Historical regulation and control strategy set driving-based distributed photovoltaic power distribution network voltage real-time control method

Abstract

The invention discloses a distributed photovoltaic power distribution network voltage real-time control method based on historical regulation and control strategy set driving. The method comprises the steps that a traditional voltage regulation and control strategy based on approximate sensitivity is formulated; constructing a historical regulation and control strategy set based on data driving and an improved BP neural network; the historical regulation and control strategy set carries out auxiliary decision making and correction on a traditional voltage regulation and control strategy; and correcting and outputting each photovoltaic output according to a historical strategy library aided decision-making regulation and control method, and completing voltage out-of-limit treatment of the power distribution network. According to the method, the photovoltaic inverter is driven to output the active power and the reactive power through the historical regulation and control strategy set, then the problem of photovoltaic access and voltage out-of-limit of the power distribution network system containing the impact load is solved, energy storage or reactive compensation equipment does not need to be additionally input, and good economical efficiency and reliability are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of voltage quality management, and in particular to a real-time voltage control method for a distributed photovoltaic distribution network driven by a historical control strategy set. Background Art

[0002] With the rapid development of energy, the proportion of new renewable energy consumption continues to rise, and the cumulative photovoltaic installed capacity has grown rapidly. Since distributed photovoltaics are connected to the distribution network in large quantities and have the characteristics of unstable output, intermittent and volatile output, they cannot be simply regarded as negative power loads. The direction of the system flow will also change, which may eventually lead to increased distribution network losses and voltage over-limit problems at some nodes. The uncertainty of photovoltaic output has brought new challenges to the reactive voltage balance of the power system. Therefore, voltage control is an urgent problem to be solved in the photovoltaic access distribution network system.

[0003] Traditional distribution networks are generally based on controlling reactive power regulation equipment such as OLTC and group switching capacitors in the distribution network to achieve the goal of stabilizing system node voltage, reducing network losses and thus achieving economic operation. The slow-action voltage regulation method based on the above-mentioned reactive power regulation resources is discrete and cannot achieve rapid regulation. In the face of sudden voltage changes in the distribution network, it is impossible to effectively regulate the distribution network in real time to ensure the voltage level. Summary of the invention

[0004] The purpose of the present invention is to address the above-mentioned deficiencies and provide a real-time voltage control method for a distributed photovoltaic distribution network driven by a historical control strategy set, which has good economy, strong reliability and high accuracy. Photovoltaic inverters with advantages such as fast response and continuous regulation are selected as regulating devices for realizing short-time scale voltage control, and rapid regulation of the system voltage is carried out. The historical control set is used for auxiliary correction to achieve better optimization effect.

[0005] In order to achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] A real-time voltage control method for a distributed photovoltaic distribution network driven by a historical control strategy set comprises the following steps:

[0007] S1, formulate a traditional voltage regulation strategy based on approximate sensitivity;

[0008] S2, construct a historical control strategy set based on data-driven and improved BP neural network;

[0009] S3, using the historical control strategy set to assist in decision-making and modify the traditional voltage control strategy;

[0010] S4, according to the historical strategy library auxiliary decision-making and control method, correct the output of each photovoltaic output and complete the voltage limit control of the distribution network.

[0011] Furthermore, the conventional voltage regulation strategy based on approximate sensitivity described in step S1 includes:

[0012] S1-1, using voltage sensitivity to measure the relationship between the changes in active and reactive power injected into each node of the distribution network and the corresponding voltage changes of the node, to solve the real-time control strategy of the distribution network voltage;

[0013] S1-2, issuing control instructions to the photovoltaic inverter according to the real-time control strategy of the distribution network voltage, driving the real-time control of the distribution network voltage.

[0014] Furthermore, the specific process of step S1-1 is as follows: first, an expression of the voltage change with respect to the active and reactive power changes of the node is obtained according to the voltage sensitivity calculation formula, and then the optimal solution, i.e., the initial control amount of active and reactive power, is iterated according to the optimization objective function, thereby obtaining a real-time control strategy for the distribution network voltage;

[0015] The voltage sensitivity calculation formula is as follows:

[0016] ΔV=[(B+Q)(GP) -1 (BQ)+(G+P) -1 ]ΔP-

[0017] [(GP)(B+Q) -1 (G+P)+(BQ) -1 ]ΔQ

[0018] In the formula, B represents the real part of the node admittance matrix; G represents the imaginary part of the node admittance matrix; P represents the diagonal element P i / V i 2 The n-order diagonal matrix of i / V i 2 The n-order diagonal element matrix of ; ΔP represents the active disturbance injected by the node; ΔQ represents the reactive disturbance injected by the node;

[0019] The optimization objective function is as follows:

[0020] F=min|VV ref |

[0021] Where V ref represents the voltage reference value, and V represents the regulated voltage value.

