Optimal control method and system for distribution network with high penetration of distributed generators, and electronic device, computer-readable storage medium and computer program product

Abstract

Provided in the embodiments of the present disclosure are an optimal control method and system for a distribution network with a high penetration of distributed generators, and an electronic device, a computer-readable storage medium and a computer program product. The optimal control method is applied to a centralized control platform. The method comprises: on the basis of an association relationship between power outputs of distributed generators at various levels, performing multi-level forecasting on historical power generation data of each distributed generator, so as to obtain a day-ahead forecasted output value of each distributed generator; for each distributed generator, determining a day-ahead output adjustment value on the basis of historical power flow data and an optimal power flow optimization model; and sending to the corresponding distributed generator the output adjustment value and forecasted output value of each distributed generator as centralized control instructions, so that the distributed generator controls an output value during each control period within the day ahead on the basis of an acquired real-time output measurement value, the acquired output adjustment value and the acquired forecasted output value.

Description

Optimal control method and system of high-proportion distributed power distribution network, electronic device, computer readable storage medium and computer program product

[0001] Cross-reference to Related Applications

[0002] The embodiments of the present disclosure are based on the Chinese patent application No. 202410819994.3, filed on June 24, 2024, entitled "Optimal control method and system of high-proportion distributed power distribution network", and the priority of the Chinese patent application is claimed. The entire contents of the Chinese patent application are hereby incorporated by reference into the present disclosure. TECHNICAL FIELD

[0003] The present disclosure relates to, but is not limited to, the technical field of operation control of high-proportion distributed power distribution network, and particularly relates to an optimal control method and system of high-proportion distributed power distribution network, an electronic device, a computer readable storage medium and a computer program product. BACKGROUND

[0004] With the development of distributed power sources, especially the rapid development of photovoltaic, gas turbine, hybrid heat supply, etc., the traditional distribution network is gradually transforming into an active distribution network with certain controllability. Under the driving factors of mature photovoltaic power generation technology and economic cost reduction, the number of distributed photovoltaic grid-connected in the medium and low voltage distribution network is rapidly increasing, and the capacity is greatly increased. The research on the safe and stable operation of the distribution network by photovoltaic power generation and other distributed power sources has attracted more and more attention from researchers. SUMMARY

[0005] The embodiments of the present disclosure provide an optimal control method and system of high-proportion distributed power distribution network, an electronic device, a computer readable storage medium and a computer program product.

[0006] The embodiments of the present disclosure provide an optimal control method of high-proportion distributed power distribution network, applied to a centralized control platform, the method comprising:

[0007] Based on the correlation relationship of the power generation of the distributed power sources at each level, the historical power generation data of each distributed power source is predicted in multiple levels to obtain the predicted output value of each distributed power source in the next day;

[0008] For each distributed power source, based on the historical power flow data and the optimal power flow optimization model, the output adjustment value in the next day is determined;

[0009] The output adjustment value and the output prediction value of each distributed power supply are sent to the corresponding distributed power supply as centralized control instructions, so that each distributed power supply controls the output value of each control period in the next day based on the obtained real-time output measurement value, the output adjustment value and the output prediction value.

[0010] The disclosure also provides a high-proportion distributed power supply power distribution network optimization control method, applied to a distributed power supply, and the method comprises the following steps:

[0011] The centralized control platform issues centralized control instructions in day-ahead, and the centralized control instructions comprise an output adjustment value and an output prediction value of the current distributed power supply in the next day, wherein the output adjustment value is obtained by the centralized control platform based on historical power flow data and an optimal power flow optimization model in day-ahead, and the output prediction value is obtained by the centralized control platform based on the correlation between the output of each level of distributed power supply and the historical power generation data of each level of distributed power supply in multi-level prediction in day-ahead;

[0012] For each control period in the day, a real-time output measurement value is obtained in the control period, and the output value of the control period is controlled based on the real-time output measurement value, the output prediction value and the output adjustment value.

[0013] The disclosure also provides a high-proportion distributed power supply power distribution network optimization control system, comprising:

[0014] A centralized control platform and a plurality of distributed power supplies of each level, wherein the centralized control platform is in communication connection with each distributed power supply;

[0015] The centralized control platform is configured to perform multi-level prediction on the historical power generation data of each distributed power supply based on the correlation between the power generation power of each level of distributed power supply, to obtain the output prediction value of each distributed power supply in the next day, to determine the output adjustment value of each distributed power supply in the next day based on historical power flow data and an optimal power flow optimization model, and to send the output adjustment value and the output prediction value of each distributed power supply to the corresponding distributed power supply as centralized control instructions;

[0016] The distributed power supply is configured to obtain a real-time output measurement value in each control period in the day, and to control the output value of the control period based on the real-time output measurement value and the output prediction value and the output adjustment value received in day-ahead.

[0017] The disclosure also provides an electronic device, comprising at least one processor and a memory;

[0018] The memory is configured to store one or more programs.

[0019] The one or more programs, when executed by the at least one processor, implement the optimization control method of the high-proportion distributed power distribution network according to any one of the above.

[0020] The disclosure also provides a readable storage medium having an execution program stored thereon, and the execution program, when executed, implements the optimization control method of the high-proportion distributed power distribution network according to any one of the above.

[0021] The disclosure provides a computer program product, which includes a computer program or instructions, and when the computer program or instructions are executed on an electronic device, the electronic device executes any one of the above methods.

[0022] Compared with the prior art, the beneficial effects of the disclosure are as follows:

[0023] The disclosure provides an optimization control method of a high-proportion distributed power distribution network, which obtains an output prediction value of each distributed power in the next day by performing multi-level prediction in the day-ahead, and improves the prediction accuracy by considering the correlation between the outputs of the distributed powers at each level. Meanwhile, the output adjustment value is obtained through a target optimization control model, so that the distributed power can obtain the output prediction value and the output adjustment value in advance, the large calculation and control instruction interaction are completed in advance, and the calculation and network bandwidth pressure of the real-time control process are reduced; during real-time regulation, each distributed power obtains the output prediction value and the output adjustment value in advance, and then performs real-time control adjustment of the current output based on the output prediction value and the output adjustment value, so as to realize local rapid control of the distributed power and improve the adjustment effect. The scheme combines the advance centralized prediction optimization and the local real-time control of the distributed power, balances the processing rate and the control effect, and has strong implementability.

[0024] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, but not limiting the disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the disclosure, the drawings needed to be used in the embodiments of the disclosure or in the background art will be described below.

[0026] The drawings herein are incorporated into the specification and form part of the specification, which show the embodiments consistent with the disclosure, and are used to illustrate the technical solutions of the disclosure together with the specification.

[0027] Fig. 1 is a flow diagram of an optimization control method of a high-proportion distributed power supply power distribution network according to an embodiment of the present disclosure;

[0028] Fig. 2 is a flow diagram of a weight combination photovoltaic output prediction according to an embodiment of the present disclosure;

[0029] Fig. 3 is a diagram of a globally reliable domain sequence quadratic programming solving process according to an embodiment of the present disclosure;

[0030] Fig. 4 is a flow diagram of an optimization control method of a high-proportion distributed power supply power distribution network according to an embodiment of the present disclosure;

[0031] Fig. 5 is a flow diagram of an optimization control method of an optimization control system of a high-proportion distributed power supply power distribution network according to an embodiment of the present disclosure;

[0032] Fig. 6 is a structural diagram of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0033] In order to make the personnel in the technical field better understand the present disclosure, the specific embodiments of the present disclosure will be further described in detail below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without making creative efforts fall within the scope of the present disclosure.

[0034] The terms "first", "second", and the like in the specification and claims of the present disclosure and the above-described drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed, or can optionally include other steps or units inherent to the process, method, product or device.

[0035] It should be understood that in the present disclosure, "at least one" refers to one or more, "multiple" refers to two or more, "at least two" refers to two or three and three or more, and "and / or" is used to describe the association relationship of the associated objects, which means that there can be three relationships, for example, "A and / or B" can mean that there are three cases of only A, only B, and A and B at the same time, where A and B can be singular or plural. The character " / " can represent that the associated objects before and after are "or" relationship, which means any combination of these items, including single item or any combination of multiple items. For example, at least one of a, b or c, can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple. The character " / " can also represent the division sign in mathematical operation, for example, a / b=a divided by b; 6 / 3=2. "At least one of the following" or similar expressions.

[0036] With the development of distributed power supply, especially the rapid development of photovoltaic, gas turbine, hybrid heat supply, etc., the traditional power distribution network is gradually changing into an active power distribution network with certain controllability. Under the driving factors of mature photovoltaic power generation technology and economic cost reduction, the number of distributed photovoltaic grid-connected in medium and low voltage distribution network is rapidly increasing, and the capacity is greatly increasing. The research on the safe and stable operation of the distribution network by photovoltaic power generation and other distributed power supply has attracted more and more attention of researchers.

