Grid-connected evaluation method and system for photovoltaic coupling model based on optimal probability power flow

Abstract

The present invention provides a photovoltaic coupling model grid-connected evaluation method and system based on optimal probability flow, which relates to the field of photovoltaic grid-connected technology. The method includes collecting photovoltaic inverter grid-connected point data; dividing the operation scenarios of photovoltaic inverters according to the output distribution of active power; building a photovoltaic coupling model based on the operation scenarios, and obtaining power probability distribution and h Subharmonic current probability distribution curve; by optimizing the probability flow, the fundamental voltage output value of each node in the distribution network is calculated based on the power probability distribution, based on h Subharmonic current probability distribution curve calculation of each node in the distribution network h Subharmonic current output value; Based on the fundamental voltage output value, fundamental voltage evaluation is performed; Based on h The subharmonic current output value is used to evaluate the harmonic current. This method takes into account the comprehensiveness and feasibility of photovoltaic model construction, which can improve the accuracy and scenario adaptability of probabilistic power flow calculation, and help to more accurately reflect the operation of the distribution network under various photovoltaic access scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic grid-connected technology, and in particular to a photovoltaic coupling model grid-connected evaluation method and system based on optimal probability power flow. Background Art

[0002] Unlike traditional power equipment, photovoltaic power generation is connected to the grid through inverters, which not only brings the risk of node voltage exceeding the limit to the distribution network, but also brings the risk of node harmonic exceeding the standard. At the same time, the volatility and randomness of photovoltaic power generation also affect the stable operation of the distribution network.

[0003] Photovoltaic output power is the basis for analyzing distribution network voltage problems using fundamental power flow calculations, and photovoltaic output harmonic current is the basis for analyzing distribution network harmonic problems using harmonic power flow calculations. Existing technologies mainly construct photovoltaic grid-connected models from the perspectives of photovoltaic grid-connected mechanism analysis and photovoltaic output data drive. The method based on mechanism analysis mainly constructs a photovoltaic power electronics simulation model and studies its internal structure and control strategy. Although it can accurately simulate the actual operation status of photovoltaics, in actual projects, photovoltaic grid-connected structures and control parameters are difficult to obtain, and model construction is difficult. On the other hand, photovoltaic grid-connected data is analyzed based on power quality monitoring devices to construct photovoltaic grid-connected models. However, this method is mainly based on a certain type of power characteristics and ignores the mutual connection of multiple power characteristics. Therefore, the corresponding calculation accuracy is limited and the analyzed power quality problems are single. Furthermore, both models ignore the mutual coupling between power and harmonic current.

[0004] Due to the instability of photovoltaic output, the existing probabilistic power flow calculation method is generally not applicable to a variety of scenarios. In other words, the existing probabilistic power flow calculation method cannot guarantee the accuracy of calculation in all scenarios. Summary of the invention

[0005] In order to solve the above technical problems existing in the prior art, the present invention provides a photovoltaic coupling model grid-connected evaluation method based on optimal probabilistic power flow, and the specific technical solution includes the following steps:

[0006] Step S1: Collecting photovoltaic inverter grid connection point data;

[0007] Step S2: dividing the operation scenarios of the photovoltaic inverter according to the output distribution of active power;

[0008] Step S3: Build a photovoltaic coupling model based on the operation scenario to obtain the power probability distribution and h Subharmonic current probability distribution curve; where: h =3,5,7,9…;

[0009] Step S4: Based on the power probability distribution, the fundamental voltage output value of each node in the distribution network is calculated by optimizing the probability flow;h The probability distribution curve of subharmonic current is used to calculate the probability distribution of each node in the distribution network through the optimal probability flow. h Subharmonic current output value;

[0010] Step S5: Based on the fundamental voltage output value, perform fundamental voltage evaluation; based on h Subharmonic current output value for harmonic current evaluation.

[0011] Preferably, step S1 specifically includes:

[0012] The active power of the photovoltaic inverter grid connection point is collected at intervals of 1 minute with a 95% probability of maximum value. P , reactive power Q and harmonic currents I h ;

[0013] ;

[0014] ;

[0015] ;

[0016] In the formula, N is the number of collections.

[0017] Preferably, Lagrange interpolation method is used to supplement the missing data.

[0018] Preferably, step S2 specifically includes:

[0019] Clustering active power using K-means clustering algorithm P ;

[0020] Step S21: Determine the number of clustering scenes k , and randomly select data points as center points for each clustering scenario, k ∈(2,3,4);

[0021] Step S22: Calculate the distance between each data point and each center point, and divide each data point into the clustering scene to which the center point closest to it belongs;

[0022] Step S23: For each clustering scene, take the mean of the data points therein as the new center point;

[0023] Step S24: Output each clustering scene.

[0024] ;

[0025] ;

[0026] In the formula,i is the sequence number of the clustering scene, P i For the i The active power sequence of the clustering scenario, i =1,2,…, k ; j is the sequence number of active power in the clustering scenario, P ij is the active power sequence P i The j Active power, n i For the i The number of data points in a clustering scenario, j =1,2,…, n i ;

[0027] Each clustering scenario also includes: the number of data points and the active power sequence P i The number of data points is the same as the reactive power series Q i and h Subharmonic current sequence I hi .

[0028] Further, repeat step S22 to step S23 to iteratively update the center point until the difference between the center points obtained by two consecutive updates is less than 10 -5 .

