A control method and system for energy storage low-frequency and low-voltage transfer discharge

Abstract

The present invention relates to the field of power control technologies, and particularly to a control method and system for energy storage low-frequency and low-voltage transfer discharge. The method respectively obtains an initial load data sequence of a power grid within at least two consecutive historical time periods before a target time period, and obtains a data fitting weight of each historical time period for the target time period; constructs a load data change prediction curve of the target time period according to the data fitting weight and the initial load data sequence, divides the load data change prediction curve into at least two sub-curves, and obtains a load change trend factor of the sub-curves; for any sampling moment within the target time period, obtains a power grid load pressure index according to the actual load data at any sampling moment and the load change trend factor of the sub-curve to perform real-time transfer discharge control on the energy storage device of the power grid. By combining the predicted load data and the actual load data, the accuracy of the energy storage device for charging and discharging the power grid is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power control, and particularly to a control method and system for energy storage low-frequency and low-voltage transfer and discharge. Background Art

[0002] Energy storage low-frequency and low-voltage transfer and discharge control can store the excess electric energy when the grid load is low and the electric energy obtained through other means through technologies such as battery packs, and provide electric energy supply during the peak period of the grid load, so as to avoid phenomena such as power interruption caused by excessive grid load or waste of electric energy caused by insufficient utilization of electric energy due to small grid load, so as to balance the grid load and reduce the grid pressure, thereby ensuring power consumption safety and improving the economic benefits and energy efficiency management level of the energy storage system.

[0003] Currently, the current energy storage transfer and discharge control usually judges the load ratio in the current power grid. When the load ratio meets the set load ratio threshold, electric energy is supplied to the power grid or the electric energy in the power grid is recovered and stored. For example, if the current load reaches more than 70% of the maximum load that the power grid can withstand, the stored electric energy is released to the power grid to cope with the insufficient power supply in the power grid. If the current load reaches less than 50% of the maximum load that the power grid can withstand, the excess electric energy in the power grid is recovered and stored. However, this method is relatively passive, lacks the ability to predict possible future problems, and is prone to missing the best response time, resulting in insufficient power compensation or untimely power energy storage, and causing waste and other situations, which is not conducive to the accurate transfer and discharge control of electric energy by the energy storage system.

[0004] Therefore, how to improve the accurate transfer and discharge control of electric energy by the energy storage system has become an urgent problem to be solved. Summary of the Invention

[0005] In view of this, the embodiments of the present invention provide a control method and system for energy storage low-frequency and low-voltage transfer and discharge to solve the problem of how to improve the accurate transfer and discharge control of electric energy by the energy storage system.

[0006] In a first aspect, the embodiments of the present invention provide a control method for energy storage low-frequency and low-voltage transfer and discharge, and the method includes the following steps:

[0007] Obtain at least two consecutive historical time period initial load data sequences of the power grid before the target time period respectively, and perform adaptive smoothing processing on each of the initial load data sequences to obtain corresponding smoothed data sequences;

[0008] Construct a load data change curve for each of the smoothed data sequences, where the horizontal axis of the load data change curve is time and the vertical axis is load data. Based on the time interval between each historical time period and the target time period, and each load data change curve, obtain the data fitting weight of each historical time period for the target time period;

[0009] Based on the data fitting weight of each historical time period for the target time period, construct a load data change prediction curve for the target time period. According to the differences between adjacent data points in the load data change prediction curve, divide the load data change prediction curve into at least two sub-curves. Based on the data differences in each sub-curve, obtain the load change trend factor of the corresponding sub-curve;

[0010] For any sampling moment within the target time period, based on the actual load data at the sampling moment and the load change trend factor of the sub-curve to which the sampling moment belongs, obtain the grid load pressure index at the sampling moment, and perform real-time charge and discharge control on the energy storage device of the grid according to the grid load pressure index at the sampling moment.

[0011] In a second aspect, an embodiment of the present invention provides an energy storage low-frequency and low-voltage charge and discharge control system, including a memory and a processor. The processor executes the computer program stored in the memory to implement the above-mentioned energy storage low-frequency and low-voltage charge and discharge control method.

