A microgrid control method and terminal based on cluster analysis

Abstract

The present invention relates to the field of microgrid control technology, and in particular to a microgrid control method and terminal based on cluster analysis, wherein hierarchical cluster analysis is performed from three dimensions: converter efficiency, total voltage mutation value, and standard deviation integral value, wherein the introduction of the converter efficiency performance indicator can reflect the working state of the converter, screen out efficiency outliers, and reduce the converter failure probability, the introduction of the total voltage mutation value can reflect the inconsistency state of battery internal resistance change, and the standard deviation integral value of the voltage reflects the difference in SOC during battery operation, and the three-dimensional indicators summarize the core performance differences of the strings, avoiding duplication caused by multiple dimensions; the hierarchical clustering method is used to reduce the calculation amount of statistical analysis on all battery cells, shorten the clustering analysis cycle, and speed up the fault judgment and control frequency; at the same time, the purpose of sorting the equilibrium condition judgment, aging condition, and DC internal resistance in a clustering method is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of microgrid control, and in particular to a microgrid control method and terminal based on cluster analysis. Background Art

[0002] With the development of energy storage technology, the intelligence of energy systems continues to evolve, and big data technology has been first put into use in areas such as power system fault monitoring, operation analysis, battery consistency judgment and early warning of faults in large-scale energy storage systems, and energy management and optimization control of various microgrids. However, with the increase in energy storage capacity and the expansion of control, the ability to store and process data has also increased accordingly, and there is a contradiction between the processing of massive data and the real-time requirements of control.

[0003] For electrochemical energy storage batteries, voltage, current, and temperature data are mainly collected to form a judgment on the battery status. However, liquid cooling temperature control technology is becoming more and more mature, and the temperature control method with safety as the goal has controlled the temperature consistency deviation of the battery within a very small range, and there is no significant difference in the average temperature during the charging, discharging, and static stages, which is not conducive to exploring the differences in the temperature change on the battery status. On the other hand, some battery management systems are still unable to collect the temperature of each battery cell, as well as the influence of factors such as heat conduction between batteries. The steady-state temperature of each battery cell at the time of data collection is not easy to obtain, which has certain limitations for algorithms that use temperature as one of the data dimensions. In scenarios with a large number of strings and a large number of battery cells in series, the amount of calculation required to collect the voltage of all batteries and perform clustering processing is large. In order to reduce the amount of calculation and increase the control frequency, the voltage data of the module or battery cluster under the same working conditions can be used instead of the single battery cell data to first screen out the modules or battery clusters with problems, and then analyze the local batteries to reduce the calculation time of the control.

[0004] In microgrids, power electronic devices and batteries are increasingly closely integrated to obtain more ideal power characteristics. Battery parameters are rarely introduced in the analysis of power system operation. On the contrary, the operating differences of power electronic equipment are rarely considered in the analysis of energy storage system performance and fault judgment. In DC or AC microgrids composed of distributed and string energy storage devices, the grid topology with a cluster of batteries as a unit and an AC or DC bus through a primary converter has gradually become a research hotspot. The performance of this type of topology battery and the converter connected in series to form a string should be considered simultaneously in the analysis; in addition, the important advantage of string application is that battery clusters with different initial capacities can be connected in parallel, so this factor should also be considered when designing the big data analysis dimension.

[0005] Therefore, considering the characteristics of string energy storage devices / systems, how to implement control based on cluster analysis is a technical problem that needs to be solved at present. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide a microgrid control method and terminal based on cluster analysis, which can achieve more effective cluster analysis-based control according to the characteristics of a string energy storage device.

[0007] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0008] A microgrid control method based on cluster analysis comprises the following steps:

[0009] S1. Define a constant power discharge period in the cyclic working condition of the microgrid. For each constant power discharge period, collect the converter efficiency, the total voltage mutation value of the battery cluster during state conversion, and the standard deviation integral value of the total voltage standard deviation of the battery cluster within a measurement period for each string;

[0010] S2, based on the three dimensions of the converter efficiency, the total voltage mutation value and the standard deviation integral value, and performing hierarchical clustering analysis according to the preset number of levels;

[0011] S3. After a preset number of rounds of cluster analysis, determine the strings corresponding to the points with the largest distance from the center point, the points with the most times the converter efficiency is classified as the lowest level, the points with the most times the total voltage mutation value is classified as the highest level, and the points with the most times the standard deviation integral value is classified as the highest level, and perform targeted processing and control on each of them.

