A Four-Layer Joint Flexible Voltage Control Method for Distributed Photovoltaic Low-Voltage Grid Connection

Abstract

The present invention discloses a four - layer combined flexible voltage control method for distributed photovoltaic low - voltage grid connection. First, obtain the predicted data of the output power of each distributed photovoltaic and the predicted data of the power of each user load for the next day, and predict the time period when the voltage exceeds the upper limit without flexible control; based on the starting time period of the upper - limit exceeding predicted the previous day and the actual situation of the voltage exceeding the upper limit on the current day, determine the starting strategy of the water - storage electric water heater; if the voltage still exceeds the upper limit on the user side, increase the reactive power consumption of the inverter to try to reduce the voltage to the qualified range; if the reactive power consumption of the photovoltaic inverter reaches the limit value and is still insufficient to achieve the control purpose, adjust the inverter to cut the active power generation of the photovoltaic to try to reduce the voltage to the qualified range; if there is still a situation where the voltage exceeds the upper limit, trip the asymmetric grouped intelligent switch to compulsorily cut the photovoltaic active power until the voltage drops below the allowable upper limit value. The present invention can solve the problem of voltage exceeding the upper limit caused by large - scale grid connection of distributed photovoltaics.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed photovoltaic flexible control, and particularly to a four-layer combined flexible voltage control method for low-voltage grid connection of distributed photovoltaics. Background Art

[0002] With the continuous construction and improvement of the new power system, large-scale new energy is developed and connected to the grid. Among them, as an important part of the new energy structure, distributed photovoltaics have developed rapidly. Due to the volatility, intermittency, and uncertainty of its power generation, the large-scale grid connection of distributed photovoltaics poses a huge challenge to the safe and stable operation of the power grid. Limited by the high cost of current electrochemical energy storage, during the period of abundant sunlight, the excess power generated cannot be stored or consumed locally. If not controlled, the power generation during the peak photovoltaic power generation period will all flow into the grid, easily causing overvoltage problems, leading to the insulation aging of electrical equipment, and further affecting the safety and power supply quality of the power grid. Therefore, during the period of large-scale reverse power transmission of photovoltaics in a certain period, the measure of curtailment of light can be taken to avoid or alleviate overvoltage problems. However, this measure will directly affect the economic benefits of photovoltaic investors. Therefore, there is an urgent need for a method that can ensure the safe and stable operation of the power grid while taking into account the economic benefits of photovoltaic investors.

[0003] When dealing with the large-scale access of photovoltaics to the distribution network, there are some deficiencies and defects in the existing technologies, such as voltage over-limit problems, unfair power distribution of distributed photovoltaics, single control strategies, high-cost energy storage requirements, insufficient prediction accuracy, and lack of flexible control means.

[0004] Currently, when large-scale photovoltaics are connected to the distribution network, the peak photovoltaic power generation period is prone to cause grid voltage over-limit problems, leading to the insulation aging of electrical equipment and affecting the safety and power supply quality of the power grid. To solve this problem, generally, methods of restricting photovoltaic power generation or installing energy storage devices are adopted to achieve the purpose of grid voltage stability. However, the existing technologies often adopt a single day-ahead control or intra-day control scheme for distributed photovoltaics, resulting in unfair power curtailment. Due to the insufficient prediction accuracy of photovoltaic output and load in the existing technologies, there are often significant deviations between the generated control scheme and the actual operating conditions. Such a control method not only has low control accuracy, wastes a large amount of photovoltaic energy, but also has a poor control effect on distributed photovoltaics. And due to the large investment cost of energy storage devices, the method of increasing energy storage devices has poor economy. Therefore, new technologies need to be proposed to solve the above problems with good economy. Summary of the Invention

[0005] In view of the above overvoltage problem caused by the large-scale access of distributed photovoltaic power generation, while taking into account the economic benefits of photovoltaic investors, the present invention provides a four-layer combined flexible voltage control method for low-voltage grid connection of distributed photovoltaic power generation, which adopts a four-layer control strategy to hierarchically regulate the photovoltaic output. By adopting multiple flexible regulations, the output of distributed photovoltaic power generation is made more stable. At the same time, it is first considered to store the excess power in water heaters, so that this part of the energy can be utilized instead of directly abandoning light, which increases the economy.

[0006] The object of the present invention is achieved by the following technical solutions.

[0007] A four-layer combined flexible voltage control method for low-voltage grid connection of distributed photovoltaic power generation includes the following steps:

[0008] S1. Obtain the daily planned output values of each distributed photovoltaic power generation in the low-voltage (380 / 220V) access distribution substation area, the historical data of each distributed photovoltaic power generation output, and the historical data of each user load power, and predict the photovoltaic output power, each user load power, and voltage characteristic curve without flexible control.

[0009] S2. Calculate the voltage reference values of each node in the low-voltage distribution substation area, compare the predicted voltage characteristic curve with the voltage reference value to judge the situation of voltage exceeding the upper limit, obtain the photovoltaic excess power based on the daily planned output value of the photovoltaic power generation and the predicted photovoltaic output power, and determine the start-up strategy of the storage water heater (such as the hot water storage pool / water tank equipped on the roof) of the photovoltaic user based on the photovoltaic excess power and the predicted voltage characteristic curve, and compare the maximum total power consumption of the storage water heater with the photovoltaic excess electric energy within the time interval to execute the first-layer control. If , continue to execute step S3; otherwise, end the calculation.

[0010] S3. Calculate and determine the power regulation mode of the photovoltaic power station, execute the second-layer control. If entering the reactive power regulation mode, execute step S4; otherwise, execute step S5.

[0011] S4. Increasing the reactive power consumption of the inverter can reduce the voltage to the qualified range. Calculate the power generation plan, judge whether the voltage of each node meets the standard, and execute the third-layer control. If it meets the standard, issue an instruction according to the current power generation plan and end the calculation; if it does not meet the standard, enter step S6.

[0012] S5. At this time, the reactive power consumption of the photovoltaic inverter reaches the limit value and is still insufficient to achieve the control purpose. Adjust the inverter to reduce the active power generation of the photovoltaic power generation to strive to reduce the voltage to the qualified range, judge whether the voltage of each node meets the standard, and execute the third-layer control. If it meets the standard, issue an instruction according to the current power generation plan and end the calculation; if it does not meet the standard, enter step S6.

[0013] S6. Execute the fourth - layer control to compulsorily reduce the PV active power by tripping the asymmetric - grouping intelligent switch until the voltage drops below the allowable upper limit value.

[0014] Further, in step S1, the specific steps are as follows:

[0015] S1.1. Input the pre - day planned output values of each distributed PV, the historical data of each distributed PV output, and the historical data of each user load power at and before a certain moment in the low - voltage access distribution sub - area, normalize the data, and divide it into a training set, a test set, and a validation set. At and before a certain moment in the low - voltage access distribution sub - area, normalize the data, and divide it into a training set, a test set, and a validation set;

[0016] S1.2. Perform variational mode decomposition (VMD) processing on the training set, the test set, and the validation set respectively to obtain 4 independent intrinsic mode functions (IMFs).

[0017] S1.3. Take the 4 corresponding intrinsic mode functions (IMFs) of the training set as the input values of the LSTM model. The output of the LSTM model is the modal components corresponding to the PV output power and each user load power without flexible control. Each intrinsic mode function is learned by an LSTM model.

[0018] S1.4. Use the trained LSTM model to predict the modal components corresponding to the future PV output power and each user load power without flexible control, and then reconstruct each modal component of the prediction result back to the original signal form to obtain the PV output data and each user load power data without flexible control.

[0019] S1.5. According to the predicted PV output data and each user load power data without flexible control, set the load and generation conditions at each time point, select the forward - backward substitution method to perform power flow calculation on the radial distribution network, and obtain the first - voltage prediction values, current prediction values, and power prediction values of each node of the power grid at each time point; use the obtained data to draw the voltage characteristic curve.

[0020] Further, in step S1.1, it is as follows:

[0021] S1.1.1. To make full use of the existing data, divide the data into a training set, a test set, and a validation set according to a set ratio; among them, the training set is used to train the LSTM model; the validation set is used to verify the generalization ability of the current LSTM model during iterative training and finally determine whether to stop further training; the test set is used to evaluate the generalization ability of the final LSTM model to ensure the reliability of the data predicted by the LSTM model.

[0022] S1.1.2. To eliminate the dimensional differences between data and accelerate the convergence speed of the LSTM model, the data is normalized and scaled to the range of [0, 1].

