Main distribution cooperative operation optimization method for complex power distribution network

Abstract

The invention discloses a complex power distribution network main distribution cooperative operation optimization method, and relates to the field of power distribution network operation optimization, and the method comprises the following steps: 1, providing power loss calculation models for different types of transformers respectively; 2, fitting the actual power loss of the transformer by adopting HE-LSSVR according to the transformer loss calculation model provided in the step 1, determining the optimal economic operation interval of the transformer, and further determining the economic operation mode of the transformer; step 3, obtaining the economic operation mode of the transformer according to the step 2; step 4, on the basis of the economic operation mode of the transformer, optimizing the main and distribution cooperative operation of the complex power distribution network according to the load balance degrees of different levels of the power distribution network; the method aims at improving the load balance degree of the complex power distribution network in a load peak scene by changing the operation mode of a transformer in a transformer substation and the state of a switch between feeder line groups.

Description

Technical Field

[0001] The present invention relates to the field of distribution network operation optimization, and in particular to a method for optimizing coordinated operation of main and distribution networks in a complex distribution network. Background Art

[0002] In the operation of the power system, due to the inconsistency of transformer parameters and technical characteristics in the substation, the active loss and reactive loss of the transformer vary nonlinearly with the load. Therefore, under the condition of supplying the same load, there must be a large difference in loss between different transformer operation modes. At the same time, if the load rate of the transformer is too low, it will cause large losses and the operation of the transformer is not economical; if the load rate is too high, it will cause the transformer to overheat and reduce its service life. Therefore, it is necessary to optimize the operation mode of the transformer. On the premise of meeting the load demand, the load rate of the transformer should be balanced and maintained between 75% and 85%, ensuring safe and reliable power supply while minimizing the operating loss. In the economic operation of transformers, there are usually differences between the nameplate parameters and the actual operating parameters, and it is impossible to accurately estimate the operating range of the transformer. At the same time, in actual applications, a power supply zone usually contains multiple substations operating in parallel at the same time. The operation mode of each substation is different. Each substation is equipped with a feeder group interconnection switch between the feeders. At peak load, some equipment may be severely overloaded, seriously threatening the safe operation of the distribution network. By adjusting the status of these interconnection switches and the operation mode of the substation, the load at different levels of the distribution network can be balanced, which is of certain significance for the safe operation of the distribution network under peak load scenarios. Summary of the invention

[0003] The purpose of the present invention is to overcome one or more deficiencies of the prior art and to provide a method for optimizing coordinated operation of main and distribution networks in a complex distribution network.

[0004] The objective of the present invention is achieved through the following technical solutions:

[0005] A method for optimizing coordinated operation of main and distribution networks in a complex distribution network, the method comprising the following steps:

[0006] Step 1: According to the difference between two-winding transformers and three-winding transformers, the active power loss and reactive power loss of different types of transformers are calculated through the four parameters of no-load current, no-load loss, short-circuit voltage and short-circuit loss, and the rated loss of two-winding transformers and three-winding transformers is further calculated to support subsequent calculation and analysis;

[0007] Step 2: The actual operation data of the transformer is screened by using the rated loss value calculated from the nameplate value, and the available data obtained after screening are fitted using HE-LSSVR with the goal of minimizing the loss function to obtain the transformer power loss curve;

[0008] Step 3: Determine the optimal economic operation range of the main transformer of the distribution network. Based on the actual transformer power loss curve, determine the optimal economic operation range of the transformer through the intersection of the loss curves of two transformers, and determine the economic operation mode of the transformer according to the optimal economic operation range;

[0009] Step 4: Based on the economic operation mode of the transformer, the coordinated operation of the main and distribution networks in the complex distribution network is optimized according to the load balance at different levels of the distribution network.

[0010] Furthermore, in step 1, a calculation method for four common transformer technical parameters, namely no-load current, no-load loss, short-circuit voltage and short-circuit loss, is proposed:

[0011] The no-load current is calculated as:

[0012]

[0013] in, is the percentage of no-load current; is the no-load current; is the rated current of the primary winding of the transformer;

[0014] The no-load loss is calculated as:

[0015]

[0016] in, represents iron loss, represents the hysteresis loss, represents the eddy current loss, represents the additional loss, where and Together they constitute the basic iron loss;

[0017] The short circuit voltage is calculated as:

[0018]

[0019] in, Indicates short circuit voltage, Indicates the short-circuit voltage percentage, Indicates the rated voltage of the transformer primary winding;

[0020] The short circuit loss is calculated as:

[0021] When analyzing the loss of a three-winding transformer, the short-circuit loss converted into the short-circuit loss at rated capacity is: , , .

[0022] Furthermore, according to the different winding capacity ratios, three-winding transformers are divided into two different types:

[0023] When the capacities of the three windings are equal to the rated capacity of the transformer, the short-circuit loss of the windings on each side is calculated as follows:

[0024]

[0025] When the three winding capacities of the transformer are different, the short-circuit loss of the windings on each side is calculated as follows:

[0026]

[0027] When the transformer is a three-winding transformer, short-circuit the secondary side and increase the primary side voltage so that the secondary side reaches the rated current. The primary side voltage at this time is , the sum of the losses between the primary and secondary sides is ; Short-circuit the tertiary side and increase the primary side voltage so that the tertiary side reaches the rated current. The primary side voltage at this time is , the sum of the losses between the primary and tertiary sides is ; Short-circuit the side with smaller rated capacity between the secondary and tertiary sides, and increase the voltage on the other side to make it reach the rated current. The voltage at this time is , the sum of the losses between the secondary and tertiary sides is , Respectively represent the short-circuit loss on each side given on the nameplate at the current rated capacity of each side, , and Indicates the rated capacity of the winding on each side given on the nameplate.