[0022] Furthermore, the specific process of step S1-2 is: according to the initial control amount calculated in step S1-1, driving the photovoltaic inverter to emit or absorb active power and reactive power to optimize the voltage of the distribution network;

[0023] The initial voltage control formula is as follows:

[0024]

[0025] In the formula, V0 represents the initial value of the system voltage; ΔV represents the change in the system voltage; S Pi represents the approximate sensitivity of the active power of the i-th distributed photovoltaic to the voltage; S Qi represents the approximate sensitivity of the reactive power of the i-th distributed photovoltaic to the voltage; ΔP PVi represents the active power regulation output of the i-th distributed photovoltaic unit; ΔQ PVi represents the reactive power control output of the i-th distributed photovoltaic unit; n is the number of distributed photovoltaic units that can be controlled by the distribution network system.

[0026] Furthermore, the step S2 constructs a historical control strategy set based on data-driven and improved BP neural network through three stages, wherein the three stages include training of historical data, initialization of the control strategy set, and online update of the control strategy set; the specific process is as follows:

[0027] S2-1, select the historical data of equipment status, fault information, operating environment and equipment control records from the power management system, data acquisition and monitoring system, and energy management system within one year as the feature quantity, and ignore other useless data;

[0028] S2-2, using the maximum-minimum normalization method to normalize the above-mentioned feature quantities, taking them as the input data set of the BP neural network model, and carrying out the training of the BP neural network model;

[0029] The training process includes:

[0030] According to the characteristic quantity obtained in step S2-1, sample data of the distribution network system after normalization is obtained in seconds as a real-time voltage characteristic input data set of the system, that is, as an input data set of the BP neural network model; the active power and reactive power of the distributed power inverter are used as output indicators to train the output end of the BP neural network model; and then a BP neural network training model is established;

[0031] S2-3, after the BP neural network model training process is completed, the bus voltage under the corresponding state is obtained by inputting any set of samples representing the active and reactive power of the node input, and the bus voltage under various scenarios is simulated by randomly generating active and reactive samples within a certain range (i.e., when the active and reactive power meet the voltage operation level), and it is used as the key to initialize the voltage control strategy set;

[0032] S2-4, updating the BP neural network model established in step 2-2 by using the latest real-time voltage control data obtained from the distribution network, i.e., the active power, reactive power and voltage status of the node, calculating the new hidden layer prediction output result and the network connection weight, and obtaining a new BP neural network model;

[0033] S2-5, repeating steps S2-3 to S2-4 until all the real-time control data of the distribution network voltage are updated, and obtaining the historical control strategy set after online update.

[0034] Furthermore, the step S3 assists decision-making and correction of the traditional voltage control strategy by using the real-time control strategy of approximate sensitivity to obtain the initial control amount of active and reactive power under the specific state of the active distribution network, and accumulates the initial value under the state (the above-mentioned corresponding specific state of the active distribution network) obtained from the historical control strategy set;

[0035] Specifically, the randomly generated distribution network voltage state is distinguished by using the over-limit state and the target state; a distribution network voltage state belonging to the over-limit state is defined as S i0 , the system bus voltage in this over-limit state is U i0 , the active and reactive power of each node in the system is expressed as (P ij0 , Q ij0 ), then the bus voltage is U i1 , the active and reactive power of each node is expressed as (P ij1 , Q ij1 ), the target state S of the distribution network in the over-limit state i1 and the distribution network voltage state S i0 Correspondingly; if both states (target state of the distribution network and voltage state of the distribution network) meet the economic requirements of regulation and the constraints of the regulation equipment, it indicates the over-limit state scenario S i0 The corresponding real-time voltage control scheme is the target state scenario S i1 The recorded node active and reactive power (P ij1 , Q ij1 ), complete the modification of the traditional voltage regulation strategy.

[0036] Furthermore, step S4 corrects the output of each photovoltaic power according to the historical strategy library auxiliary decision-making control method, as follows:

[0037] S4-1, obtaining real-time state parameters of the distribution network, the state parameters including bus voltage and active output and reactive output of the distributed photovoltaic inverter, and setting the state parameters to initial values, and resetting the number of iterations to 0;

[0038] S4-2, based on the real-time control strategy of approximate sensitivity, the initial control amount of active and reactive power of the active distribution network in this state is obtained, and the distributed photovoltaic inverter outputs active and reactive power;

[0039] S4-3, adding the initial control amount of active power and reactive power obtained in step S4-2 to the initial value obtained in S4-1, and then obtaining the historical control strategy and the distribution network bus voltage from the historical control strategy set;

[0040] S4-4. Using the distribution network bus voltage value obtained in step S4-3, combined with the historical control strategy, correct the active power and reactive power output by the distributed photovoltaic inverter in step S4-2, and output the corrected active and reactive power instructions to each regulating device on each feeder to carry out precise real-time control of the system voltage.