[0023] Further, in step S2, the specific steps are as follows:

[0024] S2.1. Calculate the voltage reference values of each node in the low-voltage distribution substation area as follows:

[0025] Use the sliding window algorithm to preprocess the voltage of each node in the low-voltage distribution substation area; calculate the average value of the voltages of each node within the time interval If the voltage of the th node does not exceed the upper limit, then use the average value of the voltage of the th node within the time interval as the first voltage reference value of the th node , where =1,2……, is the number of nodes in the distribution network, ; if the voltage of the th node exceeds the upper limit, then use the specified upper limit of voltage operation as the first voltage reference value of the th node; the formula for the first voltage reference value of the

[0026] th node is:

[0027] Among them, is the number of voltage samplings within the time interval ; is the historical measurement value of the voltage of the th node within the time interval ; is the first voltage prediction value of the th node at the moment obtained by the power flow calculation in step S1.5; is the specified upper limit of voltage operation, taken as 1.06 pu;

[0028] S2.2. Determine whether the voltage exceeds the upper limit: Compare the first voltage prediction value of each node with the first voltage reference value of each node. If the first voltage prediction value of a node is greater than the first voltage reference value, it means that the voltage of this node exceeds the upper limit; otherwise, it does not exceed the upper limit. If the voltage does not exceed the upper limit, execute step S2.3; if the voltage exceeds the upper limit, execute step S2.4;

[0029] S2.3. Set the water heater to start heating at 4:00 PM until the set temperature is reached. End the calculation.

[0030] S2.4. Subtract the PV planned output value for the day from the PV output data predicted in step S1 to obtain the PV excess power. ;

[0031] S2.5. Set the start time of the storage water heater as follows:

[0032] Compare the voltage characteristic curve obtained in step S1 with the first voltage reference value to obtain the time period when the voltage exceeds the upper limit; the water heater is set to start heating during the time period when the voltage exceeds the upper limit, and in principle, it does not heat during other time periods.

[0033] S2.6. Calculate the adjustment capacity of the storage water heater:

[0034] Assume that the storage water heater stops heating immediately after the hot water temperature reaches the set temperature Then the maximum total power consumption of the storage water heater The calculation formula is as follows:

[0035] ;

[0036] Where, Represents the specific heat capacity of water, Represents the total capacity of the water to be heated in the water heater tank, Represents the initial temperature of the water in the tank, Represents the number of storage water heaters in the low-voltage distribution substation area;

[0037] S2.7. Compare the maximum total power consumption of the storage water heater With the PV excess electrical energy within the time interval :

[0038] ;

[0039] If , that is, the PV excess electrical energy is not enough to heat the water in the water heater to the set temperature , then the storage water heater still needs to consume additional power For heating; after the time period when the voltage exceeds the upper limit ends, the storage water heater continues to heat at a constant operating power For a period of time , The calculation formula of

[0040] ;

[0041] Among them, is the operating power when the storage water heater continues to heat during the overvoltage end period. After setting the time required for continuous heating the operation is ended;

[0042] If , that is, if the storage water heater exactly completely absorbs the surplus active power during the overvoltage period, the calculation is ended and no next-level control is required;

[0043] If , that is, if the power consumption of the storage water heater has reached the threshold but the surplus photovoltaic power has not been completely consumed, the inverter reactive power optimization regulation mode of the next-level control needs to be started; at this time, the water heater heats to the set temperature at a constant operating power ; after the first-level control, the surplus photovoltaic power The calculation formula is as follows:

[0044] .

[0045] Furthermore, in step S3, when the storage water heater cannot completely absorb the surplus photovoltaic power, the inverter active and reactive powers are adjusted to try to reduce the voltage to the qualified range, and the power regulation mode of the photovoltaic power station is calculated and determined. The specific steps are as follows:

[0046] S3.1. Based on the user load power data predicted in step S1 and the surplus photovoltaic power calculated in step S2 , a new power flow calculation is performed to obtain the second voltage prediction value of each node in the low-voltage distribution substation area , is the number of nodes in the distribution network, =1,2……, ; the second voltage reference value of each node is recalculated according to step S2.1 ;

[0047] S3.2. According to step S2.5, it can be seen that at this time, there must be a node voltage exceeding the upper limit. Determine the node with the highest second voltage prediction value calculated in step S3.1 as the leading node, and the leading node number is ; the pre-adjustment value of the leading node voltage is:

[0048] ;

[0049] The power factor of the photovoltaic power station needs to be dynamically adjustable within the range of leading 0.95 to lagging of 0.95; considering the power factor limit of the photovoltaic power station, the th photovoltaic power station The first maximum reactive power output at a moment is:

[0050] ;

[0051] In the formula: is the power factor of the PV power station, with a value of 0.95; is the predicted value of the maximum active power output of the th PV power station at a moment; is the number of PV power stations in the distribution network, = 1, 2,..., ;

[0052] S3.4. By using the non-linear least squares method to fit the relationship between the second voltage prediction value of each node and the power, and combining the real-time operation data to calculate the real-time voltage sensitivity between each node;

[0053] S3.5. The reactive power can be reduced to a minimum of , and calculate the maximum adjustable amount of the dominant node voltage:

[0054] ;

[0055] In the formula: is the ( , , )-th element of the voltage-reactive power sensitivity matrix , indicating the voltage sensitivity of the th PV power station's grid-connected node power change injected into the th node. The th node is the grid-connected node of the th PV power station; is the reactive power output by the th PV power station at

[0056] S3.6. Calculate the PV reactive power evaluation coefficient of the distribution network:

[0057] ;

[0058] When , adjusting the reactive power of the power station at this time will increase the voltage amplitude of each node, and it is necessary to enter the reactive power + active power reduction mode; when the reactive power evaluation coefficient is positive, if , enter the reactive power regulation mode, if

[0059] If entering the reactive power + active power reduction mode, execute step S5; if entering the reactive power regulation mode, execute step S4.

[0060] Furthermore, the specific steps of step S3.4 are as follows:

[0061] S3.4.1: Use the historical data of the distributed photovoltaic output and the historical data of the load power of each user to perform a power flow calculation to obtain the historical injection power and historical voltage data of each node, and use the non-linear least squares method to fit the mapping relationship between the historical injection power and historical voltage data of each node;

[0062] S3.4.2: Collect the current voltage and power of each node, randomly generate a small change matrix of the injection power , define a scaling factor α, 0 < α << 1, such that ; where represents the absolute value of the current injection power of the -th node, represents the absolute value of the change in the current injection power of the -th node; keep the reactive power of each node unchanged, and obtain the voltage distribution after the power change of each node through the curve fitted in step S3.4.1, and further obtain the small voltage change matrix ;

[0063] S3.4.3: Use the least squares method to solve to obtain the current voltage-active power sensitivity matrix ; is the voltage-active power sensitivity matrix of the , )-th element, representing the voltage sensitivity of the -th node to the change in the active power injected into the -th node;

[0064] S3.4.4: Similarly, the voltage-reactive power sensitivity matrix of the node voltage to the change in the injected reactive power can be calculated ; is the voltage-reactive power sensitivity matrix of the , )-th element, representing the voltage sensitivity of the -th node to the change in the reactive power injected into the -th node.

[0065] Furthermore, in step S4, increasing the reactive power consumption of the inverter can reduce the voltage to the qualified range, and the power generation plan is calculated. The specific steps are as follows:

[0066] S4.1. Subtract the second voltage prediction value of the grid-connected node of each PV power station obtained in step S3.1 from the second voltage reference value, and combine them to obtain the voltage error matrix , at this time, the active power of the power station remains unchanged, so the active power change matrix of the PV power station ; Extract the voltage-reactive power sensitivity synthesis matrix between the grid-connected nodes of the PV power station , and we can get:

[0067] ;

[0068] In the formula: is the reactive power change matrix of the PV power station, and the ( , , ) element of the voltage-reactive power sensitivity matrix represents the voltage sensitivity of the grid-connected node of the th PV power station to the power change injected into the grid-connected node of the th PV power station,

[0069] S4.2. After calculating the reactive power change matrix of the PV power station, the reactive power output of the th PV power station is:

[0070] ;

[0071] In the formula: is the th element in the reactive power change matrix of the PV power station at time;

[0072] S4.3. Add a clipping function. If exceeds the first maximum reactive power output of the PV power station, then issue the reactive power output plan with ; The present invention performs in-day regulation on the basis of a day-ahead power generation plan. At this time, each power station generates active power according to the day-ahead power generation plan, and only needs to change the reactive power output. The first reactive power output plan of the th PV power station at time is:

[0073] ;

[0074] S4.4. Perform power flow calculation on the distribution network to obtain the third voltage prediction value of each node of the distribution network after each power station executes according to the current power generation plan, and update the third voltage reference value ; Calculate the voltage deviation of the distribution network and :

[0075] ;

[0076] S4.5. Determine whether the voltage of each node in the distribution network is up to standard. If And the third voltage prediction of all nodes Are all less than Then the node voltage of the distribution network is up to standard. Then, issue an instruction according to the current power generation plan and end the calculation. Indicates the allowable voltage deviation range, and its setting needs to be determined according to the actual situation and requirements of the power system; if the node voltage of the distribution network is not up to standard, update the photovoltaic power generation power to the current power generation plan, keep the active power output plan unchanged, and update the reactive power output plan to the first reactive power output plan , Fit the voltage sensitivity matrix between current nodes And , Execute step S6 and enter the reactive + active power regulation mode.