[0028] Furthermore, the rated loss calculation in step 1 is performed when the capacity is the same. On the premise that the four basic technical parameters of no-load current, no-load voltage, short-circuit voltage and short-circuit loss are calculated in step 1, the active power and reactive power of the transformer are calculated:

[0029] When no-load, the apparent power of the primary winding of the transformer is :

[0030]

[0031] in, is the rated voltage of the primary winding, is the rated capacity. At this time, the reactive power of the primary side of the transformer is :

[0032]

[0033] When short-circuited, the apparent power of the primary winding of the transformer is :

[0034]

[0035] Reactive power consumed by the transformer under rated load :

[0036]

[0037] For a three-winding transformer, the following are calculated based on the nameplate value, i.e. the calibration data on the nameplate: , , :

[0038] When the short-circuit voltage is equal to the rated capacity of the primary winding During calibration, the leakage magnetic power of three rated loads is calculated as follows:

[0039]

[0040] When the winding capacities on the three sides are not equal and the three short-circuit voltage values ​​are calibrated according to the rated capacity of the load-side winding, the three rated load leakage powers are calculated as follows:

[0041]

[0042] In the reactive power calculation formula of the three-winding transformer, the three leakage magnetic powers are converted into the rated power of each winding. , , , before the calculation process , , The calculation process is the same as;

[0043] According to the parameters calculated in step 1, calculate the active and reactive power losses of the transformer respectively:

[0044] Two-winding transformer: The active power loss is calculated as:

[0045]

[0046] in, is the load factor, , is the load side active power, is the reactive power on the load side;

[0047] Three-winding transformer: , and Represents the load factor of each side winding in the three-winding transformer, and its active power loss is calculated as follows:

[0048] ;

[0049] The reactive power loss of a two-winding transformer is calculated as:

[0050] ;

[0051] The reactive power consumption of a three-winding transformer is calculated as:

[0052] ;

[0053] Further calculation of transformer rated loss:

[0054] For double-winding transformers, filter the transformer nameplate parameters in the input data and input half-year historical data to calculate the rated active power loss of a single transformer. for:

[0055]

[0056] is the no-load loss of the transformer, which is a fixed value and is given on the nameplate; P is the active value of the transformer low-voltage side in historical data; is the rated capacity of the transformer, which is also the nameplate parameter value; is the rated loss of the transformer, is the nameplate parameter value;

[0057] The actual loss of a single transformer is:

[0058]

[0059] Transformer parallel calculation:

[0060]

[0061] For three-winding transformers, the loss calculation method for a single transformer is described in step 1. When two transformers are operated in parallel, the loss calculation is as follows:

[0062]

[0063] in, and are the rated no-load losses of transformer 1 and transformer 2 respectively, and is the converted transformer load loss of the second group of transformer windings 1 and 2, and is the converted transformer load loss of the second group of transformer windings 1 and 2, and is the converted transformer load loss of the third group of transformer windings 1 and 2, , , Respectively represent the actual loads carried by the windings on each side in the historical data, , , , , , They are the converted capacities of the first, second and third groups of transformer windings 1 and 2 respectively.

[0064] Furthermore, in the data fitting method based on HE-LSSVR in step 2, the loss function between the theoretical value and the observed value is minimized by solving the approximate function. The steps of the method are as follows:

[0065] SVR given a data set , we get the linear function , where the parameters are and , obtained by training with a given data set, so that the given data points fall on the function as much as possible; during training, its goal is to find a linear regression function that allows errors to exist, and limit the size of the error by introducing the loss function value as a penalty. The objective function is expressed as:

[0066]

[0067] in, represents the training error; represents the regularization parameter that controls the weight between model complexity and training error; is the loss function, which gives priority to minimizing the error and keeps Within , a penalty is imposed on errors exceeding the threshold;

[0068] In the adopted HE-LSSVR model, the Hawk Eye loss function is taken into account in the SVR model, where the loss function is defined as:

[0069]

[0070] in, is an insensitive parameter used to establish an admissible region within the loss function, Represents the model training error; Shape parameters used to eliminate curvature and steepness of the loss function; To determine the boundary parameters so that the loss does not exceed the specified boundary, a support vector regressor with robustness to outliers and ideal smoothing properties is obtained. express [ w T , b ] ,use express [ x i , 1 ] T , the HE-LSSVR model is expressed as:

[0071]

[0072] in, is the regularization term, is the empirical risk corresponding to the Hawk Eye loss function; Represents the regression error of the input, and the algorithm is:

[0073]

[0074] in, Represents a high-dimensional feature map associated with a kernel function; It is expressed as:

[0075]

[0076] in, is the coefficient vector, T is the parameter of the function after linear fitting, , For any element in the set, Bring it into the HE-LSSVR model and transform the original optimization into unconstrained optimization. The algorithm is:

[0077]

[0078] in, is the kernel function, express A coefficient in .