[0029] Preferably, step S3 specifically includes:

[0030] For each scene,

[0031] Step S31: Active power sequence P i and reactive power sequence Q i Perform joint probability modeling to obtain the power probability distribution curve f ( P i , Q i ),

[0032] ;

[0033] In the formula, h p , h q They are active power bandwidth and reactive power bandwidth respectively. K (•) is the kernel function, is the active power sequence P i The mean value of each active power in P ij is the active power sequence P i The j Active power, Reactive power sequence Q i The mean value of reactive power in Q ij Reactive power sequence Q i The j Reactive power;

[0034] Step S32: The active power sequence is transmitted through the data interval P i Divided into a Group, a =2,3,4,…;

[0035] The number of data points in each group is m i1 , m i2 , …, m ia ,

[0036] ;

[0037] The data interval is:

[0038] ;

[0039] In the formula, P imax , P imin The active power series are P i The maximum and minimum values ​​in ;

[0040] Step S33: Calculate the first b Group active power corresponding to h Subharmonic current probability distribution sub-curve f hib ( I h ),

[0041] ;

[0042] ;

[0043] In the formula, is the ordinal number of the selected model, is the number of selected models, For the Model distribution, For the The proportionality coefficient of the model distribution, For the The mean of the model distribution, For the The standard deviation of the model distribution;

[0044] in,

[0045] , ;

[0046] b =1,2,…, a ;

[0047] Step S34: Merge a strip h Subharmonic current probability distribution sub-curve, get h Subharmonic current probability distribution curve f hi ,

[0048] ;

[0049] In the formula, For the b The number of active power in the group accounts for i The weight of the amount of active power in each cluster scenario.

[0050] Preferably, the active power bandwidth is selected by minimizing the mean square error h p and reactive power bandwidth h q .

[0051] Preferably, step S4 specifically includes:

[0052] Step S41: Based on the power probability distribution curve f ( P i , Q i )and h Subharmonic current probability distribution curve f hi , respectively using methods y 1. Methods y 2. Methods y 3 and Methods y4. Calculate the probability power flow and obtain the fundamental voltage value and harmonic current value of each node in the distribution network;

[0053] Step S42: Calculation method y 1. Methods y 2. Methods y 3 and Methods y The error between the calculated value and the true value under 4 S d ;

[0054] ;

[0055] d =1,2,3,4; e =1,2,…, E ;

[0056] In the formula, d is the method number, e is the serial number of each node in the distribution network, E is the number of nodes in the distribution network, abs (•) indicates absolute value calculation, , The methods are y d The calculated e The fundamental voltage value of each node, h Subharmonic current value, , Respectively e The true value of the fundamental voltage of each node, h Subharmonic current true value;

[0057] Step S43: Select the two methods with the smallest error y d1 , y d2 Combine and construct the optimal probability power flow calculation model,

[0058] ;

[0059] In the formula, Y To optimize the probabilistic power flow calculation method, d 1∈ d , d 2∈ d , d 1≠ d2 ; , Methods y d1 ,method y d2 The weight of

[0060] With minimum comprehensive error S min is the objective function, and the solution method y d1 Weight and methods y d2 Weight , and calculate the fundamental voltage output value and h Subharmonic current output value,

[0061] ;

[0062] ;

[0063] In the formula, , Node e The fundamental voltage output value, h Subharmonic current output value, , The methods are y d1 ,method y d2 The calculated e The fundamental voltage value of each node, , The methods are y d1 ,method y d2 The calculated e Node h Subharmonic current value;

[0064] Among them, the method y 1. Methods y 2. Methods y 3 and Methods y 4 are three-point estimation method, semi-invariant method, Monte Carlo model and Latin hypercube sampling method.

[0065] Preferably, based on the fundamental voltage output value, the fundamental voltage of each node is evaluated, specifically including:

[0066] Calculate the fundamental voltage over-limit risk assessment index R u ,

[0067] ;

[0068] In the formula, U c is the supply voltage of the node in the distribution network;

[0069] According to GB / T 12325-2008 Power Quality Supply Voltage Deviation,

[0070] For power supply voltages of 35kV and above, the voltage deviation index limit range is:

[0071] ;

[0072] For three-phase power supply voltage of 20kV and below, the voltage deviation index limit range is:

[0073] ;

[0074] For 220V single-phase power supply voltage, the voltage deviation index limit range is:

[0075] ;

[0076] For Node e , if the fundamental voltage output value If the fundamental voltage output value is within the voltage deviation index limit range, the fundamental voltage is evaluated to be normal; If the fundamental voltage output value is lower than the voltage deviation index limit range, the fundamental voltage is evaluated to be too low; if the fundamental voltage output value is If it is higher than the voltage deviation index limit range, the fundamental voltage is evaluated to be too high;

[0077] Preferably, based on h Subharmonic current output value, to evaluate the harmonic current of each node, including:

[0078] calculate h Subharmonic current over-limit risk assessment indicators R h ,

[0079] ;

[0080] In the formula, I hc It is specified in GB / T 14549-1993 Power Quality Public Grid Harmonics h Nominal limit of subharmonic current;

[0081] For Node e ,like R h ≤1, then evaluate h Subharmonic current output value Normal; if R h >1, then evaluate h Subharmonic current output value Out of limit.