[0012] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0013] The present invention respectively obtains initial load data sequences of a power grid within at least two consecutive historical time periods before a target time period, and respectively performs adaptive smoothing processing on each of the initial load data sequences to obtain corresponding smoothed data sequences; constructs a load data change curve for each of the smoothed data sequences, where the horizontal axis of the load data change curve is time and the vertical axis is load data, and obtains a data fitting weight for each of the historical time periods with respect to the target time period according to the time interval between each of the historical time periods and the target time period, and each of the load data change curves; constructs a load data change prediction curve for the target time period according to the data fitting weights of each of the historical time periods with respect to the target time period, divides the load data change prediction curve into at least two sub-curves according to the differences between adjacent data points in the load data change prediction curve, and obtains a load change trend factor for the corresponding sub-curve according to the data differences in each of the sub-curves; for any sampling moment within the target time period, obtains a power grid load pressure index for the any sampling moment according to the actual load data at the any sampling moment and the load change trend factor of the sub-curve to which the any sampling moment belongs, and performs real-time charge and discharge control on the energy storage device of the power grid according to the power grid load pressure index for the any sampling moment. Among them, by analyzing the change rules and power consumption patterns of the load data in multiple historical time periods to predict the load data within the target time period, a load data change prediction curve within the target time period can be obtained. In order to perform real-time power grid load pressure analysis on each sampling moment within the target time period, according to the fluctuations of the load data change prediction curve, it is divided into multiple sub-curves with different changes, and then the load change trend factor of each sub-curve is combined with the actual load data of each sampling moment within the target time period, a more realistic power grid load pressure index can be obtained, and then the charge and discharge strategy of the energy storage device in the power grid is adjusted according to the power grid load pressure index, thereby improving the accuracy of the charge and discharge of the energy storage device to the power grid, achieving the purpose of peak shaving and valley filling of the power consumption load in the power grid, maintaining the stable operation of the power grid, and improving the utilization rate of power grid energy, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0015] Figure 1 It is a flowchart of a method for controlling charge and discharge of energy storage at low frequency and low voltage according to Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] Embodiments of the present disclosure will be described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present disclosure, and should not be construed as a limitation to the present disclosure.

[0017] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above-mentioned accompanying drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0018] In order to illustrate the technical solution of the present invention, specific embodiments will be used for illustration below.

[0019] See Figure 1 , which is a method flow chart of a control method for energy storage low-frequency and low-voltage transfer discharge provided by Embodiment 1 of the present invention. As Figure 1 shown, the method may include:

[0020] Step S101, respectively obtain at least two consecutive historical time periods before the target time period, the initial load data sequences of the power grid within each historical time period, and respectively perform adaptive smoothing processing on each initial load data sequence to obtain the corresponding smoothed data sequences.

[0021] The present invention mainly aims to obtain more accurate control decisions when controlling the charging and discharging of energy storage devices in the power grid. Therefore, in the embodiments of the present invention, a current and voltage acquisition module and other devices are used to collect load data in the power grid, and one day is used as a time period. Assuming that the target time period is the i-th day, then each day in the past month before the i-th day is used as a historical time period, and the load data within one month before the i-th day is obtained. Furthermore, the initial load data sequences corresponding to each day in the past month are obtained. It should be noted that the sampling frequency of the load data is once per second, which is not limited here and can be set according to the implementation scenario.

[0022] Since the change of load data has certain volatility in time series, which is not conducive to observing and analyzing the overall change trend. Therefore, in order to more accurately understand the change rules and patterns of the initial load data sequences of each day in the past month, the mean filtering algorithm is used to perform smoothing filtering on the initial load data sequences of each day in the past month to improve the accuracy of historical data.

[0023] Since the fluctuations of load data may vary in different time periods, when using mean filtering, the filtering windows at different positions may also be different. By determining the size of the adaptive filtering window, the overall smoothness of the load data can be improved to highlight its overall variation law over time. Therefore, for the initial load data sequence of any day, the specific process of performing adaptive mean filtering on the initial load data sequence to obtain the corresponding smoothed data sequence is as follows:

[0024] For any load data in any initial load data sequence, centering on the any load data, obtain a preset number of adjacent load data in the any initial load data sequence, and form a target data sequence with the any load data. Perform first-order difference processing on the target data sequence to obtain a first-order difference sequence, and calculate the coefficient of variation of the first-order difference sequence;

[0025] Take the negative of the coefficient of variation as the independent variable of a preset exponential function to obtain the corresponding exponential function value. Obtain the subtraction value of the constant 1 and the exponential function value, round up the product of the subtraction value and a preset hyperparameter to an odd integer to obtain the corresponding odd integer, and use the odd integer as the adaptive filtering window when performing mean filtering on the any load data;

[0026] Obtain the adaptive filtering window corresponding to each load data in the any initial load data sequence. According to the adaptive filtering window corresponding to each load data in the any initial load data sequence, perform mean filtering on the any initial load data sequence to obtain the corresponding smoothed data sequence.

