vehicles charge at a constant power, the average charging power of an electric vehicle at a certain optimization moment is a fixed value or 0 (not choosing to charge), so the regulation rate constraint does not need to be considered.

[0216] The output power of the i-th charging pile at time t can be expressed by formula (10).

[0217]

[0218] In formula (10), $P_{i}(t)$ is the output power (negative) of the $i$-th charging pile at time $t$, in kW; $S_{ij}(t)$ is the charging status matrix of the $j$-th type of electric vehicle at the $i$-th charging pile at time $t$, with a dimension of $1\times N$ j , where $N$ j is the total number of the $j$-th type of electric vehicle; $P_{ij}(t)$ is the charging power of the $j$-th type of electric vehicle at the $i$-th charging pile at time $t$, in kW, with a dimension of $N$ j $\times1$; $J$ is the total number of types of electric vehicles participating in regulation in the system.

[0219] Equation (10) describes the relationship between the output power of a certain charging pile at a certain optimization moment and the charging powers of all electric vehicles. There are some constraints formed according to actual regulations, which are reflected in the element values of the charging status matrix . For example, electric buses and electric taxis have fixed charging piles for use; electric private cars are charged more at the charging piles in residential areas, companies, and public parking lots; electric official cars are mainly charged at the charging piles in municipal buildings. Therefore, if it is stipulated that the $j$-th type of electric vehicle is not allowed to charge at the $i$-th charging pile at time $t$, then in this case, it is not necessary to set all the elements in as decision variables, but to specify them as 0. Therefore, for the optimal dispatching of electric vehicles participating in the distribution network, it is no longer necessary to separately constrain the upper and lower limits of the output power of each charging pile, but it is necessary to satisfy the output power limit constraint of each charging pile at each moment, which can be expressed by Equation (11).

[0220]

[0221] In Equation (11), $P_{i}^{max}(t)$ is the upper limit of the charging power of the $i$-th charging pile at a single optimization moment, in kW.

[0222] In addition, similar to the energy storage system, electric vehicles need to track the state of charge of the electric vehicle battery at the end of each optimization moment. However, the difference is that the charging power of the electric vehicle is a fixed value and does not participate in the optimization as a decision variable, but is used to calculate the remaining charging duration $T$ until each electric vehicle is fully charged in Equation (12) ch .

[0223]

[0224] In Equation (12), for a certain electric vehicle participating in regulation, $s$ is the current driving mileage of this electric vehicle, in km; $W$ 100 is the power consumption of this electric vehicle per 100 km, in kWh / 100 km; $P$ ch is the average charging power, in kW; $\eta$ ch is the charging efficiency.

[0225] Therefore, the charging duration constraint of the electric vehicle can be expressed by Equation (13).

[0226]

[0227] In Equation (13), is the maximum charging duration of the w-th electric vehicle starting from time t, in hours; is the actual charging duration of the w-th electric vehicle starting from time t, in hours.

[0228] Step 2: Define system constraints.

[0229] 1. Grid power balance constraint

[0230] Grid power balance means that the total power supply is equal to the total power consumption, which is reflected in the equal input and output powers of each node in the network topology. This constraint needs to be satisfied through power flow calculation. By inputting the net injection power corresponding to each type of node (the difference between the total power generation of the node and the total power load of the node), the impedance of the distribution network line, and the connectivity information of the distribution network topology, the voltage and power corresponding to each node, as well as the losses on each line, can be accurately calculated.

[0231] When performing power flow calculation, it is first necessary to determine the node type. The node type can be divided into three categories: PV nodes, PQ nodes, and Vδ nodes. A PQ node is a node with known injected active power and reactive power, and unknown voltage amplitude and phase angle; a PV node is a node with known voltage amplitude and active power, and unknown node voltage phase angle and reactive power; a Vδ node is also called a balancing node, and the voltage amplitude and phase angle of this node are known. The essence of power flow calculation is to supply power to other nodes by adjusting the active and reactive power outputs of the balancing node to maintain the power balance of the power grid.

[0232] Assume that there are n nodes in the network, where there are r PV nodes, 1 Vδ node, and n - r - 1 PQ nodes. The power flow equation can be expressed by Equation (14).

[0233]

[0234] In Equation (14), P SP and Q SP are the net injected active power vector (W) and net injected reactive power vector (Var) of all nodes respectively; U is the voltage vector of all nodes, in V; Y is the node admittance matrix, in S; U sP is the given initial node voltage of all nodes, in V; ΔP is the active power injection deviation of PQ and PV nodes, in W; ΔQ is the reactive power injection deviation of PQ nodes, in Var; ΔU is the square difference of the voltage amplitude of PV nodes, in V 2 .

[0235] 2. Grid power flow constraint

[0236] When considering the power flow constraints of the power grid, it is required that the voltage amplitude fluctuates within the allowable range, which can be expressed by Equation (15).

[0237]

[0238] In Equation (15), V g is the voltage of node g, in V; and are the upper and lower limits of the voltage amplitude of node g, respectively, in V.

[0239] At the same time, it is also required that the active power transmission and reactive power transmission of the line fluctuate within the allowable range, which can be expressed by Equation (16) and Equation (17) respectively.

[0240]

[0241]

[0242] In Equation (16), P h is the active power on line h, in W; and are the upper and lower limits of the active power on line h, respectively, in W; n b is the total number of branches in the network.

[0243] In Equation (17), Q h is the reactive power on line h, in Var; and are the upper and lower limits of the reactive power on line h, respectively, in Var.

[0244] Step 3: Define the optimization objective.

[0245] The problems of adjustable resources participating in the optimal operation of the distribution network include but are not limited to the following optimization objectives:

[0246] 1. The lowest daily operating cost

[0247] Taking the lowest operating cost of the distribution network in the whole-day scheduling cycle as the optimization objective, it mainly includes the power purchase cost of the distribution network from the superior power grid, the operation income of the load aggregator (including the calling costs of energy storage, electric vehicles, and flexible loads), the grid-connected scheduling cost of distributed power sources, and the cost of wind and light abandonment.