[0041] In order to realize a real-time voltage control method for a distributed photovoltaic distribution network driven by a historical control strategy set, the present invention also provides a real-time voltage control system for a distributed photovoltaic distribution network driven by a historical control strategy set, the system comprising:

[0042] Traditional voltage regulation strategy module, used to formulate traditional voltage regulation strategy based on approximate sensitivity;

[0043] The historical control strategy set construction module is used to construct the historical control strategy set based on data-driven and improved BP neural network;

[0044] A decision-making assistance and correction module, which is used to assist in decision-making and correct the traditional voltage control strategy by using the historical control strategy set constructed by the historical control strategy set construction module;

[0045] The voltage over-limit management module is used to correct the output of each photovoltaic power plant according to the historical strategy library auxiliary decision-making and control method to complete the voltage over-limit management of the distribution network.

[0046] Advantages of the invention:

[0047] The present invention provides a real-time voltage control method for a distributed photovoltaic distribution network driven by a historical control strategy set, which provides an important engineering solution to the voltage quality problem caused by photovoltaic access to the distribution network system.

[0048] Compared with the prior art, the present invention has the following significant advantages:

[0049] 1. No need to invest in additional energy storage or reactive power compensation equipment, which has better economic efficiency;

[0050] 2. It solves the problem that the voltage control methods of OLTC and group switching capacitors in traditional distribution networks are discrete and cannot achieve rapid control. In the face of sudden voltage changes in the distribution network, the distribution network can be effectively controlled in real time to ensure the voltage level;

[0051] 3. In order to solve the problem that the potential changes in the status of electrical equipment and network topology in the historical data of the distribution network may cause a large error between the subsequent learning results and the actual operating status, a method of online updating of the historical control strategy set is adopted to reduce this error. That is, the voltage historical control strategy set is based on the BP neural network model established by updating the latest real-time voltage control data obtained from the distribution network, and the new hidden layer prediction output results and network connection weights are calculated to obtain a new BP neural network model. Compared with the traditional voltage control method, this method of using the historical control strategy set to assist in correction is more accurate in photovoltaic control and has better voltage optimization effect.

[0052] 4. The voltage real-time control strategy driven by the historical control strategy set has good accuracy and meets the requirements of the distribution network system for real-time voltage control. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 It is a flow chart of the method for real-time voltage control of a distributed photovoltaic distribution network driven by a historical control strategy set in Example 1;

[0055] Figure 2 This is a diagram of the 380V distribution network architecture used in Example 7;

[0056] Figure 3 This is a flow chart of correcting the output of each photovoltaic power according to the historical strategy library auxiliary decision-making and control method in Example 6;

[0057] Figure 4 is a voltage curve diagram of the grid connection point when the control voltage exceeds the upper limit in Example 7;

[0058] Figure 5 is a voltage curve diagram of the grid connection point when the voltage exceeds the lower limit in Example 8;

[0059] Figure 6 This is the architecture diagram of the distributed photovoltaic distribution network voltage real-time control system driven by the historical control strategy set in Example 9. DETAILED DESCRIPTION

[0060] The following will be combined with the attached Figures 1 to 6 , the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] The purpose of the present invention is to provide a real-time voltage control method for a distributed photovoltaic distribution network driven by a historical control strategy set to solve the problems existing in the above-mentioned prior art, and to economically, reliably and accurately solve the voltage over-limit problem caused by the access of distributed energy to the distribution network and the sudden start-up of impact loads.

[0062] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0063] Example 1

[0064] Reference Figure 1 The present invention provides a real-time voltage control method for a distributed photovoltaic distribution network driven by a historical control strategy set, which specifically includes the following steps:

[0065] S1, formulate a traditional voltage regulation strategy based on approximate sensitivity;

[0066] S2, construct a historical control strategy set based on data-driven and improved BP neural network;

[0067] S3, using the historical control strategy set to assist in decision-making and modify the traditional voltage control strategy;

[0068] S4, according to the historical strategy library auxiliary decision-making and control method, correct the output of each photovoltaic output and complete the voltage limit control of the distribution network.

[0069] Example 2

[0070] Step S1 formulates a traditional voltage control strategy based on approximate sensitivity, including:

[0071] S1-1, using voltage sensitivity to measure the relationship between the changes in active and reactive power injected into each node of the distribution network and the corresponding voltage changes of the node, to solve the real-time control strategy of the distribution network voltage;

[0072] The specific process is as follows: first, the expression of voltage change with respect to the change of active and reactive power of the node is obtained according to the voltage sensitivity calculation formula, and then the optimal solution is iterated according to the optimization objective function, that is, the initial regulation amount of active and reactive power, thereby obtaining the real-time control strategy of distribution network voltage.