[0079] Furthermore, in step 2, curve fitting is performed on the non-convex HE-LSSVR model, and the specific steps include: model initialization, iterative optimization and convergence judgment;

[0080] Model initialization:

[0081] The objective function is set as , initialize the model with appropriate hyperparameter values, where the learning rate The step size used to update the model parameters; and Used to calculate the moving average of the gradient and the square of the gradient; initialize the first estimated moment and the second estimated moment The value of

[0082] Iterative optimization: including: calculating the objective function Relative to Gradient optimization, update of the first estimated moment and the second estimated moment optimization, correction of moment deviation optimization and update of model parameter optimization;

[0083] Calculate the objective function Relative to Gradient optimization: The algorithm is:

[0084]

[0085]

[0086] in, Represents the kernel matrix No. OK, Represents the objective function The gradient, Represents the loss function for The partial guide, is the coefficient vector;

[0087] Update the first estimated moment and optimize the second estimated moment: The algorithm is:

[0088]

[0089]

[0090] Corrected moment deviation optimization: The algorithm is:

[0091]

[0092]

[0093] in, and are constants used to update the gradient and the square of the gradient in the tth step;

[0094] Update model parameter optimization: The algorithm is:

[0095]

[0096] in, To avoid constants with zero denominators;

[0097] Convergence judgment:

[0098] The convergence of the model after each iteration is judged. If the maximum number of iterations is reached or the convergence criterion is met, the iteration is stopped and the output is , we can get the fitted function as: The algorithm is:

[0099]

[0100] The output after fitting is .

[0101] Furthermore, in step 4, the coordinated operation of the main distribution of the complex distribution network is optimized according to the comprehensive load balance of the distribution network, by introducing load balance indicators, which include: substation load balance, transformer load balance, and feeder load balance;

[0102] Substation load balancing degree: The formula is:

[0103]

[0104] in: and represents the upper limit of active power of substation node j and the load rate in period t, represents the active power of substation node j in period t, Indicates the load balance of the substation; is the average value of the substation load factor in the tth period; is the total number of substations;

[0105] Transformer load balance: The formula is:

[0106]

[0107] in: and is the transformer set and number of transformers in the substation at node j; is the feeder set of transformer f belonging to substation node j; and represents the active power and load rate of transformer f of substation at node j in period t; is the upper limit of transformer active power; , are the mean transformer load rate and transformer load balance degree in substation node j respectively; represents the active power of feeder k under substation j Indicates the load balance of all transformers;

[0108] Feeder load balance: The formula is:

[0109]

[0110] in: and Indicates the active power and load factor of the transformer secondary feeder, Indicates the upper limit of feeder active power; is the number of feeders of transformer f at substation node j; is the number of system transformers; and They represent the average feeder load rate and feeder load balance of transformer f in period t, Indicates the load balance of the feeder.

[0111] Furthermore, in step 4, a complex distribution network main distribution coordinated operation optimization model is established, which also includes taking the transformer operation mode and the switch state between feeder groups as decision variables, and the objective function is to maximize the global load balance of the distribution network. The algorithm is:

[0112]

[0113] in: , and Respectively represent the weights of each balance degree, which can be set according to actual conditions and meet ; Constraints include: power flow constraints, radial topology constraints, safe operation constraints, transformer power constraints, substation power constraints, and switch operation times constraints;

[0114] Power flow constraint: The formula is:

[0115]

[0116]

[0117] in, Represents the set of all branches in the distribution network; and are the active power and reactive power injected into node i during period t, respectively; and The voltage amplitudes of nodes i and j at time period t respectively; and are the conductance and susceptance of branch ij respectively; is the phase angle difference between nodes i and j;

[0118] Radial topology constraint: The formula is:

[0119]

[0120] in, Indicates the total number of nodes in the system; Indicates the number of normally closed switches in the system; Indicates the number of substations involved in the reconstruction process, Indicates the on / off state of the switch on branch ij, which is a 0-1 variable. When the switch is closed, the value is 1, otherwise it is 0. In order to avoid the island operation of the distribution network after the distributed generation is connected, power is injected into the non-substation node. , connectivity is ensured by the following formula;

[0121]

[0122] in, It represents the auxiliary power flow on branch ij during period t. It is positive when the power flows into node i and negative when it flows out. represents the set of branches connected to node i; and They represent the power and maximum capacity of the auxiliary power flow of branch ij during period t respectively; represents the auxiliary power flow power of node i;

[0123] Safe operation constraints: The formula is:

[0124]

[0125]

[0126] in, is the current of branch ij in period t, It represents the maximum current allowed to flow through branch ij during time period t; is the voltage of node i in period t; and are the maximum and minimum values ​​of the voltage allowed at node i in period t respectively;

[0127] Transformer power constraint: The formula is:

[0128]

[0129] in, Indicates the total number of feeders carried by transformer f; It represents the load of the kth feeder under transformer f during period t; It represents the load carried by transformer f in substation j during period t; and They represent the upper and lower limit constraints of the transformer output respectively;

[0130] Substation power constraint: The formula is:

[0131]

[0132] in, represents the total load carried by substation j; Indicates the maximum capacity of each transformer; represents the number of feeders of substation j, represents the number of transformers in substation j;

[0133] Constraint on the number of switch operations: The number of bus tie switch operations of a substation is controlled within 5 times in one quarter; the formula is:

[0134]

[0135]

[0136] in, and They are the maximum number of operations of the distribution network tie switch and the maximum number of operations of the substation bus tie switch, respectively. represents the switch state of branch ij at time t, represents the feeder branch switch set, represents the set of bus tie switches in the substation, and T represents the total optimization time.