[0082] On the other hand, the present invention also provides a photovoltaic coupling model grid-connected evaluation system based on optimal probabilistic power flow, comprising:

[0083] Data acquisition module, used to collect grid connection point data of photovoltaic inverters;

[0084] A scenario division module is used to divide the operation scenarios of the photovoltaic inverter according to the output distribution of active power;

[0085] A photovoltaic coupling model building module is used to build a photovoltaic coupling model based on an operation scenario. The photovoltaic coupling model is used to obtain power probability distribution and h Subharmonic current probability distribution curve;

[0086] The calculation module is used to calculate the fundamental voltage output value of each node in the distribution network based on the power probability distribution. h Subharmonic current probability distribution curve calculation of each node in the distribution network h Subharmonic current output value;

[0087] An evaluation module is used to evaluate the fundamental voltage based on the fundamental voltage output value; h Subharmonic current output value for harmonic current evaluation.

[0088] The technical solution provided by the present invention can analyze the relationship between output power and harmonic current from the perspective of photovoltaic output data drive by constructing a photovoltaic coupling model, taking into account the comprehensiveness and feasibility of model construction; on the other hand, the accuracy of probabilistic power flow calculation can be improved based on the preferred probabilistic power flow calculation method; further, the power flow calculation method that considers the fundamental harmonic coupling and probabilistic power flow accuracy helps to more accurately reflect the impact of photovoltaic output fundamental and harmonic waves on the distribution network under various photovoltaic access scenarios, and has better scenario applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] Figure 1 The figure is a schematic diagram of the steps of the photovoltaic coupling model grid-connected evaluation method based on the optimal probabilistic power flow in the present invention.

[0090] Figure 2 This is a schematic diagram of the results of the photovoltaic operation scene division in the present invention.

[0091] Figure 3 It is a schematic diagram of the corresponding relationship between the power probability curve and the harmonic current probability distribution curve in the present invention.

[0092] Figure 4 The figure is a flow chart of the calculation of the preferred probability power flow in the present invention.

[0093] Figure 5 It is a basic flow chart of the photovoltaic grid-connected evaluation stage in the present invention. DETAILED DESCRIPTION

[0094] Hereinafter, the technical solution provided by the present invention will be further elaborated in detail with reference to the accompanying drawings.

[0095] The continuous advancement of technology has deepened the complexity and precision of modern power equipment, and the impact of equipment on the power quality of distribution networks has become more diversified. Photovoltaics, represented by new energy, has developed rapidly, and its share has reached a new height. Unlike previous power equipment, photovoltaics are connected to the grid through inverters, which will not only bring node voltage limit risks to the distribution network, but also node harmonics exceed the standard risk. At the same time, the volatility and randomness of photovoltaic power generation also pose new challenges to the stable operation of the distribution network. Therefore, photovoltaic grid connection is no longer satisfied with a single fundamental voltage assessment and a single harmonic current assessment. At the same time, constructing a coupling model of photovoltaic power and harmonic current is the basis for analyzing the impact of photovoltaic grid connection on fundamental voltage and harmonic current. The present invention proposes a method for constructing a photovoltaic coupling model, which takes into account the probabilistic connection between photovoltaic power output and harmonic current output, unifies the corresponding relationship between the two, and provides model support and data support for evaluating the impact of photovoltaic grid connection on fundamental voltage and harmonic current.

[0096] The power flow calculation method used in the present invention is probabilistic power flow. At present, the probabilistic power flow calculation methods include approximate method, analytical method and simulation method. The most commonly used approximate method is the point estimation method, which can achieve fast calculation but has low accuracy when calculating high-order statistical moments of output variables; the most commonly used analytical method is the semi-invariant method, which can accept more complex scenarios but has large calculation errors in scenarios with strong nonlinearity; the most commonly used simulation methods are the model Carnot model and the Latin hypercube sampling method, which can obtain high-precision results in low-dimensional inputs but seriously reduce the calculation accuracy in scenarios containing high-dimensional input random variables. None of the above methods can guarantee the output of high-precision results in all scenarios, and there are inherent limitations on the calculation accuracy. In order to more efficiently and accurately characterize the impact of photovoltaic equipment on the distribution network, the present invention improves the accuracy of probabilistic power flow calculation through optimization and improvement.

[0097] Specifically, Figure 1 As shown, the present invention provides a photovoltaic coupling model grid-connected evaluation method based on optimal probabilistic power flow, including steps S1 to S5.

[0098] Step S1: Collecting PV inverter grid connection point data

[0099] The present invention collects data through the power quality monitoring device of the distribution network. The transmission data of photovoltaic work for 24 hours a day is collected and returned through the photovoltaic grid connection point. The data type includes active power P , reactive power Q , fundamental voltage U , fundamental current I , harmonic current Ih ( h =3,5,7,9…), etc. The data format includes the maximum value, minimum value, average value and 95% probability maximum value, and each set of data is separated by 1 minute. In the present invention, because the fundamental wave power flow and harmonic wave power flow calculation of photovoltaic grid connection are to be performed, the selected data type is active power. P , reactive power Q , harmonic current I h , the selected data form is the 95% probability maximum value,

[0100] ;

[0101] ;

[0102] , h =3,5,7,9…;

[0103] In the formula, N is the number of collections.

[0104] In the process of collecting data on site, due to some uncontrollable factors, the collected data may be missing, resulting in incomplete data. If these missing data are deleted, the accuracy of subsequent model construction will be affected. Therefore, the present invention uses Lagrange interpolation method to interpolate and supplement the missing data.