[0012] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0013] A microgrid control terminal based on cluster analysis includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the microgrid control method based on cluster analysis described above are implemented.

[0014] The beneficial effects of the present invention are as follows: a microgrid control method and terminal based on cluster analysis of the present invention performs hierarchical cluster analysis from three dimensions: converter efficiency, total voltage mutation value and standard deviation integral value. The introduction of the converter efficiency performance indicator can reflect the working state of the converter, screen out efficiency outliers, reduce the converter failure probability, and reduce operating losses. The introduction of the total voltage mutation value can reflect the inconsistency of the battery internal resistance change, and the standard deviation integral value of the voltage reflects the difference in SOC during battery operation. The three-dimensional indicators summarize the core performance differences of the strings and avoid duplication caused by multiple dimensions. The hierarchical clustering method is used to reduce the amount of calculation for statistical analysis of all battery cells, shorten the clustering analysis cycle, and speed up the fault judgment and control frequency. At the same time, the outliers in each dimension are controlled accordingly, which solves the purpose of sorting the equilibrium condition judgment, aging condition SOH and DC internal resistance DCR in a clustering method. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 An example of a string type microgrid system of a microgrid control method based on cluster analysis according to an embodiment of the present invention;

[0016] Figure 2 A flow chart of a microgrid control method based on cluster analysis according to an embodiment of the present invention;

[0017] Figure 3 A specific flow chart of a microgrid control method based on cluster analysis according to an embodiment of the present invention;

[0018] Figure 4 A structural diagram of a microgrid control terminal based on cluster analysis according to an embodiment of the present invention;

[0019] Description of labels:

[0020] 1. A microgrid control terminal based on cluster analysis; 2. A processor; 3. A memory. DETAILED DESCRIPTION

[0021] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0022] Please refer to Figures 1 to 3 , a microgrid control method based on cluster analysis, comprising the steps of:

[0023] S1. Define a constant power discharge period in the cyclic working condition of the microgrid. For each constant power discharge period, collect the converter efficiency, the total voltage mutation value of the battery cluster during state conversion, and the standard deviation integral value of the total voltage standard deviation of the battery cluster within a measurement period for each string;

[0024] S2, based on the three dimensions of the converter efficiency, the total voltage mutation value and the standard deviation integral value, and performing hierarchical clustering analysis according to the preset number of levels;

[0025] S3. After a preset number of rounds of cluster analysis, determine the strings corresponding to the points with the largest distance from the center point, the points with the most times the converter efficiency is classified as the lowest level, the points with the most times the total voltage mutation value is classified as the highest level, and the points with the most times the standard deviation integral value is classified as the highest level, and perform targeted processing and control on each of them.

[0026] From the above description, it can be seen that the beneficial effects of the present invention are: a microgrid control method based on cluster analysis of the present invention performs hierarchical cluster analysis from three dimensions: converter efficiency, total voltage mutation value and standard deviation integral value. The introduction of the converter efficiency performance indicator can reflect the working state of the converter, screen out efficiency outliers, reduce the probability of converter failure, and reduce operating losses. The introduction of the total voltage mutation value can reflect the inconsistency of the battery internal resistance change, and the standard deviation integral value of the voltage reflects the difference in SOC during battery operation. The three-dimensional indicators summarize the core performance differences of the strings and avoid duplication caused by multiple dimensions. The hierarchical clustering method is used to reduce the amount of calculation for statistical analysis of all battery cells, shorten the clustering analysis cycle, and speed up the fault judgment and control frequency. At the same time, the outliers in each dimension are controlled accordingly, which solves the purpose of sorting the equilibrium condition judgment, aging condition SOH and DC internal resistance DCR in a clustering method.