[0073] The calculation method of the approximate voltage sensitivity of the distribution network is as follows:

[0074] ΔV=[(B+Q)(GP) -1 (BQ)+(G+P) -1 ]ΔP-

[0075] [(GP)(B+Q) -1 (G+P)+(BQ) -1 ]ΔQ

[0076] Where B and G represent the real and imaginary parts of the node admittance matrix, respectively; P and Q represent the diagonal elements P i / V i 2 and Q i / V i 2 The n-order diagonal element matrix of ; ΔP and ΔQ represent the magnitude of active and reactive disturbances injected into the node respectively.

[0077] The optimization objective function is shown as follows:

[0078] F=min|VV ref |

[0079] Where V ref represents the voltage reference value, and V represents the regulated voltage value.

[0080] The constraints are:

[0081] V i,min ≤V i ≤V i,max

[0082] Where: V i,min With V i,max are the minimum and maximum voltages allowed at node i.

[0083] Power flow constraints:

[0084]

[0085] Where: P Gi , Q Gi is the active and reactive power injected into node i; P PVi and Q PVi is the distributed photovoltaic active and reactive power output at node i; P Li , Q Li is the active and reactive load at node i.

[0086] Distributed photovoltaic output constraints:

[0087]

[0088] Where P PVi.max and Q PVi.max They represent the maximum values ​​of the distributed photovoltaic active and reactive adjustable output respectively.

[0089] Example 3

[0090] The specific process of step S1-2 is: first, according to the initial control amount calculated in step S1-1, drive the photovoltaic inverter to emit or absorb active power and reactive power to optimize the distribution network voltage;

[0091] The initial voltage control formula is as follows:

[0092]

[0093] In the formula, V0 represents the initial value of the system voltage; ΔV represents the change in the system voltage; S Pi represents the approximate sensitivity of the active power of the i-th distributed photovoltaic to the voltage; S Qi represents the approximate sensitivity of the reactive power of the i-th distributed photovoltaic to the voltage; ΔP PVi represents the active power regulation output of the i-th distributed photovoltaic unit; ΔQ PVi represents the reactive power control output of the i-th distributed photovoltaic unit; n is the number of distributed photovoltaic units that can be controlled by the distribution network system.

[0094] Example 4

[0095] Step S2, constructing a historical control strategy set based on data-driven and improved BP neural network through three stages: training of historical data, initialization of control strategy set and online update of control strategy set.

[0096] The specific process is as follows:

[0097] S2-1, select the historical data of equipment status, fault information, operating environment and equipment control records from the power management system, data acquisition and monitoring system, and energy management system within one year as the feature quantity, and ignore other useless data;

[0098] S2-2, using the maximum-minimum normalization method to normalize the above-mentioned feature quantities, taking them as the input data set of the BP neural network model, and carrying out the training of the BP neural network model;

[0099] The training process includes:

[0100] According to the characteristic quantity obtained in step S2-1, sample data of the distribution network system after normalization is obtained in seconds as a real-time voltage characteristic input data set of the system, that is, as an input data set of the BP neural network model; the active power and reactive power of the distributed power inverter are used as output indicators to train the output end of the BP neural network model; and then a BP neural network training model is established;

[0101] S2-3, after the BP neural network model training process is completed, the bus voltage under the corresponding state is obtained by inputting any set of samples representing the active and reactive power of the node input, and the bus voltage under various scenarios is simulated by randomly generating active and reactive samples within a certain range (i.e., when the active and reactive power meet the voltage operation level), and it is used as the key to initialize the voltage control strategy set;

[0102] S2-4, updating the BP neural network model established in step 2-2 by using the latest real-time voltage control data obtained from the distribution network, i.e., the active power, reactive power and voltage status of the node, calculating the new hidden layer prediction output result and the network connection weight, and obtaining a new BP neural network model;

[0103] S2-5, repeating steps S2-3 to S2-4 until all the real-time control data of the distribution network voltage are updated, and obtaining the historical control strategy set after online update.

[0104] Example 5

[0105] Step S3 assists decision-making and corrections to the traditional voltage control strategy, and uses the real-time control strategy of approximate sensitivity to obtain the initial control amount of active and reactive power under the specific state of the active distribution network, and accumulates it with the initial value of the scenario obtained from the historical control strategy set.

[0106] Specifically, the randomly generated distribution network voltage state is distinguished by using the over-limit state and the target state; a distribution network voltage state belonging to the over-limit state is defined as S i0 , the system bus voltage in this over-limit state is U i0 , the active and reactive power of each node in the system is expressed as (P ij0 , Q ij0 ), then the bus voltage is U i1 , the active and reactive power of each node is expressed as (P ij1 , Q ij1 ), the target state S of the distribution network in the over-limit state i1 and the distribution network voltage state S i0 Correspondingly; if both states (target state of the distribution network and voltage state of the distribution network) meet the economic requirements of regulation and the constraints of the regulation equipment, it indicates the over-limit state scenario S i0The corresponding real-time voltage control scheme is the target state scenario S i1 The recorded node active and reactive power (P ij1 , Q ij1 ), complete the modification of the traditional voltage regulation strategy.