[0027] In an embodiment, taking the j-th load data in the a-th initial load data sequence as an example, select 10 load data before and after the j-th load data in the a-th initial load data sequence to form the target data sequence of the j-th load data, and perform first-order difference processing on the target data sequence to obtain the corresponding difference sequence. The coefficient of variation (Coeficient of Vriation, abbreviated as CV) is an index to measure the relative variability of data and is usually used to compare the dispersion degrees between different data sets. It is the ratio of the standard deviation to the mean and is usually expressed as a percentage. Therefore, by calculating the coefficient of variation of the difference sequence, the adaptive filtering window when performing mean filtering on the j-th load data is obtained, and the calculation expression for the size of the adaptive filtering window is:

[0028]

[0029] where N aj represents the size of the adaptive filtering window of the j-th load data in the a-th initial load data sequence, represents the symbol for rounding up to an odd integer, and e represents the natural constant; Yaj represents the coefficient of variation of the j-th load data in the a-th initial load data sequence, and L represents a hyperparameter with a value of 20.

[0030] It should be noted that the larger the coefficient of variation of the j-th load data, the less smooth the local data of the j-th load data and the greater the volatility. Then, the adaptive filtering window of the j-th load data should be larger to enable more effective data to be referred to during smoothing filtering.

[0031] Similarly, the adaptive filtering window of each load data in the a-th initial load data sequence is obtained. After determining the adaptive filtering window, mean filtering is performed on the a-th initial load data sequence to obtain the filtered a-th initial load data sequence, denoted as the a-th smoothed data sequence. Thus, the smoothed data sequences corresponding to the initial load data sequences of each day in the past month before the i-th day can be obtained, improving the smoothness of the power grid load data.

[0032] It is worth noting that the coefficient of variation, mean filtering, and first-order difference processing all belong to the prior art and will not be elaborated in detail here.

[0033] Step S102: Construct the load data change curve of each smoothed data sequence. The horizontal axis of the load data change curve is time, and the vertical axis is the load data. According to the time interval between each historical time period and the target time period, and each load data change curve, the data fitting weight of each historical time period for the target time period is obtained.

[0034] After obtaining the smoothed data sequences corresponding to each day in the past month before the i-th day, considering that the laws and patterns of load changes in the power grid are related to users' living habits and schedules and have certain regular characteristics, therefore, the load data of the power grid on the i-th day can be predicted through the historical load data before the i-th day (that is, the smoothed data sequences of each day in the past month), which is used to improve the accuracy of charge and discharge control of energy storage devices in combination with the actual load situation of the power grid on the i-th day.

[0035] First, for each smoothed data sequence, with time as the horizontal axis and load data as the vertical axis, construct a load data change curve corresponding to each smoothed data sequence, obtaining multiple load data change curves. Then, according to the load change patterns presented by all the load data change curves, analyze the weights of the load data for each day in the past month when predicting the load data change curve for the i-th day, that is, the data fitting weights of each day in the past month relative to the i-th day. The larger the data fitting weight, the higher the importance for predicting the load data of the i-th day. Therefore, in the embodiments of the present invention, according to the time interval between each historical time period and the target time period, and each load data change curve, obtain the data fitting weight of each historical time period for the target time period. The specific obtaining method is as follows:

[0036] (1) Respectively obtain the DTW distances between every two of the load data change curves. According to all the DTW distances and according to the time interval between each historical time period and the target time period, respectively obtain the data reference degree of each historical time period.

[0037] Among them, the step of respectively obtaining the data reference degree of each historical time period according to all the DTW distances and according to the time interval between each historical time period and the target time period includes:

[0038] For any historical time period, according to the DTW distances between the load data change curve corresponding to the any historical time period and the load data change curves corresponding to each other historical time period, use a preset Gaussian kernel function to obtain the local density value of the load data change curve corresponding to the any historical time period;

[0039] Obtain the time interval between the any historical time period and the target time period. According to the product of the reciprocal of the time interval and the local density value, obtain the data reference degree of the any historical time period.