[0027] Further, step S2 comprises the steps of:

[0028] The K-means method based on Euclidean distance is used to perform hierarchical clustering analysis, and the clustering results are divided into a preset number of levels;

[0029] Calculate the weighted distance between each point in each bin and the center point.

[0030] From the above description, it can be seen that the hierarchical clustering analysis is realized by the K-means method, and the weighted distance between each point in each file and the center point is calculated for subsequent calculations.

[0031] Furthermore, the weighted distance of the jth point in the kth file to the center point of the file is The calculation is:

[0032] ;

[0033] in, , and represents the weighted weight, , and Respectively represent data of three dimensions: the converter efficiency, the total voltage mutation value, and the standard deviation integral value, is the center point of the k-th file data, is the data of the jth point in the kth file.

[0034] From the above description, it can be seen that the weighted distance between each point in each gear and the central point is obtained by weighted calculation based on three dimensions: converter efficiency, total voltage mutation value, and standard deviation integral value.

[0035] Further, the determination of the weighted weights comprises the steps of:

[0036] Acquire data of three dimensions: initial weight and the converter efficiency of each string, the total voltage mutation value and the standard deviation integral value;

[0037] For the initial weight, the weight is adjusted based on the Hopkins statistic calculation, and the weight with the largest Hopkins statistic is obtained as the weighted weight.

[0038] From the above description, it can be seen that for the weighted weight of each dimension, an optimization algorithm is used to select the initial weight based on expert experience, so that it tends to the local optimal solution, avoiding the complexity of adaptive weight algorithms that need to combine the number of clustering gears to find the best solution.

[0039] Furthermore, it is characterized in that step S1 specifically comprises:

[0040] A constant power discharge period is defined in the cyclic working condition of the microgrid, and for each constant power discharge period:

[0041] At the beginning of the discharge from static to constant power , calculate the total voltage drop of the battery cluster of N strings, and obtain the total voltage mutation value data of each string ;

[0042] Constant power discharge period The average efficiency of the converter of N strings is calculated to obtain the converter efficiency ;

[0043] Constant power discharge period Calculate the standard deviation integral of the battery cluster of N strings to obtain the standard deviation integral value ;

[0044] ;

[0045] in, is the real-time sampling of voltage within the time period, is the average voltage during the period.

[0046] It can be seen from the above description that based on the above method, the data of three dimensions of converter efficiency, total voltage mutation value and standard deviation integral value are obtained for each string group.

[0047] Furthermore, the following steps are included between step S1 and step S2:

[0048] The collected converter efficiency, the total voltage mutation value and the standard deviation integral value are normalized:

[0049] ;

[0050] ;

[0051] ;

[0052] Make the data range of each dimension be (0,1);

[0053] in, express The minimum value in express The maximum value in express The minimum value in express The maximum value in express The minimum value in express The maximum value in .

[0054] From the above description, it can be seen that the data collected in the three dimensions are normalized respectively to eliminate the influence of dimension and order of magnitude, improve the convergence and stability of the algorithm, and enhance the clustering effect.

[0055] Further, step S3 comprises the steps of:

[0056] S31, after a preset number of rounds of cluster analysis, respectively calculating the sum of the weighted distances of the preset number of rounds of cluster analysis for each group of strings, and taking the group of strings with the largest sum of the weighted distances as the first group of strings corresponding to the point with the largest distance from the center point;

[0057] S32, obtaining, in the cluster analysis of the preset number of rounds, the second group of strings corresponding to the points where the converter efficiency is classified as the lowest level the most times, the third group of strings corresponding to the points where the total voltage mutation value is classified as the highest level the most times, and the fourth group of strings corresponding to the points where the standard deviation integral value is classified as the highest level the most times;

[0058] S33, performing targeted processing and control on the first group of strings, the second group of strings, the third group of strings, and the fourth group of strings respectively.

[0059] It can be seen from the above description that the group string corresponding to the point with the largest distance from the center point is determined according to the sum of weighted distances obtained by cluster analysis of a preset number of rounds.