[0105] Step S2: Divide PV operation scenarios

[0106] The working state of photovoltaic inverters is random and uncertain. P Varies with the intensity of sunlight. Figure 2 As shown in the figure, the present invention divides the photovoltaic operation scene into scene 1 and scene 2, that is, daytime and nighttime, according to whether there is sunlight. The present invention uses the K-means clustering algorithm to classify the active power generated by photovoltaics. P Clustering is performed to realize the division of photovoltaic operation scenarios. The following are the steps of the K-means clustering algorithm:

[0107] Step 1: Initialization. Determine the number of clustering scenes k Two data points are randomly selected as the center points of each scene.

[0108] Step 2: Assign data to the nearest center point. By calculating the distances from the data points to the center points of scene 1 and scene 2, the data points are divided into the scenes to which the nearest center points belong.

[0109] Step 3: Update the center point of the scene. For the distribution results of step 2, recalculate the center point of each scene, usually taking the mean of all data points in each scene as the new center point.

[0110] Step 4: Iterative optimization. Repeat steps 2 and 3 until the difference between the center points obtained from two consecutive updates is less than 10. -5 .

[0111] Step 5: Output the clustering results. The results include all the data points and center points that make up scene 1 and all the data points and center points that make up scene 2.

[0112] ;

[0113] ;

[0114] n 1+ n 2= N ;

[0115] In the formula, P 1. P 2 are active power datasets for scenario 1 and scenario 2 respectively; n 1. n 2 is the number of data points included in scene 1 and scene 2 respectively.

[0116] In some optional embodiments, more types of scenes may be divided according to the above method based on actual needs.

[0117] It should be noted that, because each active power corresponds to the reactive power and each harmonic current value collected at the same time, the scene division of active power also indirectly realizes the scene division of reactive power and each harmonic current.

[0118] Step S3: Constructing photovoltaic coupling model

[0119] A photovoltaic coupling model is constructed for each scenario, taking scenario 1 as an example. The total number of data points in scenario 1 is n 1. Active power is P 1, reactive power is Q1, h The subharmonic current is I h1 .

[0120] When calculating the fundamental wave power flow, the photovoltaic injection point data is P + Jb , so first of all, the active power P 1 and reactive power Q1. Joint probability modeling. The joint probability modeling function selected is a bivariate nonparametric kernel density function. The nonparametric kernel density function can estimate parameters without making prior assumptions on the input data, and is applicable to data with randomness similar to photovoltaic output.

[0121] Power probability distribution curve f ( P i , Q i )for:

[0122] ;

[0123] In the formula, h p , h q are the bandwidths of active power and reactive power respectively; K (•) is the kernel function; is the active power sequence P The mean value of each active power in 1; Reactive power sequence Q The mean value of each reactive power in 1; P 1j is the active power sequence P 1 of 1 j Active power; Q 1j Reactive power sequence Q 1 of 1 j Reactive power. Studies have shown that bandwidth has a great influence on the joint probability value. The present invention uses the method of minimizing the mean square error (MISE) to optimize the selection of bandwidth.

[0124] In the actual sampling data, Figure 3 As shown, the values ​​of each harmonic current under the same active power may be a distribution interval, and each active power corresponds to a probability distribution curve of each harmonic current. Taking the third harmonic current as an example, the active power probability distribution curve in scenario 1 corresponds to multiple third harmonic current probability distribution curves, and it is impossible to achieve a one-to-one correspondence between the active power probability distribution curve and the third harmonic current probability distribution curve. In response to this problem, the present invention innovatively proposes a method for constructing a photovoltaic coupling model, that is, by calculating the active power probability curve and the third harmonic current probability distribution curve to correspond one-to-one, and finally the two one-to-one corresponding probability curves reflect the coupling relationship between the photovoltaic fundamental wave and the harmonic wave.

[0125] In scenario 1, the active power P 1 with data interval ( P 1max -P 1min ) / 10 is divided into 10 groups, which are recorded as Z 1, Z 2, ..., Z 10 , the number of data points in each group is recorded as m 1, m 2, ..., m 10 , m 1+ m 2+…+ m 10 =n1. When the first group of active power values ​​is used, the corresponding probability distribution curve of the third harmonic current is recorded as f 311 ( I h ).in, P 1max Active power P The maximum value among 1, P 1min Active power P The minimum value among 1. The number of groups can be flexibly adjusted according to actual conditions.

[0126] In some optional schemes, the number of divided groups can be flexibly set to any positive integer other than 1 according to actual conditions.

[0127] No. b Probability distribution curve of third harmonic current f 31b ( I 3) It can be calculated through Gaussian Mixture Model (GMM).

[0128] The GMM model calculation formula is:

[0129] ;

[0130] ;

[0131] In the formula, is the number of selected models; For the Model distribution, , and They are the GMM model The coefficient of proportionality, mean, and standard deviation of the model distribution.

[0132] Among them, the proportionality coefficient Must satisfy,

[0133] and ;

[0134] b =1,2,…,9,10;

[0135] The Expectation Maximization Algorithm (EM) is used to solve the parameters.

[0136] The present invention = m b / n 1 as the first b Group active power quantity m b Active power in scenario 1 n The weight of 1 is used to merge the 10 third harmonic current probability distribution curves into one, so as to achieve a one-to-one correspondence between the active power probability curve and the third harmonic current probability distribution curve.