[0077] Furthermore, in step S5, when the reactive power consumption of the photovoltaic inverter reaches the limit value and still cannot achieve the control purpose, adjust the inverter to reduce the active power of photovoltaic power generation and strive to reduce the voltage of each node in the low-voltage distribution area to the qualified range. The specific steps are as follows:

[0078] S5.1. If the photovoltaic active power is restricted, the maximum adjustable amount of reactive power will change accordingly; calculate the second maximum output reactive power Of the Th :

[0079] ;

[0080] S5.2. Output reactive power according to To maximize the utilization of the photovoltaic power station capacity, then the change amount of output reactive power Is:

[0081] ;

[0082] S5.3. Define fairness as the equal proportion distribution of active power reduction amount by each photovoltaic power station. Therefore, each photovoltaic power station is set with the same active power reduction ratio , And perform joint active and reactive power control;

[0083] Active power reduction ratio Is specifically as follows:

[0084] ;

[0085] In the formula, is the predicted value of the maximum active power that can be output at the th photovoltaic power station at a certain moment, and is the active power curtailment of the th photovoltaic power station, which is specifically as follows:

[0086] ；

[0087] In the formula, is the planned value of the active power output of the th photovoltaic power station that needs to be calculated subsequently; The specific values of the pre-adjustment of the leading node voltage are as follows:

[0088] ;

[0089] ;

[0090] In the formula, is an unknown variable that varies between [0, 1), and the rest are constants. After simplification, a quadratic equation of can be obtained, and has a unique solution within the variation range;

[0091] S5.4. After obtaining , the active power output plan and the second reactive power output plan of each photovoltaic power station are specifically as follows:

[0092] ；

[0093] S5.5. Perform a power flow calculation on the distribution network to obtain the fourth voltage prediction value of each node in the distribution network after each photovoltaic power station executes according to the current power generation plan;

[0094] S5.6. Judge whether the voltage of each node meets the standard; in the active power curtailment mode, if the fourth voltage prediction value of each node is less than , then the voltage of the distribution network nodes meets the standard, and an instruction is issued according to the current power generation plan to end the calculation; if the voltage of the distribution network nodes does not meet the standard, then update the photovoltaic power generation power to the current power generation plan, update the active power output plan to , update the reactive power output plan to the second reactive power output plan , fit the current voltage sensitivity matrix and , execute step S6, and enter the next layer of asymmetric grouping intelligent switch control.

[0095] Further, in step S6, since the control of the first three layers has not achieved the goal at this time, a strategy of partially disconnecting PV power generation for voltage control is adopted. By tripping the externally connected asymmetric grouped intelligent switch, the reverse power flow of PV active power is forcibly reduced to mitigate the voltage rise, specifically as follows:

[0096] Most traditional intelligent switches are only equipped with one switch, and when overvoltage occurs, they can only reduce the voltage by adopting the method of 100% light curtailment, resulting in a large amount of wasted light energy. The present invention proposes to configure a flexible control switch terminal in front of each PV user meter, with 2 groups of switches. One group has a breaking capacity of 1 / 3, and the other group has a breaking capacity of 2 / 3. Light curtailment is initiated when the control effects of the water heater and PV inverter in the first three layers of control have not reduced the substation area voltage below the allowable upper limit.

[0097] Further, the specific steps of step S6 are as follows:

[0098] S6.1. First, the edge intelligent terminal issues an instruction to the flexible control switch terminals at each user end to uniformly trip the first flexible control switches with a breaking capacity of 1 / 3 for each user, fairly requiring all PV users to uniformly curtail 1 / 3 of the light. If the voltage at each node returns to the normal state, it is maintained for a period of time until the voltage at each node continues to drop by a set amount, and then the first flexible control switch is re-closed to avoid unnecessary light curtailment, and the calculation ends. According to the and instructions calculated in step S5.6 are issued, otherwise step S6.2 is executed;

[0099] S6.2. At this time, if the first flexible control switches with a breaking capacity of 1 / 3 fail to solve the problem of voltage exceeding the upper limit, then switch to uniformly tripping the second flexible control switches with a breaking capacity of 2 / 3 for each user, fairly requiring all PV users to uniformly curtail 2 / 3 of the light. If the voltage at each node returns to the normal state, it is maintained for a period of time until the voltage at each node continues to drop by a set amount, and then the second flexible control switch is re-closed to avoid unnecessary light curtailment, and the calculation ends. According to the and instructions calculated in step S5.6 are issued, otherwise step S6.3 is executed;

[0100] S6.3. At this time, if there is still a node voltage exceeding the upper limit, then both the first flexible control switch and the second flexible control switch are tripped. According to the and instructions calculated in step S5.6 are issued, and the calculation ends. Of course, this scenario generally rarely occurs.

[0101] Compared with the existing traditional switches, this method adds a group of switches, but when the voltage exceeds the upper limit, it can reduce a large amount of light curtailment compared with the configuration method of a single group of switches, and it is a more user-acceptable "flexible" control method.

[0102] Compared with the prior art, the advantages of the present invention are as follows:

[0103] By introducing a four-layer combined flexible voltage control strategy, the present invention solves the deficiencies of the prior art; firstly, a day-ahead power generation plan for distributed photovoltaic is obtained, and a coordinated control strategy integrating "day-ahead control" and "intra-day real-time regulation" is further proposed. Such a control strategy can effectively respond to emergencies encountered during the operation of photovoltaic and modify the regulation plan in a timely manner, improving the flexibility and accuracy of regulation; by introducing flexible loads and asymmetric grouping intelligent switches, making full use of existing energy storage resources and inverters, minimizing curtailment of light as much as possible, achieving fair distribution of active power based on the relative capacity and generated power of each distributed photovoltaic, and avoiding a large increase in investment in hardware equipment, with good economic efficiency.

[0104] The present invention is optimized from the perspective of ensuring the safe and stable operation of the power grid, can effectively solve the problem of over-voltage limit caused by large-scale grid connection of distributed photovoltaic, improve the power supply quality of new distribution substations, while minimizing curtailment of light as much as possible and ensuring the economic benefits of photovoltaic investors. BRIEF DESCRIPTION OF THE DRAWINGS

[0105] Figure 1 It is a step flow chart of a four-layer combined flexible voltage control method for low-voltage grid connection of distributed photovoltaic in an embodiment;

[0106] Figure 2 It is an execution flow chart of a four-layer combined flexible voltage control method for low-voltage grid connection of distributed photovoltaic in an embodiment;

[0107] Figure 3 It is a structural diagram of a low-voltage distribution network system adopted in an embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0108] The following further describes the specific implementation of the present invention in conjunction with the drawings and examples, but the implementation and protection of the present invention are not limited thereto.

[0109] Example:

[0110] A four-layer combined flexible voltage control method for low-voltage grid connection of distributed photovoltaic, as Figure 1 and Figure 2 shown, includes the following steps:

[0111] S1. In one embodiment, obtain the day-ahead planned output values of each distributed photovoltaic in the 380 / 220V low-voltage access distribution substation, the historical output data of each distributed photovoltaic, and the historical load power data of each user, and predict the photovoltaic output power, the load power of each user, and the voltage characteristic curve without flexible control. The specific steps are as follows:

[0112] S1.1. Input the 380 / 220V low-voltage access distribution substation For the pre-day planned output values of each distributed PV at and before , the historical output data of each distributed PV, and the historical load power data of each user, normalize the data and divide it into a training set, a test set, and a validation set, as follows:

[0113] S1.1.1. In one embodiment, to make full use of the existing data, divide the data into a training set, a test set, and a validation set according to a ratio of 20:4:1; among them, the training set is used to train the LSTM model; the validation set is used to verify the generalization ability of the current LSTM model during iterative training and ultimately determine whether to stop further training; the test set is used to evaluate the generalization ability of the final LSTM model to ensure the reliability of the data predicted by the LSTM model;

[0114] S1.1.2. To eliminate the dimensional differences between the data and accelerate the convergence speed of the LSTM model, normalize the data and scale the data to the range of [0,1].