[0060] Further, step S33 includes the steps of:

[0061] S331, performing secondary cluster analysis on the first group of strings and the second group of strings respectively;

[0062] The secondary cluster analysis of the first group of strings is based on two dimensions: the total voltage mutation value of each battery cell when it passes through the state transition, and the standard deviation integral value of the total voltage standard deviation of the battery cell within a measurement period, so as to obtain the battery cells with deviations in the DC internal resistance and aging degree in the first group of strings;

[0063] The secondary cluster analysis of the second group of strings is based on a single dimension of a total voltage mutation value of each cell when the state of each cell is converted, so as to obtain cells with high DC internal resistance in the second group of strings.

[0064] From the above description, it can be seen that for the group of strings with the largest sum of weighted distances, the first group of strings corresponding to the point with the largest distance from the center point, cluster analysis is further performed on the two dimensions of the total voltage mutation value of the battery cluster when the state transition occurs at the battery cell dimension and the standard deviation integral value of the total voltage standard deviation of the battery cluster within a metering period to find out the battery cells that deviate in DC internal resistance and aging degree; for the second group of strings corresponding to the point with the largest number of times the converter efficiency is classified as the lowest level, a battery cell-level cluster analysis based on the mutation value dimension is performed to find out the battery cells with high DC internal resistance.

[0065] Further, step S33 includes the steps of:

[0066] S332, performing balancing control on the battery cluster of the third string;

[0067] S333: For the fourth string group, initiate maintenance and inspection on the converter corresponding to the fourth string group.

[0068] From the above description, it can be seen that for the points with the largest number of standard deviation integrals classified as the highest level, the corresponding battery clusters are balanced and controlled to reduce the inconsistency of battery cell SOC; for the points with the largest number of efficiency classified as the lowest level, the corresponding converters are maintained and inspected to ensure that the temperature and heat dissipation conditions of the converter meet the requirements.

[0069] Please refer to Figure 4A microgrid control terminal based on cluster analysis includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the microgrid control method based on cluster analysis described above are implemented.

[0070] From the above description, it can be seen that the beneficial effects of the present invention are: a microgrid control terminal based on cluster analysis of the present invention performs hierarchical cluster analysis from three dimensions: converter efficiency, total voltage mutation value and standard deviation integral value. The introduction of the converter efficiency performance indicator can reflect the working state of the converter, screen out efficiency outliers, reduce the probability of converter failure, and reduce operating losses. The introduction of the total voltage mutation value can reflect the inconsistency of the battery internal resistance change, and the standard deviation integral value of the voltage reflects the difference in SOC during battery operation. The three-dimensional indicators summarize the core performance differences of the strings and avoid duplication caused by multiple dimensions. The hierarchical clustering method is used to reduce the amount of calculation for statistical analysis of all battery cells, shorten the clustering analysis cycle, and speed up the fault judgment and control frequency. At the same time, the outliers in each dimension are controlled accordingly, which solves the purpose of sorting the equilibrium condition judgment, aging condition SOH and DC internal resistance DCR in a clustering method.

[0071] A microgrid control method and terminal based on cluster analysis of the present invention are suitable for cluster analysis control of a string energy storage device / system / microgrid.

[0072] Please refer to Figures 1 to 3 , Embodiment 1 of the present invention is:

[0073] A microgrid control method based on cluster analysis. Figure 1 The string microgrid system shown is used as an example for subsequent description.

[0074] In this embodiment, the microgrid system contains N power supply strings, each of which is composed of a battery and a power electronic converter connected in series, and the outputs of all strings are connected in parallel to the load bus and can independently control power.

[0075] Considering that the initial voltage specifications of the batteries in each string are the same, the capacity specifications may be different. In this embodiment, when the capacity specifications are different, the power of the inverter connected in series corresponds to the battery capacity configuration, and the constant power discharge power used in the cluster analysis is converted according to the capacity. The method is that when the capacity of the ath battery is Qa, the total capacity of the remaining N-1 batteries is Qb, and the constant power discharge total power given value is Pb, then the constant power discharge given value for the ath battery is:

[0076] Pa=Qa / (Qa+Qb)×Pb.