[0137] Finally, the probability distribution curve of the third harmonic current in scenario 1 is obtained f 31 :

[0138] ;

[0139] Since the third harmonic current probability distribution curve incorporates the weight based on power data, it can reflect the mutual relationship between power and third harmonic current. The photovoltaic coupling model constructed by the present invention takes into account the feasibility of model construction and the diversity of photovoltaic power quality analysis.

[0140] Among them, the bivariate nonparametric kernel density function used in power joint probability modeling can be replaced by a Gaussian mixture function, a Copula function, and a combination of defining multiple distribution functions.

[0141] Calculate the harmonic current probability distribution curves corresponding to the remaining harmonic currents.

[0142] Step S4: Calculate the power flow based on the optimal probability

[0143] Existing research methods for probabilistic power flow calculation include approximate methods represented by point estimation, analytical methods represented by semi-invariant methods, and simulation methods represented by Monte Carlo. These methods all have inherent limitations in calculation accuracy in certain scenarios. The present invention proposes an improved method based on optimized probabilistic power flow to achieve the purpose of improving power flow calculation accuracy and scenario adaptability.

[0144] The present invention provides four probability power flow calculation methods, namely, three-point estimation method y 1. Semi-invariant method y2. Monte Carlo Model y 3 and Latin Hypercube Sampling y 4. Select two methods from the four methods to construct the optimal power flow calculation model and calculate the final probabilistic power flow calculation results.

[0145] like Figure 4 As shown, the preferred probabilistic power flow calculation method specifically includes the following four steps.

[0146] Step 1: Use each method to perform probability power flow calculation independently. When inputting the power probability curve and harmonic current probability distribution curve in sequence, use the method y 1. Methods y 2. Methods y 3 and Methods y 4 Calculate the fundamental voltage value and harmonic current value of each node in the system.

[0147] Step 2: Calculate each probability power flow error. The distribution network will also install power quality monitoring devices at some nodes other than the photovoltaic grid-connected nodes. The nodes where the power quality monitoring devices are installed are recorded as node 1, node 2, ..., node E . Real-time data collection is performed on it, and the collected data is used as the node voltage and h The real values ​​of subharmonic currents are recorded as and , e =1,2,…, E By analyzing the error between the calculated value and the true value of the probabilistic power flow method at the node with power quality monitoring device under each method, the error is recorded as S 1. S 2. S 3. S 4.

[0148] ;

[0149] d =1,2,3,4;

[0150] In the formula, d is the method number, e is the serial number of each node in the distribution network, E is the number of nodes in the distribution network, abs (•) indicates absolute value calculation, , The methods are y d The calculated e The fundamental voltage value of each node, h Subharmonic current value, , Respectivelye The true value of the fundamental voltage of each node, h Subharmonic current true value.

[0151] Step 3: Choose the two methods with the smallest error y d1 , y d2 , build an optimal probability power flow calculation model,

[0152] ;

[0153] In the formula, Y To optimize the probabilistic power flow calculation method, d 1∈ d , d 2∈ d , d 1≠ d 2; , Methods y d1 ,method y d2 The weight of

[0154] Solve the parameters of the optimal probability power flow calculation model, where the unknown parameters are , , with the minimum comprehensive error S min is the objective function, weight , The selection range and method y d1 and methods y d2 The usage conditions are constraint solving parameters.

[0155] Specifically, the objective function is:

[0156] ;

[0157] In the formula, , The methods are y d1 ,method y d2 The calculated e The fundamental voltage value of each node, , The methods are y d1 ,method y d2 The calculated e Node h Subharmonic current value;

[0158] The particle swarm optimization algorithm is used to solve the model parameters. Among them, the particle swarm optimization algorithm can be replaced by genetic algorithm, ant colony optimization algorithm, firefly optimization algorithm and simulated annealing optimization algorithm.

[0159] Step 4: Output the final power flow calculation results. If the optimal combination is y 1 and y 2. The final output power flow calculation results are:

[0160] ;

[0161] In the formula, , Node e The fundamental voltage output value, h Subharmonic current output value; , The methods are y d1 ,method y d2 The calculated e The fundamental voltage value of each node; , The methods are y d1 ,method y d2 The calculated e Node h Subharmonic current value.

[0162] By optimizing the probabilistic power flow method, more complex and changeable scenarios can be adapted and the flexibility and accuracy of probabilistic power flow calculation can be improved.

[0163] Step S5: PV grid-connected impact assessment

[0164] like Figure 5 As shown, the evaluation object is photovoltaic. The present invention constructs a photovoltaic coupling model, and then uses the optimal probability flow calculation method to bring the model into the distribution network to perform fundamental wave probability flow and harmonic wave probability flow calculation, and obtains the voltage value and each harmonic current value of each node of the distribution network. By analyzing the flow calculation results of each node of the distribution network, the impact of photovoltaic grid connection on the distribution network is evaluated.

[0165] Fundamental voltage evaluation: By inputting P + Jb The power probability distribution f ( P , Q), the final output node voltage value of each node of the system is calculated by using the fundamental wave probability flow. The calculated final output node voltage value is compared with the voltage deviation index limit value. The voltage deviation index limit value refers to "GB / T 12325-2008 Power Quality Supply Voltage Deviation".

