Energy scheduling and energy-saving optimization method for smart power grid

Abstract

The invention relates to the technical field of data processing, in particular to an energy scheduling and energy-saving optimization method of a smart power grid. Obtaining information richness according to the frequency spectrum data of the generating capacity time sequence; obtaining a self-adaptive neighborhood window according to the spectrum data and the information richness; obtaining a power generation reference weight according to the time interval characteristics; obtaining power generation similarity according to the power generation quantity difference characteristics and the power generation reference weight in the adaptive neighborhood window at any moment and the same moment of other dates; and obtaining a change difference degree according to the difference characteristic and the time interval characteristic of the generating capacity change at any moment and other moments in the adaptive neighborhood window. According to the power generation similarity and the change difference degree, the robustness weight at any moment is obtained, time sequence decomposition is carried out, prediction and fusion are carried out according to the decomposition result, electric energy scheduling is carried out according to the power generation prediction data, and the accuracy of data decomposition and the reasonability of electric energy scheduling are improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method for energy scheduling and energy saving optimization of a smart grid. Background Art

[0002] As the scale of clean energy power stations continues to expand, although they have the advantages of low cost and environmental protection in power supply, there are intermittent and fluctuating conditions in the power generation process. For example, wind power generation is affected by changes in wind direction and wind speed, and photovoltaic power generation is affected by sunlight intensity. The instability of clean energy brings certain challenges to the energy scheduling of the power grid.

[0003] To improve the matching stability between the power supply and load sides of the power grid and avoid energy shortages and waste, existing technologies decompose power plant power generation data using the STL time series decomposition algorithm. LSTM long short-term memory neural networks are then used to predict the seasonal, trend, and residual terms after decomposition. The prediction results are then integrated to obtain more accurate future power generation forecasts. Power scheduling is then performed based on these forecasts, improving the matching between the power supply and load sides. The robustness weight of each data point in the STL decomposition directly influences the final decomposition results. However, clean energy generation is often affected by environmental factors, resulting in significant fluctuations in power generation data at certain times. If the robustness weight at any given moment is poorly aligned with the power generation characteristics, the decomposition results can be inaccurate, ultimately impacting the power generation forecast and the rationality of power scheduling. Summary of the Invention

[0004] In order to solve the technical problem that the robustness weights of each moment in the STL decomposition of the above-mentioned clean energy power generation time series affect the accuracy of the decomposition results, thereby resulting in low rationality of power generation prediction results and power scheduling, the purpose of the present invention is to provide an energy scheduling and energy-saving optimization method for a smart grid. The technical solutions adopted are as follows:

[0005] Obtain the power generation time series of the power station;

[0006] Obtaining information richness based on amplitude difference characteristics of different frequencies in the spectrum data of the power generation time series; obtaining an adaptive neighborhood window based on the spectrum data and the information richness;

[0007] Obtain a power generation reference weight based on the time interval characteristics between any moment in the power generation time series and the same moment on other dates; obtain power generation similarity based on the power generation difference characteristics between any moment and the same moment on other dates within the adaptive neighborhood window and the power generation reference weight at the same moment on other dates; obtain the degree of change difference based on the difference characteristics and time interval characteristics of the power generation changes between any moment and other moments in the adaptive neighborhood window;

[0008] The robustness weight at any moment is obtained according to the power generation similarity and the degree of change difference; the robustness weights of all moments in the power generation time series are decomposed through a time series decomposition algorithm, and prediction and fusion are performed according to the decomposition results to obtain power generation prediction data; and power scheduling is performed according to the power generation prediction data.

[0009] Furthermore, the step of obtaining information richness based on amplitude difference characteristics of different frequencies in the spectrum data of the power generation time series includes:

[0010] In the formula, R represents the information richness, G represents the number of amplitudes in the spectrum data that are not constant 0, e represents the natural constant, and F max Indicates the maximum amplitude in the spectrum data, F g Indicates the gth amplitude in the spectrum data that is not a constant 0 and is not the maximum value. Indicates the information contribution of the g-th amplitude, H max Indicates the frequency corresponding to the maximum amplitude in the spectrum data, H g Indicates the frequency corresponding to the g-th amplitude, |H max -H g | indicates frequency difference.

[0011] Furthermore, the step of obtaining an adaptive neighborhood window according to the spectrum data and the information richness includes:

[0012] The ratio of the preset sampling frequency of the power generation time series to the frequency corresponding to the maximum amplitude value in the spectrum data is calculated to obtain a range reference; the information richness is negatively correlated with the information richness through a preset function, and the product with the range reference is calculated and the nearest odd number is taken to obtain the data length; any analysis moment in the power generation time series is used as the center of the adaptive neighborhood window, and the data length is used as the window length of the adaptive neighborhood window to obtain the adaptive neighborhood window at the arbitrary analysis moment.