[0248] 2. The highest comprehensive energy efficiency

[0249] Taking the highest comprehensive energy efficiency of the distribution network in a daily scheduling cycle as the optimization goal, it is mainly reflected in the ratio of the total electricity consumption of the total load in the distribution network to the total power supply within a day. Among them, the total load in the distribution network includes the electrical loads at each node, the total load of the energy storage system, the total load of electric vehicles, and other adjustable loads (such as temperature-controlled cooling-type load that can be reduced); a part of the total power supply is the total absorbed power of photovoltaic and wind power in the distribution network, and the other part is the power purchased by the distribution network from the superior grid, which is the amount obtained by converting the power generation efficiency of the superior grid. Therefore, the comprehensive energy efficiency of the distribution network is the result of integrating the energy efficiency within the distribution network and the energy efficiency of the superior grid.

[0250] Step Four: Initialize the decision variables.

[0251] Taking the genetic algorithm in the intelligent algorithm as an example, it is necessary to specify the population size participating in the operation, and regard all decision variables participating in the optimization as genes. During the process of initializing the genes, although the adjustment rate constraint has been specified in Step Two, if all the sets of genes in the first-generation population cannot meet all the constraints required for optimization after initialization, the genetic algorithm will not be able to continue the operation. Therefore, during the process of initializing the genes, it is necessary to consider the adjustment rate constraint and specify the initialization method for different adjustable resources.

[0252] Step Five: Formulate the optimization strategy.

[0253] 1. Formulate the optimization method

[0254] (1) Renewable energy

[0255] In many existing multi-objective optimization problems of distribution networks containing renewable energy, in order to ensure the maximum absorption of renewable energy, it is assumed that renewable energy is fully generated and utilized, while ignoring that in a certain operating scenario, the demand on the load side is not so large and the remaining capacity or adjustment rate of the energy storage system is insufficient, resulting in the fact that the generated power of renewable energy cannot be fully absorbed and stored in the energy storage system at the optimization moment. The present invention takes into account the above operating scenario and proposes the concept of "downward adjustment" of the output power, that is, by means of power electronic components, the output of renewable energy is intentionally reduced at each optimization moment, and on the basis of meeting the power balance, optimization goals such as economic optimality and the highest comprehensive energy efficiency are achieved. The "downward adjustment" of the output power of renewable energy will inevitably cause problems such as wind and light abandonment. In order to reflect the interactive relationship between the source, grid, load, and storage, and to achieve power balance not only by relying on the "downward adjustment" of the output power of renewable energy, the present invention adds the costs of wind and light abandonment to the expression of the daily operating cost of the objective function, so that the resources of renewable energy, the grid side, the load side, and the storage side will play a game according to the different costs of wind and light abandonment and scheduling costs and the optimization goal of the highest comprehensive energy efficiency, and an optimal adjustable resource scheduling plan can be obtained.

[0256] (2) Temperature-controlled load

[0257] For a single temperature-controlled flexible load that can be curtailed, its adjustable potential is not large. Therefore, in practice, the scenario mainly involves the collective control of temperature-controlled loads for regulation. Considering this optimization scenario, the temperature-controlled units participating in regulation should be of the same model and under the same working conditions. Therefore, taking large-space buildings such as shopping malls and cinemas as the main cooling loads, a method for unified scheduling of all temperature-controlled units providing cooling for this cooling load is proposed.

[0258] (3) Electric vehicles

[0259] From Equation (12), the maximum charging duration of a certain electric vehicle starting from a certain optimization moment can be calculated. However, since the electric vehicle does not necessarily start charging at the start time of the scheduling period and does not necessarily stop charging at the end time of the scheduling period, it is very likely to start charging or stop charging at the middle time of the scheduling period. Since the optimal scheduling only focuses on the configuration of each resource at a certain optimization moment, the configuration of resources in the previous scheduling period will not change. Therefore, in the case of disordered charging before the electric vehicle participates in regulation, the situation of starting charging at the middle time of a certain scheduling period is equivalent to starting charging from the start time of this period, and the situation of stopping charging at the middle time of a certain scheduling period is equivalent to stopping charging until the end time of this period. And the charging power corresponding to the actual charging duration within the scheduling period where the charging start time or end time is located is converted into the average charging power throughout the time, so as to calculate the load magnitude when the electric vehicle charges disorderly before participating in regulation. For the case of orderly charging after the electric vehicle participates in optimal regulation, in order to facilitate the regulation and configuration of resources, it is assumed that the optimal scheduling result is that the electric vehicle starts charging at the start time of the scheduling period and stops charging at the end time of the scheduling period.

[0260] After formulating the methods for each adjustable resource to participate in optimization, the output magnitudes of each adjustable resource at each optimization moment can be obtained. Since the adjustable resources are distributed at different nodes in the network topology, the net injection power of the nodes can be calculated by combining the output of the adjustable resources at each optimization moment and the load magnitudes of each node, and substituting it into the power flow calculation to satisfy the node power balance constraints and network constraints expressed by Equations (14) to (17).

[0261] 2. Formulate an optimization algorithm

[0262] Due to the introduction of translatable loads of electric vehicle types, discrete integer variables appear in the multi-objective optimization problem. The present invention proposes to adopt a hierarchical optimization method to reduce the complexity of system optimization control and avoid the occurrence of the curse of dimensionality. Among them, the lower-layer optimization first performs combinatorial optimization on the electric vehicles that may be charged at each charging pile node in the network topology at each optimization moment, with the lowest total daily charging cost and the lowest total daily net load growth ratio as the optimization objectives, and solves the optimal charging state matrix described in Equation (10). That is, the optimal combined charging plan of the jth type of electric vehicle at the ith charging pile at time t; the upper-layer optimization then solves the output power of each charging pile at each optimization moment according to the optimal combined charging plan of the electric vehicle obtained by the lower-layer optimization, and uses it as a constant load to participate in the solution of the optimal scheduling plan of other adjustable resources in the distribution network.