[0115] S1.2. Perform variational mode decomposition (VMD) processing on the training set, the test set, and the validation set respectively to obtain 4 independent intrinsic mode functions (IMFs);

[0116] S1.3. Use the 4 corresponding intrinsic mode functions (IMFs) of the training set as the input values of the LSTM model. The output of the LSTM model is the modal components corresponding to the PV output power and the load power of each user without flexible control. Each intrinsic mode function is learned by an LSTM model;

[0117] S1.4. Use the trained LSTM model to predict the modal components corresponding to the future PV output power and the load power of each user without flexible control, and then reconstruct each modal component of the prediction result back to the original signal form to obtain the PV output data and the load power data of each user without flexible control;

[0118] S1.5. According to the predicted PV output data and the load power data of each user without flexible control, set the load and generation conditions at each time point, select the forward-backward substitution method to perform power flow calculation on the radial distribution network, and obtain the first voltage prediction value, current prediction value, and power prediction value of each node of the power grid at each time point; use the obtained data to draw the voltage characteristic curve.

[0119] S2. Calculate the voltage reference values of each node in the low-voltage distribution substation area, compare the predicted voltage characteristic curve with the voltage reference values to determine the situation of voltage exceeding the upper limit, obtain the photovoltaic surplus power based on the photovoltaic planned output value for the day ahead and the predicted photovoltaic output power, and determine the startup strategy of the storage water heater of the photovoltaic user based on the photovoltaic surplus power and the predicted voltage characteristic curve. The total maximum power consumption of the storage water heater is compared with the photovoltaic surplus electric energy within the time interval to perform the first-level control. If , then continue to execute step S3; otherwise, end the calculation. The specific steps are as follows:

[0120] S2.1. Calculate the voltage reference values of each node in the low-voltage distribution substation area as follows:

[0121] Preprocess the voltages of each node in the low-voltage distribution substation area using a sliding window algorithm; calculate the average value of the voltages of each node within the time interval . If the voltage of the -th node does not exceed the upper limit, then use the average value of the voltage of the -th node within the time interval as the first voltage reference value of the =1,2……, -th node, where is the number of nodes in the distribution network; If the voltage of the -th node exceeds the upper limit, then use the specified upper limit of voltage operation as the first voltage reference value of the

[0122] -th node; The calculation formula for the first voltage reference value

[0123] of the -th node is: where is the number of voltage samplings within the time interval ; is the -th node obtained from the power flow calculation in step S1.5 at the first voltage prediction value

[0124] S2.2. Determine whether the voltage exceeds the upper limit: Compare the first voltage prediction values of each node Compare with the first voltage reference value of each node. If the first voltage prediction value of the node is greater than the first voltage reference value, it indicates that the voltage of this node exceeds the upper limit; otherwise, it does not exceed the upper limit. If the voltage does not exceed the upper limit, execute step S2.3; if the voltage exceeds the upper limit, execute step S2.4;

[0125] S2.3. Set the water heater to start heating at 4 pm until the set temperature and end the calculation;

[0126] In one embodiment, the set temperature is set to 75 °C;

[0127] S2.4. Subtract the photovoltaic power output data predicted in step S1 from the photovoltaic daily planned power output value to obtain the photovoltaic excess power ;

[0128] S2.5. Set the start time of the storage water heater as follows:

[0129] Compare the voltage characteristic curve obtained in step S1 with the first voltage reference value to obtain the time period when the voltage exceeds the upper limit; the water heater is set to start heating during the time period when the voltage exceeds the upper limit, and in principle, it does not heat during other time periods;

[0130] S2.6. Calculate the regulation ability of the storage water heater:

[0131] Set the storage water heater to stop heating immediately after the hot water temperature reaches 75 °C, then the maximum total power consumption of the storage water heater The calculation formula is as follows:

[0132] ;

[0133] Among them, represents the specific heat capacity of water, represents the total capacity of the water to be heated in the water heater tank, represents the initial temperature of the water in the tank, represents the number of storage water heaters in the low-voltage distribution substation area;

[0134] S2.7. Compare the maximum total power consumption of the storage water heater with the photovoltaic excess electric energy within the time interval :

[0135] ;

[0136] If , that is, the photovoltaic excess electric energy is not enough to heat the water in the water heater to 75 °C, then the storage water heater still needs to consume additional electric energy Heat up; after the end of the time period when the voltage exceeds the upper limit, the storage water heater heats up continuously at a constant operating power for a period of time , The calculation formula is as follows:

[0137] ;

[0138] Among them, is the operating power when the storage water heater continues to heat up during the overvoltage end period. After setting the time required for continuous heating , the operation ends;

[0139] If , that is, the storage water heater exactly fully absorbs the surplus active power during the overvoltage period, the calculation ends and no further layer of control is required;

[0140] If , that is, the electricity consumption of the storage water heater has reached the threshold but the excess photovoltaic power has not been fully consumed, the inverter reactive power optimization regulation mode of the next layer of control needs to be started; at this time, the water heater heats up to 75°C at a constant operating power ; after the first layer of control, the excess photovoltaic power The calculation formula is as follows:

[0141] .

[0142] S3. Calculate and determine the power regulation mode of the photovoltaic power station, execute the second layer of control. If the reactive power regulation mode is entered, execute step S4, otherwise execute step S5;

[0143] The storage water heater cannot fully absorb the excess photovoltaic power. Adjust the active and reactive power of the inverter to try to reduce the voltage to the qualified range. Calculate and determine the power regulation mode of the photovoltaic power station. The specific steps are as follows:

[0144] S3.1. Based on the predicted user load power data in step S1 and the excess photovoltaic power calculated in step S2 , perform a power flow calculation again to obtain the second voltage prediction value of each node in the low-voltage distribution substation area , is the number of nodes in the distribution network, =1,2……, ; recalculate the second voltage reference value of each node according to step S2.1 ;

[0145] S3.2. According to step S2.5, it can be seen that at this time, there must be a node voltage exceeding the upper limit. Determine the second voltage prediction value calculated in step S3.1 The highest node is the leading node, and the leading node number is ; The pre-adjusted value of the leading node voltage is:

[0146] ;

[0147] S3.3. The power factor of the photovoltaic power station needs to be dynamically adjustable within the range of leading 0.95 to lagging 0.95; Considering the power factor limit of the photovoltaic power station, the th photovoltaic power station at the moment, the first maximum reactive power that can be output

[0148] is:

[0149] In the formula: is the power factor of the photovoltaic power station, and the value is 0.95; is the predicted value of the maximum active power that can be output at the th photovoltaic power station at the moment; is the number of photovoltaic power stations in the distribution network, = 1, 2,...,

[0150] S3.4. By using the non-linear least squares method to fit the relationship between the second voltage prediction value of each node and the power, and combining the real-time operation data to calculate the real-time voltage sensitivity between each node, the specific steps are as follows:

[0151] S3.4.1. Use the historical data of the distributed photovoltaic output and the historical data of the power of each user load to perform power flow calculation to obtain the historical injection power and historical voltage data of each node, and use the non-linear least squares method to fit the mapping relationship between the historical injection power and historical voltage data of each node;

[0152] S3.4.2. Collect the current voltage and power of each node, randomly generate a matrix of small changes in the injection power, define a scaling factor α, 0 < α << 1, so that it satisfies ; Among them, represents the absolute value of the current injection power of the th node, represents the absolute value of the change in the current injection power of the th node; Keep the reactive power of each node unchanged, and obtain the voltage distribution after the power change of each node through the curve obtained by fitting in step S3.4.1, and further obtain the matrix of small voltage changes;

[0153] S3.4.3. Solve using the least squares method to obtain the current voltage - active power sensitivity matrix ; is the voltage - active power sensitivity matrix the , ) - th element, representing the voltage sensitivity of the - th node to the change in active power injected into the - th node;

[0154] S3.4.4. Similarly, the voltage - reactive power sensitivity matrix of the node voltage to the change in injected reactive power can be calculated ; is the voltage - reactive power sensitivity matrix the , ) - th element, representing the voltage sensitivity of the - th node to the change in reactive power injected into the - th node.