[0137] The beneficial effects of the present invention are:

[0138] (1) A curve fitting method based on HE-LSSVR is proposed, which can more accurately estimate the loss curve of the transformer and thus determine the economic operation range of the transformer;

[0139] (2) It helps to ensure that substations, transformers and feeders in the distribution network operate as evenly as possible while ensuring reliable power supply for peak loads and reducing power losses at the transformer level, thereby improving the economy and stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0140] Figure 1 It is a schematic diagram of the overall method of the present invention;

[0141] Figure 2 It is the HE-LSSVR curve fitting schematic diagram;

[0142] Figure 3 Schematic diagram a for the critical load power of a single transformer and two transformers in parallel operation;

[0143] Figure 4 Schematic diagram b of critical load power of transformers running in single and two units in parallel;

[0144] Figure 5 Schematic diagram c of critical load power of transformers running in single and two units in parallel;

[0145] Figure 6 Determine the flow chart for the economic operation range of the transformer. DETAILED DESCRIPTION

[0146] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0147] refer to Figure 1 , Figure 1 The overall method of the present invention is schematically shown, comprising the following steps:

[0148] Step 1. Propose power loss calculation models for different types of transformers. Starting from the four common technical parameters of transformers, namely no-load current, no-load loss, short-circuit voltage and short-circuit loss, respectively propose calculation methods for active power loss and reactive power loss of two-winding transformers and three-winding transformers, and then propose calculation methods for their rated losses to support subsequent analysis and calculation;

[0149] Step 2. The actual operation data of the transformer is screened based on the rated loss value calculated from the nameplate value, and the available data obtained after the screening is fitted using HE-LSSVR with the goal of minimizing the sum of squared errors to obtain the transformer power loss curve;

[0150] Step 3. Determine the optimal economic operation range of the main transformer of the distribution network. Based on the actual power loss fitting curve of the transformer, find the intersection of the loss curves of the two transformers to determine the optimal economic operation point and operation range.

[0151] Step 4. Based on the optimal economic operation range of a single transformer, the comprehensive load balance of the distribution network is optimized to construct the coordinated operation optimization of the main and distribution networks.

[0152] The specific implementation is:

[0153] Step 1. Propose the calculation method of four common transformer technical parameters: no-load current, no-load loss, short-circuit voltage, and short-circuit loss:

[0154] 1) The no-load current is calculated as follows:

[0155]

[0156] in, is the no-load current percentage; is the no-load current; is the rated current of the primary winding of the transformer.

[0157] 2) No-load loss is calculated as follows:

[0158]

[0159] in, represents iron loss, represents the hysteresis loss, represents the eddy current loss, represents the additional loss, where and Together they constitute the basic iron loss.

[0160] 3) The short-circuit voltage is calculated as follows:

[0161]

[0162] When the transformer is a three-winding transformer, short-circuit the secondary side and increase the primary side voltage so that the secondary side reaches the rated current. The primary side voltage at this time is , the sum of the losses between the primary and secondary sides is ; Short-circuit the tertiary side and increase the primary side voltage so that the tertiary side reaches the rated current. The primary side voltage at this time is , the sum of the losses between the primary and tertiary sides is ; Short-circuit the side with smaller rated capacity between the secondary and tertiary sides, and increase the voltage on the other side to make it reach the rated current. The voltage at this time is , the sum of the losses between the secondary and tertiary sides is .

[0163] 4) Short-circuit loss is calculated as follows:

[0164] According to the winding capacity ratio, three-winding transformers can be divided into two different types:

[0165] (1) When the capacities of the three windings are equal to the rated capacity of the transformer, the short-circuit loss of the windings on each side is calculated as follows:

[0166]

[0167] (2) When the three winding capacities of the transformer are different, the short-circuit loss of the windings on each side is calculated as follows:

[0168]

[0169] Then calculate according to the calculation method when the capacity is the same:

[0170] The active power loss and reactive power loss of the transformer are obtained by conversion based on the four parameters described in step 1:

[0171] 1) When no-load, the apparent power of the primary winding of the transformer is :

[0172]

[0173] in, is the rated voltage of the primary winding, Rated capacity

[0174] At this time, the reactive power of the transformer primary winding is :

[0175]

[0176] 2) When short-circuited, the apparent power of the primary winding of the transformer is :

[0177]

[0178] Reactive power consumed by the transformer under rated load :

[0179]

[0180] For a three-winding transformer, according to the data marked on its nameplate, the , , :

[0181] (1) When the short-circuit voltage is equal to the rated capacity of the primary winding During calibration, the leakage magnetic power of three rated loads is calculated as follows:

[0182]

[0183] (2) When the three-side winding capacities are not equal and the three short-circuit voltage values ​​are calibrated according to the rated capacity of the load-side winding, the three rated load leakage powers are calculated as follows:

[0184]

[0185] In the reactive power calculation formula of the three-winding transformer, the three leakage magnetic powers need to be converted into the rated power of each winding. , , , the calculation process is as follows , , The calculation process is the same.

[0186] According to the parameters calculated in the above steps, the active and reactive power losses of the transformer are calculated respectively:

[0187] (1) If it is a double-winding transformer, its active power loss is calculated as follows:

[0188]

[0189] in, is the load factor, , is the load side active power, Reactive power on the load side

[0190] If it is a three-winding transformer, its active power loss is calculated as follows:

[0191]

[0192] (2) The reactive power loss of a two-winding transformer is calculated as follows:

[0193]

[0194] The reactive power consumption of a three-winding transformer is calculated as follows:

[0195]

[0196] According to the above formula, the transformer rated loss calculation model is further proposed:

[0197] For double-winding transformers, filter the transformer nameplate parameters in the input data and input half-year historical data to calculate the rated active power loss of a single transformer. for:

[0198]

[0199] is the no-load loss of the transformer, which is a fixed value and is given on the nameplate; P is the active value of the transformer low-voltage side in historical data; is the rated load loss, which is also the nameplate parameter value; is the rated capacity of the transformer, is the nameplate parameter value;

[0200] The actual loss of a single transformer is:

[0201]

[0202] Transformer parallel calculation:

[0203]

[0204] For a three-winding type transformer, when two transformers are operated in parallel, the loss is calculated as follows:

[0205]

[0206] in, , are the rated no-load losses of transformer 1 and transformer 2 respectively, and is the transformer load loss after conversion of transformer winding 1 and transformer winding 2 , the rest are the same.