[0263] The reason for the present invention to consider hierarchical optimization is that since the multi-objective optimization problem mixes discrete integer decision variables and continuous decision variables (due to using the charging state matrix as the decision variable, it contains 0-1 programming), the formed optimization problem is too complex. If stochastic optimization algorithms such as genetic algorithms are used to handle such multi-objective optimization problems with mixed discrete (integer) and continuous decision variables, several initial charging state matrices need to be generated in the stage of decision variable initialization. Even if the more efficient implicit enumeration method is adopted, due to the high complexity of the distribution network model (including power flow calculation), the time-consuming for multiple solutions of the distribution network optimization objective is huge. And due to reasons such as the population-based optimization of genetic algorithms, its own optimization rate is slow, and the optimization rate will be even slower when dealing with such multi-objective optimization problems with mixed discrete and continuous decision variables, and it is prone to the situation of combinatorial explosion. Therefore, the present invention introduces a hierarchical optimization method, and the lower layer separately considers the 0-1 programming problem of the orderly charging of electric vehicles containing only discrete integer variables.

[0264] The reason why the present invention considers selecting the lowest total daily charging cost as one of the optimization objectives in the lower-layer optimization is that it enables the electric vehicle load to respond to the incentives of time-of-use charging electricity prices, which is beneficial to peak shaving and valley filling of the daily load curve of the distribution network. And this optimization objective belongs to the sub-optimization objective of the lowest daily operating cost of the upper-layer optimization objective (Equation (28) is part of Equation (18)), which can, to a certain extent, avoid the situation where the optimal solution of the lower-layer optimization cannot meet the global optimum of the upper-layer optimization due to hierarchical optimization. However, due to the incentives of time-of-use electricity prices, if only considering the lowest total daily charging cost as the sole optimization objective, it may cause electric vehicles to concentrate on charging at low electricity price moments, which may result in the emergence of another electricity consumption peak. Therefore, it is necessary to introduce another optimization objective that conflicts with the lowest total daily charging cost, that is, the lowest total growth ratio of the daily net load of electric vehicles. The growth ratio of the net load of electric vehicles at the current optimization moment can reflect the relationship between the charging power of electric vehicles and the total active load of the system at this moment. If the lowest growth ratio of the net load of electric vehicles at the current optimization moment is taken as the optimization direction, it can ensure that a charging peak is avoided as much as possible at the current optimization moment. Taking the lowest total growth ratio of the net load at all optimization moments as the optimization objective can ensure that a charging peak is avoided as much as possible during the valley moments of the total active load of the system. However, if only taking the lowest total growth ratio of the daily net load as the sole optimization objective, since the total active load of the system is lower at the valley moment, it may cause the charging load of electric vehicles at the valley moment of the total active load of the system to be lower than that at the peak moment. Therefore, two linear objective functions are linearly superimposed to obtain the expression shown in Equation (30). According to Equation (30), in order to minimize the sum of the normalized objective functions f sublayer to be minimized, it is necessary to make C EVA as small as possible while making the value of R ΔLoad as large as possible. During peak electricity price moments, due to the higher charging unit price, electric vehicles tend to not charge more. Compared with the values of each parameter in the unordered charging state before scheduling, after scheduling, it will make have a smaller value, and then make the value of C EVA smaller, which is in the same optimization direction as f sublaye r, but at the same time it will also make the value of R ΔLoad smaller, which is in the opposite optimization direction to f sublayer . Therefore, there is a conflict between the two sub-objective functions, and the lower-layer optimization is committed to finding a compromise solution in the optimal solution; during low electricity price moments, due to the lower charging unit price, electric vehicles tend to charge more. Compared with the values of each parameter in the unordered charging state before scheduling, after scheduling, it will make have a larger value, and then make the value of C EVA larger, which is in the opposite optimization direction to f sublayer , but at the same time it will also make the value of R ΔLoad larger, which is in the same optimization direction as f sublayerThe optimization directions are the same, so there is a contradiction between the two sub-objective functions. The lower-layer optimization aims to find a compromise solution in the optimal solution. According to the lower-layer dual-optimization objective model and the total optimization objective model after linear superposition proposed by the present invention, it can, to a certain extent, not only ensure peak shaving and valley filling of the total active load of the system (including the charging load), but also ensure that no new charging peak will be formed.

[0265] Since the decision variables of the lower-layer optimization are discrete integer variables with a value range of {0, 1}, and the objective function of the lower-layer optimization is a linear function. Suppose there are N SOCV decision variables in the lower-layer optimization. Then, even if the idea of hierarchical optimization is adopted, if the "exhaustive method" is used, linear programming needs to be executed times, and for each additional variable, the solution time will double. Therefore, considering the issue of time complexity, the present invention uses the branch and bound method applicable to solving mixed-integer linear programming problems for solution.

[0266] The essence of the branch and bound method is an improved exhaustive method. Its core idea is to decompose the original mixed-integer linear programming problem into solving individual linear programming problems, and continuously update the upper bound (optimal feasible solution) and lower bound (optimal linear relaxation solution) of the original problem during the solution process, avoid the execution process of sub-optimal solutions, and reduce the calculation time. Its execution block diagram is as Figure 1 shown.

[0267] (2) Upper-layer optimization and algorithm selection

[0268] The upper-layer optimization considers choosing the genetic algorithm as the optimization algorithm. The genetic algorithm is a stochastic optimization algorithm simulated and developed based on the genetic principle and natural selection in nature, including self-adaptability and random search. Its advantages are as follows:

[0269] I. Starting from multiple initial values for optimization, it has a stronger ability to obtain the global optimal solution.

[0270] II. It does not require the objective function to be continuously differentiable and has stronger robustness.

[0271] III. It is convenient for dealing with complex non-linear programming problems with non-linear objective functions and a large number of non-linear constraints.

[0272] For the multi-objective problem of optimizing the operation of the adjustable resource-assisted distribution network, since it includes both complex non-linear constraints (power grid power flow constraints) and non-linear objective functions (the calculation result of the power grid power flow "the exchanged power with the external network connection line" participates in the calculation of the objective function), and also includes integer decision variables (belonging to a non-convex optimization problem, and it is easy to obtain a local optimal solution using numerical optimization algorithms), the genetic algorithm is selected for solution.

