Power distribution network reactive power compensation control method and system based on multi-source data analysis

Abstract

The application discloses a power distribution network reactive power compensation control method and system based on multi-source data analysis, and relates to the technical field of power distribution network reactive power compensation. Through obtaining multi-source data of a to-be-controlled console area at the same moment within a monitoring period, the operation state of the current reactive power compensation mechanism is classified by using an operation state judgment function value, a recurrent neural network model taking pretreated time series data as input is constructed, the historical data features of the influence factor data and the real-time data features of the real-time data are obtained by using the trained neural network model, the operation state of the current reactive power compensation mechanism is determined by performing feature matching, the instantaneous power and compensation capacity of the reactive power compensation mechanism under the current operation state and the difference between the instantaneous power and the instantaneous power sampling value under the current compensation capacity are respectively calculated, and the power distribution network reactive power compensation adjustment strategy is determined according to the comparison result of the difference calculation result and the given value of the reactive power under the current voltage.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of reactive power compensation of power distribution network, and particularly relates to a reactive power compensation control method and system of power distribution network based on multi-source data analysis. BACKGROUND

[0002] With the current distribution network distributed power installation types being various, the power installation structure proportion of the power source mainly including photovoltaic, supplemented by energy storage, hydroelectricity, wind power and biomass power generation is increasing, and the complexity of the distribution network system is also increasing, which puts forward higher requirements for grid reactive power regulation and power quality control.

[0003] Due to the increasing proportion of distributed photovoltaic access in local areas, the grid load sent down is greatly reduced, which causes great changes in the load characteristics of many distribution network lines, and the fluctuation of the operation characteristics of the distribution network is strong. The traditional reactive power compensation control strategy of the distribution network does not dynamically match the operation characteristics of the reactive power compensation mechanism under high-frequency variable working conditions. During the operation of the reactive power compensation mechanism of the distribution network, the power output may be continuously over-compensated, step-up and step-down, which is difficult to realize the coordinated control of the reactive power compensation of the distribution network and the operation power, causes the deviation between the control result and the required target, and has low control accuracy and intelligent degree, which cannot meet the increasingly complex distribution network scheduling requirements. Therefore, the present application provides a reactive power compensation control method and system of power distribution network based on multi-source data analysis. SUMMARY

[0004] The main purpose of the present application is to provide a reactive power compensation control method and system of power distribution network based on multi-source data analysis, which can effectively solve the problems in the background art.

[0005] To achieve the above purpose, the technical scheme adopted by the present application is,

[0006] The reactive power compensation control method of power distribution network based on multi-source data analysis comprises the following steps:

[0007] Step one: obtaining the node operation data, main user side voltage data and power factor λ of the to-be-controlled station area at the same time t in the monitoring period T t , constructing operation state judgment functions f1(λ t ) and f2(λ t ) according to the obtained power factor λ t , and classifying the operation state of the reactive power compensation mechanism in the period from t to t+1 in the current monitoring period T by using the function value;

[0008] The expression of the operation state judgment function f1(λ t ) is:

[0009]

[0010] The expression of the running state judgment function f2(λ t ) is as follows:

[0011]

[0012] In the formula, t, t+1∈T; N is the number of data sampling in the monitoring period T;

[0013] The running state includes:

[0014] A first running state in which the power factor is steadily increasing;

[0015] A second running state in which the power factor is steadily decreasing;

[0016] A third running state in which the power factor remains unchanged;

[0017] A fourth running state in which the power factor is fluctuantly increasing;

[0018] A fifth running state in which the power factor is fluctuantly decreasing.

[0019] The classification principle of the running state is:

[0020] When f1(λ t )=-1 and f2(λ t )=1, the running state is the first running state in which the power factor is steadily increasing;

[0021] When f1(λ t )=1 and f2(λ t )=1, the running state is the second running state in which the power factor is steadily decreasing;

[0022] When f1(λ t )=0, the running state is the third running state in which the power factor remains unchanged;

[0023] When f1(λ t )=-1 and f2(λ t )=-1, the running state is the fourth running state in which the power factor is fluctuantly increasing;

[0024] When f1(λ t )=1 and f2(λ t )=-1, the running state is the fifth running state in which the power factor is fluctuantly decreasing.