[0107] Example 6

[0108] See attached Figure 3 In step S4, the output of each photovoltaic power is corrected according to the auxiliary decision-making control method based on the historical strategy library to complete the voltage over-limit management of the distribution network. The specific process is as follows:

[0109] S4-1, set the initial bus voltage, load, active power and reactive power, and reset the number of iterations to 0;

[0110] S4-2, using the approximate sensitivity control strategy to calculate the node active and reactive power changes;

[0111] S4-3, update controllable photovoltaic active and reactive output;

[0112] S4-4, bring in the historical control strategy set for auxiliary correction;

[0113] S4-5: Check whether the voltage exceeds the limit. If so, return to step S4-2.

[0114] Example 7

[0115] In order to verify the effect of the method of the present invention on the control of voltage exceeding the upper limit, refer to the attached Figure 4 ,in accordance with Figure 2 The simulation model is established. The active distribution network system has 13 nodes and 12 feeder branches. The upper input voltage of the distribution system is 10kV, which is connected to the system through a transformer and reduces the voltage on the low-voltage side of the distribution network to 380V. Line parameters: Z L =0.642+j0.083Ω / km, L=0.8km is the distance between each node, the power factor of the three distributed photovoltaics is limited to 0.95, the adjustable range is 500kW, 200kW, 200kW, and the upper and lower limits of the bus voltage per unit value are [0.95,1.05]. This scenario selects a situation where the weather changes from cloudy to sunny. In this situation, the active output of each distributed photovoltaic increases suddenly due to weather changes, and the load changes slightly. The final result is that the bus voltage rises above the upper limit of 1.05pu.

[0116] Compared with the traditional voltage control strategy that uses approximate sensitivity for rough regulation, the method of the present invention can use the auxiliary correction of the historical strategy set to correct the photovoltaic regulation error caused by the error in the sensitivity itself, while preventing the bus voltage from exceeding the upper limit (exceeding 1.05pu), so as to finally make the regulation result better and the voltage level lower. The specific process is as follows:

[0117] 1. Starting from time t1, the active output of photovoltaic power increases sharply, so the bus voltage rises accordingly. At time t2, the voltage tends to be stable at 1.055pu;

[0118] 2. Starting from time t2, the optimization methods of historical strategy library assisted decision-making control and approximate sensitivity rough control are used to adjust the active and reactive output of photovoltaics. By reducing the active output and emitting inductive reactive power, the voltage level is reduced. At time t3, both real-time control strategies reduce the bus voltage to a stable value, where the purple curve bus voltage is adjusted to 1.041pu, the blue curve voltage is adjusted to 1.043pu, and the red curve voltage, which forms a reference for these two strategies, is maintained at 1.054pu. The traditional voltage control strategy is substituted into the rough calculation model of photovoltaic control quantity to obtain the active and reactive command values ​​of each photovoltaic. The historical strategy library assisted decision-making control modifies the former result through the auxiliary decision-making model to obtain the active and reactive commands of photovoltaics, making the control result better.

[0119] Example 8

[0120] In order to verify the effect of the method of the present invention on the control of voltage exceeding the lower limit, refer to the attached Figure 5 This scenario is selected from the heavy load situation of the line after the impact load is connected in actual operation. The network is not reactive and the bus voltage drops below the lower limit. When the bus voltage fluctuation is detected, the control method of the present invention immediately adopts the real-time voltage control method of the distributed photovoltaic distribution network driven by the historical control strategy set to solve the problem of the bus voltage exceeding the lower limit. The specific process is as follows:

[0121] 1. When the load level is high, the impact load is connected, and the bus voltage starts to drop from time t1. When it stabilizes at time t2, the voltage drops to 0.945pu;

[0122] 2. The traditional voltage control strategy and the real-time voltage control strategy driven by the historical control strategy are respectively used for control. The active and reactive power optimization instructions are calculated by the control algorithm, and the instructions are input into the simulation system to adjust the photovoltaic output. At t3, the voltage tends to be stable. Under the control strategy assisted by the historical strategy set, the voltage is adjusted to 0.986pu. Under the traditional voltage control strategy, the voltage is adjusted to 0.973pu, which verifies the superiority of the invention in the scenario of controlling the voltage exceeding the lower limit.