[0155] S3.5. The reactive power can be reduced to a minimum of . Calculate the maximum adjustable amount of the dominant node voltage:

[0156] ;

[0157] In the formula: is the voltage - reactive power sensitivity matrix the , ) - th element, representing the voltage sensitivity of the - th node to the change in the grid - connected node power of the - th photovoltaic power station. The - th node is the grid - connected node of the - th photovoltaic power station; is at the - th moment, the reactive power output by the

[0158] S3.6. Calculate the photovoltaic reactive power evaluation coefficient :

[0159] ;

[0160] When , adjusting the reactive power of the power station at this time will increase the voltage amplitude of each node, and it is necessary to enter the reactive power + active power reduction mode; when the reactive power evaluation coefficient is positive, if , enter the reactive power adjustment mode, if

[0161] If entering the reactive power + active power reduction mode, execute step S5. If entering the reactive power regulation mode, execute step S4.

[0162] S4. Increase the reactive power consumption of the inverter to reduce the voltage to the qualified range. Calculate the power generation plan, and determine whether the voltage of each node meets the standard. Execute the third-layer control. If it meets the standard, issue an instruction according to the current power generation plan and end the calculation. If it does not meet the standard, enter step S6. The specific steps are as follows:

[0163] S4.1. Subtract the second voltage prediction value of the grid-connected node of each photovoltaic power station calculated in step S3.1 from the second voltage reference value, and merge them to obtain the voltage error matrix , at this time, the active power of the power station remains unchanged, so the active power change matrix of the photovoltaic power station ; Extract the voltage-reactive power sensitivity synthesis matrix between the grid-connected nodes of the photovoltaic power station , and we can get:

[0164] ;

[0165] In the formula: is the reactive power change matrix of the photovoltaic power station, and the ( , , )-th element of the voltage-reactive power sensitivity matrix represents the voltage sensitivity of the grid-connected node of the -th photovoltaic power station to the power change injected into the grid-connected node of the -th photovoltaic power station,

[0166] S4.2. After calculating the reactive power change matrix of the photovoltaic power station, the reactive power output of the -th photovoltaic power station is:

[0167] ;

[0168] In the formula: is the -th element in the reactive power change matrix of the photovoltaic power station at ;

[0169] S4.3. Add a clipping function. If exceeds the first maximum reactive power output of the photovoltaic power station, then issue the reactive power output plan with ; At this time, each power station generates active power according to the daily power generation plan, and only needs to change the reactive power output. The first reactive power output plan of the -th photovoltaic power station at for:

[0170] ;

[0171] S4.4. Calculate the power flow of the distribution network and obtain the third voltage prediction value of each node in the distribution network after each power station executes the current power generation plan. , update the third voltage reference value ; Calculate the voltage deviation and :

[0172] ;

[0173] S4.5, determine whether the voltage of each node of the distribution network meets the standard. If And the third voltage prediction value of all nodes All less than If the voltage of the distribution network node meets the standard, the command will be issued according to the current power generation plan and the calculation will be ended. Indicates the allowable voltage deviation range, which needs to be determined according to the actual situation and requirements of the power system; if the voltage of the distribution network node does not meet the standard, the photovoltaic power generation power is updated to the current power generation plan, the active power output plan remains unchanged, and the reactive power output plan is updated to the first reactive power output plan , fitting the voltage sensitivity matrix between the current nodes and , execute step S6 and enter the reactive power + active power regulation mode.

[0174] S5. At this time, the reactive power consumption of the photovoltaic inverter reaches the limit value but is still insufficient to achieve the control purpose. The inverter is adjusted to reduce the active power of photovoltaic power generation in an effort to reduce the voltage to a qualified range. The voltage of each node is judged to be up to standard. The third level of control is executed. If the standard is met, the command is issued according to the current power generation plan and the calculation ends. If the standard is not met, the process proceeds to step S6.

[0175] At this point, the reactive power consumption of the PV inverter has reached the limit, but it is still insufficient to achieve the control purpose. The inverter is adjusted to reduce the active power of PV power generation in an effort to reduce the voltage of each node in the low-voltage distribution station area to an acceptable range. The specific steps are as follows:

[0176] S5.1. If the photovoltaic active power is limited, the maximum adjustable reactive power will change accordingly; calculate the photovoltaic power stations The second maximum output reactive power at the moment :

[0177] ;

[0178] S5.2, press Output reactive power to maximize the utilization of the PV power station capacity, and the change in output reactive power is as follows:

[0179]

[0180] S5.3. Define fairness as the equal - proportion distribution of active power curtailment among PV power stations according to their maximum power generation. Therefore, each PV power station is set with the same active power curtailment ratio , and perform combined active - reactive power control;

[0181] The active power curtailment ratio is specifically as follows:

[0182] ;

[0183] In the formula, is the predicted value of the maximum available active power output of the th PV power station at time , and is the active power curtailment of the

[0184] ；

[0185] th PV power station, specifically as follows: is the planned active power output value of the th PV power station at time

[0186] that needs to be calculated subsequently;

[0187] ;

[0188] In the formula, is an unknown variable that varies between [0, 1), and the rest are all constants. After simplification, a quadratic equation of can be obtained, and has a unique solution within the variation range;

[0189] S5.4. After obtaining , the active power output plan and the second reactive power output plan of each PV power station are specifically as follows:

[0190] ；

[0191] S5.5. Perform power flow calculation on the distribution network to obtain the fourth voltage prediction value of each node in the distribution network after each PV power station executes according to the current power generation plan ;

[0192] S5.6. Determine whether the voltage of each node meets the standard; in the active power reduction mode, if the fourth voltage prediction value of each node is less than then the voltage of the distribution network nodes meets the standard, issue an instruction according to the current power generation plan, and end the calculation; if the voltage of the distribution network nodes does not meet the standard, then update the photovoltaic power generation power to the current power generation plan, update the active power output plan to , update the reactive power output plan to the second reactive power output plan , fit the current voltage sensitivity matrix between each node and , execute step S6, and enter the control of the next layer of asymmetric grouped intelligent switches.

[0193] S6. Execute the fourth layer of control, and forcefully reduce the photovoltaic active power by tripping the asymmetric grouped intelligent switches until the voltage drops below the allowable upper limit;

[0194] At this time, the control of the first three layers has not achieved the goal, adopt the strategy of partially cutting off the power of photovoltaic power generation, and forcefully reduce the voltage rise caused by the reverse power transmission of photovoltaic active power by tripping the externally connected asymmetric grouped intelligent switches, as follows:

[0195] Most traditional intelligent switches are only equipped with one switch, and when the voltage is too high, they can only reduce the voltage by means of 100% light abandonment, resulting in a large waste of light energy. The present invention proposes to configure a flexible control switch terminal in front of each photovoltaic user meter, configure 2 groups of switches, one group with a breaking capacity of 1 / 3 and the other group with a breaking capacity of 2 / 3, and start light abandonment when the control effects of the water storage type electric water heater and the photovoltaic inverter in the control of the first three layers have not reduced the substation area voltage below the allowable upper limit.

[0196] The specific steps of step S6 are as follows:

[0197] S6.1. First, the edge intelligent terminal issues an instruction to the flexible control switch terminals of each user end, and uniformly trips the first flexible control switches with a breaking capacity of 1 / 3 of each user, fairly requiring all photovoltaic users to uniformly abandon 1 / 3 of the light. If the voltage of each node returns to the normal state, maintain it for a period of time until the voltage of each node continues to drop by a set amount and then close the first flexible control switch again to avoid unnecessary light abandonment, end the calculation, and issue an instruction according to and calculated in step S5.6, otherwise execute step S6.2;

[0198] S6.2. If the first flexible control switch with a breaking capacity of 1 / 3 fails to solve the problem of voltage exceeding the upper limit at this time, then replace it with the second flexible control switch that uniformly trips each user with a breaking capacity of 2 / 3, and fairly require all PV users to uniformly abandon 2 / 3 of the light. If the voltage at each node returns to the normal state, maintain it for a period of time until the voltage at each node continues to drop by a set amount, and then reconnect the second flexible control switch to avoid unnecessary light abandonment, end the calculation, and issue the and issued instructions, otherwise execute step S6.3;

[0199] S6.3. If there is still a node voltage exceeding the upper limit at this time, then both the first flexible control switch and the second flexible control switch are tripped, and issue the and issued instructions to end the calculation. Of course, this scenario generally rarely occurs.

[0200] Compared with the existing traditional switches, this method adds a group of switches, but it can reduce a large amount of light abandonment compared with the configuration method of a single group of switches when the voltage exceeds the upper limit, and it is a more user-acceptable "flexible" control method.