[0025] Step two: taking the node running data and the main user-side voltage data as influencing factors, performing correlation analysis on the influencing factors and the power factor of the reactive power compensation mechanism, and obtaining the correlation coefficient r of the ith influencing factor and the power factor of the reactive power compensation mechanism under different running statesi ;

[0026] Step 3: Extract the time series data of various influencing factors of the reactive compensation mechanism under different operating conditions for preprocessing, build a recurrent neural network model with the preprocessed time series data as input, train the neural network model, and use the trained neural network model to obtain the historical data characteristics of the influencing factor data It is expressed as the jth historical data feature of the i-th influencing factor under the k-th operating state;

[0027] Among them, the output layer of the trained neural network model is removed and the last hidden layer is retained to obtain the historical data characteristics of the i-th influencing factor data.

[0028] Step 4: Obtain the real-time data of the influencing factors and use the neural network model to extract the real-time data features x of the real-time data ij , x ij The jth real-time data feature of the i-th influencing factor is represented, the real-time data feature is matched with the historical data feature, and the current operating state of the reactive power compensation mechanism is determined according to the matching result;

[0029] The determination process includes:

[0030] Step S41: Calculate the real-time data features x respectively ij and the historical data characteristics of the kth operating state Feature matching between The calculation formula is:

[0031] Step S42: Get the matching degree of each feature The maximum value in the calculation result in, k=1,2,...,K; K is the type of operating state;

[0032] Step S43: Taking the maximum value of feature matching The corresponding operating state is used as the current operating state of the reactive power compensation mechanism.

[0033] Step 5: Calculate the instantaneous power s and compensation capacity Q of the reactive compensation mechanism in the current operating state respectively c , calculate the difference △s between the instantaneous power under the compensation capacity in the current operating state and the instantaneous power sampling value s', where △s = s-s', compare the calculated difference △s with the reactive power set value Q under the current voltage, and determine the distribution network reactive compensation adjustment strategy based on the comparison result;

[0034] The principle for determining the reactive power set value Q is:

[0035]

[0036] Q max is the maximum value of the reactive power output of the reactive power compensation mechanism; U1, U2, U3 and U4 are all voltage unit values, and U1

[0037] compensation capacity Q c is calculated by the following formula:

[0038]

[0039] In the formula, β is the average load rate of the transformer area in the monitoring period T; P max is the average power sum of the electrical equipment in the to-be-controlled transformer area; is the load power factor angle before reactive power compensation; is the load power factor angle after reactive power compensation.

[0040] The determination principle of the reactive power compensation adjustment strategy of the power distribution network is:

[0041] If Δs>Q, the reactive power compensation power of the power distribution network is adjusted in the direction of reduction;

[0042] If ΔsQ, the reactive power compensation power of the power distribution network is adjusted in the direction of increase.

[0043] The reactive power compensation control system of the power distribution network based on multi-source data analysis comprises:

[0044] a data acquisition module for acquiring node operation data, main user-side voltage data, power factor λ t of the to-be-controlled transformer area at the same time t in the monitoring period T;

[0045] an operation state classification module for constructing operation state judgment functions f1(λ t ) and f2(λ t ) according to the acquired power factor λ t and classifying the operation state of the reactive power compensation mechanism in the period from t to t+1 in the current monitoring period T by using the function values;

[0046] a data correlation analysis module for acquiring correlation coefficients r i between the i-th influencing factor and the power factor of the reactive power compensation mechanism under different operation states;

[0047] a feature extraction module for acquiring historical data features and real-time data features x ij of the influencing factor data;

[0048] a running state determination module for determining the current running state of the reactive power compensation mechanism according to the matching result;

[0049] a current state quantity acquisition module for calculating the instantaneous power s and the compensation capacity Q of the reactive power compensation mechanism under the current running state; c

[0050] a reactive power compensation adjustment strategy formulation module for calculating the difference △s between the instantaneous power under the compensation capacity and the instantaneous power sampling value s' under the current running state, comparing the difference △s with the given value Q of the reactive power under the current voltage, and determining the reactive power compensation adjustment strategy of the power distribution network according to the comparison result.