[0123] Example 9

[0124] This embodiment provides a real-time voltage control system for a distributed photovoltaic distribution network driven by a historical control strategy set, the system comprising:

[0125] Traditional voltage regulation strategy module, used to formulate traditional voltage regulation strategy based on approximate sensitivity;

[0126] The historical control strategy set construction module is used to construct the historical control strategy set based on data-driven and improved BP neural network;

[0127] A decision-making assistance and correction module, which is used to assist in decision-making and correct the traditional voltage control strategy by using the historical control strategy set constructed by the historical control strategy set construction module;

[0128] The voltage over-limit management module is used to correct the output of each photovoltaic power plant according to the historical strategy library auxiliary decision-making and control method to complete the voltage over-limit management of the distribution network.

[0129] Furthermore, the conventional voltage regulation strategy module formulates a conventional voltage regulation strategy based on approximate sensitivity, comprising the following steps:

[0130] S1-1, using voltage sensitivity to measure the relationship between the changes in active and reactive power injected into each node of the distribution network and the corresponding voltage changes of the node, to solve the real-time control strategy of the distribution network voltage;

[0131] S1-2, issuing control instructions to the photovoltaic inverter according to the real-time control strategy of the distribution network voltage, driving the real-time control of the distribution network voltage.

[0132] Furthermore, the historical control strategy set construction module constructs a historical control strategy set based on data-driven and improved BP neural network through three stages, including historical data training, control strategy set initialization and control strategy set online update; the specific process is as follows:

[0133] S2-1, select the historical data of equipment status, fault information, operating environment and equipment control records from the power management system, data acquisition and monitoring system, and energy management system within one year as the feature quantity, and ignore other useless data;

[0134] S2-2, using the maximum-minimum normalization method to normalize the above-mentioned feature quantities, taking them as the input data set of the BP neural network model, and carrying out the training of the BP neural network model;

[0135] The training process includes:

[0136] According to the characteristic quantity obtained in step S2-1, sample data of the distribution network system after normalization is obtained in seconds as a real-time voltage characteristic input data set of the system, that is, as an input data set of the BP neural network model; the active power and reactive power of the distributed power inverter are used as output indicators to train the output end of the BP neural network model; and then a BP neural network training model is established;

[0137] S2-3, after the BP neural network model training process is completed, the bus voltage under the corresponding state is obtained by inputting any set of samples representing the active and reactive power of the node input, and the bus voltage under various scenarios is simulated by randomly generating active and reactive samples within a certain range (i.e., when the active and reactive power meet the voltage operation level), and it is used as the key to initialize the voltage control strategy set;

[0138] S2-4, updating the BP neural network model established in step 2-2 by using the latest real-time voltage control data obtained from the distribution network, i.e., the active power, reactive power and voltage status of the node, calculating the new hidden layer prediction output result and the network connection weight, and obtaining a new BP neural network model;

[0139] S2-5, repeating steps S2-3 to S2-4 until all the real-time control data of the distribution network voltage are updated, and obtaining the historical control strategy set after online update.

[0140] Furthermore, the auxiliary decision and correction module assists in decision-making and correction of the traditional voltage control strategy by utilizing the real-time control strategy of approximate sensitivity to obtain the initial control quantities of active and reactive power under the specific state of the active distribution network, and accumulates them with the initial values ​​under this state (the above-mentioned corresponding specific state of the active distribution network) obtained from the historical control strategy set.

[0141] Furthermore, the voltage over-limit management module corrects and outputs each photovoltaic output according to the historical strategy library auxiliary decision-making control method, as follows:

[0142] S4-1, obtaining real-time state parameters of the distribution network, the state parameters including bus voltage and active output and reactive output of the distributed photovoltaic inverter, and setting the state parameters to initial values, and resetting the number of iterations to 0;

[0143] S4-2, based on the real-time control strategy of approximate sensitivity, the initial control amount of active and reactive power of the active distribution network in this state is obtained, and the distributed photovoltaic inverter outputs active and reactive power;

[0144] S4-3, adding the initial control amount of active power and reactive power obtained in step S4-2 to the initial value obtained in S4-1, and then obtaining the historical control strategy and the distribution network bus voltage from the historical control strategy set;

[0145] S4-4. Using the distribution network bus voltage value obtained in step S4-3, combined with the historical control strategy, the output of the distributed photovoltaic inverter in step S4-2 is corrected, and the corrected active and reactive power instructions are output to each regulating device on each feeder to carry out accurate real-time control of the system voltage.

[0146] The above part uses examples to illustrate the principles and implementation methods of the present invention. The description of the above examples is only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A real-time voltage control method for a distributed photovoltaic distribution network driven by a historical control strategy set, characterized in that: The steps include: S1, formulate a traditional voltage regulation strategy based on approximate sensitivity; S2, construct a historical control strategy set based on data-driven and improved BP neural network; S3, using the historical control strategy set to assist in decision-making and modify the traditional voltage control strategy; S4, according to the historical strategy library auxiliary decision-making and control method, correct the output of each photovoltaic output and complete the voltage limit control of the distribution network.