[0207] Step 2. According to the calculation formula of active power loss and reactive power loss of the above transformer, a basis is provided for the subsequent determination of the economic operation area of ​​the transformer. To determine the economic operation area of ​​the transformer, it is also necessary to accurately grasp the actual power loss characteristics of the transformer. In this embodiment, HE-LSSVR is used to fit the actual power loss curve of the transformer. The specific steps are as follows:

[0208] Use the transformer nameplate parameters in the input data to filter the original data, set the original historical data array as Data, and the rated loss value array corresponding to the low-voltage side active value of each original historical data as Data2. If the following conditions are met: actual loss value>0.3*rated loss value&&actual loss value<1.7*rated loss value&&transformer high-voltage side data is not 0. Determine that the data is available, and add the absolute value, actual loss and rated loss of the transformer low-voltage side load (double winding) or low-voltage side plus medium-voltage side load (three windings) into the new array Data3. The specific form is as follows:

[0209]

[0210] HE-LSSVR is used to fit the data. The actual curve is used for fitting. HE-LSSVR is used to fit the data with the actual load in Data3 as the horizontal axis and the rated loss as the vertical axis.

[0211] See also Figure 2 , Figure 2 The following is a schematic diagram of the HE-LSSVR curve fitting method. The core idea of ​​this method is to solve the approximate function so that the loss function between the theoretical value and the observed value is minimized. The following are the detailed steps and key contents of this method:

[0212] SVR is a regression algorithm based on support vector machines. Its basic idea is to find an optimal hyperplane in the feature space so that the distance from the sample point to the hyperplane is as small as possible and the distance from the hyperplane is less than or equal to a given threshold. The core of SVR is to give a data set , estimates a linear function , where the parameters are and , obtained by training with a given data set, so that the given data points fall on the function as much as possible. During training, its goal is to find a linear regression function that allows a certain error, and to limit the size of the error by introducing the loss function value as a penalty. The objective function can be expressed as:

[0213]

[0214] in, represents the training error; represents the regularization parameter that controls the weight between model complexity and training error; is the loss function. This objective function gives priority to minimizing the error, which means keeping A larger penalty is imposed on errors exceeding the threshold.

[0215] In the HE-LSSVR model adopted in this embodiment, the Hawk Eye loss function is taken into account in the SVR model, where the loss function is defined as:

[0216]

[0217] in, is an insensitive parameter used to establish an admissible region within the loss function; is a shape parameter used to eliminate the curvature and steepness of the loss function; is a boundary parameter used to ensure that the loss does not exceed the specified boundary.

[0218] This results in a support vector regressor that is robust to outliers and has ideal smoothing properties. express [ w T , b ] ,use express [ x i , 1 ] T , the HE-LSSVR model is expressed as:

[0219]

[0220] in, is the regularization term, is the empirical risk corresponding to the proposed Hawk Eye loss function; Represents the regression error of the input, that is:

[0221]

[0222] in, Represents a high-dimensional feature map associated with a kernel function.

[0223] Since the above model has strong non-convexity, in order to enhance its nonlinear adaptive ability and make it applicable to practical scenarios, the performance theorem is used to transform It is expressed as:

[0224]

[0225] in, is the coefficient vector. Since the above minimization problem operates in the reproducing kernel Hilbert space and the penalty term is non-decreasing, the performance theorem can be applied to the optimization problem. Bringing it into the HE-LSSVR model, the original optimization problem is transformed into an unconstrained optimization problem:

[0226]

[0227] in, is the kernel function.

[0228] Therefore, the non-convex HE-LSSVR model has a smooth characteristic. The parameters of the model are optimized using a gradient-based algorithm, and then curve fitting is performed. In this embodiment, adaptive moment estimation is used to optimize the parameters. The specific method is as follows:

[0229] Model initialization:

[0230] In this step, the objective function is set as , initialize the model with appropriate hyperparameter values, where the learning rate The step size used to update the model parameters; and Used to calculate the moving average of the gradient and the square of the gradient; initialize the first estimated moment and the second estimated moment The value of

[0231] Iterative Optimization:

[0232] (1) Calculate the objective function Relative to The gradient is:

[0233]

[0234]

[0235] in, Represents the kernel matrix No. OK.

[0236] (2) Update the first estimated moment and the second estimated moment:

[0237]

[0238]

[0239] (3) Correction moment deviation:

[0240]

[0241]

[0242] (4) Update model parameters:

[0243]

[0244] Convergence judgment:

[0245] In this step, the convergence of the model after each iteration is judged. If the maximum number of iterations is reached or the convergence criterion is met, the iteration is stopped and the output is .

[0246] From this we can get the fitted function as:

[0247]

[0248] Then we can get the fitted curve, such as Figure 2 shown.

[0249] Step 3. Determine the optimal operating point of the transformer based on the power fitting of the transformer in the above steps. The specific confirmation process is as follows:

[0250] See also Figure 3 —5, Figure 3 —5 is the critical load power diagram of a single transformer and two transformers in parallel operation. Figure 3 Indicates the critical load power between transformers A and B It is an imaginary root, and the power loss curve has no intersection; Figure 4 Indicates the critical load power between transformers A and B , the power loss curve intersects point; Figure 5 Indicates the critical load power between transformers A and B Between and The power loss curve intersects point.

[0251] See also Figure 6 First, input the transformer nameplate parameters, calculate the rated loss value of the transformer through the model described in step 1, then input the actual operating parameters of the transformer, and filter out the available data by comparing with the calculated rated loss value. Then, fit the available data through the method described in step 2 to obtain the transformer power loss curve, and finally determine the optimal economic operation range of the main transformer of the distribution network according to the power loss curve.