[0051] The system comprises a memory, a processor, and a computer program stored on the memory and executable on the processor.

[0052] The present application has the following advantages,

[0053] Compared with the prior art, by acquiring the multi-source data of the to-be-controlled board area at the same time within the monitoring period, including the node running data, the main user-side voltage data, and the power factor, the running state of the current reactive power compensation mechanism is classified by using the function value of the running state determination, the node running data and the main user-side voltage data are used as the influencing factors, the correlation analysis is performed on the influencing factors and the power factor of the reactive power compensation mechanism, the correlation coefficients of the influencing factors and the power factor of the reactive power compensation mechanism under different running states are acquired, the recurrent neural network model taking the pretreated time series data as the input is constructed, the historical data features of the influencing factor data and the real-time data features of the real-time data are acquired by using the trained neural network model, the running state of the current reactive power compensation mechanism is determined by performing feature matching, the instantaneous power and the compensation capacity of the reactive power compensation mechanism under the current running state and the difference between the instantaneous power under the current compensation capacity and the instantaneous power sampling value are calculated, the reactive power compensation adjustment strategy of the power distribution network is determined according to the comparison result of the difference calculation result and the given value of the reactive power under the current voltage, the coordination control of the reactive power compensation and the running power of the power distribution network can be realized, the deviation between the result quantity and the required target quantity of the control is reduced, the control accuracy and the intelligent degree are improved, and the increasingly complex power distribution network scheduling requirements are met. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 The flowchart of the power distribution network reactive power compensation control method based on multi-source data analysis of the present application;

[0055] Figure 2 The structural diagram of the power distribution network reactive power compensation control system based on multi-source data analysis of the present application. DETAILED DESCRIPTION​

[0056] The present application is further illustrated in conjunction with the specific embodiments, wherein the accompanying drawings are only used for exemplary illustration, the representations are only schematic diagrams, not physical diagrams, and cannot be understood as the limitation of the present application. In order to better illustrate the specific embodiments of the present application, some components of the drawings are omitted, enlarged or reduced, and do not represent the actual product size.

[0057] The specific implementation process of the technical scheme of the present application comprises the following steps:

[0058] Step 1: Obtain the node running data, main user side voltage data and power factor λ of the to-be-controlled console area at the same time t within the monitoring period T t .

[0059] Step 2: According to the obtained power factor λ t , construct the running state judgment function f1(λ t ) and f2(λ t ), and use the function value to classify the running state of the reactive power compensation mechanism in the period from t to t+1 within the current monitoring period T. The running state includes: the first running state of the power factor showing a steady increase; the second running state of the power factor showing a steady decrease; the third running state of the power factor remaining unchanged; the fourth running state of the power factor showing a fluctuating increase; and the fifth running state of the power factor showing a fluctuating decrease.

[0060] Among them, the expression of the running state judgment function f1(λ t ) is:

[0061]

[0062] The expression of the running state judgment function f2(λ t ) is:

[0063]

[0064] In the formula, t, t+1∈T; N is the data sampling number within the monitoring period T;

[0065] The classification principle of the running state is:

[0066] When f1(λ t )=-1 and f2(λ t )=1, the running state is the first running state of the power factor showing a steady increase;

[0067] When f1(λ t )=1 and f2(λ t )=1, the running state is the second running state of the power factor showing a steady decrease;

[0068] When f1(λ t ) = 0, the operation state is the third operation state of keeping power factor unchanged;

[0069] When f1(λ t ) = -1 and f2(λ t ) = -1, the operation state is the fourth operation state of fluctuating and increasing power factor;

[0070] When f1(λ t ) = 1 and f2(λ t ) = -1, the operation state is the fifth operation state of fluctuating and decreasing power factor.