[0013] Furthermore, the step of obtaining a power generation reference weight according to the time interval characteristics between any moment in the power generation time series and the same moment on other dates includes:

[0014] Calculate the inverse of the time interval between the arbitrary moment and the same moment on any other date to obtain a first value; calculate the sum of all first values corresponding to the arbitrary moment to obtain a second value; calculate the ratio of the first value to the second value to obtain the power generation reference weight of the same moment on any other date for the arbitrary moment.

[0015] Furthermore, the step of obtaining power generation similarity based on power generation difference characteristics between the arbitrary time and the same time on other dates within the adaptive neighborhood window and the power generation reference weights at the same time on other dates includes:

[0016] Where Q represents the power generation similarity at any time, N represents the number of the same time on other dates at any time, e represents a natural constant, Y represents the power generation time series segment within the adaptive neighborhood window at any time, X n represents the time series fragment of power generation within the adaptive neighborhood window at the same time on the nth other date, DTW(Y,X n ) represents Y and X n The dynamic time warping distance between Indicates the change similarity; T n It represents the reference weight of power generation at the same time on the nth other date.

[0017] Furthermore, the step of obtaining the degree of difference in change based on the difference characteristics and time interval characteristics of the power generation change at any moment and other moments in the adaptive neighborhood window includes:

[0018] Where W represents the degree of change difference, M represents the number of moments in the adaptive neighborhood window, e represents a natural constant, and K represents the sum of the power generation differences between any moment and the adjacent moments before and after it; L m represents the sum of the power generation differences between the mth other moment and the adjacent moments before and after within the adaptive neighborhood window at any moment; Indicates the change difference characteristic value; D m Represents the reciprocal of the time interval between the arbitrary moment and the mth other moment.

[0019] Furthermore, the step of obtaining the robustness weight at any moment according to the power generation similarity and the degree of change difference includes:

[0020] Calculate the sum of the degree of change difference at all moments within the adaptive neighborhood window at the arbitrary moment to obtain a fourth value; calculate the ratio of the degree of change difference at the arbitrary moment to the fourth value to obtain a change difference reference value; calculate the difference between a constant 1 and the change difference reference value to obtain a change normality; calculate the product of the power generation similarity and the change normality and normalize them to obtain the robustness weight at the arbitrary moment.

[0021] The present invention has the following beneficial effects:

[0022] In the present invention, obtaining information richness can characterize the frequency information richness of the power generation time series. Furthermore, based on the information richness and frequency data, the adaptive neighborhood window for power generation data analysis can be accurately determined. This allows for more accurate analysis of local characteristic variations in the power generation data within the time series, improving the accuracy of the robustness weights at each moment during the decomposition process. Since clean energy power stations exhibit seasonality during power generation, obtaining power generation reference weights can determine the credibility of similar power generation characteristics between different moments, further improving the accuracy of robustness weights. Obtaining power generation similarity can characterize the degree to which the power generation characteristics at any moment conform to normal power generation, thereby determining the magnitude of the robustness weights. Obtaining the degree of variation can characterize the degree of variation in local power generation characteristics at different moments within the adaptive neighborhood window, reducing errors in robustness weight acquisition. Finally, the power generation time series is decomposed based on the robustness weights at all moments, improving decomposition accuracy and enhancing the accuracy of power generation forecast data. Power scheduling based on power generation forecast data improves the rationality of power scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0024] Figure 1 A flow chart of a method for energy scheduling and energy saving optimization of a smart grid provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0025] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of a smart grid energy scheduling and energy-saving optimization method proposed by the present invention. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0026] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0027] The following describes in detail a specific solution of a smart grid energy scheduling and energy-saving optimization method provided by the present invention with reference to the accompanying drawings.

[0028] See also Figure 1 , which shows a flow chart of a method for energy scheduling and energy saving optimization of a smart grid provided by an embodiment of the present invention, the method comprising the following steps:

[0029] Step S1, obtaining the power generation time series of the power station.

[0030] In this embodiment of the present invention, the implementation scenario is to dispatch power from clean energy power stations to improve the energy efficiency and rationality of power dispatch. First, the power station's power generation time series is obtained. The historical power generation time series of any clean energy power station up to the current time is obtained. In this embodiment of the present invention, 30 days of power generation data are collected, and the collection frequency is once per second; implementers can determine this frequency based on the implementation scenario.