2. The method for real-time voltage control of a distributed photovoltaic distribution network driven by a historical control strategy set according to claim 1 is characterized in that: The conventional voltage regulation strategy based on approximate sensitivity described in step S1 includes: S1-1, using voltage sensitivity to measure the relationship between the changes in active and reactive power injected into each node of the distribution network and the corresponding voltage changes of the node, to solve the real-time control strategy of the distribution network voltage; S1-2, issuing control instructions to the photovoltaic inverter according to the real-time control strategy of the distribution network voltage, driving the real-time control of the distribution network voltage.

3. The method for real-time voltage control of a distributed photovoltaic distribution network driven by a historical control strategy set according to claim 2 is characterized in that: The specific process of step S1-1 is: first, according to the voltage sensitivity calculation formula, the expression of the voltage change amount with respect to the active and reactive changes of the node is obtained, and then according to the optimization objective function, the optimal solution is iterated, that is, the initial control amount of active and reactive power, thereby obtaining the real-time control strategy of the distribution network voltage; The voltage sensitivity calculation formula is as follows: ΔV=[(B+Q)(G-P) -1 (B-Q)+(G+P) -1 ]ΔP- [(G-P)(B+Q) -1 (G+P)+(B-Q) -1 ]ΔQ In the formula, B represents the real part of the node admittance matrix; G represents the imaginary part of the node admittance matrix; P represents the diagonal element P i / V i 2 The n-order diagonal matrix of i / V i 2 The n-order diagonal element matrix of ; ΔP represents the active disturbance injected by the node; ΔQ represents the reactive disturbance injected by the node; The optimization objective function is as follows: F=min|VV ref | Where V ref represents the voltage reference value, and V represents the regulated voltage value.

4. The method for real-time voltage control of a distributed photovoltaic distribution network based on a historical control strategy set drive according to claim 2 is characterized in that: The specific process of step S1-2 is: according to the initial control amount calculated in step S1-1, drive the photovoltaic inverter to emit or absorb active power and reactive power to optimize the voltage of the distribution network; The initial voltage control formula is as follows: In the formula, V0 represents the initial value of the system voltage; ΔV represents the change in the system voltage; S Pi represents the approximate sensitivity of the active power of the i-th distributed photovoltaic to the voltage; S Qi represents the approximate sensitivity of the reactive power of the i-th distributed photovoltaic to the voltage; ΔP PVi represents the active power regulation output of the i-th distributed photovoltaic unit; ΔQ PVi represents the reactive power control output of the i-th distributed photovoltaic unit; n is the number of distributed photovoltaic units that can be controlled by the distribution network system.

5. The method for real-time voltage control of a distributed photovoltaic distribution network driven by a historical control strategy set according to claim 1 or 2, characterized in that: The step S2 constructs a historical control strategy set based on data-driven and improved BP neural network through three stages, wherein the three stages include training of historical data, initialization of the control strategy set, and online update of the control strategy set. The specific process is as follows: S2-1, select the historical data of equipment status, fault information, operating environment and equipment control records from the power management system, data acquisition and monitoring system, and energy management system within one year as the feature quantity, and ignore other useless data; S2-2, using the maximum-minimum normalization method to normalize the above-mentioned feature quantities, taking them as the input data set of the BP neural network model, and carrying out the training of the BP neural network model; The training process includes: According to the characteristic quantity obtained in step S2-1, sample data of the distribution network system after normalization is obtained in seconds as a real-time voltage characteristic input data set of the system, that is, as an input data set of the BP neural network model; the active power and reactive power of the distributed power inverter are used as output indicators to train the output end of the BP neural network model; and then a BP neural network training model is established; S2-3, after the BP neural network model training process is completed, the bus voltage under the corresponding state is obtained by inputting any set of samples representing the active and reactive power of the node input, and the active and reactive samples are randomly generated within a certain range to simulate the bus voltage conditions under various scenarios, and it is used as the key to initialize the voltage control strategy set; S2-4, updating the BP neural network model established in step 2-2 by using the latest real-time voltage control data obtained from the distribution network, i.e., the active power, reactive power and voltage status of the node, calculating the new hidden layer prediction output result and the network connection weight, and obtaining a new BP neural network model; S2-5, repeating steps S2-3 to S2-4 until all the real-time control data of the distribution network voltage are updated, and obtaining the historical control strategy set after online update.