[0040] In one embodiment, first, the DTW algorithm is used to calculate the DTW distance between each pair of load data change curves to characterize the change similarity between the load data change curves. Then, a Gaussian kernel function with a kernel bandwidth parameter of 2 is obtained, and using the Gaussian kernel function, the local density value of each load data change curve among all load data change curves is calculated. The higher the local density value, the closer the overall change pattern of the load data change curve is to that of other load data change curves, and the more it can represent the daily load change pattern and mode in the power grid. Conversely, the smaller the local density value, the more certain changes have occurred in the load usage of the load data change curve on the corresponding day. Therefore, in order to more accurately describe the prediction accuracy of the load data of the past month for the load data on the i-th day, the length of the time interval between the corresponding day of each load data change curve and the i-th day is further combined. The closer the time interval, the greater the reference value of the load data. It should be noted that both the DTW algorithm and the Gaussian kernel function belong to the prior art and will not be described in detail here.

[0041] The calculation expression for the reference degree of the data on the (i - j)-th day within the past month is:

[0042]

[0043] where, μ i-b represents the reference degree of the data on the (i - b)-th day within the past month, ρ i-b represents the local density value of the load data change curve on the (i - b)-th day within the past month, t i-b represents the number of days between the (i - b)-th day and the i-th day within the past month, that is, the periodic interval of the time period, and 1 represents a constant.

[0044] Similarly, the reference degree of the data for each day within the past month can be obtained.

[0045] (2) Obtain the reference degree of each of the historical time periods, obtain the sum of the reference degrees, and for any historical time period, calculate the ratio of the reference degree of the any historical time period to the sum of the reference degrees to obtain the data fitting weight of the any historical time period for the target time period.

[0046] In one embodiment, the calculation expression for the data fitting weight of the (i - b)-th day within the past month relative to the i-th day is:

[0047]

[0048] where, q i-b represents the data fitting weight of the (i - b)-th day within the past month relative to the i-th day, μ i-bIt represents the data reference degree of the (i - b)-th day within the past month, and m represents the number of days in the past month, which is also the total number of periods of the historical time period.

[0049] It should be noted that the greater the data reference degree, the greater the weight of the load data of the corresponding day in predicting the load data of the i-th day.

[0050] Thus, the data fitting weights of each day within the past month relative to the i-th day can be obtained.

[0051] Step S103: Based on the data fitting weights of each historical time period for the target time period, construct a predicted curve of the load data change for the target time period. According to the differences between adjacent data points in the predicted curve of the load data change, divide the predicted curve of the load data change into at least two sub-curves. Based on the data differences in each sub-curve, obtain the load change trend factor corresponding to the sub-curve.

[0052] After determining the data fitting weights of each day within the past month relative to the i-th day, the load change of the i-th day can be predicted by combining the smoothed data sequences of each day within the past month, and correspondingly, a predicted curve of the load data change for the i-th day can be obtained. The method for obtaining the predicted curve of the load data change for the i-th day is as follows:

[0053] Take the data fitting weight of each historical time period for the target time period as the weight of each load data in the corresponding smoothed data sequence. For any sampling moment within the target time period, obtain the load data corresponding to the same sampling moment in the smoothed data sequences corresponding to each historical time period to form the historical load data sequence of the any sampling moment, and perform weighted average processing on the historical load data sequence to obtain the load prediction value of the any sampling moment;

[0054] Obtain the load prediction values of each sampling moment within the target time period, and construct a predicted curve of the load data change composed of the load prediction values of each sampling moment within the target time period. The horizontal axis of the predicted curve of the load data change is time, and the vertical axis is the load prediction value.

[0055] In one embodiment, taking the k-th sampling moment on the i-th day as an example, in the smoothed data sequences corresponding to each day within the past month, the load data corresponding to the k-th sampling moment of each day are respectively obtained to form the historical load data sequence of the k-th sampling moment on the i-th day. Meanwhile, the data fitting weights of each day within the past month relative to the i-th day are respectively used as the weights of each load data in the corresponding smoothed data sequence. Then, the historical load data sequence of the k-th sampling moment on the i-th day is weighted and summed, and then the mean value is calculated. This mean value is used as the load prediction value of the k-th sampling moment on the i-th day. At this time, the load prediction of the k-th sampling moment on the i-th day is completed. Similarly, the load prediction values of each sampling moment on the i-th day are obtained, and then a load data change prediction curve with time on the horizontal axis and load prediction values on the vertical axis is formed.