[0037] High proportion of distributed photovoltaic grid-connected changes the direction of traditional power flow, and sometimes the reverse power flow is greater than the conventional power flow, and there is a possibility of reverse overload or even overload of lines and transformers in local areas. At the same time, the time sequence characteristics of photovoltaic and load determine the mismatch between the peak period of photovoltaic output and the heavy load period, and the high proportion of distributed photovoltaic access brings the "duck curve" phenomenon, and the network's lowest load period changes from night to day, and the risk of network overvoltage increases. In addition, photovoltaic power generation is greatly affected by factors such as light intensity and temperature, and its output has the characteristics of intermittency and randomness, which makes the voltage fluctuation problem in the network obvious. Therefore, it is of great significance to study the dynamic optimization control method of high proportion of photovoltaic distribution network to promote the access and consumption of photovoltaic and improve the flexible response ability of power system.

[0038] At present, the coordinated optimization control problem of intelligent distribution network of distributed power supply is mainly based on the prediction of neural network model for optimization control. However, the existing prediction method mostly uses a single model for prediction, and the prediction performance is very limited, and the correlation of distributed power supply power under the multi-level distribution network structure is ignored, resulting in low prediction accuracy.

[0039] For a high proportion, large-scale distributed power distribution network, the calculation speed, instruction issuing and other processes are difficult to meet the real-time regulation requirements due to the large scale of distributed power and high model complexity; if the regulation is directly based on the obtained prediction results, the prediction regulation amount may not match the actual situation in extreme or sudden situations, and the regulation accuracy is low.

[0040] Embodiments of the present disclosure:

[0041] The method provided by the embodiments of the present disclosure for optimizing control of a high proportion distributed power distribution network is applied to a centralized control platform, and FIG. 1 is a flowchart of the method for optimizing control of a high proportion distributed power distribution network, as shown in FIG. 1, the method can include the following steps S110-S130:

[0042] In step S110, based on the correlation of the power generation of distributed power at each level, the historical power generation data of each distributed power is predicted at multiple levels to obtain the output prediction value of each distributed power in the next day.

[0043] In step S120, for each distributed power, based on the historical power flow data and the optimal power flow optimization model, the output adjustment value in the next day is determined.

[0044] In step S130, the output adjustment value and the output prediction value of each distributed power are sent to the corresponding distributed power as centralized control instructions, so that each distributed power controls the output value of each control period in the next day based on the obtained real-time output measurement value, output adjustment value and output prediction value.

[0045] In the example embodiment, each control period can be determined based on a control frequency or a control time interval. Each level of the distributed power supply includes a distribution area, a feeder, and a transformer, and the distributed power supply at different levels corresponds to different voltage levels, and the distributed power supply can be a distributed photovoltaic or other energy forms such as wind power generation. The embodiments of the present disclosure improve the prediction accuracy of the output value of the distributed power supply by considering the correlation between the distributed power supplies at different levels. The centralized control platform performs multi-level prediction and optimal power flow optimization on the distributed power supplies at each level of the next day based on the correlation of the power generation of the distributed power supplies at each level, obtains the output prediction value and the output adjustment value of the next day, and issues them to the corresponding distributed power supply; each distributed power supply controls the output value of each control period in the current prediction period based on real-time measurement data and the output prediction value and the output adjustment value received in advance. The centralized control platform performs one multi-level photovoltaic prediction and one day-ahead centralized optimization control on the next day in advance, and the distributed power supply corrects the output based on the real-time measurement value in each time period, realizes the intra-day local rapid control of the distributed power supply, and achieves good control effect and improves the control speed.

[0046] The existing high-proportion distributed photovoltaic output prediction method mostly uses a single model for prediction, and the prediction performance is very limited, and the correlation of the distributed photovoltaic power under the multi-level distribution network structure is ignored. Therefore, it is necessary to consider the adaptive combination prediction method of multiple prediction models, and to correct the power prediction results of the multi-level distributed photovoltaic power depending on the correlation of the distribution area-feeder-transformer.

[0047] The embodiments of the present disclosure provide an optimization control method for a high-proportion distributed power supply distribution network, which obtains the output prediction value of each distributed power supply in the next day by performing multi-level prediction in advance, considers the correlation between the outputs of the distributed power supplies at each level, and improves the prediction accuracy. At the same time, the output adjustment value is obtained through the target optimization control model, so that the distributed power supply can obtain the output prediction value and the output adjustment value in advance, realize the completion of large calculation and control instruction interaction in advance, and reduce the calculation and network bandwidth pressure of real-time control process; during real-time regulation, each distributed power supply can obtain the real-time measurement value of its own output, and then perform real-time control adjustment of the current output based on the output prediction value and the output adjustment value received in advance, thereby realizing the local rapid control of the distributed power supply and improving the regulation effect. The present scheme combines the advance centralized prediction optimization and local real-time control of the distributed power supply, balances the processing rate and control effect, and has strong implementability.

[0048] In an example embodiment, considering that there are many factors affecting the distributed photovoltaic power generation, the distributed photovoltaic outputs at different voltage levels in the same area have certain correlation, and photovoltaic power generation has intermittency and randomness, the upper limit of the power prediction accuracy is low when a single algorithm is used.

[0049] The implementation process of the multi-level prediction of the historical power generation data of the plurality of distributed power sources of each level based on the correlation of the power generation of the distributed power sources of each level in step S110 includes:

[0050] The historical power generation data of the distributed power sources of each level in a preset historical period is obtained, and the historical power generation data includes historical output data of the distributed power sources of each level and corresponding historical environmental data.

[0051] The historical output data of the distributed power sources of each level and the corresponding historical environmental data are input into a plurality of different neural network models as input data, and the output data of each neural network model is obtained.

[0052] The output data of each neural network model is fused to obtain the output prediction value of each distributed power source in the next day.

[0053] In the example embodiment, the historical environmental data can include at least one of air pressure, humidity, wind speed, total radiation, etc., and other related weather information can be added as historical environmental data as needed. The plurality of different neural network models include support vector machines (SVM), genetic algorithm optimized back propagation neural network (Genetic Algorithm Back Propagation, GA-BP), and radial basis neural network (Radial Basis Function, RBF), and the plurality of different neural network models can also include XGBoost (eXtreme Gradient Boosting), which is a machine learning system under the Gradient Boosting framework, and other artificial intelligence algorithms such as ensemble learning. The fusion processing can be weighted averaging processing.

[0054] For example, the implementation process of the fusion processing of the output data of each neural network model to obtain the output prediction value of each distributed power source in the next day is as follows: for each distributed power source, based on the prediction error of each neural network model, the prediction weight corresponding to each neural network model is determined; based on the output data of each neural network model and the corresponding prediction weight, the output prediction value of the distributed power source in the next day is determined.

[0055] In the embodiment, the prediction error of each neural network model can be determined based on the actual value and the predicted value, the variance corresponding to the prediction sample can be determined based on the prediction error, and the prediction weight of each neural network model can be determined based on the variance of the prediction sample of each neural network model and the variance fusion result of the plurality of neural network models. The variance fusion result of the plurality of neural network models can be the sum of the reciprocals of the variances. The determination of the prediction weight of each neural network model can be based on the test sample, or the prediction weight of each neural network model can be determined in the prediction process, or the prediction weight of each neural network model determined by the test sample can be updated in the prediction process. The present example does not limit the above.

[0056] For example, FIG. 2 is a schematic diagram of a weight combination photovoltaic power output prediction process provided by an embodiment of the present disclosure. As shown in FIG. 2, the output prediction value can be determined by the following steps:

[0057] First step: constructing an original data set 21: obtaining historical data of photovoltaic power 2102 at different photovoltaic access voltage levels 2103 (different levels) and corresponding data such as air pressure, humidity, wind speed, and total radiation 2101.

[0058] Second step: normalizing the data in the original data set, i.e., linearly changing to the [-1, 1] interval, and then dividing the normalized data set into a training set and a test set (for example, dividing according to 8:2).

[0059] Third step: inputting the training set into the SVM, GA-BP, and RBF models in parallel for training, and then inputting the test set into the trained SVM 22, GA-BP 23, and RBF 24 models in parallel for testing, to obtain the photovoltaic power prediction value P SVM 25 of the support vector machine, the photovoltaic power prediction value P GA-BP 26 of the genetic algorithm optimized BP neural network, and the photovoltaic power prediction value P RBF 27 of the radial basis neural network, respectively.