6. The method for real-time voltage control of a distributed photovoltaic distribution network driven by a historical control strategy set according to claim 1 or 2, characterized in that: The step S3 assists decision-making and correction of the traditional voltage control strategy by using the real-time control strategy of approximate sensitivity to obtain the initial control amount of active and reactive power under the specific state of the active distribution network, and accumulates it with the initial value under the state obtained from the historical control strategy set; Specifically, the randomly generated distribution network voltage state is distinguished by using the over-limit state and the target state; a distribution network voltage state belonging to the over-limit state is defined as S i0 , the system bus voltage in this over-limit state is U i0 , the active and reactive power of each node in the system is expressed as (P ij0 , Q ij0 ), then the bus voltage is U i1 , the active and reactive power of each node is expressed as (P ij1 , Q ij1 ), the target state S of the distribution network in the over-limit state i1 and the distribution network voltage state S i0 Correspondingly; if the target state of the distribution network and the voltage state of the distribution network both meet the economic constraints of the regulation and the constraints of the regulation equipment, it means the over-limit state scenario S i0 The corresponding real-time voltage control scheme is the target state scenario S i1 The recorded node active and reactive power (P ij1 , Q ij1 ), complete the correction of the traditional voltage regulation strategy.

7. The method for real-time voltage control of a distributed photovoltaic distribution network based on a historical control strategy set drive according to claim 1 or 2, characterized in that: Step S4 corrects the output of each photovoltaic power according to the historical strategy library auxiliary decision-making control method, as follows: S4-1, obtaining real-time state parameters of the distribution network, the state parameters including bus voltage and active output and reactive output of the distributed photovoltaic inverter, and setting the state parameters to initial values, and resetting the number of iterations to 0; S4-2, based on the real-time control strategy of approximate sensitivity, the initial control amount of active and reactive power of the active distribution network in this state is obtained, and the distributed photovoltaic inverter outputs active and reactive power; S4-3, adding the initial control amount of active power and reactive power obtained in step S4-2 to the initial value obtained in S4-1, and then obtaining the historical control strategy and the distribution network bus voltage from the historical control strategy set; S4-4. Using the distribution network bus voltage value obtained in step S4-3, combined with the historical control strategy, the output of the distributed photovoltaic inverter in step S4-2 is corrected, and the corrected active and reactive power instructions are output to each regulating device on each feeder to carry out accurate real-time control of the system voltage.

8. A control system for implementing the real-time voltage control method of a distributed photovoltaic distribution network driven by a historical control strategy set as described in claim 1 or 2, characterized in that: include: Traditional voltage regulation strategy module, used to formulate traditional elegant regulation strategy based on approximate sensitivity; The historical control strategy set construction module is used to construct the historical control strategy set based on data-driven and improved BP neural network; A decision-making assistance and correction module, which is used to assist in decision-making and correct the traditional voltage control strategy by using the historical control strategy set constructed by the historical control strategy set construction module; The voltage over-limit management module is used to correct the output of each photovoltaic power plant according to the historical strategy library auxiliary decision-making and control method to complete the voltage over-limit management of the distribution network.

9. The control system according to claim 8, characterized in that: The traditional voltage regulation strategy module formulates a traditional voltage regulation strategy based on approximate sensitivity, including the following steps: S1-1, using voltage sensitivity to measure the relationship between the changes in active and reactive power injected into each node of the distribution network and the corresponding voltage changes of the node, to solve the real-time control strategy of the distribution network voltage; S1-2, issuing control instructions to the photovoltaic inverter according to the real-time control strategy of the distribution network voltage, driving the real-time control of the distribution network voltage.

10. The control system according to claim 8, characterized in that: The historical control strategy set construction module constructs a historical control strategy set based on data-driven and improved BP neural network through three stages, including historical data training, control strategy set initialization and control strategy set online update; the specific process is as follows: S2-1, select the historical data of equipment status, fault information, operating environment and equipment control records from the power management system, data acquisition and monitoring system, and energy management system within one year as the feature quantity, and ignore other useless data; S2-2, using the maximum-minimum normalization method to normalize the above-mentioned feature quantities, taking them as the input data set of the BP neural network model, and carrying out the training of the BP neural network model; The training process includes: According to the characteristic quantity obtained in step S2-1, sample data of the distribution network system after normalization is obtained in seconds as a real-time voltage characteristic input data set of the system, that is, as an input data set of the BP neural network model; the active power and reactive power of the distributed power inverter are used as output indicators to train the output end of the BP neural network model; and then a BP neural network training model is established; S2-3, after the BP neural network model training process is completed, the bus voltage under the corresponding state is obtained by inputting any set of samples representing the active and reactive power of the node input, and the active and reactive samples are randomly generated within a certain range to simulate the bus voltage conditions under various scenarios, and it is used as the key to initialize the voltage control strategy set; S2-4, updating the BP neural network model established in step 2-2 by using the latest real-time voltage control data obtained from the distribution network, i.e., the active power, reactive power and voltage status of the node, calculating the new hidden layer prediction output result and the network connection weight, and obtaining a new BP neural network model; S2-5, repeating steps S2-3 to S2-4 until all the real-time control data of the distribution network voltage are updated, and obtaining the historical control strategy set after online update.

Images