[0252] The actual fitting is used to find the optimal operating point of the transformer, and the intersection point of the main transformer A and the main transformer B. The solution range is limited to 0 to the minimum value of the rated capacity of the main transformer A and the main transformer B. If there is no solution, the economic operation point A is empty. If a solution is obtained , then the economic operation point 1 is , if two solutions are obtained, take the largest one. Find the intersection point of the main transformer A (B) and the parallel operation, and limit the range of the solution to 0 to the rated capacity of the main transformer A (B). If there is a solution , economic operation point 2 is If there are two solutions and , then take the larger value of the two solutions.

[0253] Substation and Two transformers, one of which ,set up For transformer and transformer Critical power between parallel operation; For transformer and transformer Critical power between parallel operation; For transformer and transformer The three critical powers have the following three relationships:

[0254] Relation 1: From the above transformer and Critical load power between (virtual root), that is Figure 3 middle and The two curves do not have an intersection. In this case, the transformer Significantly better than transformers , so the transformer There is no economic operation mode, only transformer With transformer Economical operation mode between parallel operation Switch to .

[0255] Relation 2: From the above transformer and Critical load power between ,Right now Figure 4 middle and The two curves intersect at point, Between and Between transformers There is still no economic operation mode, only transformers With transformer Economical operation mode between parallel operation Switch to .

[0256] Relation 3: From the above transformer and Critical load power between ,Right now Figure 5 middle and The two curves are compared with point, Between and There is a transformer between ,transformer and transformer Three economic operation modes running in parallel, namely When the transformer Running economy; When the transformer Running economy; When the transformer Parallel operation economy.

[0257] Step 4. According to the transformer economic operation range obtained above, the economic operation mode of the transformer in a single substation can be determined. However, in the actual power system, a power supply zone contains multiple substations, which operate in parallel and supply power to the supply zone together. If each substation is strictly operated in an economic operation mode, it may cause an unbalanced transformer load rate and increase the system operation risk. Considering the safety of system operation, it is necessary to take the load rate of equipment including transformers into consideration. Therefore, the coordinated operation of the main distribution of the complex distribution network with load balancing is optimized to achieve balanced operation of all equipment in the distribution network under peak load scenarios and ensure safe and reliable power supply of the distribution network. The specific steps are as follows:

[0258] Introduce the balance indicator. The specific steps are as follows:

[0259] Substation load balancing degree:

[0260]

[0261] in: and represents the upper limit of active power of substation node j and the load rate in the tth period; is the average value of the substation load factor in the tth period; is the total number of substations.

[0262] Transformer load balance:

[0263]

[0264] in: and is the transformer set and number of transformers in the substation at node j; is the feeder set of transformer f belonging to substation node j; and represents the active power and load rate of transformer f of substation at node j in period t; is the upper limit of transformer active power; , They represent the mean transformer load rate and transformer load balance degree in substation node j respectively.

[0265] Feeder load balance:

[0266]

[0267] in: and Indicates the active power and load factor of the transformer secondary feeder; is the number of feeders of transformer f at substation node j; is the number of system transformers; and They represent the average feeder load rate and feeder load balance of transformer f in period t respectively.

[0268] Establish a complex distribution network main and distribution coordinated operation optimization model. The specific steps are as follows:

[0269] The transformer operation mode and the switch status between feeder groups are used as decision variables, and the objective function is to maximize the global load balance of the distribution network:

[0270]

[0271] in: , and Respectively represent the weights of each balance degree, which can be set according to actual conditions and meet .

[0272] The constraints are:

[0273] Power flow constraints:

[0274]

[0275]

[0276] in, Represents the set of all branches in the distribution network; and are the active power and reactive power injected into node i during period t, respectively; and The voltage amplitudes of nodes i and j at time period t respectively; and are the conductance and susceptance of branch ij respectively; is the phase angle difference between nodes i and j.

[0277] Radial topology constraints:

[0278]

[0279] in, Indicates the total number of nodes in the system; Indicates the number of normally closed switches in the system; Indicates the number of substations involved in the reconstruction process; in order to avoid the island operation of the distribution network after the distributed generation is connected, power is injected into the non-substation nodes , the connectivity is ensured by the following formula.

[0280]

[0281] in, It represents the auxiliary power flow on branch ij during period t. It is positive when the power flows into node i and negative when it flows out. represents the set of branches connected to node i; and They represent the power and maximum capacity of the auxiliary power flow of branch ij in period t respectively.

[0282] Safe operation constraints:

[0283]

[0284]

[0285] in, is the current of branch ij in period t, It represents the maximum current allowed to flow through branch ij during time period t; is the voltage of node i in period t; and are respectively the maximum and minimum values ​​of the voltage allowed at node i in period t.

[0286] Transformer power constraints:

[0287]

[0288] in, Indicates the total number of feeders carried by transformer f; It represents the load of the kth feeder under transformer f during period t; It represents the load carried by transformer f in substation j during period t; and They respectively represent the upper and lower limit constraints of the transformer output.

[0289] Substation power constraints:

[0290]

[0291] in, represents the total load carried by substation j; Indicates the maximum capacity of each transformer.

[0292] Switching action times constraint:

[0293] Frequent opening and closing of switches will cause the topology of the distribution network to change constantly, which will bring great challenges to the stable operation of the distribution network. Excessive operation times will also shorten the life of the switch, so it is necessary to impose certain restrictions on the number of switch operations. For the bus tie switch in the substation, since each operation may generate an operation overvoltage, it should be subject to strict restrictions. Generally speaking, the number of bus tie switch operations in a substation within a quarter is controlled within 5 times.