[0060] Fourth step: according to the power prediction results of the three models, first calculating the variance δ a (a = 1, 2, 3) corresponding to the prediction result of each model, and then calculating the weight coefficient η a (a = 1, 2, 3) of the corresponding single prediction model according to the variance, as shown in the following formula (1):

[0061] In the formula, δ a represents the variance corresponding to the prediction sample of the a-th model; η a is the weight of the prediction result of the a-th model, and a = 1, 2, 3 (corresponding to the weight η1 28, the weight η2 29, and the weight η3 210 in FIG. 2); denotes the error between the actual value and the predicted value of the a-th model in the n-th sample; n s denotes the error between the actual value and the predicted value of the a-th model in the n-th sample; n s denotes the number of samples, denotes the average error of the a-th model.

[0062] The fifth step is to multiply the calculated weight by the predicted value of the corresponding single prediction model respectively, and finally obtain the photovoltaic power prediction result P out 211: P out = η1P SVM + η2P GA-BP + η3P RBF (2)

[0063] In this embodiment, support vector machine (SVM), genetic algorithm optimized BP neural network (GA-BP) and radial basis neural network (RBF) algorithms are combined, a multi-level resource prediction method based on variable weight combination of multiple models is used, the correlation characteristics of the distributed power generation power of each level are considered, compared with the prediction algorithm of a single model, the prediction accuracy of the node power generation power is improved, the complexity of the model is reduced, and high-precision prediction of multi-level distributed photovoltaic power is realized.

[0064] In terms of optimal power flow (OPF) optimization objectives of a distribution network, there are currently single-objective optimization (SOP) and multi-objective optimization (MOP). SOP selects one objective as the optimization objective, such as maximum distributed photovoltaic output, and considers other objectives or requirements as constraint conditions. MOP requires several objectives to be simultaneously optimized, which is more in line with actual application requirements. Unlike single-objective optimization, which has a limited solution, multi-objective optimization usually has a set of equilibrium solutions, known as Pareto solutions. In terms of OPF solving algorithms, since the OPF problem of a source distribution network is essentially a constrained nonlinear programming problem, how to linearize the power flow equation with high precision or convex the problem is the key to solving the OPF of the source distribution network.

[0065] The existing OPF optimization objective has not taken into account the maximum photovoltaic output and the economic and safe operation of the network, and has not considered the photovoltaic coordination constraint under high penetration rate access and light load extreme scenarios, often resulting in an unreasonable optimization scheme with excessive single photovoltaic adjustment and small or no adjustment of other photovoltaic adjustment. At the same time, the existing OPF algorithm is difficult to achieve accurate and fast solution under large-scale variable participation.

[0066] In some example embodiments, before determining the output adjustment value of the next day based on the historical power flow data and the optimal power flow optimization model, the optimal power flow optimization model is constructed by the following process:

[0067] With the maximization of the output of the distributed power source, the minimization of the network voltage deviation of the power distribution network, and the minimization of the network line loss of the power distribution network as the target, the target optimization function is determined;

[0068] Considering the operation safety of the power distribution network, the output range of each distributed power source, and the output adjustment proportion of each distributed power source, the constraint condition of the target optimization function is determined;

[0069] Based on the target optimization function and the constraint condition, the optimal power flow optimization model is constructed.

[0070] For example, first, the objective function of the optimal power flow optimization model of the high-proportion distributed photovoltaic power distribution network is established, and the process is as follows:

[0071] (1) Maximize the output of the distributed photovoltaic

[0072] In order to maximize the efficiency of the distributed photovoltaic as much as possible, such as improving the utilization rate of wind energy and solar energy, the distributed photovoltaic needs to transmit active power output to the power system as much as possible under the condition allowed by the system. Therefore, the first optimization target is to maximize the active power output of the distributed photovoltaic:

[0073] In the formula, N PV is the number of distributed photovoltaics in the system, is the active power output value of the nth distributed photovoltaic.

[0074] (2) Minimize the network voltage deviation

[0075] The distributed photovoltaic connected to the system will cause the change of the voltage of each node of the system, and the voltage amplitude of the system is an important indicator to measure the safety and power quality of the system. Low voltage amplitude will lead to poor operation of the load, and high voltage amplitude will make the load and the system work in an unsafe state. The network voltage deviation index at the system level is defined as follows:

[0076] In the formula, I is the number of network nodes, U i , are the voltage amplitude of node i, the reference voltage amplitude, the upper limit of the voltage amplitude, and the lower limit of the voltage amplitude, respectively.

[0077] (3) Minimize the network line loss

[0078] In order to make the network run in a more economical way, the active network loss should be minimized, that is:

[0079] where B is the set of network branches, i, j are the first and last end node numbers of the branch, g ij is the branch conductance between node i, j, θ ij is the phase angle difference of node i, j voltage.

[0080] (4) Optimize the combination of the above objective functions

[0081] For the above three objective functions, the weighted method is used to convert the multi-objective optimization model into a single-objective optimization problem. The objective function is as follows:

[0082] where, respectively represent the initial value of the distributed power output, the initial value of the network voltage deviation of the distribution network, and the initial value of the network line loss of the distribution network, and α1, α2, α3 represent the corresponding weights of the distributed power output, the corresponding weights of the network voltage deviation of the distribution network, and the corresponding weights of the network line loss of the distribution network, and their values are positive; F1(x), F2(x), F3(x) represent the distributed power output, the network voltage deviation of the distribution network, and the network line loss of the distribution network, respectively, and N PV is the number of distributed power sources in the distribution network, is the active power output value of the nth distributed power source, I is the number of network nodes of the distribution network, U i represents the voltage amplitude of network node i, represents the reference voltage amplitude of network node i, represents the upper limit of the voltage amplitude of network node i, represents the lower limit of the voltage amplitude of network node i, B is the set of network branches, and the subscripts i, j represent the branch formed by network node i and network node j as the first and last end nodes, g ij is the branch conductance between node i, j, θ ij is the phase angle difference of node i, j voltage. The first term takes a negative sign because the original target of the first term is to maximize and the fused target is to minimize, so a negative sign is added.

[0083] Then, the equality constraints and inequality constraints of the optimal power flow optimization model of the high-proportion distributed photovoltaic distribution network are established. Through the power flow power constraint, the voltage amplitude constraint of the distributed power source, and the branch power constraint, the operation safety of the distribution network is satisfied; the output prediction value of each distributed power source is taken as the upper limit of the output of the corresponding distributed power source to determine the output range constraint of each distributed power source; the equal adjustment proportion of the output of each distributed power source is taken as the output adjustment proportion constraint. The implementation process is as follows:

[0084] (1) Equality and inequality constraints related to power flow

[0085] The network operation constraints should be satisfied while optimizing the network objective function, such as power flow equation constraints, node voltage amplitude constraints, line power constraints and control variable constraints. The power flow equation constraints are shown in equation (7):

[0086] where, Pdis(i) and Qdis(i) are the active power and reactive power of the distributed generator connected at network node i, Pload(i) and Qload(i) are the active power and reactive power of the load connected at network node i, U i U(i) is the voltage amplitude of network node i, U j U(j) is the voltage amplitude of network node j, I is the number of network nodes of the distribution network, G ij B ij Y(i,j) and Y(j,i) are the elements in the admittance matrix between network nodes i and j, θ ij is the phase angle difference between the voltages of network nodes i and j.

[0087] The upper and lower limits of the voltage amplitude of each node of the network are shown in equation (8), and the line power constraints are shown in equation (9): U i,min ≤U i ≤U i,max (8)

[0088] where, U i,min U(i) is the lower limit of the voltage amplitude of network node i, U i,max U(i) is the upper limit of the voltage amplitude of network node i; S is the apparent power of line l, P and Q are the active power and reactive power flowing through line l, Smax is the upper limit of the apparent power of line l.

[0089] (2) Output regulation constraints of photovoltaic groups

[0090] In order to fully utilize renewable energy, distributed photovoltaic and centralized photovoltaic generally work at rated power, and the adjustment of the active power is generally to reduce the output or keep it, and the reactive power output is zero, as shown in the following equation:

[0091] where, Pmin(i) and Pmax(i) are the minimum active power and maximum active power of the photovoltaic connected at node i.

[0092] In the high proportion distributed photovoltaic access scene, in order to avoid the down-regulation of the output of a certain distributed photovoltaic exceeding its own adjustment capacity, the quasi-equal curtailing ratio of PV (QCR) constraint of distributed photovoltaic output regulation is introduced, the curtailing ratio of PV (CR) refers to the ratio of the actual active power down-regulation value of the photovoltaic power station to the maximum active power value that can be generated, and the quasi-equal curtailing ratio output is to realize the similar curtailing ratio output of each distributed photovoltaic, and the constraint is mathematically expressed as follows:

[0093] In the formula, is the output regulation value of the distributed power source connected to the network node i after optimization, ε is the maximum output adjustment ratio of the distributed power source in the distribution network, r QCR is the preset output adjustment ratio of the distributed power source, that is, the ratio of the actual active power adjustment value of the distributed power source to the maximum active power value that can be generated; R represents the set of network nodes. By considering the synergy of photovoltaic, the optimization is carried out from the whole photovoltaic level, and the approximate proportional uniform down-regulation of all photovoltaics is realized.