[0077] All battery strings and converters are temperature-controlled by independent or centralized liquid cooling thermal management, so that the temperature fluctuations of each battery cell and converter radiator are within a smaller range. Therefore, this embodiment defines it as a constant temperature state.

[0078] S1. Define a constant power discharge period in the cyclic working condition of the microgrid. For each constant power discharge period, collect the converter efficiency of each string, the total voltage mutation value of the battery cluster during state transition, and the standard deviation integral value of the total voltage standard deviation of the battery cluster within a measurement period.

[0079] Step S1 is specifically as follows:

[0080] A constant power discharge period is defined in the cyclic working condition of the microgrid, and for each constant power discharge period:

[0081] At the beginning of the discharge from static to constant power , calculate the total voltage drop of the battery cluster of N strings, and obtain the total voltage mutation value data of each string ;

[0082] Constant power discharge period The average efficiency of the converter of N strings is calculated to obtain the converter efficiency ;

[0083] Constant power discharge period Calculate the standard deviation integral of the battery cluster of N strings to obtain the standard deviation integral value ;

[0084] ;

[0085] in, is the real-time sampling of voltage within the time period, is the average voltage value during the period.

[0086] In this embodiment, a constant power discharge period is defined in the cyclic working condition of the microgrid. , each time entering the corresponding time period, cluster analysis is performed on the core performance of all strings. The microgrid controller implements an average distribution strategy with equal power output for all N strings, and collects the total voltage of N battery clusters, the input power and output power of N converters in real time (the amount of data collected to calculate the converter input and output power is not described here), and calculates the converter efficiency, total voltage mutation value, and standard deviation integral value in real time.

[0087] The steps between step S1 and step S2 include:

[0088] The collected converter efficiency, the total voltage mutation value and the standard deviation integral value are normalized:

[0089] ;

[0090] ;

[0091] ;

[0092] Make the data range of each dimension be (0,1);

[0093] in, express The minimum value in express The maximum value in express The minimum value in express The maximum value in express The minimum value in express The maximum value in .

[0094] S2. Based on the three dimensions of the converter efficiency, the total voltage mutation value and the standard deviation integral value, hierarchical clustering analysis is performed according to a preset number of levels.

[0095] In this embodiment, at the end of each x-time constant power discharge under the same working condition, cluster analysis is performed on all N groups of data collected and processed, and the data is divided into a preset number of gears, which are three gears in this embodiment. The distance of the data of each gear from the center of the gear is calculated. The distances of the x-times clusters are summed, and the battery cluster corresponding to the group of distances with the largest summation result is clustered at the cell level.

[0096] Specifically, step S2 includes the steps of:

[0097] The K-means method based on Euclidean distance is used to perform hierarchical clustering analysis, and the clustering results are divided into a preset number of levels;

[0098] Calculate the weighted distance between each point in each file and the center point;

[0099] The weighted distance of the jth point in the kth file to the center point of the file The calculation is:

[0100] ;

[0101] in, , and represents the weighted weight, , and Respectively represent data of three dimensions: the converter efficiency, the total voltage mutation value, and the standard deviation integral value, is the center point of the k-th file data, is the data of the jth point in the kth file.

[0102] In this embodiment, based on the three-level division, k is 1, 2, and 3.

[0103] Because each dimension of the data represents a type of unbalanced feature of the group, dividing the features into three categories of "poor, average, and good" is helpful to distinguish the poor points of the group and implement subsequent control, so the data is clustered and analyzed according to three levels.

[0104] In this embodiment, based on the above steps, for each of the x times, the processed N groups of data are subjected to cluster analysis divided into three levels, and the weighted distance of the jth group of data in each level to the center point of the level is calculated. The clustering method adopts the K-means method based on Euclidean distance. In this method, when K=3, the clustering results are divided into three levels.

[0105] The determination of weighted weights includes the following steps:

[0106] Acquire data of three dimensions: initial weight and the converter efficiency of each string, the total voltage mutation value and the standard deviation integral value;

[0107] For the initial weight, the weight is adjusted based on the Hopkins statistic calculation, and the weight with the largest Hopkins statistic is obtained as the weighted weight.