[0166] (1) For supply voltages of 35 kV and above, the sum of the absolute values ​​of positive and negative deviations should not exceed 10% of the system nominal voltage. If the upper and lower deviations of the supply voltage have the same sign (i.e. both are positive deviations or both are negative deviations), the larger absolute value of the deviation shall be used as the basis for measurement.

[0167] (2) For three-phase power supply voltage of 20 kV and below, the voltage deviation should be ±7% of the nominal voltage.

[0168] (3) For 220V single-phase power supply, the voltage deviation should be +7%, -10% of the nominal voltage.

[0169] Based on the fundamental voltage limit range specified in the standard, in order to assess the risk of photovoltaic grid connection to the fundamental voltage of the distribution network exceeding the limit, the fundamental voltage exceeding the limit risk assessment index is defined. R u ,

[0170] ;

[0171] In the formula, U c is the supply voltage of the node in the distribution network.

[0172] Fundamental voltage over-limit risk assessment indicators R u Compared with the upper and lower limits specified in the indicator, the fundamental voltage over-limit risk assessment indicator R u When the voltage is within the upper and lower limits, the node has no fundamental voltage over-limit problem, and the fundamental voltage over-limit risk assessment index R u The closer to the upper or lower limit, the greater the risk of exceeding the limit. When the fundamental voltage exceeds the limit risk assessment index R u When it is not within the upper and lower limits, it means that there is a risk that the fundamental voltage of the evaluation node is too low or too high.

[0173] For fundamental voltage over-limit risk assessment indicators R uNodes that are not within the specified range of the indicators have problems with too low or too high fundamental voltage, and fundamental voltage adjustment is required to make them comply with the standard requirements. The size of the reactive power affects the phase difference of the fundamental voltage, and the fundamental voltage is adjusted. The existing fundamental voltage adjustment method is mainly adjusted through professional equipment, such as reactive power adjustment devices, adding reactors to absorb reactive power, adding capacitors to emit reactive power, etc. The main governance process is: governance node demand analysis; selection of appropriate governance equipment; determination of equipment installation location; equipment parameter setting according to risk indicators; dynamic adjustment of equipment parameters to make the fundamental voltage exceed the limit risk assessment indicator. R u Stable within the standard range.

[0174] Harmonic current evaluation: By inputting h Subharmonic current probability distribution curve f hi , using the harmonic probability power flow calculation to obtain the h Subharmonic current value. Compare the calculated harmonic current with the index limit of each subharmonic current. The limit of each subharmonic current is shown in Table 1 with reference to GB / T 14549-1993 Power Quality Public Grid Harmonics.

[0175] Table 1 Public grid harmonic current limits

[0176] .

[0177] In order to assess the risk of PV grid-connected harmonic current exceeding the limit of distribution network, we define h Subharmonic current over-limit risk assessment indicators R h ,

[0178] ;

[0179] In the formula, I hc for h Nominal limit of subharmonic current;

[0180] For Node e ,like R h ≤1, then evaluate h Subharmonic current output value Normal; if R h >1, then evaluate h Subharmonic current output value Out of limit.

[0181] for h Subharmonic current over-limit risk assessment indicators R hFor nodes with a harmonic current limit greater than 1, there is a risk of harmonic current exceeding the limit, and harmonic current adjustment is required to make the hth harmonic current limit risk assessment index R h Less than 1. Harmonic current content exceeding the limit will cause output current waveform distortion, affect power supply quality, and even damage sensitive equipment. The existing idea of ​​managing harmonic current exceeding the limit is mainly to filter out the harmonic current exceeding the limit or use equipment to increase the reverse harmonic current exceeding the limit to achieve the purpose of offset. The management equipment used is mostly filters and converters. The filter management process is: determine the target level of harmonic management based on the harmonic risk index; select appropriate filtering equipment; determine the filter installation location; set the filter parameters; real-time monitoring and optimization of filter parameters. h Subharmonic current over-limit risk assessment indicators R h The control process of the converter is as follows: determine the amplitude and frequency according to the harmonic current over-limit value; select the control converter; set the converter parameters to make it emit the reverse harmonic current value; dynamically observe h Subharmonic current over-limit risk assessment indicators R h Make it stable less than 1.

[0182] The present invention also provides a photovoltaic coupling model grid-connected evaluation system based on optimal probability power flow, comprising:

[0183] Data acquisition module, used to collect grid connection point data of photovoltaic inverters;

[0184] A scenario division module is used to divide the operation scenarios of the photovoltaic inverter according to the output distribution of active power;

[0185] A photovoltaic coupling model building module is used to build a photovoltaic coupling model based on an operation scenario. The photovoltaic coupling model is used to obtain power probability distribution and h Subharmonic current probability distribution curve;

[0186] The calculation module is used to calculate the fundamental voltage output value of each node in the distribution network based on the power probability distribution. h Subharmonic current probability distribution curve calculation of each node in the distribution network h Subharmonic current output value;

[0187] An evaluation module is used to evaluate the fundamental voltage based on the fundamental voltage output value; h Subharmonic current output value for harmonic current evaluation.

[0188] The technical solution provided by the present invention can analyze the relationship between output power and harmonic current from the perspective of photovoltaic output data drive by constructing a photovoltaic coupling model, taking into account the comprehensiveness and feasibility of model construction; on the other hand, the accuracy of probabilistic power flow calculation can be improved based on the preferred probabilistic power flow calculation method; further, the power flow calculation method that considers the fundamental harmonic coupling and probabilistic power flow accuracy helps to more accurately reflect the impact of photovoltaic output fundamental and harmonic waves on the distribution network under various photovoltaic access scenarios, and has better scenario applicability.