[0094] In some other examples, the constructed optimal power flow optimization model needs to be optimized and solved, and the implementation process of determining the output regulation value of the next day in step S120 based on the historical power flow data and the optimal power flow optimization model is as follows:

[0095] Based on the sequence quadratic programming algorithm, the optimal power flow optimization model is converted into a corresponding quadratic programming subproblem;

[0096] In each iteration process of the sequence quadratic programming algorithm, based on the historical output data and the historical load data, the corresponding displacement of the current iteration is searched by using the trust region to determine the new iteration point, until the iteration termination condition is reached, and the global optimal iteration point is obtained;

[0097] Based on the global optimal iteration point, the output regulation value of the next day is determined.

[0098] In the example embodiment, the Trust-Region Sequential Quadratic Program (TRSQP) is used to solve the optimal power flow optimization model of the high proportion distributed photovoltaic distribution network, and the output regulation value of the next day of each distributed power source (such as distributed photovoltaic) is obtained. The solving process is as follows:

[0099] (1) Subproblem conversion based on SQP (Sequential Quadratic Program)

[0100] Firstly, the SQP method is used to obtain the quadratic approximation model of the original problem as the subproblem of the trust region method. For the original optimization problem expressed by the objective function (6) and the constraint conditions (7)-(11), it is simplified as follows:

[0101] In the formula, X is the control variable of the current optimization problem, that is, the active power output of the distributed photovoltaic; Z is the dependent variable of the optimization problem, that is, the voltage state quantity of each node. h(Z, X) = 0 is the equality constraint of the optimization problem, that is, the power flow equation constraint, which is also the function expression of the dependent variable Z with respect to the control variable X; g(Z, X) ≤ 0 is the inequality constraint of the optimization problem, that is, the node voltage and line power constraint; X is the control variable constraint, X, X is the control variable constraint, X,

[0102] For the optimization problem in the form of formula (12), the Taylor expansion is used to expand the objective function at the kth iteration point X k into a quadratic function, and the constraint condition at the iteration point X k is expanded into a linear function, and the sub-quadratic programming problem model is obtained as follows:

[0103] Then the subproblem of the kth iteration is as follows:

[0104] In the formula, the superscript T represents the matrix transpose, S is the optimal step length of the current iteration to be solved in the subproblem; H is the Hessian matrix of the objective function, and its approximate symmetric matrix can be used to replace it during iteration, so that the Hessian matrix does not need to be solved again when solving the subproblem of the next iteration point.

[0105] (2) Trust region search algorithm

[0106] The trust region method is an important technique for global convergence in optimization algorithm, which determines the new iteration point by solving the displacement of each iteration in the optimization algorithm. The basic idea is as follows: first, a "trust region radius △" is given as the upper bound of the displacement length, and a closed loop region is determined with the current iteration point as the center and the upper bound as the radius, which is called "trust region". Then, the "candidate displacement" is determined by solving the optimal point of the "objective function sub-problem" in this region. If the candidate displacement can make the objective function value have sufficient decrease, then accept the candidate displacement as the new displacement, and keep or expand the trust region radius, continue the new iteration. Otherwise, it is indicated that the approximation degree of the quadratic model and the objective function is not ideal, and the trust region radius needs to be reduced, and then a new candidate displacement is obtained by solving the sub-problem in the new trust region. Repeat the above process until the iteration termination condition is met.

[0107] Therefore, it is necessary to select a suitable evaluation function as the guide of the global convergence of TRSQP algorithm to increase or decrease the trust region radius of the next search. The evaluation function P(X) is defined as shown in formula (16): P(X) = f(X) + σ × ||h(X)| ∞ +σ×||g + (X)| ∞ (16)

[0108] In the formula, σ is a constraint penalty coefficient; g + (X) represents that there is

[0109] Φ(S) is the second-order Taylor expansion of the original objective function formula (12):

[0110] The parameter r k is defined to represent the ratio of the improvement value of the evaluation function P(X) and the improvement value of the second-order model function Φ(S) under the current step length S:

[0111] In the formula, P(X k ) represents the evaluation function value at the iteration point X k , Φ(0) represents the initial second-order model function value, and Φ(S) represents the second-order model function value under the step length S.

[0112] If rk is closer to 1, it indicates that the second-order model function can well fit the actual objective function under the current step length, and in this case, the current step length can be accepted and the trust region radius and the penalty parameter can be increased; otherwise, the step length is considered to be rejected and the trust region radius is reduced. The TRSQP guide method can be established as follows:

[0113] In the formula, σ0is the preset maximum penalty parameter and the radius of the trust region k σkis the penalty parameter of the kth iteration k is the radius of the trust region of the kth iteration

[0114] When the trust region Δk k at the iteration point Xk k , the global optimal point is considered to be reached when the infinite norm of the step vector S obtained by solving the sub-quadratic programming problem is less than the set precision value ζ and the penalty parameter reaches the specified maximum value.

[0115] For example, FIG. 3 is a schematic diagram of a global trust region sequence quadratic programming solving process provided by an embodiment of the present disclosure, as shown in FIG. 3, solving a high-proportion distributed photovoltaic power distribution network dynamic optimization problem by using TRSQP includes the following steps:

[0116] Step S31, input parameters X0, Δ0, ζ1, ζ2, σ0,

[0117] Step S31 specifically includes: 1) setting the initial value X0of each control variable; setting the maximum radius of the trust region of the variable X in the optimization problem setting the initial radius of the trust region setting the control variable solving precision ζ1, the constraint precision ζ2, the initial penalty parameter σ0and the maximum penalty parameter setting the iteration number k = 0.

[0118] Step S32, performing power flow calculation according to the current control variable value Xk k ;

[0119] Step S32 specifically includes: 2) performing power flow calculation according to the current control variable value, to obtain the values of the state variables such as voltage nodes and branch power in the power distribution network.

[0120] Step S33, forming a current point trust region quadratic programming sub-problem, obtaining a step S, and updating Xk+1 k+1 = Xk k + S, σk+1 k+1 = 2σk k ;

[0121] Step S33 specifically includes: 3) forming the sub-problem formula (12) and solving it to obtain the optimal step S, and updating Xk+1 k+1 = Xk k + S.

[0122] Step S34, performing power flow calculation, and judging whether the following condition is met

[0123] Step S35, judge whether the penalty parameter and step length satisfy the optimal condition;

[0124] Here, if yes, go to step S37, if no, go to step S36;

[0125] Step S36, calculate rk;

[0126] Step S37, end the output of the optimal result;

[0127] Steps S34 to S37 are specifically: according to the current control variable value X k+1 The state quantity information such as network voltage and branch power flow is obtained by carrying out the power flow calculation, if and satisfies ||S||<min(ζ1,△ k ), the calculation is terminated, and the calculation result is output; otherwise, calculate rk, and go to step 5).

[0128] Step S38, judge whether r k <0.25;

[0129] Here, if yes, go to step S39, if no, go to step S310;

[0130] Step S39, continue to iterate X k+1 =X k ,△ k+1 =||S|| ∞ / 2;

[0131] Step S310, update the trust region;

[0132] Steps S38 to S310 are specifically: 5) if r k <0.25, reject to update the step length, set X k+1 =X k ,△ k+1 =||S|| ∞ / 2, and go to step 2); if r k ≥0.25, accept the step length, and adjust the trust region and the penalty parameter accordingly, and go to step 3).

[0133] The present embodiment can make the solving process converge quickly and improve the model solving speed by solving the optimal power flow optimization model based on the sequence quadratic programming algorithm guided by the trust region (TRSQP).

[0134] The embodiment of the present disclosure aims at the uniform and rapid optimization control problem of the distribution network under large-scale distributed photovoltaic access, based on the high-precision prediction result of the distributed photovoltaic, taking the maximum output of the distributed photovoltaic, the minimum voltage deviation of the network and the minimum line loss of the network as the objective function, introducing the photovoltaic output standby constraint to guide the uniform reduction of the photovoltaic cluster, further considering the equality and inequality constraints related to power flow, the output adjustment constraints of the photovoltaic cluster and other constraint conditions, and establishing an optimal power flow-based optimization control model of the high-proportion distributed photovoltaic distribution network; the multi-objective optimization problem is converted into a single-objective optimization problem through linear weighting, and a reliable domain sequence quadratic programming algorithm is used for solving.