[0071] Step 3: Taking the node operation data and the main user side voltage data as the influence factors, the correlation analysis between the influence factors and the power factor of the reactive power compensation mechanism is performed to obtain the correlation coefficient r i between the ith influence factor and the power factor of the reactive power compensation mechanism under different operation states.

[0072] The steps of the correlation analysis include:

[0073] Step S31: determining the mother sequence and the characteristic sequence;

[0074] Generally, the mother sequence is composed of the target performance and is the data sequence reflecting the characteristic factors of the target performance; the characteristic sequence is composed of all the influence factors and is the data sequence influencing the characteristic factors of the target performance; therefore, the time sequence data of the power factor of the reactive power compensation mechanism is taken as the mother sequence and the time sequence data of the influence factors is taken as the characteristic sequence.

[0075] Step S32: performing the standardization processing on each data in the data sequence;

[0076] The processing formula is: wherein, e h is the hth data value in the mother sequence; E h is the standardized value of the hth data in the mother sequence; is the mean value of the data in the mother sequence; f qh is the hth data value of the qth data in the characteristic sequence; F qh is the standardized value of the hth data of the qth data in the characteristic sequence; is the mean value of the qth data in the characteristic sequence.

[0077] Step S33: calculating the correlation coefficient value ξ qh of all the data points in the data sequence.

[0078] The calculation formula is:

[0079]

[0080] wherein,

[0081] is taken h -F qh the minimum value of all the calculated minimum values;

[0082] is taken h -F qh the maximum value of all the calculated maximum values; p is a resolution coefficient, and the value range is (0, 1];

[0083] Step S34: calculating the correlation degree value according to the calculation result of the correlation coefficient, and the calculation formula is:

[0084]

[0085] wherein, R q is the correlation coefficient of the qth influence factor and the power factor of the reactive power compensation mechanism; Q is the type of influence factor.

[0086] Step 4: extracting the time series data of each influence factor of the reactive power compensation mechanism in different operating states for pretreatment; the pretreatment method includes:

[0087] Linear normalization

[0088] scaling the time series data range linearly to a specific interval, usually [0, 1] or [-1, 1], and the calculation formula is: wherein X is the original data, X min and X max are the minimum value and the maximum value of the wave band respectively.

[0089] Z-score standardization

[0090] by standardizing the time series data, the mean value is 0 and the variance is 1. The calculation formula is: wherein μ is the mean value of the wave band, and σ is the standard deviation of the wave band.

[0091] Step 5: constructing a recurrent neural network model with the pretreated time series data as input, training the neural network model, and obtaining the historical data characteristics of the influence factor data by using the trained neural network model denotes the jth historical data characteristic of the ith influence factor under the kth operating state;

[0092] Among them, long short-term memory networks or gated recurrent control units are usually used as the basic units of the recurrent neural network model, and the preprocessed time series data are passed to the RNN layer. During the training process, the parameters of the model are adjusted, and a number of data are selected to verify the adjusted model to improve the accuracy and generalization ability of the model. On the trained recurrent neural network model, by removing the output layer of the trained neural network model and retaining the last hidden layer, the historical data characteristics of the i-th influencing factor data can be obtained.

[0093] Step 6: Obtain real-time data of influencing factors and use the neural network model to extract real-time data features x ij , x ij It is represented as the jth real-time data feature of the i-th influencing factor;

[0094] Step 7: Match the real-time data features with the historical data features, and determine the current operating status of the reactive power compensation mechanism based on the matching results;

[0095] The specific process includes:

[0096] Step S71: Calculate the real-time data features x respectively ij and the historical data characteristics of the kth operating state Feature matching between The calculation formula is:

[0097] Step S72: Get the matching degree of each feature The maximum value in the calculation result in, k=1,2,...,K; K is the type of operating state;

[0098] Step S73: Taking the maximum value of feature matching The corresponding operating state is used as the operating state of the current reactive power compensation mechanism.