[0273] Since the genetic algorithm belongs to the stochastic optimization algorithm and has disadvantages such as large computational complexity and long operation time, it is necessary to improve the algorithm. After evaluating the computational rates of various algorithms, the fast non-dominated sorting genetic algorithm (Non-dominated Sorting Genetic Algorithm-II, NSGA-II) is selected to handle the multi-objective problem of optimizing the operation of the adjustable resource-assisted distribution network. This algorithm is based on the classical genetic algorithm, and the differences from the traditional genetic algorithm are as follows:

[0274] I. Add a fast non-dominated sorting process

[0275] The NSGA-II algorithm performs fast non-dominated sorting in the population composed of the first-generation parents and each subsequent generation of offspring and the previous-generation parents to generate a non-dominated solution set, which is used to update the new parent population with the same population size. Suppose there are a total of M obj objective functions and the population size is N pop where the maximum time complexity of the fast non-dominated sorting process is O(M obj (2N pop )) 2 ).

[0276] II. Add the estimation of crowding distance

[0277] The NSGA-II algorithm defines the concept of crowding distance before the process of updating the new parent population, which ensures the diversity of the population. The maximum time complexity of the crowding distance estimation process is O(M obj (2N pop )lg(2N pop ))

[0278] III. Add the crowding degree sorting with the elite retention mechanism

[0279] The NSGA-II algorithm adopts the crowding degree sorting method with the elite retention mechanism in the process of updating the new parent population. It sorts the non-dominated solution set in ascending order of Pareto rank and assigns it to the new parent population until the individuals in the non-dominated solution set corresponding to a certain Pareto rank cannot all be put into the new parent population (the size of the new parent population reaches N). Then, the non-dominated solution set corresponding to this Pareto rank is arranged in descending order of crowding degree and put into the new parent population in turn until the size of the new parent population reaches N

[0280] Adding the crowding degree sorting with the elite retention mechanism can ensure that the obtained local optimal solutions are not lost. While retaining excellent individuals to improve the evolutionary level of the population, it accelerates the convergence speed towards the Pareto front, greatly reducing the time required for algorithm optimization. The maximum time complexity of the crowding degree sorting process with the elite retention mechanism is O(2N poplg(2N pop ))。

[0281] The execution block diagram of the NSGA-II algorithm is as Figure 2 shown.

[0282] Step Six: Execute the optimization of the operation process.

[0283] Application embodiments of the present invention:

[0284] 1. Basic parameters of the embodiment

[0285] The embodiment adopts the network topology structure of IEEE 33 nodes, as Figure 3 shown.

[0286] There are 32 branches in this network, the reference voltage at the head end of the power supply network is 12.66 kV, the three-phase power reference value is 10 MVA, and the total network load is 6220.5 + j2332.7 kVA. The embodiment network includes five typical adjustable resources: distributed photovoltaic, distributed wind power, energy storage, electric vehicles, and temperature-controlled loads.

[0287] (1) Distributed photovoltaic

[0288] In the embodiment network, distributed photovoltaic is mainly distributed in large shopping malls (node 11), large cinemas (node 14), and the rooftops of residential communities (nodes 21, 24, 28), and its parameters are shown in Table 1.

[0289] Table 1 Distributed photovoltaic parameters

[0290] Location Node 11 14 21 24 28 Capacity (kWh) 1200 800 300 200 500 Ramp Rate (kW / h) 120 80 30 20 50 Maintenance Cost (yuan / kWh) 0.01 0.01 0.01 0.01 0.01 Penalty for Curtailed Photovoltaic Power (yuan / kWh) 0.55 0.55 0.55 0.55 0.55

[0291] (2) Distributed wind power

[0292] In the embodiment network, distributed wind power is mainly distributed in the suburbs (nodes 1, 17, 32), and its parameters are shown in Table 2.

[0293] Table 2 Distributed wind power parameters

[0294] Location Node 17 32 1 Capacity (kWh) 1200 700 1100 Ramp Rate (kW / h) 120 70 110 Maintenance Cost (yuan / kWh) 0.035 0.035 0.035 Penalty for Curtailed Wind Power (yuan / kWh) 0.45 0.45 0.45

[0295] (3) Electrical energy storage

[0296] In the embodiment network, electrical energy storage is mainly close to the distribution of high-power distributed energy, and is set to be at the same node as high-power distributed photovoltaic and distributed wind power, and its parameters are shown in Table 3.

[0297] Table 3 Electrical energy storage parameters

[0298] Location Node 11 14 17 1 32 Capacity (kwh) 1600 1000 1000 1000 1200 Ramp Rate (kW / h) 150 150 150 150 150 Maintenance Cost (yuan / kwh) 0.003 0.003 0.003 0.003 0.003 Initial SOC 0.5 0.5 0.5 0.5 0.5

[0299] (4) Electric vehicles

[0300] The embodiments use the electric vehicle data statistically collected in Shenzhen in 2019, covering four types of electric vehicles: buses, taxis, private cars, and official cars. Among them, all buses are of the "BYD K9" type, and all taxis, private cars, and official cars are of the "BYD E6" type. The starting charging times of the four types of electric vehicles all follow a normal distribution, and the daily average driving mileage all follows a lognormal distribution. The parameters corresponding to each type of electric vehicle are shown in Table 4.

[0301] Table 4 Electric Vehicle Parameters

[0302]

[0303] For electric vehicles to participate in the optimal operation of the distribution network as adjustable resources, the embodiments make the following assumptions:

[0304] a. The starting charging times, daily driving mileage, and charging power of each type of electric vehicle are independent random variables.

[0305] b. The charging power of each type of electric vehicle is regarded as a constant power model, without the two stages of constant voltage charging and constant current charging.

[0306] c. After adjustment, each type of electric vehicle will only charge during its main charging period (with the user's charging demand as the first constraint).

[0307] d. All vehicles are fully charged each time.

[0308] e. Assume that the initial SOC of all vehicles at time 0 is 0.5.

[0309] According to Equation (10) and the constraints of the electric vehicle charging specification, the charging status indication matrix shown in Table 5 is obtained.

[0310] Table 5 Charging Status Indication Matrix

[0311]

[0312] (5) Temperature Control Load

[0313] The temperature control load participates in the regulation based on the group control technology. Typical flexible temperature control loads are concentrated in the refrigeration air conditioners of large shopping malls and large cinemas. Large shopping malls need to maintain a low temperature during business hours and a relatively low temperature during non-business hours, while large cinemas require a low temperature throughout the day, representing the difference in adjustable time. There are also differences between them in terms of the user satisfaction coefficient, representing the difference in adjustable potential. The parameters are shown in Table 6.