[0135] In view of the problems of prediction difficulty, large operation risk of the distribution network, complex operation adjustment and multi-resource coordination and the like caused by the distributed photovoltaic, the embodiment of the present disclosure carries out short-term power prediction of the distributed photovoltaic through the station-line-transformer multi-level distributed photovoltaic, carries out distribution network operation optimization control in combination with the distribution network operation optimization control of the distributed photovoltaic power prediction data in important special scenes such as grid maintenance, holidays, load peak and rainy and dusty weather, provides an operation risk prevention and control scheme for the control personnel, realizes efficient local consumption of the distributed photovoltaic under the premise of guaranteeing the safety and stability of the distribution network and the power supply reliability, breaks through the limitation of the insufficient existing distribution network control verification means, and provides auxiliary decision support for the safe, reliable and economic operation of the distribution network with large-scale distributed photovoltaic access. The method of the embodiment of the present disclosure can support not less than 10,000 distributed power sources to access, the average response time of the optimization control is less than 15s, and the simulation accuracy is not less than 95%.

[0136] The embodiment of the present disclosure has the following advantages:

[0137] In another embodiment, considering that there is a large deviation between the current photovoltaic prediction value and the actual output, but the equal standby rate reduction control method of the high-proportion photovoltaic active power according to the real-time measurement value of the photovoltaic involves many links, has large calculation amount and long time, and one high-proportion photovoltaic reduction control is performed every hour in 12 hours of the day (assuming that the photovoltaic does not generate power and does not perform control from 18:00 to 06:00 at night), 12 measurements are required and transmitted to the centralized control platform, 12 control strategies are calculated, and 12 control instructions are issued, which is not conducive to realizing the rapid control of the high-proportion photovoltaic by the new distribution system. Based on this, the embodiment of the present disclosure provides a kind of optimization control method of high proportion distributed power distribution network, applied to distributed power supply, Fig. 4 is a kind of optimization control method of high proportion distributed power distribution network provided by the embodiment of the present disclosure The flowchart of the second, as shown in Fig. 4, the method comprises the following steps S410 and S420:

[0138] In step S410, the centralized control instruction issued by the centralized control platform is received, the centralized control instruction including an output adjustment value and an output prediction value of the current distributed power in the next day, the output adjustment value being obtained by the centralized control platform based on historical power flow data of the distribution network and an optimal power flow optimization model in the day-ahead, and the output prediction value being obtained by the centralized control platform based on the correlation between the outputs of the distributed power at each level and the historical power generation data of the distributed power at each level in the day-ahead.

[0139] In step S420, for each control period in the day, a real-time output measurement value is obtained in the control period, and the output value of the control period is controlled based on the real-time output measurement value, the output prediction value and the output adjustment value.

[0140] The present implementation is based on the distributed power side, and the technical implementation details are referred to the embodiments of the centralized control platform side, which will not be described here.

[0141] In some example embodiments, the implementation process of controlling the output value of the control period based on the real-time output measurement value, the output prediction value and the output adjustment value in step S420 is as follows:

[0142] The real-time output measurement value is compared with the output prediction value.

[0143] If the real-time output measurement value is not less than the output prediction value, or the real-time output measurement value is less than the output prediction value and the real-time output measurement value is greater than or equal to the predicted adjusted output value, the output adjustment value is used to control and adjust the output value of the control period.

[0144] If the real-time output measurement value is less than the predicted adjusted output value, the real-time output measurement value is taken as the output value of the control period.

[0145] The predicted adjusted output value is obtained by adjusting the output prediction value using the output adjustment value.

[0146] For example, taking the distributed power as photovoltaic, the output value of photovoltaic 06:00-18:00 of the next day is predicted once in the day-ahead, the down-regulation amount (output adjustment value) of the photovoltaic is calculated based on the predicted value, the active power down-regulation value of photovoltaic 06:00-18:00 of the next day is obtained, and the photovoltaic is issued with the down-regulation value.

[0147] The output value of each distributed photovoltaic is measured in real time every hour in the day, involving three cases:

[0148] ① The real-time output measurement value of the photovoltaic is greater than or equal to the day-ahead output prediction value , and the output adjustment value calculated in the day-ahead is followed The control is performed, i.e. the predicted regulated output value The operation can meet the requirements of economy and safety.

[0149] The predicted regulated output value ≤ the photovoltaic measured value < the day-ahead output prediction value The control is also performed according to the day-ahead calculated output regulation value, i.e. the output control value is

[0150] The real-time output measured value of the photovoltaic power < the predicted regulated output value i.e. the regulated photovoltaic output value, then the output control value is the actual photovoltaic output value The output is not controlled.

[0151] The formula of the above output regulation process is as follows:

[0152] In the formula, represents the real-time output measured value of the distributed power supply at the network node i, represents the day-ahead output prediction value of the distributed power supply at the network node i, represents the day-ahead output regulation value of the distributed power supply at the network node i.

[0153] In one aspect, the embodiment of the present disclosure considers the correlation characteristics of the distributed photovoltaic power at different levels of "district feeder transformer", and uses support vector machines, BP neural networks optimized by genetic algorithms, radial basis neural networks, etc. to obtain the preliminary prediction value of the multi-level photovoltaic power generation. Further, the variance-covariance weight dynamic allocation method is used to combine the single prediction algorithm to predict the initial value, and a high-precision distributed photovoltaic power prediction model based on variable weight combination of "district feeder transformer" is constructed. On the other hand, based on the high-precision prediction result of the distributed photovoltaic power, the maximum output of the distributed photovoltaic power, the minimum deviation of the network voltage, and the minimum network line loss are taken as the objective function, the photovoltaic output standby constraint is introduced to guide the uniform down-regulation of the photovoltaic cluster, and further considering the equality and inequality constraints related to power flow, the output regulation constraints of the photovoltaic cluster and other constraint conditions, a high-proportion distributed photovoltaic distribution network optimization control model based on optimal power flow is established. Through linear weighting, the multi-objective optimization problem is converted into a single-objective optimization problem, and a reliable domain sequence quadratic programming algorithm is used for solving, so as to realize the uniform and rapid optimization control of the distribution network under the high proportion and large-scale distributed photovoltaic access.

[0154] Those skilled in the art can understand that, in the above method of the specific embodiment, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process, and the specific execution order of each step should be determined according to its function and possible internal logic.

[0155] The embodiments of the present disclosure provide a high-proportion distributed power distribution network optimization control method.

[0156] Based on the same inventive concept, the embodiments of the present disclosure also provide a high-proportion distributed power distribution network optimization control system. Since the principle of the system in the embodiments of the present disclosure for solving the problem is similar to the high-proportion distributed power distribution network optimization control method described above, the implementation of the system can be referred to the implementation of the method. The high-proportion distributed power distribution network optimization control system comprises a centralized control platform and a plurality of distributed power sources at each level, and the centralized control platform and each distributed power source are in communication connection;

[0157] The centralized control platform is configured to perform multi-level prediction on the historical power generation data of each distributed power source based on the correlation of the power generation of the distributed power sources at each level, to obtain the output prediction value of each distributed power source in the next day; determine the output adjustment value of each distributed power source in the next day based on the historical power flow data and the optimal power flow optimization model; and send the output adjustment value and the output prediction value of each distributed power source as a centralized control instruction to the corresponding distributed power source;

[0158] The distributed power source is configured to, for each control period within a day, obtain a real-time output measurement value at the control period, and control the output value of the control period based on the real-time output measurement value and the output prediction value and the output adjustment value received in advance.

[0159] In a possible implementation, the levels of the distributed power source include a transformer, a feeder and a transformer.

[0160] In a possible implementation, the centralized control platform comprises a multi-level prediction part, and the multi-level prediction part is configured to:

[0161] obtain historical power generation data of the distributed power sources at each level in a preset historical period, the historical power generation data comprising historical output data of the distributed power sources at each level and corresponding historical environmental data;

[0162] input the historical output data of the distributed power sources at each level and the corresponding historical environmental data as input data into a plurality of different neural network models in parallel, and obtain output data of each neural network model correspondingly;

[0163] fuse the output data of each neural network model to obtain the output prediction value of each distributed power source in the next day.

[0164] In a possible implementation, the multi-level prediction part is further configured to:

[0165] For each distributed power source, a prediction weight corresponding to each neural network model is determined based on a prediction error of each neural network model; and an output prediction value of the distributed power source in the next day is determined based on output data of each neural network model and the corresponding prediction weight.

[0166] In a possible implementation, the plurality of different neural network models include a support vector machine, a genetic algorithm-optimized back propagation neural network, and a radial basis neural network.