[0294]

[0295]

[0296] in, and They respectively represent the maximum number of operations of the distribution network tie switch and the maximum number of operations of the substation bus tie switch.

[0297] In order to reduce the operation loss at the substation level and ensure the power supply safety during peak periods and improve the stability of the power system, a method for optimizing the coordinated operation of the main and distribution of complex distribution networks is proposed, so that the power supply partition can operate economically while the load of different equipment in the distribution network is as balanced as possible under the premise of meeting the load demand. The method includes firstly proposing a calculation model for the power loss of different types of transformers respectively. On this basis, considering the difference between the actual parameters of the transformer during operation and the parameters given on the nameplate, a curve fitting method based on HE-LSSVR is proposed to accurately determine the economic operation range of the transformer power loss curve, and then determine its economic operation mode. Then, under the premise of the transformer economic operation mode, the coordinated operation optimization method of the main and distribution of complex distribution networks according to the load balance degree is used to achieve balanced operation of equipment in the distribution network in the peak load scenario and ensure safe and reliable power supply in the supply area. According to the power loss calculation model of different types of transformers, the transformers are divided into two-winding transformers and three-winding transformers. Starting from the four common technical parameters of transformers, namely no-load current, no-load loss, short-circuit voltage and short-circuit loss, a power loss calculation model of the transformer is constructed; based on the curve fitting method of HE-LSSVR, according to the difference between the actual operating parameters and the nameplate parameters of the transformer, the power loss of the transformer is accurately fitted with the minimum sum of squares of errors as the goal; according to the load balance of all equipment in the distribution network, under the premise of economical operation of the transformer, the load balance of different levels of the distribution network is made as much as possible.

[0298] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art do not deviate from the spirit and scope of the present invention, and should be within the scope of protection of the claims attached to the present invention.

Claims

1. A method for optimizing coordinated operation of main and distribution networks in a complex distribution network, characterized in that: The method comprises the following steps: Step 1: According to the difference between two-winding transformers and three-winding transformers, the active power loss, reactive power loss and rated loss of two-winding transformers and three-winding transformers are calculated through the four parameters of no-load current, no-load loss, short-circuit voltage and short-circuit loss; Step 2: Calculate the rated loss value through the nameplate value to screen the actual operation data of the transformer, and use HE-LSSVR to fit the available data obtained after screening with the goal of minimizing the loss function to obtain the transformer power loss curve; Step 3: Determine the optimal economic operation range of the main transformer of the distribution network. Based on the transformer power loss curve, determine the optimal economic operation range of the transformer through the intersection of the loss curves of two transformers, and determine the economic operation mode of the transformer according to the optimal economic operation range; Step 4: Based on the economic operation mode of the transformer, the coordinated operation of the main and distribution networks in the complex distribution network is optimized according to the load balance at different levels of the distribution network.

2. A complex distribution network main distribution coordinated operation optimization method according to claim 1, characterized in that: In step 1, the calculation method of four transformer technical parameters, namely no-load current, no-load loss, short-circuit voltage and short-circuit loss, is proposed: The no-load current is calculated as: ； in, is the percentage of no-load current; is the no-load current; is the rated current of the primary winding of the transformer; The no-load loss is calculated as: ； in, represents iron loss, represents the hysteresis loss, represents the eddy current loss, represents the additional loss, where and Together they constitute the basic iron loss; The short circuit voltage is calculated as: ； in, Indicates short circuit voltage, Indicates the short-circuit voltage percentage, Indicates the rated voltage of the transformer primary winding; The short circuit loss is calculated as: When analyzing the loss of a three-winding transformer, the short-circuit loss converted into the short-circuit loss at rated capacity is: , , .

3. A complex distribution network main distribution coordinated operation optimization method according to claim 2, characterized in that: In step 1, according to different winding capacity ratios, three-winding transformers are divided into two different types: When the capacities of the three windings are equal to the rated capacity of the transformer, the short-circuit loss of the windings on each side is calculated as follows: ； When the three winding capacities of the transformer are different, the short-circuit loss of the windings on each side is calculated as follows: ； When the transformer is a three-winding transformer, short-circuit the secondary side and increase the primary side voltage so that the secondary side reaches the rated current. The primary side voltage at this time is , the sum of the losses between the primary and secondary sides is ; Short-circuit the tertiary side and increase the primary side voltage so that the tertiary side reaches the rated current. The primary side voltage at this time is , the sum of the losses between the primary side and the tertiary side is ; Short-circuit the side with smaller rated capacity between the secondary and tertiary sides, and increase the voltage on the other side to make it reach the rated current. The voltage at this time is , the sum of the losses between the secondary and tertiary sides is , Respectively represent the short-circuit loss on each side given on the nameplate at the current rated capacity of each side, , and Indicates the rated capacity of the winding on each side given on the nameplate.