[0314] Table 6 Temperature Control Load Parameters

[0315] Large Shopping Mall Large Cinema Location Node 11 14 Number of Temperature Control Loads 100 50 Minimum Satisfaction Coefficient 0.9 0.8 Coefficient of Performance (COP) of Refrigeration 3 2.6 Equivalent Thermal Resistance (℃ / kW) 2.5 1.5 Equivalent Heat Capacity (kW / ℃) 2.5 1.5 Ramp Rate (kw / h) Maximum Output Power / 3 Maximum Output Power / 3

[0316] Among them, when calculating the upper limit of the output power value of the temperature control load, boundary conditions need to be considered, that is, when considering the user comfort level of 1 and reaching the lower boundary of the temperature control load dead zone, the maximum output power required at each moment.

[0317] When calculating the lower limit of the output power value of the temperature control load, boundary conditions need to be considered, that is, when considering the user comfort level of 0.8 / 0.9 and reaching the upper boundary of the temperature control load dead zone, the maximum output power required at each moment.

[0318] (6) Others

[0319] The price parameters of the example system are shown in Table 7.

[0320] Table 7 Price Parameters

[0321]

[0322] 2. Results of the Example

[0323] First, referring to Equations (28) to (30), the lower-layer optimization is performed with the lowest total daily charging cost of electric vehicles and the lowest total daily net load growth ratio as the optimization objectives, and a comparison chart of the unordered charging load and the ordered charging load of electric vehicles is obtained, as shown in Figure 4 shown. A comparison chart of the total active power load of the power grid during the ordered charging of electric vehicles and the total active power load of the power grid during the unordered charging is shown in Figure 5 shown.

[0324] From Figure 5 it can be seen that under the premise of taking the user's charging demand as the first constraint, that is, the user charges during the main charging time, and at the same time considering the two optimization objectives of the lowest total daily charging cost of electric vehicles and the lowest total daily net load growth ratio, by using the lower-layer optimization method and optimization algorithm proposed in the present invention, it is possible to avoid the emergence of a new charging load peak while flattening the peak and filling the valley of the total active power load of the power grid.

[0325] Then, referring to Equations (18) to (24), the upper-layer optimization is performed with the lowest daily operating cost and the highest comprehensive energy efficiency as the optimization objectives, where the population size is taken as 500 and the number of evolutionary generations is taken as 15, and all feasible solutions of the double-objective optimization feasible region are obtained as shown in Figure 6 shown. The optimal values of the objective functions corresponding to the Pareto front are shown in Figure 7 shown.

[0326] Figure 6 Among them, there are 474 groups of feasible solutions and 13 Pareto levels in the feasible region. The fitness function values of the first optimization objective of the feasible solution set are concentrated in the range of [-60, -57.5] and [1.330×10 5 , 1.365×10 5within the range of the fitness function value of the second optimization objective. Since the optimization direction is minimization, the daily operating cost of the first optimization objective is the same as the value of the first fitness function, and the daily comprehensive energy efficiency of the second optimization objective is the opposite of the value of the second fitness function.

[0327] Figure 7 Among them, there are 17 groups of solutions on the Pareto front, all of which are non-dominated solutions with a Pareto rank of 1 in the feasible region, indicating that other solutions in the feasible region are dominated by these 17 groups of solutions and these 17 groups of solutions do not dominate each other.

[0328] When the two optimized objectives take the same weight, according to equations (31) to (32), the two optimized objective functions are normalized and then superimposed to obtain the daily optimal allocation plan of the adjustable resources in the park, as Figure 8 shown.

[0329] Finally, the daily optimal allocation plan of the adjustable resources and the corresponding optimal daily operating cost of 133,840 yuan and the daily comprehensive energy efficiency of 59.164% are obtained.

[0330] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific embodiments of the present invention, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered by the protection scope of the claims of the present invention.