[0201] In one embodiment, a power system network architecture as shown in Figure 3 is used for simulation analysis. The low-voltage distribution network includes 14 nodes (9 load nodes and 5 power supply nodes), 20 branches, and 3 transformers. Among them, the first node 1, the second node 2, and the third node 3 are connected to PV generating sets, corresponding to the first PV power station 1, the second PV power station 2, and the third PV power station 3 respectively. At the same time, to simulate the impact of other power generation methods on the power system, a thermal power generating set and a hydropower generating set are connected to the sixth node 6 and the eighth node 8 respectively, and the remaining nodes are all load nodes. The system base capacity is 150 MVA, and the node types include PV nodes and PQ nodes.

[0202] Set three low-voltage distribution network operation scenarios. In scenario 1, the actual value of PV power generation is 1.1 times the day-ahead prediction value, and the actual value of the load is 0.9 times the day-ahead prediction value; in scenario 2, the actual value of PV power generation is 1.3 times the day-ahead prediction value, and the actual value of the load is 0.8 times the day-ahead prediction value; the actual value of PV power generation is 1.4 times the day-ahead prediction value, and the actual value of the load is 0.7 times the day-ahead prediction value. All three scenarios are verified by simulation at 13:00 noon.

[0203] For Scenario 1, since neither the actual PV output nor the load exceeds the predicted values by much, the timed regulation mode of the water storage type electric water heater can be used to control the voltage within an appropriate range. Therefore, the reactive power regulation mode and the active power curtailment mode do not need to be considered. Two solutions are set for simulation comparison, namely, the PV power station shuts down when the voltage exceeds the upper limit and only the timed regulation of the water storage type electric water heater. The specific implementation of the two solutions is as follows:

[0204] Solution 1: Take extreme measures. During the peak power generation period (from 10:00 am to 15:00 pm), the overvoltage protection acts and the power station shuts down to ensure that the grid connection point voltage is within the safe range. The power station attempts to restart after 1 minute. If the voltage still exceeds the upper limit, it continues to shut down.

[0205] Solution 2: Adopt the timed regulation mode of the water storage type electric water heater. According to the overall operation data within the PRMC of the distribution network, formulate a timed regulation plan for the water storage type electric water heater and issue instructions to the control center of the water storage type electric water heater for execution.

[0206] The voltage quality evaluation indicators of each control scheme are shown in the following table. Among them, r Q represents the voltage qualification rate, D D represents the voltage discrete degree, and Watt% and Var% respectively represent the ratios of the total active power and reactive power injected by the PV power station in the distribution network to the maximum available active power and reactive power. Comparing the evaluation indicators under the two control schemes with the results without control, the voltage qualification rate reaches 100%, and the voltage discrete degree decreases. Compared with Solution 1, for Solution 2, the actual total output of the PV power station is 100%, the total reactive power output of the power station is 25.42%, and the voltage discrete degree of the distribution network is reduced by 11.31%, significantly improving the voltage quality. The example results show that when only a small part of the PV output is excessive, adopting the timed regulation mode of the water storage type electric water heater can better solve the voltage quality problem and ensure the safe and stable operation of the system.

[0207] Table 1 Voltage Quality Evaluation Table for Scenario 1

[0208]

[0209] In Scenario 2, the PV output is excessive to a certain extent, and the situation of the voltage exceeding the upper limit is more serious than that in Scenario 1. Three solutions are set for simulation comparison, namely, the PV power station shuts down when the voltage exceeds the upper limit, only the timed regulation of the water storage type electric water heater, and the timed regulation of the water storage type electric water heater + reactive power regulation mode. The specific implementation of the three solutions is as follows:

[0210] Solution 3: When the voltage at the grid connection node of the PV power station exceeds the upper limit, the power station shuts down. The voltages at the grid connection nodes of the first PV power station 1, the second PV power station 2, and the third PV power station 3 did not exceed the upper limit during the simulation period, and all generated electricity at the maximum power.

[0211] Solution 4: Only implement the timed adjustment mode of the storage water heater. Due to limited adjustment ability, the storage water heater heats at a certain power at each moment and cannot consume all the remaining PV capacity. Calculate the voltage distribution after the water heater consumption and compare it with other solutions.

[0212] Solution 5: Implement the combined operation mode of timed adjustment and reactive power adjustment of the storage water heater. According to the previous method, when the ability of the water heater to consume PV reaches the limit, start the reactive power adjustment mode, re - formulate the PV power generation plan, and issue commands to the PV inverter control center for execution, thereby adjusting the system reactive power.

[0213] The following table lists the voltage quality and power distribution evaluation indicators of the distribution network under different control solutions in Scenario 2. Under Solution 3, when the voltage exceeds the upper limit, the PV power station shuts down, and the PV curtailment rate is as high as 43.87%, with too low economic benefits and is not considered. Compared with Solution 4, in Solution 5, not only does the active power output of the PV reach 100%, but there is no voltage exceeding the upper limit throughout the period, the reactive power output exceeds 50%, and the voltage deviation degree is reduced by 18.68%, meeting the basic requirements of voltage quality and achieving full consumption of PV output. The example results show that when there is a certain degree of excess PV output, only using the timed adjustment of the storage water heater cannot fully consume it. At this time, a reactive power adjustment mode can be added to achieve full consumption of PV output within the reactive power adjustment range of the inverter and ensure voltage quality.

[0214] Table 2 Voltage Quality Evaluation Table for Scenario 2

[0215]

[0216] In Scenario 3, there is a large excess of PV output, and the voltage exceeding the upper limit situation is relatively serious. A total of three solutions are set for simulation comparison, namely, the PV power station shuts down when the voltage exceeds the upper limit, the timed adjustment of the storage water heater + reactive power adjustment mode, and the timed adjustment of the storage water heater + reactive power adjustment mode + active power curtailment mode. The specific implementation of the three solutions is as follows:

[0217] Solution 6: When the grid connection node of the PV power station exceeds the upper limit, the power station shuts down. The grid connection nodes of the first PV power station 1, the second PV power station 2, and the third PV power station 3 did not exceed the upper limit during the simulation period, and all generated electricity at the maximum power.

[0218] Scenario 7: Implement an operation mode that combines the timed regulation of a storage water heater with reactive power regulation. According to the method described above, when the ability of the water heater to absorb photovoltaic power reaches its limit, activate the reactive power regulation mode, re - formulate the photovoltaic power generation plan, and issue an order to the photovoltaic inverter control center for execution, thereby regulating the reactive power of the system. Since the reactive power regulation ability of the inverter is limited, calculate the voltage distribution after the reactive power regulation ability reaches the maximum regulation margin and compare it with other scenarios.

[0219] Scenario 8: Implement an operation mode that combines the timed regulation of a storage water heater with reactive power regulation and active power curtailment. When the ability of the inverter to regulate reactive power reaches its limit, activate the active power curtailment mode, re - formulate the photovoltaic power generation plan, and issue an order to the photovoltaic inverter control center for execution, and achieve the voltage quality requirements by curtailing some active power.

[0220] The following table lists the voltage quality and power distribution evaluation indicators of the distribution network under different control scenarios in Scenario 3. Under Scenario 6, when the voltage exceeds the upper limit, the photovoltaic power station shuts down, and the curtailment rate of the photovoltaic power station is as high as 49.89%, and the economic benefit is too low to be considered. Compared with Scenario 7, although Scenario 8 fails to fully absorb photovoltaic power, it ensures the voltage quality, reduces the voltage deviation by 26.42%, and better reduces the voltage fluctuation risk. The calculation example results show that the proposed power optimization strategy makes full use of the reactive power regulation function of the power station. After the reactive power regulation amount of the photovoltaic power station reaches the threshold, it coordinates the control of the active and reactive power of the power station and improves the voltage quality of the photovoltaic grid - connection point.