[0189] Furthermore, the above data-driven photovoltaic coupling model and power flow optimization scheme have achieved the following beneficial effects on the basis of the above-mentioned beneficial effects: using Lagrange interpolation method to supplement missing data, optimizing the scene center point, and using the method of minimizing the mean square error to select the active power bandwidth and reactive power bandwidth all help to improve the evaluation accuracy.

Claims

1. A photovoltaic coupling model grid-connected evaluation method based on optimal probabilistic power flow, characterized in that: The following steps are involved: Step S1: Collecting photovoltaic inverter grid connection point data; Step S2: dividing the operation scenarios of the photovoltaic inverter according to the output distribution of active power; Step S3: Build a photovoltaic coupling model based on the operation scenario to obtain the power probability distribution and h Subharmonic current probability distribution curve; where: h =3,5,7,9…; Step S4: Based on the power probability distribution, the fundamental voltage output value of each node in the distribution network is calculated by optimizing the probability flow; h The probability distribution curve of subharmonic current is used to calculate the probability distribution of each node in the distribution network through the optimal probability flow. h Subharmonic current output value; Step S5: Based on the fundamental voltage output value, perform fundamental voltage evaluation; Based on h Subharmonic current output value, for harmonic current evaluation; Step S4 specifically includes: Step S41: Based on the power probability distribution curve f ( P i , Q i )and h Subharmonic current probability distribution curve f hi , respectively using methods y 1. Methods y 2. Methods y 3 and Methods y 4. Calculate the probability power flow and obtain the fundamental voltage value and harmonic current value of each node in the distribution network; Step S42: Calculation method y 1. Methods y 2. Methods y 3 and Methods y The error between the calculated value and the true value under 4 S d ; ； d ＝1，2，3，4； e ＝1，2，…， E ； In the formula, d is the method number, e is the serial number of each node in the distribution network, E is the number of nodes in the distribution network, abs (•) indicates absolute value calculation, , The methods are y d The calculated e The fundamental voltage value of each node, h Subharmonic current value, , Respectively e The true value of the fundamental voltage of each node, h Subharmonic current true value; Step S43: Select the two methods with the smallest error y d1 , y d2 Combine and construct the optimal probability power flow calculation model, ； In the formula, Y To optimize the probabilistic power flow calculation method, d 1∈ d , d 2∈ d , d 1≠ d 2; , Methods y d1 ,method y d2 The weight of With minimum comprehensive error S min is the objective function, and the solution method y d1 Weight and methods y d2 Weight , and calculate the fundamental voltage output value and h Subharmonic current output value, ； ； In the formula, , Node e The fundamental voltage output value, h Subharmonic current output value, , The methods are y d1 ,method y d2 The calculated e The fundamental voltage value of each node, , The methods are y d1 ,method y d2 The calculated e Node h Subharmonic current value; Among them, the method y 1. Methods y 2. Methods y 3 and Methods y 4 are three-point estimation method, semi-invariant method, Monte Carlo model and Latin hypercube sampling method.

2. The photovoltaic coupling model grid-connected evaluation method based on the optimal probability power flow according to claim 1, characterized in that: Step S1 specifically includes: The active power of the photovoltaic inverter grid connection point is collected at intervals of 1 minute with a 95% probability of maximum value. P , reactive power Q and harmonic currents I h ; ； ； ； In the formula, N is the number of collections.

3. The photovoltaic coupling model grid-connected evaluation method based on the optimal probability power flow according to claim 2, characterized in that: Lagrange interpolation method was used to supplement missing data.

4. The photovoltaic coupling model grid-connected evaluation method based on the optimal probability power flow according to claim 2, characterized in that: Step S2 specifically includes: Clustering active power using K-means clustering algorithm P ; Step S21: Determine the number of clustering scenes k , and randomly select data points as center points for each clustering scenario, k ∈(2,3,4); Step S22: Calculate the distance between each data point and each center point, and divide each data point into the clustering scene to which the center point closest to it belongs; Step S23: For each clustering scene, take the mean of the data points therein as the new center point; Step S24: Output each clustering scene. ； ； In the formula, i is the sequence number of the clustering scene, P i For the i The active power sequence of the clustering scenario, i =1,2,…, k ; j is the sequence number of active power in the clustering scenario, P ij is the active power sequence P i The j Active power, n i For the i The number of data points in a clustering scenario, j =1,2,…, n i ; Each clustering scenario also includes: the number of data points and the active power sequence P i The number of data points is the same as the reactive power series Q i and h Subharmonic current sequence I hi .

5. The photovoltaic coupling model grid-connected evaluation method based on the optimal probability power flow according to claim 4, characterized in that: After step S23, the following steps are also included: Repeat steps S22 to S23 to iteratively update the center point until the difference between the center points obtained from two consecutive updates is less than 10 -5 .