[0031] Step S2: obtaining information richness according to the amplitude difference characteristics of different frequencies in the spectrum data of the power generation time series; and obtaining an adaptive neighborhood window according to the spectrum data and the information richness.

[0032] When power grids perform energy scheduling and energy conservation optimization, they typically forecast power plant generation data. These forecasts are then matched and dispatched based on the predicted power output and grid load demand. Accurate power generation forecasts are crucial for ensuring the rationality of power scheduling. Clean energy power stations exhibit low power stability. For example, photovoltaic power stations experience fluctuations and periodicity in power generation due to varying sunlight intensity at different times of the day, making power generation forecasting more difficult. To improve forecast accuracy, the power generation time series can be decomposed using the existing STL time series decomposition algorithm to produce seasonal, trend, and residual terms. These terms are then predicted separately and integrated to produce the final power generation forecast. During the STL decomposition process, each data point has a corresponding robust weight, which can reduce the impact of outliers on the decomposition results. However, the power generation of clean energy is easily affected by environmental factors, resulting in mutation characteristics in different situations in the power generation time series. The existing method of obtaining the robust weight of data points based on the residual method has low accuracy. Therefore, it is necessary to combine the data characteristics of the power generation time series to improve the rationality of the robust weight of each data point and improve the accuracy of decomposition and prediction.

[0033] Furthermore, when decomposing time series data, the change characteristics of the time series data need to be linked to the previous and next data, and analyzed in combination with the change patterns of the previous and next data. The robustness weights of the data points should also be determined in combination with the data change patterns. Since the power generation change patterns and trends of different power stations are different, when performing STL decomposition on the power generation time series of different power stations, the reference range for determining the robustness weights of the data points also differs. In order to more accurately obtain the robustness weights of the power station during decomposition, first, a suitable analysis window range is selected based on the information characteristics in the power generation time series of the power station; therefore, the information richness is obtained based on the amplitude difference characteristics of different frequencies in the spectrum data of the power generation time series; it should be noted that the spectrum data of the power generation time series is obtained by fast Fourier transform, and the specific steps will not be repeated; preferably, in an embodiment of the present invention, the step of obtaining information richness includes:

[0034]

[0035] In the formula, R represents the information richness, G represents the number of amplitudes in the spectrum data that are not constant 0, e represents the natural constant, and F max Indicates the maximum amplitude in the spectrum data, F g Indicates the gth amplitude in the spectrum data that is not a constant 0 and is not the maximum value. Indicates the information contribution of the g-th amplitude, H max Indicates the frequency corresponding to the maximum amplitude in the spectrum data, H g Indicates the frequency corresponding to the g-th amplitude, |H max -H g | indicates the frequency difference. H max is the main frequency of the current signal, representing the overall change characteristics, while the remaining frequencies represent local fluctuations superimposed on the overall change characteristics; if the amplitudes of the remaining frequency signals except the main frequency signal are higher and the difference from the main frequency is greater, that is, the information contribution and frequency difference are greater, then it means that the frequency information of the current signal is richer and the fluctuation characteristics are more obvious. Therefore, when analyzing the robustness weight, a smaller window range should be selected to enable a more detailed analysis of the local characteristic changes of the power generation time series.

[0036] Furthermore, when the information richness is greater, it means that the frequency information of the power generation time series is richer, so when analyzing the robustness weight, a smaller window range should be selected, so that the local characteristic changes of the power generation time series can be analyzed more carefully. Therefore, an adaptive neighborhood window is obtained according to the spectrum data and information richness; preferably, in an embodiment of the present invention, the step of obtaining an adaptive neighborhood window includes: calculating the ratio of the preset sampling frequency of the power generation time series to the frequency corresponding to the maximum amplitude value in the spectrum data to obtain a range reference; in an embodiment of the present invention, the preset sampling frequency is 1 Hz, and the range reference represents the number of power generation data points contained in a periodic signal, and then the range reference is used as a reference value for range determination. After the information richness is negatively correlated with the preset function, the product with the range reference is calculated and the nearest odd number is taken to obtain the data length; in an embodiment of the present invention, the preset function is In the formula, S is the mapping value, e represents a natural constant, and R represents information richness. The mapping range is (0, 3). Implementers can determine this value based on their implementation scenario. The greater the information richness, the smaller the window range and the shorter the data length. The adaptive neighborhood window at any analysis moment in the power generation time series is used as the center of the adaptive neighborhood window. The adaptive neighborhood window at any analysis moment is used as the time for robustness weight analysis. The data length is used as the window length of the adaptive neighborhood window to obtain the adaptive neighborhood window at that arbitrary analysis moment. This obtains the adaptive neighborhood window for each moment in the power generation time series. Subsequent robustness weight analysis can be performed based on the data within this adaptive neighborhood window.