[0167] In a possible implementation, the centralized control platform further includes an optimization control part, and the optimization control part is configured to:

[0168] maximize output of the distributed power source, minimize network voltage deviation of the power distribution network, and minimize network line loss of the power distribution network, determine a target optimization function;

[0169] considering operation safety of the power distribution network, output range of each distributed power source, and output adjustment proportion of each distributed power source, determine a constraint condition of the target optimization function;

[0170] based on the target optimization function and the constraint condition, build an optimal power flow optimization model.

[0171] In a possible implementation, an expression of the target optimization function is:

[0172] in which, respectively represent an initial value of output of the distributed power source, an initial value of network voltage deviation of the power distribution network, and an initial value of network line loss of the power distribution network, α1, α2, and α3 respectively represent corresponding weights of the output of the distributed power source, the network voltage deviation of the power distribution network, and the network line loss of the power distribution network, and the values of α1, α2, and α3 are positive; F1(x), F2(x), and F3(x) respectively represent the output of the distributed power source, the network voltage deviation of the power distribution network, and the network line loss of the power distribution network, N PV is the number of distributed power sources in the power distribution network, is an active output value of the nth distributed power source, I is the number of network nodes of the power distribution network, U i represents a voltage amplitude of the network node i, represents a reference voltage amplitude of the network node i, represents an upper limit of the voltage amplitude of the network node i, represents a lower limit of the voltage amplitude of the network node i, B is a network branch set, and subscripts i and j represent a branch formed with the network node i and the network node j as first and last end nodes, g ijis the branch conductance between network nodes i, j, θ ij is the phase angle difference of the voltages of network nodes i, j.

[0173] In a possible implementation, the optimization control part is further configured to:

[0174] satisfy the operation safety of the distribution network through the power flow constraint, the voltage amplitude constraint of the distributed power supply, and the branch power constraint;

[0175] take the output prediction value of each distributed power supply as the output upper limit of the corresponding distributed power supply to determine the output range constraint of each distributed power supply;

[0176] take the equal output adjustment proportion of each distributed power supply as the output adjustment proportion constraint.

[0177] In a possible implementation, the formula of the power flow constraint is as follows:

[0178] wherein, are the active output and the reactive output of the distributed power supply connected to network node i, respectively, are the active power and the reactive power of the load at network node i, respectively, U i is the voltage amplitude of network node i, U j is the voltage amplitude of network node j, I is the number of network nodes of the distribution network, G ij ,B ij are elements in the admittance matrix of the branch between network nodes i, j, respectively, θ ij is the phase angle difference of the voltages of network nodes i, j.

[0179] The formula of the voltage amplitude constraint of each distributed power supply is as follows: U i,min ≤U i ≤U i,max

[0180] wherein, U i,min is the lower limit of the voltage amplitude of network node i, U i,max is the upper limit of the voltage amplitude of network node i;

[0181] The formula of the branch power constraint is as follows:

[0182] wherein, is the apparent power of branch l, are the active power and the reactive power flowing through branch l, respectively, is the upper limit of the apparent power of branch l;

[0183] The formula of the output range constraint of each distributed power supply is as follows:

[0184] wherein, Pmin and Pmax are the minimum and maximum active power output of the distributed generator connected at the network node i, respectively;

[0185] The formula of the output adjustment ratio constraint is as follows:

[0186] wherein, Popt is the output adjustment value of the distributed generator connected at the network node i after optimization, ε is the maximum output adjustment ratio of the distributed generator in the distribution network, rQCR is the preset output adjustment ratio of the distributed generator, i.e., the ratio of the actual active power adjustment value of the distributed generator to the maximum active power value that can be generated by the distributed generator; and R represents the set of network nodes.

[0187] In a possible implementation, the historical load data includes historical output data and historical load data, and the optimization control part is further configured to:

[0188] Based on the sequential quadratic programming algorithm, the optimal power flow optimization model is converted into a corresponding quadratic programming sub-problem;

[0189] In each iteration process of the sequential quadratic programming algorithm, based on the historical output data and the historical load data, the corresponding displacement of the current iteration is searched and guided by the trust region to determine a new iteration point until the iteration termination condition is reached, and a global optimal iteration point is obtained;

[0190] Based on the global optimal iteration point, the output adjustment value of the next day is determined.

[0191] In a possible implementation, each distributed generator is further configured to:

[0192] Compare the real-time output measurement value and the output prediction value;

[0193] If the real-time output measurement value is not less than the output prediction value, or the real-time output measurement value is less than the output prediction value and the real-time output measurement value is greater than or equal to the predicted adjusted output value, the output value of the control period is controlled and adjusted by using the output adjustment value;

[0194] If the real-time output measurement value is less than the predicted adjusted output value, the real-time output measurement value is taken as the output value of the control period;

[0195] wherein, the predicted adjusted output value is obtained by adjusting the output prediction value by using the output adjustment value.

[0196] In the present implementation, the optimization control process of the system is divided into three stages, which are multi-level photovoltaic prediction and day-ahead centralized control of the centralized control platform side and day-ahead local control of the distributed power side. Fig. 5 is a schematic diagram of an optimization control process of an optimization control system of a high-proportion distributed power distribution network provided by the embodiment of the present disclosure. As shown in Fig. 5, the centralized control platform includes a multi-level prediction part 54 and an optimization control part 57, and the distributed power includes distributed photovoltaic 59. The barometric pressure, temperature, total radiation 51, photovoltaic power generation power 52, and photovoltaic access voltage level 53 are input into a hybrid prediction model 541 (composed of multiple single models) of the multi-level prediction part 54, the outputs of the multiple single models are weighted and combined to obtain a photovoltaic prediction value, the photovoltaic prediction value, a load prediction result 55 (active and reactive power prediction of the load can be based on a single model), and network topology information 56 of the distribution network are input into the optimization control part 57, the adjustment amount optimization control is performed by an optimization model 571 in the optimization control part 57 in combination with a constraint condition, a photovoltaic adjustment amount 58 is obtained, the photovoltaic adjustment amount is issued to each distributed photovoltaic 59, based on the comparison result of the photovoltaic output after adjustment and the actual photovoltaic output, the photovoltaic adjustment amount is updated to obtain an updated photovoltaic adjustment amount 510, local control is realized, and based on the updated photovoltaic adjustment amount 510, a control signal 513 is output by an automatic control system 511 or a dispatcher 512, each photovoltaic output is controlled, output control is realized to obtain measurement data 514. Then the measurement data 514 is input into the high-proportion photovoltaic distribution network 515 to obtain the actual active power output 516 of the photovoltaic, and the actual active power output 516 of the photovoltaic is input into the optimization control part 57.

[0197] In the embodiment of the present disclosure, the multi-level photovoltaic prediction based on model training combination, the day-ahead centralized optimization control based on OPF, and the day-to-day local rapid control based on real-time measurement data are sequentially implemented to realize effective prediction of photovoltaic power generation power, and approximate proportional uniform adjustment of all photovoltaic active power outputs and day-to-day rapid feedback correction control are realized from the photovoltaic whole level.

[0198] In the embodiment of the present disclosure and other embodiments, the "part" can be a part of circuit, a part of processor, a part of program or software, etc., and of course can also be a unit, and can also be a module or non-modular.

[0199] The embodiment of the present disclosure has the following advantages:

[0200] FIG. 6 is a structural schematic diagram of an electronic device provided by the embodiment of the present disclosure. As shown in FIG. 6, the embodiment of the present disclosure further provides an electronic device, which can be a computer device, a single-chip microcomputer device, a smart mobile device, etc. The electronic device 60 in the embodiment can include a processor 61, a memory 62, a transceiver component 63, etc. The memory 62, the processor 61 and the transceiver component 63 are connected through a bus 64; the memory 62 can be used to store an execution program, and the exemplary execution program can include instructions; the processor 61 is used to execute the instructions stored in the memory. The memory 62 can also be used to store data, which can be called and / or modified when the instructions are executed.

[0201] The processor 61 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, and are suitable for implementing one or more instructions, loading and executing one or more instructions in the storage medium to realize the corresponding method flow or corresponding function, so as to realize the steps of the high-proportion distributed power distribution network optimization control method in the above embodiment.

[0202] The embodiment of the present disclosure has the following advantages:

[0203] Based on the same inventive concept, the embodiment of the present disclosure further provides a readable storage medium, which is an electronic device readable storage medium (Memory). The electronic device readable storage medium is a memory device in the electronic device, and is used to store programs and data. It can be understood that the storage medium herein can include a built-in storage medium in the electronic device, and of course can also include an expansion storage medium supported by the electronic device. The storage medium provides a storage space, and the storage space stores an operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and the instructions can be one or more execution programs (including program codes). It should be noted that the storage medium herein can be a high-speed RAM memory, or a non-volatile memory such as at least one disk memory. The processor loads and executes one or more instructions stored in the storage medium, and the steps of the high-proportion distributed power distribution network optimization control method in the above embodiment can be realized.