[0056] After predicting the power grid load usage on the i-th day, that is, after obtaining the load data change prediction curve of the i-th day, the change rules and trends of the load data change prediction curve are analyzed. Specifically, first, according to the inflection points in the load data change prediction curve, the load data change prediction curve is divided into multiple sub-curves, and one sub-curve represents one load change trend. Then, the load change trend factors of each sub-curve are further analyzed in detail.

[0057] Among them, the method of dividing the load data change prediction curve into multiple sub-curves according to the inflection points in the load data change prediction curve includes:

[0058] For any data point on the load data change prediction curve, K data points on the left side of the any data point are obtained on the load data change prediction curve to form a left adjacent data sequence, and the left adjacent data sequence is subjected to first-order difference processing, and the mean value of the difference values in the corresponding first-order difference sequence is recorded as the left difference value mean;

[0059] K data points on the right side of the any data point are obtained on the load data change prediction curve to form a right adjacent data sequence, and the right adjacent data sequence is subjected to first-order difference processing, and the mean value of the difference values in the corresponding first-order difference sequence is recorded as the right difference value mean;

[0060] If the left difference value mean or the right difference value mean does not exist for the any data point, the confidence level that the any data point is an inflection point is set to 0; if the left difference value mean and the right difference value mean exist for the any data point, the absolute value of the difference between the left difference value mean and the right difference value mean is calculated as the confidence level that the any data point is an inflection point;

[0061] Obtain the confidence that the left adjacent data point of any of the data points is an inflection point and the confidence that the right adjacent data point is an inflection point. If the confidence that any of the data points is an inflection point is greater than both the confidence that the left adjacent data point is an inflection point and the confidence that the right adjacent data point is an inflection point, then mark any of the data points as an inflection point;

[0062] Obtain at least one inflection point on the predicted curve of load data change. Use the inflection point as a curve segmentation point to divide the predicted curve of load data change into at least two sub-curves.

[0063] In an embodiment, set K = 10. Taking the x-th data point on the predicted curve of load data change as an example, on the predicted curve of load data change, obtain 10 data points before the x-th data point to form a left adjacent data sequence, perform a first-order difference process on the left adjacent data sequence, and correspondingly obtain the mean value of the difference values in the first-order difference sequence, denoted as the left difference value mean of the x-th data point. Similarly, obtain 10 data points after the x-th data point to form a right adjacent data sequence, perform a first-order difference process on the right adjacent data sequence, and correspondingly obtain the mean value of the difference values in the first-order difference sequence, denoted as the right difference value mean of the x-th data point. Then, based on the left difference value mean and the right difference value mean of the x-th data point, obtain the confidence that the x-th data point is an inflection point. The calculation expression of the confidence is:

[0064]

[0065] where P i represents the confidence that the x-th data point on the predicted curve of load data change is an inflection point, represents the right difference value mean of the x-th data point, represents the left difference value mean of the x-th data point, and || represents the absolute value symbol.

[0066] It should be noted that the larger the value of, the greater the difference in the differential data of the selected data points on the left and right sides of the x-th data point, indicating that the curve change difference on both sides of the x-th data point is greater, and the corresponding confidence that the x-th data point is an inflection point is greater.

[0067] Considering that not every data point on the predicted curve of load data change has a left adjacent data sequence and a right adjacent data sequence, and such data points are generally the first few data points or the last few data points on the predicted curve of load data change, and are generally data points where the predicted curve of load data change is in a stable stage. Therefore, for data points that only have a left adjacent data sequence or only have a right adjacent data sequence, directly set the confidence that these data points belong to inflection points to 0.

[0068] After determining the confidence level of each data point on the load data change prediction curve as an inflection point, compare the confidence level of the x-th data point as an inflection point with the confidence levels of the (x - 1)-th and (x + 1)-th data points as inflection points respectively. If the confidence level of the x-th data point as an inflection point is greater than the confidence level of the (x - 1)-th data point as an inflection point and also greater than the confidence level of the (x + 1)-th data point as an inflection point, then determine the x-th data point as an inflection point. According to the method of determining the x-th data point as an inflection point, perform the same traversal on each data point on the load data change prediction curve to obtain at least one inflection point on the load data change prediction curve. Further, using the inflection point as a curve segmentation point, the load data change prediction curve can be divided into at least two sub-curves.