Claims

1. A multi-objective optimal operation method for distribution networks considering the adjustable resource regulation rate, It is characterized in that it includes the following steps: Step 1: Obtain several objectives that can assist in the optimal operation of the distribution network; The several objectives include renewable energy, energy storage systems, temperature-controlled loads, and electric vehicle loads; Step 2: According to the several objectives in Step 1, set the corresponding decision variables and their constraints; When renewable energy participates in regulation, select the output power of each grid-connected renewable energy unit in each optimization moment after participating in the optimized dispatch plan as the decision variable, and the regulation of the decision variable does not need to consider the regulation rate constraint, and its constraint condition is that the grid-connected power is greater than zero and less than the predicted output power; When the energy storage system participates in regulation, select the output power of each energy storage system in each optimization moment as the decision variable, and both the regulation rate constraint and the energy storage output constraint need to be considered; When the temperature-controlled load participates in regulation, select the total output power of all electric refrigeration units supplying cooling to each cold load in each optimization moment as the decision variable, and the regulation rate constraint of the temperature-controlled load needs to be considered while considering the user satisfaction constraint; When the electric vehicle charging load participates in regulation, the main body participating in power regulation is the output power of the charging device at the network topology node in each optimization moment, and the output power of a certain charging device at a certain moment is the opposite of the sum of the charging powers of all electric vehicles at that place at that moment; Step 3: According to the decision variables and their constraints in Step 2, construct system constraints; The system constraints include grid power balance constraints and grid power flow constraints; The grid power balance constraint is used to control the total power supply to be equal to the total power consumption, which is reflected in the equality of the input and output powers of each node in the network topology. This constraint needs to be satisfied through power flow calculation; by inputting the net injection power corresponding to each type of node, the impedance of the distribution network line, and the connectivity information of the distribution network topology, the voltage and power corresponding to each node, as well as the losses on each line, are calculated; The grid power flow constraint is used to control the voltage amplitude to fluctuate within the allowable range, and at the same time control the active power transmission and reactive power transmission on the line to fluctuate within the allowable range; Step 4: According to the system constraints in Step 3, construct an optimization objective with the lowest daily operating cost and the highest comprehensive energy efficiency; The lowest daily operating cost means the lowest operating cost of the distribution network during the entire day's scheduling cycle, which includes the cost of purchasing electricity from the superior grid by the distribution network, the operating income of the load aggregator, the grid-connected scheduling cost of distributed power sources, and the cost of wind and light abandonment; The highest comprehensive energy efficiency means the highest comprehensive energy efficiency of the distribution network during the entire day's scheduling cycle, which is used to reflect the ratio of the total load power consumption in the distribution network to the total power supply within one day; Step 5: According to the optimization objective in Step 4, formulate an optimization strategy for adjustable resources; The optimization strategy includes the following contents: For the power generation of renewable energy, construct a downward regulation model to control its output power; The downward regulation model intentionally reduces the output power of renewable energy through power electronic components at each optimization moment, and achieves the optimization objectives of economic optimality and the highest comprehensive energy efficiency on the basis of meeting power balance; At the same time, the curtailment costs of wind and light are added to the expression of the daily operating cost of the objective function. As a loss of power sales revenue caused by reducing power generation, it enables renewable energy to compete with the resources on the grid side, load side, and storage side according to the different curtailment costs of wind and light, dispatching costs, and the optimization goal of the highest comprehensive energy efficiency, so as to obtain the optimal adjustable resource dispatching plan; For temperature-controlled loads, the temperature-controlled units participating in regulation are uniformly dispatched; For electric vehicle loads, in the case of unordered charging before the electric vehicles participate in regulation, the situation of starting charging at the middle moment of a certain dispatching period is equivalent to starting charging from the starting moment of this period, and the situation of stopping charging at the middle moment of a certain dispatching period is equivalent to stopping charging at the end moment of this period. And the charging power corresponding to the actual charging duration within the dispatching period where the charging start moment or end moment is located is converted into the average charging power within the whole moment, so as to calculate the load size when the electric vehicles are unordered charging before participating in regulation; Obtain the output power of each adjustable resource at each optimization moment. Combine the output of the adjustable resources at each optimization moment and the load size of each node to calculate the net injection power of the node, and substitute it into the power flow calculation to meet the node power balance constraint and network constraint; Step Six: Optimize the operation process according to the optimization strategy in Step Five; Use the optimization algorithm to find the optimal solution to the problem of optimizing the operation of the adjustable resource-assisted distribution network, and execute according to the optimal solution to achieve the multi-objective optimal operation of the distribution network.

2. A method for multi-objective optimal operation of a distribution network considering the adjustment rate of adjustable resources according to claim 1, characterized in that In the second step described above, the calculation formula for the adjustment rate constraint of the energy storage system is as follows: Among them, is the output power of the th energy storage system at time , which is negative during charging and positive during discharging, ; is the output power of the th energy storage system at time , ; is the upper limit of the regulation rate of the th energy storage system, ; The calculation formula for the capacity constraint of the energy storage system is as follows: Among them, is the output electrical energy of the th energy storage system at moment, ; is the total capacity of the th energy storage system, ; is the initial state of charge of the th energy storage system, is the number of cumulative charge and discharge moments.

3. A method for multi-objective optimal operation of a distribution network considering the adjustment rate of adjustable resources according to claim 2, characterized in that The calculation formula for the cooling load is as follows: wherein, is the indoor temperature of the th cooling load within the time period; is the indoor temperature of the th cooling load within the time period; is the equivalent heat capacity of the th cooling load, ; is the equivalent thermal resistance of the th cooling load, ; is the outdoor ambient temperature of the cooling load at the time; is the total output power of all the electric refrigeration units for cooling the th cooling load at the time, The user satisfaction is related to the indoor temperature of the cooling load, and its value range is , which reflects the adjustable ability and potential of the cooling load; the calculation formula for the user satisfaction constraint of the cooling load is as follows: Among them, is the expected indoor temperature of the th user with the cold load at a certain moment, ; is the user satisfaction; if the minimum boundary value of the user satisfaction is specified, the difference between the indoor temperature value deduced inversely and the expected temperature of the user is the adjustable range corresponding to the cold load; The set temperature of the electric refrigeration unit should be the same as the expected temperature of the user. The relationship between the upper and lower limits of the fluctuation of the actual user's expected temperature and the set temperature of the electric refrigeration unit is as follows: Among them, is the set temperature of all electric refrigeration units for cooling the th cooling load within the time period, , and numerically it is equal to ; is the hysteresis width between the upper and lower limits of the operating temperature of the electric refrigeration unit, ; and are respectively the upper and lower limits of the fluctuation of the actual user's desired temperature, , indicating that the indoor temperature within the range meets the user's expectations and also increases the adjustable potential of the cooling load to a certain extent; The mathematical model of the electric refrigeration unit is as follows: Among them, is the coefficient of performance of all electric refrigeration units for cooling the th cooling load; is the total power consumption of all electric refrigeration units for cooling the th cooling load at time, For the th chilled load, the calculation formula for the regulation rate constraint of all electric refrigeration units for cooling is as follows: Among them, For all the electric refrigeration units providing cooling for the th cooling load, the total output power at moment, ; For the upper limit of the regulation rate of all the electric refrigeration units providing cooling for the th cooling load, ; For the electric refrigeration machine, the constraint on the upper limit of the power consumption value also needs to be considered, as shown in the following formula: Among them, is the upper limit of the total output power of all electric refrigeration units for cooling the th cooling load, ; Whether an electric vehicle chooses to charge at a charging pile at a certain moment belongs to a 0-1 programming problem. The decision variable is defined as the charging status matrix of the type of electric vehicle at the th charging pile, and all elements are 0-1 binary integer variables, taking 1 when choosing to charge and 0 when choosing not to charge; At the moment of the output power of the th charging pile is calculated as follows: Used to reflect the relationship between the output power of a charging pile and the charging powers of all electric vehicles at a certain optimization moment, the existing constraints are reflected in the element values of the charging state matrix ; Among them, is the output power of the th charging pile at time ; is the charging status matrix of the th type of electric vehicle at the th charging pile at time , where is the total number of the th type of electric vehicle; is the charging power of the th type of electric vehicle at the th charging pile at time ; is the total number of types of electric vehicles participating in regulation in the system; The calculation formula for satisfying the output power limit constraint of each charging pile at each moment is as follows: Among them, is the upper limit of the charging power of the th charging pile at a single optimization moment, ; The remaining charging duration until each electric vehicle is fully charged The calculation formula is as follows: Among them, for a certain electric vehicle participating in the regulation, is the current driving mileage of the electric vehicle, ; is the power consumption per of driving of the electric vehicle, ; is the average charging power, ; is the charging efficiency; The calculation formula for the charging duration constraint of electric vehicles is as follows: Among them, is the maximum charging duration of the th electric vehicle starting from moment; ; is the actual charging duration of the th electric vehicle starting from moment. .