6. The photovoltaic coupling model grid-connected evaluation method based on the optimal probability power flow according to any one of claims 4 or 5, characterized in that: Step S3 specifically includes: For each scene, Step S31: Active power sequence P i and reactive power sequence Q i Perform joint probability modeling to obtain the power probability distribution curve f ( P i , Q i ), ； In the formula, h p , h q They are active power bandwidth and reactive power bandwidth respectively. K (•) is the kernel function, is the active power sequence P i The mean value of each active power in P ij is the active power sequence P i The j Active power, Reactive power sequence Q i The mean value of reactive power in Q ij Reactive power sequence Q i The j Reactive power; Step S32: The active power sequence is transmitted through the data interval P i Divided into a Group, a =2,3,4,…; The number of data points in each group is m i1 , m i2 , …, m ia , ； The data interval is: ； In the formula, P imax , P imin The active power series are P i The maximum and minimum values ​​in ; Step S33: Calculate the first b Group active power corresponding to h Subharmonic current probability distribution sub-curve f hib ( I h ), ； ； In the formula, is the ordinal number of the selected model, is the number of selected models, For the Model distribution, For the The proportionality coefficient of the model distribution, For the The mean of the model distribution, For the The standard deviation of the model distribution; in, ， ； b ＝1,2，…， a ； Step S34: Merge a strip h Subharmonic current probability distribution sub-curve, get h Subharmonic current probability distribution curve f hi , ； In the formula, For the b The number of active power in the group accounts for i The weight of the amount of active power in each cluster scenario.

7. The photovoltaic coupling model grid-connected evaluation method based on the optimal probability power flow according to claim 6, characterized in that: Selecting active power bandwidth by minimizing mean square error h p and reactive power bandwidth h q .

8. The photovoltaic coupling model grid-connected evaluation method based on the optimal probability power flow according to claim 6, characterized in that: (1) Based on the fundamental voltage output value, the fundamental voltage of each node is evaluated, including: Calculate the fundamental voltage over-limit risk assessment index R u , ； In the formula, U c is the supply voltage of the node in the distribution network; According to GB / T 12325-2008 Power Quality Supply Voltage Deviation, For power supply voltages of 35kV and above, the voltage deviation index limit range is: ； For three-phase power supply voltage of 20kV and below, the voltage deviation index limit range is: ； For 220V single-phase power supply voltage, the voltage deviation index limit range is: ； For Node e , if the fundamental voltage output value If the fundamental voltage output value is within the voltage deviation index limit range, the fundamental voltage is evaluated to be normal; If the fundamental voltage output value is lower than the voltage deviation index limit range, the fundamental voltage is evaluated to be too low; if the fundamental voltage output value is If it is higher than the voltage deviation index limit range, the fundamental voltage is evaluated to be too high; (2) Based on h Subharmonic current output value, to evaluate the harmonic current of each node, including: calculate h Subharmonic current over-limit risk assessment indicators R h , ； In the formula, I hc It is specified in GB / T 14549-1993 Power Quality Public Grid Harmonics h Nominal limit of subharmonic current; For Node e ,like R h ≤1, then evaluate h Subharmonic current output value Normal; if R h >1, then evaluate h Subharmonic current output value Out of limit.

9. A photovoltaic coupling model grid-connected evaluation system based on optimal probability power flow, characterized in that: include: Data acquisition module, used to collect grid connection point data of photovoltaic inverters; A scenario division module is used to divide the operation scenarios of the photovoltaic inverter according to the output distribution of active power; A photovoltaic coupling model building module is used to build a photovoltaic coupling model based on an operation scenario. The photovoltaic coupling model is used to obtain power probability distribution and h Subharmonic current probability distribution curve; The calculation module is used to calculate the fundamental voltage output value of each node in the distribution network based on the power probability distribution. h Subharmonic current probability distribution curve calculation of each node in the distribution network h Subharmonic current output value; An evaluation module, used for evaluating the fundamental voltage based on the fundamental voltage output value; based on h Subharmonic current output value, for harmonic current evaluation; The calculation process of the calculation module includes: Based on the power probability distribution curve f ( P i , Q i )and h Subharmonic current probability distribution curve f hi , respectively using methods y 1. Methods y 2. Methods y 3 and Methods y 4. Calculate the probability power flow and obtain the fundamental voltage value and harmonic current value of each node in the distribution network; Separate calculation method y 1. Methods y 2. Methods y 3 and Methods y The error between the calculated value and the true value under 4 S d ; ； d ＝1，2，3，4； e ＝1，2，…， E ； In the formula, d is the method number, e is the serial number of each node in the distribution network, E is the number of nodes in the distribution network, abs (•) indicates absolute value calculation, , The methods are y d The calculated e The fundamental voltage value of each node, h Subharmonic current value, , Respectively e The true value of the fundamental voltage of each node, h Subharmonic current true value; Choose the two methods with the smallest error y d1 , y d2 Combine and construct the optimal probability power flow calculation model, ； In the formula, Y To optimize the probabilistic power flow calculation method, d 1∈ d , d 2∈ d , d 1≠ d 2; , Methods y d1 ,method y d2 The weight of With minimum comprehensive error S min is the objective function, and the solution method y d1 Weight and methods y d2 Weight , and calculate the fundamental voltage output value and h Subharmonic current output value, ； ； In the formula, , Node e The fundamental voltage output value, h Subharmonic current output value, , The methods are y d1 ,method y d2 The calculated e The fundamental voltage value of each node, , The methods are y d1 ,method y d2 The calculated e Node h Subharmonic current value; Among them, the method y 1. Methods y 2. Methods y 3 and Methods y 4 are three-point estimation method, semi-invariant method, Monte Carlo model and Latin hypercube sampling method.

Images