[0108] Specifically, in this embodiment, the initial weights of the three dimensions are set to Ws=(converter efficiency weight, total voltage mutation value weight, standard deviation integral value weight)=(W1, W2, W3)=(0.2, 0.4, 0.4) according to the expert experience value. The selection of the initial weights is based on the effect of the parameters on the inconsistency of each group string of the microgrid. The converter efficiency reflects the power electronic loss of the converter under working conditions, and has a certain impact on the inconsistency of the group string and the aging of the converter; the total voltage mutation value weight reflects the difference in the DC internal resistance of the battery cluster, which has a greater impact on the battery loss and aging; the standard deviation integral value reflects the voltage value characteristics of the battery cluster, and can also reflect the battery consistency to a greater extent, so the above initial weights are given.

[0109] Represent the three-dimensional performance data of the i-th string as an array , for all N arrays, calculate the Hopkins statistic H(t) with the initial weight Ws. Keep W1 = 0.2 unchanged in the string performance weight. Initially, set W2 = W3, and with as the step size, decrease W2 and increase W3, and update the calculation of H(t + 1). When H(t + 1) > H(t), it indicates that the new weight is better than the initial weight, then continue to decrease W2 with W1 as the lower limit, and iteratively calculate until H reaches the maximum value in this adjustment direction; when H(t + 1) < H(t), then with as the step size, with W1 as the lower limit, decrease W3 and increase W2 until H reaches the maximum value in this direction as well, and compare the maximum values of H in the two directions to select the set of weights that makes H the largest.

[0110] The calculation method of the Hopkins statistic is not elaborated here. In this embodiment, the difference is that the optimization iteration method proposed here is based on the empirical data of determining W1 = 0.2 and W2 and W3 not less than 0.2. When the importance of the three groups of data cannot be judged in advance, a two-dimensional optimization method of equalizing weights is adopted, that is, the optimization algorithm of H requires two nested iterations.

[0111] S3. After clustering analysis for a preset number of rounds, determine the strings corresponding to the point with the largest distance from the center point, the point with the most occurrences of the converter efficiency classified into the lowest grade, the point with the most occurrences of the total voltage mutation value classified into the highest grade, and the point with the most occurrences of the standard deviation integral value classified into the highest grade, and perform targeted processing and control respectively.

[0112] In this embodiment, further sum the weighted distances for x times to obtain .

[0113] Step S3 includes the steps:

[0114] S31. After clustering analysis for a preset number of rounds, calculate the sum of the weighted distances of the preset number of rounds of clustering analysis for each string respectively, and take the string with the largest sum of the weighted distances as the first string corresponding to the point with the largest distance from the center point;

[0115] S32. Obtain the second string corresponding to the point with the most occurrences of the converter efficiency classified into the lowest grade, the third string corresponding to the point with the most occurrences of the total voltage mutation value classified into the highest grade, and the fourth string corresponding to the point with the most occurrences of the standard deviation integral value classified into the highest grade in the clustering analysis for a preset number of rounds;

[0116] S33. Perform targeted processing and control on the first string, the second string, the third string, and the fourth string respectively;

[0117] Step S33 includes the steps:

[0118] S331, performing secondary cluster analysis on the first group of strings and the second group of strings respectively;

[0119] The secondary cluster analysis of the first group of strings is based on two dimensions: the total voltage mutation value of each battery cell when it undergoes state transition, and the standard deviation integral value of the total voltage standard deviation of the battery cell within a metering period, so as to obtain the battery cells with deviations in DC internal resistance and aging degree in the first group of strings.

[0120] The largest point indicates that this point has a large deviation in the x-times distance calculations. Since the efficiency of the converter is only given a small weight in the weight design, this deviation is more caused by the inconsistency of the battery.

[0121] In this embodiment, for each gear The largest points and mutation values ​​are classified as the points with the largest number of the highest levels, and the corresponding battery clusters are subjected to cell-level cluster analysis. The cluster analysis method can refer to the steps in the previous article. The difference is that only the total voltage mutation value during state transition and the standard deviation integral value of the total voltage standard deviation of the battery cell within a measurement period are used (excluding the converter efficiency dimension) to find out the battery cells that deviate in DC internal resistance and aging degree.