[0069] Among them, the method for analyzing the load change trend factor of each sub-curve in detail is as follows:

[0070] For any sub-curve, obtain the difference between each data point on the any sub-curve and its left adjacent data point on the load data change prediction curve, use each difference as the independent variable of a preset exponential function to obtain the corresponding exponential function result, and obtain the load change trend factor of the any sub-curve according to the mean value of all exponential function results.

[0071] In one embodiment, taking the y-th sub-curve as an example, the expression for calculating the load change trend factor of the y-th sub-curve is:

[0072]

[0073] Among them, X y represents the load change trend factor of the y-th sub-curve, n y represents the total number of data points on the y-th sub-curve, exp() represents the exponential function with the natural constant as the base, F y(s+1) represents the load prediction value corresponding to the (s + 1)-th data point on the y-th sub-curve, F ys represents the load prediction value corresponding to the s-th data point on the y-th sub-curve.

[0074] It should be noted that by calculating the difference between the load prediction values of the subsequent data point and the previous data point on the y-th sub-curve to analyze the change trend of the y-th sub-curve, if the corresponding value is greater than 1, it means that the y-th sub-curve shows an upward trend, that is, the load pressure of the y-th sub-curve is relatively large during the time period corresponding to the i-th day; on the contrary, if the value is less than 1, it indicates that the y-th sub-curve shows a downward trend, and the load pressure of the y-th sub-curve is relatively small during the time period corresponding to the i-th day.

[0075] So far, the load change trend factor of each sub-curve in the load data change prediction curve for the i-th day can be obtained. The larger the load change trend factor, the greater the power grid load pressure during the corresponding time period.

[0076] Step S104, for any sampling moment within the target time period, based on the actual load data at any sampling moment and the load change trend factor of the sub-curve to which any sampling moment belongs, obtain the power grid load pressure index at any sampling moment, and perform real-time charge and discharge control on the energy storage devices of the power grid according to the power grid load pressure index at any sampling moment.

[0077] When the actual load data at any sampling moment on the i-th day is collected, combined with the load data change prediction curve obtained by predicting the load on the i-th day using historical load data, analyze the power grid load pressure index at this sampling moment. The specific method is as follows: Obtain the maximum load capacity that the power grid can withstand, calculate the ratio between the actual load data at any sampling moment and the maximum load capacity that the power grid can withstand, obtain the product between the ratio and the load change trend factor of the sub-curve to which any sampling moment belongs, and perform normalization processing on the product to obtain the power grid load pressure index at any sampling moment.

[0078] In an embodiment, the calculation expression of the power grid load pressure index at any sampling moment on the i-th day is:

[0079]

[0080] Among them, Y0 represents the power grid load pressure index at any sampling moment on the i-th day, norm() represents the normalization function, F0 represents the actual load data at any sampling moment on the i-th day, F max represents the maximum load capacity that the power grid can withstand, and X0 represents the load change trend factor of the sub-curve to which any sampling moment on the i-th day belongs.

[0081] It should be noted that represents the ratio between the actual power grid load at the current sampling moment and the maximum load capacity that the power grid can currently withstand. The larger this ratio, the greater the pressure on the power grid load at this moment. Similarly, the larger the load change trend factor of the sub-curve corresponding to the current sampling moment, the greater the load pressure on the power grid in the next period of time. Therefore, the larger the power grid load pressure index at the current sampling moment.

[0082] After determining the power grid load pressure index at any sampling moment on the i-th day, real-time charge and discharge control can be performed on the energy storage devices in the power grid. The specific control method is as follows:

[0083] Obtain a preset first load pressure index threshold and a second load pressure index threshold, and the first load pressure index threshold is less than the second load pressure index threshold. If the grid load pressure index at any sampling moment is less than the first load pressure index threshold, charge the energy storage device of the grid. If the grid load pressure index at any sampling moment is greater than the second load pressure index threshold, discharge the energy storage device of the grid.