[0221] Table 3 Voltage Quality Evaluation Table for Scenario 3

[0222]

[0223] Obviously, the above - mentioned embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limiting the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation manners here. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. A four-layer combined flexible voltage control method for distributed photovoltaic low-voltage grid connection, characterized in that, It includes the following steps: S1. Obtain the daily planned output values of each distributed photovoltaic (PV) in the low-voltage access distribution substation area, the historical PV output data, and the historical load power data of each user, and predict the PV output power, the load power of each user, and the voltage characteristic curve without flexible control: According to the predicted PV output data and the load power data of each user without flexible control, set the load and generation conditions at each time point, select the forward-backward substitution method to perform power flow calculation on the radial distribution network, and obtain the first voltage prediction value, current prediction value, and power prediction value of each node in the power grid at each time point; Use the obtained data to draw the voltage characteristic curve; S2. Calculate the voltage reference values of each node in the low-voltage distribution substation area, compare the predicted voltage characteristic curve with the voltage reference values to judge the situation of voltage exceeding the upper limit. Based on the photovoltaic planned output value for the day ahead and the predicted photovoltaic output power, obtain the photovoltaic excess power. And based on the photovoltaic excess power and the predicted voltage characteristic curve, determine the startup strategy of the storage water heater of photovoltaic users, and take the maximum total power consumption of the storage water heater and the time interval to compare with the photovoltaic excess electric energy during this period, and execute the first-level control to obtain the photovoltaic excess power : If , then continue to execute step S3, otherwise end the calculation; The calculation of the voltage reference value of each node in the low-voltage distribution substation area is as follows: Preprocess the voltages of each node in the low-voltage distribution substation using a sliding window algorithm; calculate the time interval The average value of the voltages of each node within, if the voltage of the th node does not exceed the upper limit, then use the time interval The average value of the voltage of the th node within as the first voltage reference value of the th node , is the number of nodes in the distribution network, =1,2……, ; If the voltage of the th node exceeds the upper limit, the specified upper limit of the operating voltage is used as the first voltage reference value of the th node; the calculation formula for the first voltage reference value of the th node is as follows: Among them, is the number of voltage samplings within the time interval ; is the historical measured value of the voltage of the th node within the time interval obtained from the power flow calculation in step S1.5 ; is the first voltage prediction value of the th node at the moment; is the specified upper limit of voltage operation; S3. Calculate and determine the PV power plant power regulation mode. The PV power plant power regulation mode includes a reactive power regulation mode and a reactive power + active power reduction mode, and perform the second-layer control: If entering the reactive power + active power reduction mode, execute step S5; if entering the reactive power regulation mode, execute step S4; Among them, the water storage type electric water heater cannot fully absorb the excess PV electric energy. Adjust the active and reactive power of the inverter to try to reduce the voltage to the qualified range, and calculate and determine the PV power plant power regulation mode. The specific steps are as follows: S3.

1. Based on the user load power data predicted in step S1 and the PV excess power after the first - layer control calculated in step S2 , perform a power flow calculation again to obtain the second - voltage prediction values of each node in the low - voltage distribution sub - station area , where \(n\) is the number of nodes in the distribution network =1,2……, ; recalculate the second - voltage reference value of each node according to step S2 ; S3.

2. Determine the second voltage prediction value calculated in step S3.1 The node with the highest is the leading node, and the leading node number is ; The pre-adjustment value of the leading node voltage ； S3.

3. The power factor of the PV power station is dynamically adjustable within the range from leading 0.95 to lagging 0.95; Considering the power factor limitation of the PV power station, the first maximum reactive power output for the th PV power station at time is: Wherein: is the power factor of the PV power station, with a value of 0.95; is the th predicted value of the maximum available active power output at the th moment of the PV power station; is the number of PV power stations in the distribution network, = 1, 2,..., S3.

4. Fit the relationship between the predicted second voltage values of each node and the power by using the non - linear least - squares method, and calculate the real - time voltage sensitivity between each node in combination with the real - time operation data; ​ S3.

5. The reactive power can be reduced to as low as , and calculate the maximum adjustable amount of the leading node voltage : Wherein: is the voltage-reactive power sensitivity matrix of the , )-th element, indicating the voltage sensitivity of the -th node to the change in the grid-connected node power of the -th photovoltaic power station. The -th node is the grid-connected node of the -th photovoltaic power station; is the reactive power output by the -th photovoltaic power station at time ; S3.

6. Calculate the photovoltaic reactive power evaluation coefficient of the distribution network : At this time, adjusting the reactive power of the power station will increase the voltage amplitude of each node, and it is necessary to enter the reactive power + active power reduction mode; when the reactive power evaluation coefficient is positive, if , enter the reactive power regulation mode, if , then enter the reactive power + active power reduction mode; If entering the reactive power + active power reduction mode, execute step S5; if entering the reactive power regulation mode, execute step S4; S4. Increase the reactive power consumption of the inverter to reduce the voltage to the qualified range, calculate the power generation plan, and judge whether the voltage of each node meets the standard, and perform the third-layer control: If it meets the standard, issue an instruction according to the current power generation plan and end the calculation; if it does not meet the standard, enter step S6; S5. At this time, the reactive power consumption of the PV inverter reaches the limit value and is still insufficient to achieve the control purpose. Adjust the inverter to reduce the active power of PV power generation to try to reduce the voltage to the qualified range, and judge whether the voltage of each node meets the standard, and perform the third-layer control: If it meets the standard, issue an instruction according to the current power generation plan and end the calculation; if it does not meet the standard, enter step S6; S6. Perform the fourth-layer control: Forcefully reduce the active power of PV by tripping the asymmetric grouped intelligent switch until the voltage drops below the allowable upper limit value.

2. The distributed photovoltaic low-voltage grid-connected four-layer combined flexible voltage control method according to claim 1, wherein, In step S1, the specific steps are as follows: S1.

1. Input the low-voltage access distribution transformer area For the planned output values of each distributed PV for the day before and at the time , the historical output data of each distributed PV, and the historical load power data of each user, normalize the data and divide it into a training set, a test set, and a validation set; S1.

2. Perform variational mode decomposition (VMD) processing on the training set, test set, and validation set respectively to obtain 4 independent intrinsic mode components; S1.

3. Use the 4 corresponding intrinsic mode components of the training set as the input values of the long short-term memory (LSTM) model. The output of the LSTM model is the modal components corresponding to the PV output power and the load power of each user without flexible control. Each intrinsic mode component is learned by an LSTM model; S1.

4. Use the trained LSTM model to predict the modal components corresponding to the future PV output power and the load power of each user without flexible control, and then reconstruct each modal component of the prediction result into the original signal form to obtain the PV output data and the load power data of each user without flexible control; S1.

5. Set the load and generation conditions at each time point according to the predicted PV output data without flexible control and the load power data of each user, select the forward-backward substitution method to perform power flow calculation on the radial distribution network, and obtain the first voltage prediction value, current prediction value and power prediction value of each node in the power grid at each time point; use the obtained data to draw the voltage characteristic curve.

3. A four - layer combined flexible voltage control method for distributed photovoltaic low - voltage grid connection according to claim 2, characterized in that, In step S1.1, it is specifically as follows: S1.1.

1. Divide the data into a training set, a test set and a validation set according to a set ratio; among them, the training set is used to train the LSTM model; the validation set is used to verify the generalization ability of the current LSTM model during iterative training and finally determine whether to stop further training; the test set is used to evaluate the generalization ability of the final LSTM model to ensure the reliability of the data predicted by the LSTM model. S1.1.

2. Perform normalization processing on the data and scale the data to the range of [0, 1].

4. A four - layer combined flexible voltage control method for distributed photovoltaic low - voltage grid connection according to claim 1, characterized in that, In step S2, the specific steps are as follows: S2.

1. Calculate the voltage reference value of each node in the low-voltage distribution area. S2.

2. Determine whether the voltage exceeds the upper limit: Compare the first voltage prediction value of each node with the first voltage reference value of each node. If the first voltage prediction value of a node is greater than the first voltage reference value, it indicates that the voltage of this node exceeds the upper limit; otherwise, it does not exceed the upper limit. If the voltage does not exceed the upper limit, execute step S2.3; if the voltage exceeds the upper limit, execute step S2.4; S2.

3. Set the water heater to start heating at 4:00 p.m. until the set temperature is reached , and end the calculation; S2.

4. Subtract the PV power output value in the PV daily plan from the PV power output data predicted in step S1 to obtain the PV excess power ; S2.

5. Set the start time of the storage water heater, specifically as follows: Compare the voltage characteristic curve obtained in step S1 with the first voltage reference value to obtain the time period when the voltage exceeds the upper limit; the water heater is set to start heating during the time period when the voltage exceeds the upper limit, and in principle, it does not heat during other time periods. S2.

6. Calculate the regulation ability of the storage water heater. Set the storage water heater to stop heating immediately after the hot water temperature reaches the set temperature Then the maximum total power consumption of the storage water heater The calculation formula is as follows: Among them, represents the specific heat capacity of water, represents the total capacity of the water to be heated in the water heater tank, represents the initial temperature of the water in the tank, represents the number of storage water heaters in the low-voltage distribution area; S2.