4. A complex distribution network main distribution coordinated operation optimization method according to claim 3, characterized in that: The rated loss calculation in step 1 is performed when the capacity is the same. The active power and reactive power of the transformer are calculated based on the calculation of the four basic technical parameters of no-load current, no-load voltage, short-circuit voltage and short-circuit loss in step 1: When no-load, the apparent power of the primary winding of the transformer is : ； in, is the rated voltage of the primary winding, is the rated capacity. At this time, the reactive power of the primary side of the transformer is : ； When short-circuited, the apparent power of the primary winding of the transformer is : ； Reactive power consumed by the transformer under rated load : ； For a three-winding transformer, the following are calculated based on the nameplate value, i.e. the calibration data on the nameplate: , , : When the short-circuit voltage is equal to the rated capacity of the primary winding During calibration, the leakage magnetic power of three rated loads is calculated as follows: ； When the winding capacities on the three sides are not equal and the three short-circuit voltage values ​​are calibrated according to the rated capacity of the load-side winding, the three rated load leakage powers are calculated as follows: ； In the reactive power calculation formula of the three-winding transformer, the three leakage magnetic powers are converted into the rated power of each winding. , , , before the calculation process , , The calculation process is the same as; According to the parameters calculated in step 1, calculate the active and reactive power losses of the transformer respectively: Two-winding transformer: The active power loss is calculated as: ;in, is the load factor, , is the load side active power, is the reactive power on the load side; Three-winding transformer: , and Represents the load factor of each side winding in the three-winding transformer, and its active power loss is calculated as follows: ； The reactive power loss of a two-winding transformer is calculated as: ； The reactive power consumption of a three-winding transformer is calculated as: ； Further calculation of transformer rated loss: For double-winding transformers, filter the transformer nameplate parameters in the input data and input half-year historical data to calculate the rated active power loss of a single transformer. for: ； is the no-load loss of the transformer, which is a fixed value; P is the active value of the transformer low-voltage side in the historical data; is the rated capacity of the transformer, which is also the nameplate parameter value; is the rated loss of the transformer, is the nameplate parameter value; The actual loss of a single transformer is: ； Transformer parallel calculation: ； For three-winding transformers, the loss calculation method for a single transformer is described in step 1. When two transformers are operated in parallel, the loss calculation is as follows: ； in, and are the rated no-load losses of transformer 1 and transformer 2 respectively, and is the converted transformer load loss of the second group of transformer windings 1 and 2, and is the converted transformer load loss of the second group of transformer windings 1 and 2, and is the converted transformer load loss of the third group of transformer windings 1 and 2, , , Respectively represent the actual loads carried by the windings on each side in the historical data, , , , , , They are the converted capacities of the first, second and third groups of transformer windings 1 and 2 respectively.

5. The method for optimizing coordinated operation of main and distribution networks in a complex distribution network according to claim 1, characterized in that: The data fitting method based on HE-LSSVR in step 2 minimizes the loss function between the theoretical value and the observed value by solving the approximate function. The steps of the method are as follows: SVR given a data set , we get the linear function , where the parameters are and , so that the given data points fall on the function as much as possible; during training, find a linear regression function that allows errors to exist, and limit the size of the error by introducing the loss function value as a penalty. The objective function is expressed as: ； in, represents the training error; represents the regularization parameter that controls the weight between model complexity and training error; is the loss function, which gives priority to minimizing the error and keeps Within.

6. A complex distribution network main distribution coordinated operation optimization method according to claim 5, characterized in that: In step 2, curve fitting is performed on the non-convex HE-LSSVR model, and the specific steps include: model initialization, iterative optimization and convergence judgment; Model initialization: The objective function is set as , initialize the model with appropriate hyperparameter values, where the learning rate The step size used to update the model parameters; and Used to calculate the moving average of the gradient and the square of the gradient; initialize the first estimated moment and the second estimated moment The value of Iterative optimization: including: calculating the objective function Relative to Gradient optimization, update of the first estimated moment and the second estimated moment optimization, correction of moment deviation optimization and update of model parameter optimization; Calculate the objective function Relative to Gradient optimization: The algorithm is: ； ； in, Represents the kernel matrix No. OK, Represents the objective function The gradient, Represents the loss function for The partial guide, is the coefficient vector; Update the first estimated moment and optimize the second estimated moment: The algorithm is: ； ； Corrected moment deviation optimization: The algorithm is: ； ； in, and are constants used to update the gradient and the square of the gradient in the tth step; Update model parameter optimization: The algorithm is: ； in, To avoid constants with zero denominators; Convergence judgment: The convergence of the model after each iteration is judged. If the maximum number of iterations is reached or the convergence criterion is met, the iteration is stopped and the output is , we can get the fitted function as: The algorithm is: ； The output after fitting is .

7. A complex distribution network main distribution coordinated operation optimization method according to claim 5, characterized in that: In step 4, the coordinated operation of the main distribution network in the complex distribution network is optimized according to the comprehensive load balance of the distribution network, by introducing load balance indicators, which include: substation load balance, transformer load balance, and feeder load balance; Substation load balancing degree: The formula is: ； in: and represents the upper limit of active power of substation node j and the load rate in period t, represents the active power of substation node j in period t, Indicates the load balance of the substation; is the average value of the substation load factor in the tth period; is the total number of substations; Transformer load balance: The formula is: ； in: and is the transformer set and number of transformers in the substation at node j; is the feeder set of transformer f belonging to substation node j; and represents the active power and load rate of transformer f of substation at node j in period t; is the upper limit of transformer active power; , are the mean transformer load rate and transformer load balance degree in substation node j respectively; represents the active power of feeder k under substation j Indicates the load balance of all transformers; Feeder load balance: The formula is: ； in: and Indicates the active power and load factor of the transformer secondary feeder, Indicates the upper limit of feeder active power; is the number of feeders of transformer f at substation node j; is the number of system transformers; and They represent the average feeder load rate and feeder load balance of transformer f in period t, Indicates the load balance of the feeder.

8. A complex distribution network main distribution coordinated operation optimization method according to claim 5, characterized in that: In step 4, a complex distribution network main distribution coordinated operation optimization model is established, which also includes taking the transformer operation mode and the switch state between feeder groups as decision variables, and the objective function is to maximize the global load balance of the distribution network. The algorithm is: ； in: , and Represent the weights of each balance degree respectively, and satisfy .

Images