[0204] Those skilled in the art will appreciate that embodiments of the disclosure can be supplied as a method, a system, or a computer program product. Accordingly, embodiments of the disclosure can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the disclosure can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer-readable program code.

[0205] Embodiments of the disclosure are described herein with reference to the drawings, in which are shown flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the disclosure. It will be understood that each flow and / or block of the flowcharts and / or block diagrams, and combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing device, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowcharts and / or block diagrams block or blocks.

[0206] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowcharts and / or block diagrams block or blocks.

[0207] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowcharts and / or block diagrams block or blocks.

[0208] Finally, it should be noted that the above-described embodiments are merely intended for describing and illustrating, but not limiting, the technical solutions of the present disclosure, and that even though the present disclosure has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present disclosure, various modifications, replacements, or equivalent replacements of the specific embodiments of the present disclosure can be made by those skilled in the art, and these modifications, replacements, or equivalent replacements are all within the scope of protection of the claims of the present disclosure. Industrial applicability

[0209] The embodiment of the present disclosure provides a kind of high proportion distributed power distribution network optimization control method and system, electronic equipment, computer readable storage medium and computer program product, wherein the high proportion distributed power distribution network optimization control method is applied to centralized control platform, including: based on the correlation of the power generation of each level of distributed power, the historical power generation data of each distributed power is multi-level predicted, obtains the output prediction value of each distributed power in next day;For each distributed power, based on historical power flow data and optimal power flow optimization model, determine the output adjustment value in next day;The output adjustment value and output prediction value of each distributed power are sent to the corresponding distributed power as centralized control instruction, so that the distributed power is controlled based on the real-time output measurement value, output adjustment value and output prediction value obtained in each control period in next day, to control output value.Through multi-level prediction in day, the output prediction value of each distributed power in next day is obtained, the correlation between the output of each level of distributed power is considered, and the prediction accuracy is improved.Meanwhile, the output adjustment value is obtained by target optimization control model, so that the distributed power can obtain the output prediction value and output adjustment value in advance, realize the completion of large amount of calculation and control instruction interaction in advance, to reduce the calculation and network bandwidth pressure of real-time control process;In real-time regulation, each distributed power obtains the output measurement value by real-time, and then based on the output prediction value and output adjustment value received in advance, real-time control adjustment of current output is carried out, to realize the local rapid control of distributed power, and improve the adjustment effect.The scheme combines the advance centralized prediction optimization of distributed power and local real-time control, balances processing rate and control effect, and has strong implementability.

Claims

1. An optimal control method of a high-proportion distributed power supply power distribution network, applied to a centralized control platform, the method comprising: based on the correlation of the power generation of the distributed power supply at each level, performing multi-level prediction on the historical power generation data of each distributed power supply to obtain the output prediction value of each distributed power supply in the next day; for each distributed power supply, determining the output adjustment value in the next day based on historical power flow data and an optimal power flow optimization model; sending the output adjustment value and the output prediction value of each distributed power supply to the corresponding distributed power supply as a centralized control instruction, so that each distributed power supply controls the output value of each control period in the next day based on the obtained real-time output measurement value, the output adjustment value and the output prediction value.

2. The method of claim 1, wherein, The levels of the distributed power supply include a transformer, a feeder and a transformer.

3. The method of claim 2, wherein, Based on the correlation of the power generation of the distributed power supply at each level, the historical power generation data of multiple distributed power supplies at each level is predicted to obtain the output prediction value of each distributed power supply in the next day, comprising: obtaining the historical power generation data of the distributed power supply at each level in the preset historical period, the historical power generation data including the historical output data of the distributed power supply at each level and the corresponding historical environmental data; inputting the historical output data of the distributed power supply at each level and the corresponding historical environmental data as input data into multiple different neural network models in parallel to obtain the output data of each neural network model; fuse the output data of each neural network model to obtain the output prediction value of each distributed power supply in the next day.

4. The method of claim 3, wherein, The output data of each neural network model is fused to obtain the output prediction value of each distributed power supply in the next day, comprising: for each distributed power supply, based on the prediction error of each neural network model, determine the prediction weight of each neural network model corresponding to the prediction weight; based on the output data of each neural network model and the corresponding prediction weight, determine the output prediction value of the distributed power supply in the next day.

5. The method of claim 4, wherein, The multiple different neural network models include support vector machines, genetic algorithm optimized back propagation neural networks and radial basis neural networks.

6. The method of claim 1, wherein, Before the optimal power flow optimization model is determined based on the historical power flow data and the optimal power flow optimization model, the method further comprises: determining a target optimization function with the maximum output of the distributed power supply, the minimum network voltage deviation of the power distribution network and the minimum network line loss of the power distribution network as the target; considering the operation safety of the power distribution network, the output range of each distributed power supply and the output adjustment proportion of each distributed power supply, determining the constraint condition of the target optimization function; based on the target optimization function and the constraint condition, constructing the optimal power flow optimization model.

7. The method of claim 6, wherein, An expression of the target optimization function is: In the formulae, respectively represent the initial value of the output of the distributed power supply, the initial value of the network voltage deviation of the distribution network, and the initial value of the network line loss of the distribution network, and a1, a2, a3 respectively represent the corresponding weight of the output of the distributed power supply, the corresponding weight of the network voltage deviation of the distribution network, and the corresponding weight of the network line loss of the distribution network, and the values of which are all positive; F1(x), F2(x), and F3(x) respectively represent the output of the distributed power supply, the network voltage deviation of the distribution network, and the network line loss of the distribution network, and N PV is the number of the distributed power supply in the distribution network, Pn is the active power value of the nth distributed power source, I is the number of network nodes of the distribution network, U i denotes the voltage amplitude of the network node i, denotes the reference voltage amplitude of the network node i, Vmax(i) denotes the upper voltage amplitude limit of the network node i, Vmin,i represents the lower limit of the voltage magnitude of network node i, B is the set of network branches, the subscripts i, j represent the branch formed by taking network node i and network node j as the first and last end nodes, g ij is the branch conductance between network nodes i, j, θ ij is the phase angle difference of the voltages of network nodes i, j.

8. The method of claim 6, wherein, The constraint condition of the target optimization function is determined by considering the operation safety of the power distribution network, the output range of each distributed power supply and the output adjustment proportion of each distributed power supply, comprising: The operation safety of the power distribution network is satisfied by a tidal power constraint, a voltage amplitude constraint of the distributed power supply and a branch power constraint; The output prediction value of each distributed power supply is taken as an output upper limit of the corresponding distributed power supply to determine an output range constraint of each distributed power supply; An output adjustment proportion of each distributed power supply is taken as an output adjustment proportion constraint.

9. The method of claim 8, wherein, The tidal current power constraint is formulated as follows: wherein PQI, PQJ are the active and reactive power output of the distributed generator at the network node i, j respectively, respectively the active and reactive power of the load at network node i, U i is the voltage magnitude of network node i, U j is the voltage magnitude of network node j i, I is the number of network nodes of the power distribution network, G ij , B ij are respectively the elements of the admittance matrix of the branch between network nodes i, j, θ ij is the phase angle difference of the voltages of network nodes i, j; The voltage amplitude constraint of each distributed power supply is as follows: U i,min ≤U i ≤U i,max where U i,min is the lower voltage amplitude limit for network node i, U i,max is the upper voltage amplitude limit for network node i. The formula of the branch power constraint is as follows: wherein, for the branch l the apparent power, respectively the active and reactive power flowing through the branch l, is an apparent power upper limit of branch l; The formula of the output range constraint of each distributed power source is as follows: wherein is a minimum value and a maximum value of active power output of the distributed power supply at network node i, respectively; The output adjustment ratio constraint formula is as follows: wherein to optimize the output adjustment value of the distributed power source at the subsequent network node i, ε is a maximum output adjustment proportion of the distributed power supply in the power distribution network, rQCR is a preset output adjustment proportion of the distributed power supply, that is, a ratio of an actual active power adjustment value of the distributed power supply to a maximum active power value that can be output by the distributed power supply, and R represents a network node set.

10. The method of any one of claims 1-9, wherein, The historical tidal data includes historical output data and historical load data, and the output adjustment value of the next day is determined based on the historical tidal data and the optimal tidal optimization model, including: The optimal tidal optimization model is converted into a corresponding quadratic programming sub-problem based on a sequential quadratic programming algorithm; In each iteration process of the sequential quadratic programming algorithm, a corresponding displacement of the current iteration is searched based on the historical output data and the historical load data by using a trust region to determine a new iteration point until an iteration termination condition is reached to obtain a global optimal iteration point; The output adjustment value of the next day is determined based on the global optimal iteration point.