7. Compare the total maximum power consumption of the storage water heater with the excess photovoltaic power within the time interval as follows: ； If , that is, the surplus photovoltaic power is not enough to heat the water in the water heater to the set temperature , then the storage water heater still needs to consume additional power for heating; after the end of the time period when the voltage is above the upper limit, the storage water heater continues to heat at a constant operating power for a period of time , The calculation formula of Among them, is the operating power when the storage water heater continues to heat during the overvoltage end period. After setting the time required for continuous heating the operation ends; If , that is, if the storage water heater exactly and completely absorbs the surplus active power during the overvoltage period, the calculation ends and there is no need to perform the next-level control; If , that is, when the electricity consumption of the storage water heater has reached the threshold but the excess PV power has not been fully consumed, the inverter reactive power optimization regulation mode of the next layer of control needs to be activated; at this time, the water heater heats up to the set temperature at a constant operating power ; after the first layer of control, the excess PV power is calculated as follows: 。 5. A four-layer combined flexible voltage control method for distributed photovoltaic low-voltage grid connection according to claim 1, characterized in that, The specific steps of step S3.4 are as follows: S3.4.

1. Use the historical data of each distributed PV output and the historical data of each user load power to perform power flow calculation to obtain the historical injection power and historical voltage data of each node, and use the nonlinear least squares method to fit the mapping relationship between the historical injection power and historical voltage data of each node. S3.4.

2. Collect the current voltages and powers of each node, and randomly generate a matrix of small changes in the injection power , define a scaling factor , 0 < << 1, such that ; where represents the absolute value of the current injection power of the -th node, represents the absolute value of the change in the current injection power of the -th node; keep the reactive power of each node unchanged, and obtain the voltage distribution after the power change of each node through the curve obtained by fitting in step S3.4.1, and further obtain the matrix of small voltage changes ; S3.4.

3. Solve using the least squares method to obtain the current voltage - active power sensitivity matrix ; is the voltage - active power sensitivity matrix The , ) - th element of represents the voltage sensitivity of the -th node to the change in active power injected into the -th node; S3.4.

4. Similarly, calculate the voltage-reactive power sensitivity matrix of the nodal voltage with respect to the change in the injected reactive power. ; Let it be the voltage-reactive power sensitivity matrix . The element at the ( , ) position represents the voltage sensitivity of the -th node to the change in the reactive power injected into the -th node.

6. A four - layer combined flexible voltage control method for distributed photovoltaic low - voltage grid connection according to claim 5, characterized in that, In step S4, increase the reactive power consumption of the inverter to reduce the voltage to the qualified range, and calculate the power generation plan. The specific steps are as follows: S4.

1. Subtract the second voltage prediction value of the grid connection node of each photovoltaic power station obtained in step S3.1 from the second voltage reference value, and combine them to obtain a voltage error matrix , at this time, the active power of the power station remains unchanged, so the active power change matrix of the photovoltaic power station ; Extract the voltage-reactive power sensitivity synthesis matrix between the grid connection nodes of the photovoltaic power station , and obtain: In the formula: is the reactive power change matrix of the PV power station, the voltage-reactive power sensitivity matrix The (( , )) element, indicating the voltage sensitivity of the grid connection node of the th PV power station to the power change of the grid connection node of the th PV power station, is the voltage error matrix of the grid connection node; S4.

2. After obtaining the reactive power change matrix of the PV power station, the reactive power output of the th PV power station is as follows: In the formula: is the reactive power change matrix of the PV power station at the moment in the th element; S4.

3. Add a clipping function. If exceeds the first maximum reactive power output of the PV power station , then issue a reactive power output plan; At this time, each power station generates active power according to the daily power generation plan, and only needs to change the reactive power output. The th PV power station 's first reactive power output plan at is: ； S4.

4. Perform power flow calculation on the distribution network to obtain the predicted third voltage values of each node in the distribution network after each power station executes according to the current power generation plan. , update the third voltage reference value ; Calculate the voltage deviation of the distribution network and :[[]]END]] ； S4.

5. Determine whether the voltage of each node in the distribution network meets the standard. If and the third voltage prediction value of all nodes are all less than then the voltage of the distribution network nodes meets the standard, and an instruction is issued according to the current power generation plan to end the calculation. represents the allowable voltage deviation range; if the voltage of the distribution network nodes does not meet the standard, update the photovoltaic power generation to the current power generation plan, keep the active power output plan unchanged, and update the reactive power output plan to the first reactive power output plan. , fit the current voltage sensitivity matrix between each node and , execute step S6, and enter the reactive + active power reduction mode.

7. A four-layer combined flexible voltage control method for distributed photovoltaic low-voltage grid connection according to claim 6, characterized in that In step S5, when the reactive power consumption of the PV inverter reaches the limit value and is still insufficient to achieve the control purpose, adjust the inverter to reduce the active power generation of the PV to strive to reduce the voltage of each node in the low-voltage distribution area to the qualified range. The specific steps are as follows: S5.

1. If the active power of the PV is restricted, the maximum adjustable reactive power will change accordingly; calculate the second maximum reactive power that can be output at the th PV power station at the moment: S5.

2. According to the output reactive power to maximize the utilization of the PV power station capacity, the change in the output reactive power is as follows: S5.

3. Define fairness as the equal - proportion distribution of active power curtailment among PV power stations according to their maximum power generation. Therefore, each PV power station is set with the same active power curtailment ratio , and perform coordinated active - reactive power control; Active power reduction ratio The details are as follows: In the formula, is the predicted value of the maximum available active power output at the th photovoltaic power station, and is the active power reduction of the th photovoltaic power station, which is specifically as follows: In the formula, is the planned active power output value of the th photovoltaic power station at the subsequent moment to be calculated; The specific pre-adjustment value of the voltage of the main node is as follows: In the formula, is an unknown quantity that varies between [0, 1), are all constants. As can be seen from S5.1 and S5.2 can be expressed by an expression containing . After simplification, a quadratic equation of one variable can be obtained, and has a unique solution within the range of variation; S5.

4. Calculate After that, the active power output plan and the second reactive power output plan are as follows: S5.

5. Perform a power flow calculation on the distribution network to obtain the fourth voltage prediction values of each node in the distribution network after each PV power station executes according to the current power generation plan. ; S5.

6. Determine whether the voltage of each node meets the standard; in the active power reduction mode, if the fourth voltage prediction value of each node is less than then the voltage of the distribution network nodes meets the standard, issue an instruction according to the current power generation plan, and end the calculation; if the voltage of the distribution network nodes does not meet the standard, update the photovoltaic power generation to the current power generation plan, update the active power output plan to , update the reactive power output plan to the second reactive power output plan , and issue an instruction according to the calculated and ; fit the current voltage sensitivity matrix between each node and , execute step S6, and enter the next layer of asymmetric grouping intelligent switch control.

8. A four - layer combined flexible voltage control method for distributed photovoltaic low - voltage grid connection according to claim 7, characterized in that, In step S6, when the first three layers of control do not achieve the purpose, adopt the strategy of partial PV power cut-off, and forcefully reduce the voltage rise caused by the reverse power transmission of PV active power by tripping the external asymmetric grouping intelligent switch. Specifically as follows: Configure a flexible control switch terminal for each PV user before the meter, and configure 2 groups of switches, one group with a breaking capacity of 1 / 3 and the other group with a breaking capacity of 2 / 3; when the control effects of the storage water heater and the PV inverter in the first three layers of control have not reduced the voltage of the distribution area below the allowable upper limit, start light curtailment.

9. A distributed photovoltaic low-voltage grid-connected four-layer combined flexible voltage control method according to claim 7 or 8, characterized in that The specific steps of step S6 are as follows: S6.

1. First, the edge intelligent terminal issues instructions to the flexible control switch terminals of each client, uniformly tripping the first flexible control switches of each user with a breaking capacity of 1 / 3, fairly requiring all PV users to uniformly discard 1 / 3 of the light. If the voltage of each node returns to the normal state, maintain it for a period of time until the voltage of each node continues to drop by a set amount, then reconnect the first flexible control switch to avoid unnecessary light curtailment, end the calculation, and issue instructions according to and calculated in step S5.

6. Otherwise, execute step S6.2; S6.

2. If the first flexible control switch with a breaking capacity of 1 / 3 fails to solve the problem of voltage exceeding the upper limit at this time, then replace it with the second flexible control switch that uniformly trips all users with a breaking capacity of 2 / 3, and fairly require all PV users to uniformly discard 2 / 3 of the light; if the voltage of each node returns to the normal state, then maintain it for a period of time until the voltage of each node continues to drop by a set amount, and then reconnect the second flexible control switch to avoid unnecessary light curtailment, end the calculation, and issue the instructions obtained according to the calculation in step S5.6 and otherwise, execute step S6.3; S6.

3. If there is still a node voltage exceeding the upper limit at this time, both the first flexible control switch and the second flexible control switch trip, and the calculation ends.

Images