[0084] In an embodiment, set the first load pressure index threshold T1 to 0.3 and the second load pressure index threshold T2 to 0.65. When Y0 < T1, it indicates that the current sampling moment does not belong to the peak period of the grid load, and the excess electric energy in the grid can be recovered and stored, that is, charge the energy storage device of the grid. When Y0 > T2, it indicates that the current sampling moment belongs to the peak period of the grid load, and electric energy supply is required to avoid phenomena such as power interruption caused by excessive grid load or waste of electric energy caused by insufficient utilization of electric energy due to small grid load. Therefore, discharge the energy storage device of the grid to balance the grid load and relieve the grid pressure. When T1 ≤ Y0 ≤ T2, keep the grid operation state unchanged, and neither charge nor discharge the energy storage device of the grid.

[0085] Thus, through the above method for obtaining the grid load pressure index, it is possible to judge the charging and discharging of the grid at each sampling moment of the i-th day, so as to improve the response ability of the energy storage device of the grid, ensure power consumption safety, and enhance the economic benefits and energy efficiency management level of the energy storage system.

[0086] Based on the same inventive concept as the above method, an embodiment of the present invention also provides an energy storage low-frequency and low-voltage conversion discharge control system, including a memory and a processor. The processor executes the computer program stored in the memory to implement the steps of any one of the above methods for an energy storage low-frequency and low-voltage conversion discharge control method.

[0087] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A control method for energy storage low-frequency and low-voltage conversion and discharge, characterized in that The method includes: Obtaining at least two consecutive historical time period's initial load data sequences of the power grid before a target time period respectively, and performing adaptive smoothing processing on each of the initial load data sequences to obtain corresponding smoothed data sequences; Constructing a load data change curve for each of the smoothed data sequences, where the horizontal axis of the load data change curve is time and the vertical axis is load data. According to the time interval between each historical time period and the target time period, and each load data change curve, obtaining the data fitting weight of each historical time period for the target time period; Constructing a load data change prediction curve for the target time period according to the data fitting weight of each historical time period for the target time period. According to the difference between adjacent data points in the load data change prediction curve, dividing the load data change prediction curve into at least two sub-curves, and obtaining the load change trend factor of the corresponding sub-curve according to the data difference in each sub-curve; For any sampling moment within the target time period, obtaining the power grid load pressure index of the any sampling moment according to the actual load data of the any sampling moment and the load change trend factor of the sub-curve to which the any sampling moment belongs, and performing real-time charge-discharge control on the energy storage device of the power grid according to the power grid load pressure index of the any sampling moment; The constructing the load data change prediction curve for the target time period according to the data fitting weight of each historical time period for the target time period includes: Taking the data fitting weight of each historical time period for the target time period as the weight of each load data in the corresponding smoothed data sequence. For any sampling moment within the target time period, obtaining the load data corresponding to the same sampling moment as the any sampling moment in the smoothed data sequence corresponding to each historical time period to form the historical load data sequence of the any sampling moment, and performing weighted mean processing on the historical load data sequence to obtain the load prediction value of the any sampling moment; Obtaining the load prediction values of each sampling moment within the target time period, and constructing a load data change prediction curve composed of the load prediction values of each sampling moment within the target time period, where the horizontal axis of the load data change prediction curve is time and the vertical axis is the load prediction value; The dividing the load data change prediction curve into at least two sub-curves according to the difference between adjacent data points in the load data change prediction curve includes: For any data point on the load data change prediction curve, obtaining K data points on the left side of the any data point on the load data change prediction curve to form a left adjacent data sequence, and performing first-order difference processing on the left adjacent data sequence, and correspondingly obtaining the mean value of the difference values in the first-order difference sequence, denoted as the left difference value mean; Obtain K data points on the right side of the any data point on the predicted load data change curve to form a right adjacent data sequence, perform a first-order difference process on the right adjacent data sequence, and correspondingly obtain the mean value of the difference values in the first-order difference sequence, denoted as the right difference value mean; If there is no left difference value mean or right difference value mean for the any data point, set the confidence level of the any data point as an inflection point to 0; if there are both the left difference value mean and the right difference value mean for the any data point, calculate the absolute value of the difference between the left difference value mean and the right difference value mean as the confidence level of the any data point as an inflection point; Obtain the confidence level of the left adjacent data point of the any data point as an inflection point and the confidence level of the right adjacent data point of the any data point as an inflection point. If the confidence level of the any data point as an inflection point is greater than both the confidence level of the left adjacent data point as an inflection point and the confidence level of the right adjacent data point as an inflection point, mark the any data point as an inflection point; Obtain at least one inflection point on the predicted load data change curve, use the inflection point as a curve segmentation point, and divide the predicted load data change curve into at least two sub-curves; The obtaining of the load change trend factor corresponding to each sub-curve according to the data difference in each sub-curve includes: For any sub-curve, obtain the difference between each data point on the any sub-curve and its left adjacent data point on the predicted load data change curve, use each difference as the independent variable of a preset exponential function to obtain the corresponding exponential function result, and obtain the load change trend factor of the any sub-curve according to the mean value of all exponential function results.