[0122] The secondary cluster analysis of the second group of strings is based on a single dimension of a total voltage mutation value of each cell when the state of each cell is converted, so as to obtain cells with high DC internal resistance in the second group of strings.

[0123] In this embodiment, for the points with the largest number of total voltage mutation values ​​classified as the highest level, a battery cell level clustering analysis is performed on the corresponding battery clusters (the data is one-dimensional, containing only the dimension of total voltage mutation values) to find out the battery cells with high DC internal resistance.

[0124] S332: Perform balancing control on the battery cluster of the third string.

[0125] In this embodiment, the points with the largest number of standard deviation integrals classified as the highest level are balanced and the corresponding battery clusters are controlled to reduce the inconsistency of the battery cell SOC.

[0126] S333: For the fourth string group, initiate maintenance and inspection on the converter corresponding to the fourth string group.

[0127] In this embodiment, for the points with the largest number of points classified as the lowest efficiency level, the corresponding converters are maintained and inspected to ensure that the temperature and heat dissipation conditions of the converters meet the requirements.

[0128] Please refer to Figure 4 , Embodiment 2 of the present invention is:

[0129] A microgrid control terminal 1 based on cluster analysis includes a processor 2, a memory 3, and a computer program stored in the memory 3 and executable on the processor 2. When the processor 2 executes the computer program, the steps of the microgrid control method based on cluster analysis described in the above embodiment 1 are implemented.

[0130] In summary, the present invention provides a microgrid control method and terminal based on cluster analysis:

[0131] (1) In a DC or AC microgrid composed of multiple strings, solve the problem that the strings composed of batteries of different capacities cannot evenly distribute the output power and cannot perform DC internal resistance and consistency judgment.

[0132] (2) The main performance characteristics of each string in the system composed of multiple battery clusters are distinguished, and the performance indicator of converter efficiency is added to reflect the working status of the converter, screen out efficiency outliers, reduce the probability of converter failure, and reduce operating losses. The clustering of voltage drop performance at the moment of sudden load can reflect the inconsistency of battery internal resistance change, and the voltage standard deviation integral reflects the difference in SOC during battery operation. The three-dimensional indicators summarize the core performance differences of the strings and avoid duplication caused by multiple dimensions.

[0133] (3) The weight of each dimension is selected using an optimization algorithm based on expert experience, so that it tends to the local optimal solution, avoiding the complexity of adaptive weight algorithms that need to combine the number of clustering gears to find the optimal solution.

[0134] (4) The hierarchical clustering method is used to reduce the amount of calculation required for statistical analysis of all battery cells, shorten the clustering analysis cycle, and speed up the frequency of fault judgment and control.

[0135] (5) The outliers in each dimension are controlled accordingly. The purpose of sorting the equilibrium condition judgment, aging condition SOH and DC internal resistance DCR in a clustering method is solved.