[0099] Step 8: Calculate the instantaneous power s and compensation capacity Q of the reactive compensation mechanism under the current operating state respectively c ;

[0100] Compensation capacity Q c The calculation formula is:

[0101]

[0102] Where, β is the average load rate of the substation within the monitoring period T; P max is the average power of the electrical equipment in the control area; is the load power factor angle before reactive power compensation; Power factor angle of the load after reactive power compensation.

[0103] Step 9: calculate the difference △s between the instantaneous power under the compensation capacity in the current operating state and the instantaneous power sample value s', wherein △s = s - s';

[0104] Step 10: compare the difference △s with the given value Q of the reactive power under the current voltage, and determine the adjustment strategy of the power distribution network reactive power compensation according to the comparison result;

[0105] The determination principle of the given value Q of the reactive power is:

[0106]

[0107] Q max is the maximum value of the reactive output of the reactive power compensation mechanism; U1, U2, U3 and U4 are voltage unit values, and U1 < U2 < U3 < U4; U is the voltage value of the reactive power compensation mechanism;

[0108] The determination principle of the adjustment strategy of the power distribution network reactive power compensation is:

[0109] If △s > Q, the reactive power compensation power of the power distribution network is adjusted in the direction of reduction;

[0110] If △s < Q, the reactive power compensation power of the power distribution network is adjusted in the direction of increase.

[0111] The above shows and describes the basic principles and main features of the present application and the advantages of the present application. It should be understood by those skilled in the art that the present application is not limited to the above-mentioned embodiments, and the above-mentioned embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A distribution network reactive power compensation control method based on multi-source data analysis, characterized in that: include: Step 1: Obtain the node operation data, main user side voltage data, and power factor λ of the control area at the same time t within the monitoring period T. t , according to the power factor λ obtained t Construct the running state judgment function f1(λ t ) and f2(λ t ), using the function value to classify the operating status of the reactive power compensation mechanism from t to t+1 in the current monitoring period T; Step 2: Using the node operation data and the main user side voltage data as influencing factors, perform correlation analysis on the influencing factors and the power factor of the reactive compensation mechanism, and obtain the correlation coefficient r between the i-th influencing factor and the power factor of the reactive compensation mechanism under different operating conditions. i ; Step 3: Extract the time series data of various influencing factors of the reactive compensation mechanism under different operating conditions for preprocessing, build a recurrent neural network model with the preprocessed time series data as input, train the neural network model, and use the trained neural network model to obtain the historical data characteristics of the influencing factor data It is expressed as the jth historical data feature of the i-th influencing factor under the k-th operating state; Step 4: Obtain the real-time data of the influencing factors and use the neural network model to extract the real-time data features x of the real-time data ij , x ij The jth real-time data feature of the i-th influencing factor is represented, the real-time data feature is matched with the historical data feature, and the current operating state of the reactive power compensation mechanism is determined according to the matching result; Step 5: Calculate the instantaneous power s and compensation capacity Q of the reactive compensation mechanism in the current operating state respectively c , calculate the difference △s between the instantaneous power under the compensation capacity in the current operating state and the instantaneous power sampling value s', where △s = s-s', compare the calculated difference △s with the reactive power set value Q under the current voltage, and determine the distribution network reactive compensation adjustment strategy based on the comparison result.

2. The distribution network reactive power compensation control method based on multi-source data analysis according to claim 1, characterized in that: Operation state judgment function f1(λ t ) is: Operation state judgment function f2(λ t ) is: Where, t, t+1∈T; N is the number of data sampling times within the monitoring period T.

3. The distribution network reactive compensation control method based on multi-source data analysis according to claim 1, characterized in that: The operating status includes: The first operating state in which the power factor increases steadily; The second operating state in which the power factor decreases steadily; A third operating state in which the power factor remains unchanged; a fourth operating state in which the power factor fluctuates and increases; The fifth operating state is one in which the power factor fluctuates and decreases.