4. A method for multi-objective optimal operation of a distribution network considering the adjustment rate of adjustable resources according to claim 1, characterized in that In the third step described above, the method of power flow calculation is as follows: First, it is necessary to judge the node type; the node type is divided into three categories: PV nodes, PQ nodes, and Vδ nodes; PQ nodes are nodes with known injected active power and reactive power, and unknown voltage amplitude and phase angle; PV nodes are nodes with known voltage amplitude and active power, and unknown node voltage phase angle and reactive power; Vδ nodes are also called balance nodes, and the voltage amplitude and phase angle of this node are known. The essence of power flow calculation is to supply power to other nodes by adjusting the active and reactive power output of the balance node to maintain the power balance of the power grid; There are a total of nodes in the network, among which there are PV nodes, 1 Vδ node, and PQ nodes. The power flow equation is obtained, and its calculation formula is as follows: wherein, and are the full-node net injected active power vector and the full-node net injected reactive power vector respectively; is the full-node voltage vector, ; is the nodal admittance matrix, ; is the given initial nodal voltage of the full node, ; is the active power injection deviation of PQ and PV nodes, ; is the reactive power injection deviation of PQ nodes, ; is the square difference of PV node voltage magnitudes, ; The calculation formula for voltage is as follows: Among them, is the voltage of the node ; ; and are respectively the upper and lower limits of the voltage amplitude of the node ; ; The calculation formulas for active power transmission and reactive power transmission in the line are as follows: Among them, is the active power on the line ; ; and are the upper and lower limits of the active power on the line respectively; ; is the total number of branches in the network; For the reactive power on the line Var; and are respectively the upper and lower limits of the reactive power on the line .​ 5. A multi-objective optimal operation method for a distribution network considering the adjustable resource regulation rate according to claim 1, characterized in that In the fourth step, the calculation formula with the lowest operation cost of the distribution network in the whole-day scheduling period as the optimization objective is as follows: Among them, is the electricity purchase cost of the distribution network from the superior power grid at time, yuan; is the operating income of the load aggregator at time, yuan; is the grid connection scheduling cost of the distributed power source at time, yuan; is the penalty cost for wind curtailment at time, yuan; is the cost of light curtailment at time, yuan, is the total number of optimized times; Among them, the power purchase cost of the distribution network from the superior power grid at each moment is calculated by the following formula: Among them, is the unit price of purchasing electricity from the superior power grid at time ; is the power exchange value of the connection line with the superior power grid at time ; is the duration of a single optimization time ; Operating revenue of the momentary load aggregator The calculation formula is as follows: Wherein, is the output power of the nth energy storage system at time ; is the energy storage scheduling price of the nth energy storage system at time ; is the total number of energy storage systems participating in regulation; is the output power of the nth electric vehicle at time ; is the charging electricity price of the nth electric vehicle at time ; is the total number of electric vehicles participating in regulation; is the response power of the nth flexible load participating in regulation at time ; is the calling cost of the nth flexible load participating in regulation at time ; is the total number of electric vehicles participating in regulation; Grid connection scheduling cost of distributed power sources at each moment The calculation formula is as follows: Among them, is the output power of the nth grid-connected distributed energy source at time ; is the maintenance cost of the nth grid-connected distributed energy source at time ; is the total number of grid-connected distributed energy sources; Penalty cost of curtailed wind power at a certain moment and curtailed PV power cost are calculated as follows: Among them, is the curtailment penalty unit price at time ; is the predicted output of the th grid-connected wind turbine at time ; is the scheduled output of the th grid-connected wind turbine after participating in the optimization at time ; is the total number of grid-connected wind turbines; is the curtailment price of PV power at time ; is the day-ahead predicted output of the th grid-connected PV unit at time ; is the output of the th grid-connected PV unit after participating in the optimized dispatch plan at time ; is the total number of grid-connected PV units; The specific expression with the highest comprehensive energy efficiency of the distribution network in the whole-day scheduling period as the optimization objective is as follows: Among them, is the load magnitude at the node within the time instant ; ; is the total number of nodes in the distribution network; is the renewable energy penetration rate of the upstream power grid; is the renewable energy power generation efficiency of the upstream power grid; is the non-renewable energy power generation efficiency of the upstream power grid.

6. A multi-objective optimal operation method for a distribution network considering the adjustable resource regulation rate according to claim 1, characterized in that In the fifth step, for the electric vehicle load, a lower-layer dual-optimization objective model and a total optimization objective model after linear superposition are constructed for optimization; The lower-layer dual-optimization objective model includes the following: First, a combined optimization is performed on the electric vehicles charging at each charging pile node in the network topology at each optimization time, with the minimum total daily charging cost and the minimum total daily net load growth ratio as the optimization goal, to solve the optimal charging state matrix ,Right now Moment Electric vehicles in the The optimal combined charging plan for each charging pile is calculated; then, the output power of each charging pile at each optimization moment is solved according to the optimal combined charging plan for electric vehicles obtained by the lower optimization, and it is used as a constant load to participate in the solution of the optimal scheduling plan for other adjustable resources in the distribution network; The objective function expression for the total daily charging cost of electric vehicles is as follows: Among them, is the total daily charging cost of electric vehicles, in yuan; is at the output power of the th electric vehicle; is at the charging electricity price of the th electric vehicle; is the total number of electric vehicles participating in regulation; is the duration of a single optimization moment, ; is the total number of optimization moments; The objective function expression for the total daily net load growth ratio of electric vehicles is as follows: Among them, is the daily net load growth ratio of electric vehicles in the network topology; is the total active load in the network topology before electric vehicle charging at time ; Normalize the two objective functions first and values so that the and values obtained from each optimization calculation are of the same order of magnitude, and then specify the weights and occupied by the two optimization objectives, linearly superimpose the two objective functions into a single objective function to form a total optimization objective model, and its calculation formula is as follows: Among them, is the sum of the normalized objective functions optimized for the lower layer; and are the weights of the two optimization objectives, namely the total daily charging cost and the total daily net load growth ratio, respectively; and are the normalization coefficients of the two objective function values, namely the total daily charging cost and the total daily net load growth ratio, respectively; Since the definition of the sum of daily net load growth ratios is that for all optimization moments within a day, calculate the ratio of the sum of the charging powers of all electric vehicles at each moment to the total active load in the network at that optimization moment, and then calculate the sum of the net load growth ratios for all optimization moments; assuming there are optimization moments within a day, then can take values , then the value range of the normalized objective function, the sum of daily net load growth ratios, is ; Therefore, can take values , where is the total charging cost when all electric vehicles charge at the maximum power during peak electricity price moments, then the value range of the normalized objective function, the sum of daily charging costs, is also ; The weights and of the two optimization objectives should be specified according to the actual emphasis of the optimization objectives, satisfying .