[0136] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A microgrid control method based on cluster analysis, characterized in that: Includes steps: S1. Define a constant power discharge period in the cyclic working condition of the microgrid. For each constant power discharge period, collect the converter efficiency, the total voltage mutation value of the battery cluster during state conversion, and the standard deviation integral value of the total voltage standard deviation of the battery cluster within a measurement period for each string; S2, based on the three dimensions of the converter efficiency, the total voltage mutation value and the standard deviation integral value, and performing hierarchical clustering analysis according to the preset number of levels; Step S2 comprises the steps of: The K-means method based on Euclidean distance is used to perform hierarchical clustering analysis, and the clustering results are divided into a preset number of levels; Calculate the weighted distance between each point in each file and the center point; S3. After a preset number of rounds of cluster analysis, determine the strings corresponding to the points with the largest distance from the center point, the points with the largest number of times the converter efficiency is classified as the lowest level, the points with the largest number of times the total voltage mutation value is classified as the highest level, and the points with the largest number of times the standard deviation integral value is classified as the highest level, and perform targeted processing and control on each of them; Step S3 includes the steps of: S31, after a preset number of rounds of cluster analysis, respectively calculating the sum of the weighted distances of the preset number of rounds of cluster analysis for each group of strings, and taking the group of strings with the largest sum of the weighted distances as the first group of strings corresponding to the point with the largest distance from the center point; S32, obtaining, in the cluster analysis of the preset number of rounds, the second group of strings corresponding to the points where the converter efficiency is classified as the lowest level the most times, the third group of strings corresponding to the points where the total voltage mutation value is classified as the highest level the most times, and the fourth group of strings corresponding to the points where the standard deviation integral value is classified as the highest level the most times; S33, performing targeted processing and control on the first group of strings, the second group of strings, the third group of strings, and the fourth group of strings respectively; Step S33 includes the steps of: S331, performing secondary cluster analysis on the first group of strings and the second group of strings respectively; The secondary cluster analysis of the first group of strings is based on two dimensions: the total voltage mutation value of each battery cell when it passes through the state transition, and the standard deviation integral value of the total voltage standard deviation of the battery cell within a measurement period, so as to obtain the battery cells with deviations in the DC internal resistance and aging degree in the first group of strings; The secondary cluster analysis of the second group of strings is based on a single dimension of a total voltage mutation value of each cell when the state of each cell is switched, so as to obtain cells with high DC internal resistance in the second group of strings; S332, performing balancing control on the battery cluster of the third string; S333. For the fourth string group, initiate maintenance and inspection on the converter corresponding to the fourth string group.

2. A microgrid control method based on cluster analysis according to claim 1, characterized in that: The weighted distance d of the jth point in the kth file to the center point of the file kj The calculation is: Among them, w1, w2 and w3 represent weighted weights, η, ΔV and I VSD Respectively represent the data of the three dimensions of the converter efficiency, the total voltage mutation value and the standard deviation integral value, (η k_S , ΔV k_S , I VSD_k_S ) is the center point of the k-th file data, (η kj_S , ΔV kj_S , I VSD_kj_S ) is the data of the jth point in the kth file.

3. A microgrid control method based on cluster analysis according to claim 1, characterized in that: The determination of weighted weights includes the following steps: Acquire data of three dimensions, namely, initial weight and the converter efficiency of each string, the total voltage mutation value and the standard deviation integral value; For the initial weight, the weight is adjusted based on the Hopkins statistic calculation, and the weight with the largest Hopkins statistic is obtained as the weighted weight.

4. The microgrid control method based on cluster analysis according to claim 1, characterized in that: Step S1 is specifically as follows: A constant power discharge period is defined in the cyclic working condition of the microgrid, and for each constant power discharge period: At the time t0-t1 from static to constant power discharge start, calculate the total voltage drop of the battery cluster of N strings, and obtain the total voltage mutation value data ΔV of each string i ∈(ΔV1,ηV N ); During the constant power discharge period t1-t2, the average efficiency of the converters of N strings is calculated to obtain the converter efficiency η i ∈(η1,η N ); In the constant power discharge period t1-t2, the standard deviation integral of the battery cluster of N strings is calculated to obtain the standard deviation integral value I VSD_i ∈(I VSD_1 , I VSD_N ); Among them, V i (t) is the real-time sampling of voltage during the period, V avg (t) is the average voltage during the period.

5. A microgrid control method based on cluster analysis according to claim 4, characterized in that: The steps between step S1 and step S2 include: The collected converter efficiency, the total voltage mutation value and the standard deviation integral value are normalized: Make the data range of each dimension be (0, 1); Where, ΔV min Indicates △V i ∈(ΔV1, ΔV N ), ΔV max Indicates ΔV i ∈(ΔV1, ΔV N ), η i_min Represents η i ∈(η1,η N ), η i_max Represents η i ∈(η1,η N ), I VSD_i_min Indicates I VSD_i ∈(I VSD_1 , I VSD_N ), I VSD_i_max Indicates I VSD_i ∈(I VSD_1 , I VSD_N ) is the maximum value among .

6. A microgrid control terminal based on cluster analysis, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps in the microgrid control method based on clustering analysis described in any one of claims 1 to 5 are implemented.

Images