4. The distribution network reactive power compensation control method based on multi-source data analysis according to claim 1, characterized in that: The classification principles of the operating status are: When f1(λ t )=-1, and f2(λ t )=1, the operating state is the first operating state in which the power factor increases steadily; When f1(λ t )=1, and f2(λ t )=1, the operating state is the second operating state in which the power factor decreases steadily; When f1(λ t )=0, the operating state is the third operating state in which the power factor remains unchanged; When f1(λ t )=-1, and f2(λ t )=-1, the operating state is the fourth operating state in which the power factor fluctuates and increases; When f1(λ t )=1, and f2(λ t )=-1, the operating state is the fifth operating state in which the power factor fluctuates and decreases.

5. The method for reactive power compensation control of distribution network based on multi-source data analysis according to claim 1, characterized in that: The principle for determining the reactive power set value Q is: Q max is the maximum reactive output of the reactive compensation mechanism; U1, U2, U3, and U4 are all per-unit voltage values, and U1<U2<U3<U4; U is the voltage value of the reactive compensation mechanism.

6. The distribution network reactive power compensation control method based on multi-source data analysis according to claim 1, characterized in that: Compensation capacity Q c The calculation formula is: Where, β is the average load rate of the substation within the monitoring period T; P max is the average power of the electrical equipment in the control area; is the load power factor angle before reactive power compensation; is the load power factor angle after reactive power compensation.

7. The distribution network reactive power compensation control method based on multi-source data analysis according to claim 1, characterized in that: In step 4, the process of determining the operating status of the current reactive power compensation mechanism includes: Step S41: Calculate the real-time data features x respectively ij and the historical data characteristics of the kth operating state Feature matching between The calculation formula is: Step S42: Get the matching degree of each feature The maximum value in the calculation result in, k=1,2,...,K; K is the type of operating state; Step S43: Taking the maximum value of feature matching The corresponding operating state is used as the current operating state of the reactive power compensation mechanism.

8. The distribution network reactive power compensation control method based on multi-source data analysis according to claim 1, characterized in that: The principles for determining the reactive power compensation adjustment strategy of the distribution network are: If △s>Q, the reactive compensation power of the distribution network is adjusted in the direction of reduction; If △s<Q, the reactive compensation power of the distribution network will be adjusted in the direction of increasing.

9. A distribution network reactive power compensation control system based on multi-source data analysis, characterized in that: The system is used to implement the steps of the distribution network reactive compensation control method based on multi-source data analysis according to any one of claims 1 to 8, including: Used to obtain the node operation data, main user side voltage data, power factor λ of the control area at the same time t within the monitoring period T t Data acquisition module; Used to obtain the power factor λ t Construct the running state judgment function f1(λ t ) and f2(λ t ), an operating state classification module that uses a function value to classify the operating state of the reactive power compensation mechanism in the period t to t+1 within the current monitoring period T; Used to obtain the correlation coefficient r between the i-th influencing factor and the power factor of the reactive compensation mechanism under different operating conditions i Data association analysis module; Historical data features used to obtain influencing factor data and real-time data features x ij Feature extraction module; An operating state determination module for performing feature matching between the real-time data features and the historical data features, and determining the current operating state of the reactive power compensation mechanism according to the matching result; Used to calculate the instantaneous power s and compensation capacity Q of the reactive compensation mechanism in the current operating state c Current state quantity acquisition module; The reactive compensation adjustment strategy formulation module is used to calculate the difference △s between the instantaneous power under the compensation capacity in the current operating state and the instantaneous power sampling value s', and compare the calculated difference △s with the reactive power set value Q under the current voltage, and determine the reactive compensation adjustment strategy of the distribution network based on the comparison result.

10. The distribution network reactive power compensation control system based on multi-source data analysis according to claim 9, characterized in that: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the distribution network reactive compensation control method based on multi-source data analysis according to any one of claims 1 to 8 can be implemented.

Images