7. A multi-objective optimal operation method for a distribution network considering the adjustable resource regulation rate according to claim 6, characterized in that The overall optimization objective model aims to minimize the sum of the normalized objective functions optimized at the lower layer and needs to make as small as possible while making as large as possible; during peak electricity price hours, since the charging unit price is high, electric vehicles tend not to charge more. Compared with the values of each parameter in the unordered charging state before scheduling, after scheduling, it will make smaller, and thus make smaller, which is in the same optimization direction as , but at the same time it will also make smaller, which is in the opposite optimization direction to . Therefore, a contradiction arises between the two sub-objective functions, and the lower layer optimization is committed to finding a compromise solution in the optimal solution; during off-peak electricity price hours, since the charging unit price is low, electric vehicles tend to charge more. Compared with the values of each parameter in the unordered charging state before scheduling, after scheduling, it will make larger, and thus make larger, which is in the opposite optimization direction to , but at the same time it will also make larger, which is in the same optimization direction as . Therefore, a contradiction arises between the two sub-objective functions, and the lower layer optimization is committed to finding a compromise solution in the optimal solution.

8. A multi-objective optimal operation method for a distribution network considering the adjustable resource regulation rate according to claim 1, characterized in that In the sixth step, the optimization algorithm is the NSGA-II optimization algorithm, which specifically includes the following: Before optimization, it is necessary to set the population size and the number of evolutions ; then perform the optimization; After the optimization is completed, a set of solutions and the Pareto rank corresponding to each set of solutions will be obtained; after all the solutions with a Pareto rank of 1 are extracted, a set of solutions will be obtained; subsequently, all the calculated values of the objective functions corresponding to the set of solutions will be normalized, with the aim of converting all the calculated values of the objective functions into the same order of magnitude; the normalization method for the calculated value of the th objective function is expressed as follows: Among them, is the -th vector composed of the normalized values of the -th objective function; is the maximum value among the calculated values of the -th objective function; is the minimum value among the calculated values of the -th objective function; is the vector composed of the calculated values of the -th objective function. After normalizing objective functions, the next step is to specify weights for the optimization objectives , satisfying ; then, calculate, through the following formula, the sum of the normalized values of the objective functions corresponding to each set of solutions. The specific calculation formula is as follows: Among them, is the vector composed of the sum of the normalized values of each objective function corresponding to each set of solutions in a set of solutions, and is the weight occupied by the -th optimization objective; Finally, the set of solutions corresponding to the maximum value found in is the optimal solution to the problem of optimizing the operation of the adjustable resource-assisted distribution network.

9. A multi-objective optimal operation method for a distribution network considering the adjustable resource regulation rate according to any one of claims 1-8, characterized in that It also includes the initialization of decision variables; It is necessary to set the population size participating in the operation and regard all decision variables participating in the optimization as genes; in the process of initializing genes, considering the regulation rate constraint, initialization methods are set for different adjustable resources, including a unified generation method and a special generation method; The unified generation method is divided into the unified generation method for the initial value of genes at the first optimization moment and the unified generation method for the initial value of genes at non-first optimization moments; The unified generation method for the initial value of genes at the first optimization moment specifically includes the following: The generation of the initial value of genes of different adjustable resources at the first optimization moment is not restricted by the regulation rate, as shown in the following formula: Among them, is the initial value of the th gene in the population; is the lower limit of the value of the th gene; is the upper limit of the value of the th gene; is a random real number uniformly distributed within; The unified generation method for the initial value of genes at non-first optimization moments specifically includes the following: The generation of the initial value of genes of different adjustable resources at non-first optimization moments is restricted by the value size and regulation rate in the previous moment, as shown in the following formula: Among them, and expressions respectively represent taking the larger one of the elements and the element and taking the smaller one of the elements and the element ; is the initial value of the -th gene in the population ; is the upper limit of the adjustment rate of the adjustable resource represented by the -th gene in the population at the unit optimization time ; The special generation method updates the upper and lower limits of the output power value of the energy storage system at the next optimization moment based on the state of charge and regulation rate constraint of the energy storage system at the end of the previous scheduling period; The generation method of the initial value of genes at the first optimization moment of the special generation method is the same as the unified generation method of the initial value of genes at the first optimization moment of the unified generation method; In the generation method of the initial value of genes at non-first optimization moments of the special generation method, further constraints on the upper and lower limits of gene values are added; for the generation method of the initial value of genes at non-first optimization moments of the energy storage system, as shown in the following formula: Among them, is the population from the th gene value to the th gene value cumulative value, ; is the th gene in the population initial value, ; is the capacity of the energy storage system, ; is the initial state of charge of the energy storage system.

10. A multi-objective optimal operation system for a distribution network considering the adjustable resource regulation rate, characterized in that It includes: One or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors are caused to implement a multi-objective optimal operation method for a distribution network considering an adjustable resource adjustment rate as described in any one of claims 1-9.

Images