2. The energy storage low-frequency and low-voltage conversion and discharge control method according to claim 1, wherein The respectively performing adaptive smoothing processing on each initial load data sequence to obtain the corresponding smoothed data sequence includes: For any load data in any initial load data sequence, take the any load data as the center, obtain a preset number of adjacent load data in the any initial load data sequence, form a target data sequence with the any load data, perform a first-order difference process on the target data sequence to obtain a first-order difference sequence, and calculate the coefficient of variation of the first-order difference sequence; Use the negative value of the coefficient of variation as the independent variable of a preset exponential function to obtain the corresponding exponential function value, obtain the subtraction value of 1 and the exponential function value, round up the product of the subtraction value and a preset hyperparameter to an odd integer, and use the odd integer as the adaptive filtering window when performing mean filtering on the any load data; Obtain the adaptive filtering window corresponding to each load data in the any initial load data sequence, and perform mean filtering on the any initial load data sequence according to the adaptive filtering window corresponding to each load data in the any initial load data sequence to obtain the corresponding smoothed data sequence.

3. A method for controlling energy storage low-frequency and low-voltage transfer discharge according to claim 1, characterized in that, The obtaining of the data fitting weight of each historical time period for the target time period according to the time interval between each historical time period and the target time period, and each load data change curve includes: Obtain the DTW distance between every two of the load data change curves respectively. According to all the DTW distances and the time interval between each historical time period and the target time period, obtain the data reference degree of each historical time period respectively. Obtain the data reference degree of each historical time period, and get the sum of the data reference degrees. For any historical time period, calculate the ratio of the data reference degree of the any historical time period to the sum of the data reference degrees to obtain the data fitting weight of the any historical time period for the target time period.

4. The energy storage low-frequency and low-voltage transfer discharge control method according to claim 3, wherein, The obtaining the data reference degree of each historical time period according to all the DTW distances and the time interval between each historical time period and the target time period respectively includes: For any historical time period, according to the DTW distances between the load data change curve corresponding to the any historical time period and the load data change curves corresponding to each other historical time period respectively, use a preset Gaussian kernel function to obtain the local density value of the load data change curve corresponding to the any historical time period. Obtain the time interval between the any historical time period and the target time period. According to the product of the reciprocal of the time interval and the local density value, obtain the data reference degree of the any historical time period.

5. A method for controlling energy storage low-frequency and low-voltage transfer discharge according to claim 1, characterized in that The obtaining the grid load pressure index of any sampling moment according to the actual load data of any sampling moment and the load change trend factor of the sub-curve to which any sampling moment belongs includes: Obtain the maximum load capacity that the grid can bear, calculate the ratio between the actual load data of any sampling moment and the maximum load capacity that the grid can bear, obtain the product of the ratio and the load change trend factor of the sub-curve to which any sampling moment belongs, and perform normalization processing on the product to obtain the grid load pressure index of any sampling moment.

6. A method for controlling energy storage low-frequency and low-voltage transfer discharge according to claim 1, characterized in that The performing real-time charge and discharge control on the energy storage device of the grid according to the grid load pressure index of any sampling moment includes: Obtain a preset first load pressure index threshold and a second load pressure index threshold, and the first load pressure index threshold is less than the second load pressure index threshold. If the grid load pressure index of any sampling moment is less than the first load pressure index threshold, charge the energy storage device of the grid. If the grid load pressure index of any sampling moment is greater than the second load pressure index threshold, discharge the energy storage device of the grid.

7. A control system for energy storage low-frequency and low-voltage transfer and discharge, comprising a memory and a processor, characterized in that, The processor executes the computer program stored in the memory to implement a method for controlling charge and discharge of energy storage at low frequency and low voltage as described in any one of claims 1-6.

Images