11. An optimization control method of a high-proportion distributed power supply power distribution network, applied to a distributed power supply, the method comprising: receiving a centralized control instruction issued by a centralized control platform in a day-ahead manner, the centralized control instruction including an output adjustment value and an output prediction value of the current distributed power supply in the next day, the output adjustment value being obtained by the centralized control platform in a day-ahead manner based on historical tidal data of the power distribution network and an optimal tidal optimization model, and the output prediction value being obtained by the centralized control platform in a day-ahead manner based on an association relationship of outputs of distributed power supplies at each level and historical power generation data of the distributed power supplies at each level; for each control period in a day, obtaining a real-time output measurement value at the control period and controlling an output value of the control period based on the real-time output measurement value, the output prediction value and the output adjustment value.

12. The method of claim 11, wherein, The controlling of the output value of the control period based on the real-time output measurement value, the output prediction value and the output adjustment value includes: comparing the real-time output measurement value with the output prediction value; if the real-time output measurement value is not less than the output prediction value, or the real-time output measurement value is less than the output prediction value and the real-time output measurement value is greater than or equal to a predicted adjusted output value, then the output value of the control period is controlled and adjusted by using the output adjustment value; if the real-time output measurement value is less than the predicted adjusted output value, then the real-time output measurement value is taken as the output value of the control period. The predicted adjusted output value is obtained by adjusting the output prediction value by the output adjustment value.

13. An optimal control system for a high penetration distributed generation power distribution network, the system comprising: The centralized control platform is in communication connection with each of the distributed power sources. The centralized control platform is configured to perform multi-level prediction on historical power generation data of each distributed power source based on the correlation of power generation of the distributed power sources at each level, to obtain an output prediction value of each distributed power source for the next day, to determine an output adjustment value of each distributed power source for the next day based on historical power flow data and an optimal power flow optimization model, and to send the output adjustment value and the output prediction value of each distributed power source to the corresponding distributed power source as a centralized control instruction. The distributed power source is configured to, for each control period within a day, obtain a real-time output measurement value at the control period, and control the output value of the control period based on the real-time output measurement value and the output prediction value and the output adjustment value received in advance.

14. The system of claim 13, wherein, The levels of the distributed power source include a transformer, a feeder line, and a transformer.

15. The system of claim 14, wherein, The centralized control platform includes a multi-level prediction part configured to obtain historical power generation data of the distributed power sources at each level during a preset historical period, the historical power generation data including historical output data of the distributed power sources at each level and corresponding historical environmental data, to input the historical output data of the distributed power sources at each level and the corresponding historical environmental data as input data into a plurality of different neural network models in parallel, to obtain output data of each neural network model correspondingly, and to fuse the output data of each neural network model to obtain an output prediction value of each distributed power source for the next day.

16. The system of claim 15, wherein, The multi-level prediction part is further configured to, for each distributed power source, determine a prediction weight of each neural network model based on a prediction error of each neural network model, and to determine the output prediction value of the distributed power source for the next day based on the output data of each neural network model and the corresponding prediction weight.

17. The system of claim 16, wherein, The plurality of different neural network models include a support vector machine, a back propagation neural network optimized by a genetic algorithm, and a radial basis neural network.

18. The system of claim 13, wherein, The centralized control platform further includes an optimization control part configured to determine a target optimization function with the maximum output of the distributed power source, the minimum network voltage deviation of the power distribution network, and the minimum network line loss of the power distribution network as targets. The optimization control part is further configured to determine a constraint condition of the target optimization function considering the operation safety of the power distribution network, the output range of each distributed power source, and the output adjustment proportion of each distributed power source. The optimization control part is further configured to construct the optimal power flow optimization model based on the target optimization function and the constraint condition.

19. The system of claim 18, wherein, An expression of the target optimization function is: In the formulae, respectively represent the initial value of the output of the distributed power supply, the initial value of the network voltage deviation of the distribution network, and the initial value of the network line loss of the distribution network, and a1, a2, a3 respectively represent the corresponding weight of the output of the distributed power supply, the corresponding weight of the network voltage deviation of the distribution network, and the corresponding weight of the network line loss of the distribution network, and the values of which are all positive; F1(x), F2(x), and F3(x) respectively represent the output of the distributed power supply, the network voltage deviation of the distribution network, and the network line loss of the distribution network, and N PV is the number of the distributed power supply in the distribution network, Pn is the active power value of the nth distributed power source, I is the number of network nodes of the distribution network, U i Vn represents the voltage amplitude of the network node i, Vrefi represents the reference voltage amplitude of the network node i, Vmax(i) denotes the upper voltage amplitude limit of the network node i, Vmin,i represents the lower limit of the voltage magnitude of network node i, B is the set of network branches, the subscripts i, j represent the branch formed by taking network node i and network node j as the first and last end nodes, g ij is the branch conductance between network nodes i, j, θ ij is the phase angle difference of the voltages of network nodes i, j.

20. The system of claim 18, wherein, The optimization control part is further configured to meet the operation safety of the power distribution network by a power flow constraint, a voltage amplitude constraint of the distributed power supply, and a branch power constraint; take the output prediction value of each distributed power supply as an output upper limit of the corresponding distributed power supply to determine an output range constraint of each distributed power supply; and take equal output adjustment proportions of each distributed power supply as an output adjustment proportion constraint.

21. The system of claim 20, wherein, The tidal current power constraint is formulated as follows: wherein, PQI, PQJ are the active and reactive power output of the distributed generator at the network node i, j respectively, respectively the active and reactive power of the load at network node i, U i is the voltage magnitude of network node i, U j is the voltage magnitude of network node j, I is the number of network nodes of the power distribution network, G ij ,B ij are respectively the elements of the admittance matrix of the branch between network nodes i, j, θ ij is the phase angle difference of the voltages of network nodes i, j; The voltage amplitude constraint of each distributed power supply is as follows: U i,min ≤U i ≤U i,max where U i,min is the lower voltage amplitude limit for network node i, U i,max is the upper voltage amplitude limit for network node i. The formula of the branch power constraint is as follows: wherein for the branch l the apparent power, respectively the active and reactive power flowing through the branch l, an apparent power upper limit of branch l; The formula of the output range constraint of each distributed power source is as follows: wherein, a minimum active output and a maximum active output of the connected distributed power supply at network node i, respectively; The output adjustment ratio constraint formula is as follows: wherein, to optimize the output adjustment value of the distributed power source at the subsequent network node i, ε is the maximum output adjustment ratio of the distributed power supply in the power distribution network, r QCR is the preset output adjustment ratio of the distributed power supply, i.e., the ratio of the actual active power adjustment value of the distributed power supply to the maximum active power value that can be generated by the distributed power supply; and R represents the network node set.

22. The system of any of claims 13-21, wherein, The historical power flow data includes historical output data and historical load data, and the optimization control part is further configured to convert the optimal power flow optimization model into a corresponding quadratic programming sub-problem based on a sequential quadratic programming algorithm. In each iteration process of the sequential quadratic programming algorithm, a corresponding displacement of the current iteration is searched for by using a trust region based on the historical output data and the historical load data to determine a new iteration point until an iteration termination condition is reached to obtain a global optimal iteration point. Based on the global optimal iteration point, an output adjustment value of the next day is determined.

23. The system of claim 13, wherein, The distributed power supply is further configured to compare the real-time output measurement value and the output prediction value; if the real-time output measurement value is not less than the output prediction value, or the real-time output measurement value is less than the output prediction value and the real-time output measurement value is greater than or equal to a predicted adjusted output value, then the output value of the control period is controlled and adjusted by using the output adjustment value; if the real-time output measurement value is less than the predicted adjusted output value, then the real-time output measurement value is taken as the output value of the control period; and the predicted adjusted output value is obtained by adjusting the output prediction value by using the output adjustment value. at least one processor and a memory; 24. An electronic device, the device comprising: The memory and the processor are connected through a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the optimization control method of the high-proportion distributed power supply power distribution network as claimed in any one of claims 1 to 12 is implemented. The computer program or instructions are stored on the computer readable medium, and when the computer program or instructions are executed, the optimization control method of the high-proportion distributed power supply power distribution network as claimed in any one of claims 1 to 12 is implemented.

25. A readable storage medium, wherein, 26. A computer program product, comprising a computer program or instructions, which, when executed on an electronic device, cause the electronic device to perform the method of any one of claims 1 to 12. ​

Images