[0037] Step S3, obtaining a power generation reference weight based on the time interval characteristics between any moment in the power generation time series and the same moment on other dates; obtaining a power generation similarity based on the power generation difference characteristics between any moment and the same moment on other dates within the adaptive neighborhood window and the power generation reference weight at the same moment on other dates; obtaining a degree of change difference based on the difference characteristics and time interval characteristics of the power generation change between any moment and other moments in the adaptive neighborhood window.

[0038] Because the power generation characteristics of clean energy power stations at the same time of day have a strong correlation, for example, in photovoltaic power generation, the angle of sunlight at noon is similar every day, so the power generation and change trends at noon are also relatively similar. This can be used to analyze the power generation characteristics of the same time period each day to determine whether the change patterns are similar. However, due to certain seasonal variations in the time series caused by factors such as the natural environment, two identical time periods with a close time interval will have a higher reference weight when analyzing power generation characteristics. Therefore, the power generation reference weight is obtained based on the time interval characteristics between any time in the power generation time series and the same time on other days.

[0039] Preferably, in an embodiment of the present invention, the step of obtaining a power generation reference weight includes: calculating the inverse of the time interval between any moment and the same moment on any other date to obtain a first value; the closer the time interval, the larger the first value, indicating that the closer the time interval between the same moment on any other date and the arbitrary moment, the higher the credibility of the power generation characteristic analysis results between the two. Calculating the sum of all first values corresponding to the arbitrary moment to obtain a second value; the purpose is to ensure that the sum of the power generation reference weights for the same moment on all other dates is 1. Calculating the ratio of the first value to the second value to obtain the power generation reference weight for the same moment on any other date with respect to the arbitrary moment, the sum of all power generation reference weights being 1.

[0040] Furthermore, if the similarity between the power generation data change trends of the adaptive neighborhood window at any moment and the adaptive neighborhood windows at the same moment on all other dates is high, it means that the power generation characteristics at that moment are relatively close to the data during normal power generation. Therefore, the power generation similarity can be obtained based on the power generation difference characteristics within the adaptive neighborhood window at any moment and the same moment on other dates, and the power generation reference weights at the same moment on other dates; preferably, in the implementation of the present invention, the step of obtaining the power generation similarity includes:

[0041]

[0042] Where Q represents the power generation similarity at any moment, N represents the number of the same moments on other dates at any moment, e represents a natural constant, Y represents the power generation time series segment within the adaptive neighborhood window at any moment, X represents the power generation time series segment within the adaptive neighborhood window at any moment, n represents the time series fragment of power generation within the adaptive neighborhood window at the same time on the nth other date, DTW(Y,X n ) represents Y and X n It should be noted that the dynamic time warping distance is obtained through the existing dynamic time warping algorithm. The more similar the changing trends of the two sequences are, the smaller the dynamic time warping distance is. Indicates the change similarity. The dynamic time warping distance is negatively correlated with each other through the exponential function. The greater the change similarity, the more similar the power generation change trends in the two periods are. n Represents the reference weight of power generation at the same time on the nth other date. A larger reference weight indicates a closer time interval between two identical periods, and a greater confidence level in the similarity of changes. Furthermore, a greater similarity indicates that the power generation characteristics at that particular moment are more consistent with normal power generation conditions.

[0043] Furthermore, the greater the power generation similarity at any moment, the more normal the power generation characteristics at any moment, and thus the greater the robustness weight of the power generation data at any moment during STL decomposition; but since the result of power generation similarity is obtained based on the power generation data at all moments within the adaptive neighborhood window, if the data at other moments within the adaptive neighborhood window at any moment have mutation characteristics, the power generation similarity at any moment will be low, thereby affecting the accuracy of the robustness weight. Therefore, it is necessary to obtain the degree of difference between the local change characteristics of power generation at any moment and the local change characteristics of power generation at other moments, and further obtain the robustness weight based on the degree of difference; and then obtain the degree of change difference based on the difference characteristics and time interval characteristics of the power generation changes at any moment and other moments within the adaptive neighborhood window; preferably, in an embodiment of the present invention, the step of obtaining the degree of change difference includes:

[0044]

[0045] In the formula, W represents the degree of change difference, M represents the number of moments in the adaptive neighborhood window, e represents a natural constant, and K represents the sum of the differences between the power generation at any moment and the adjacent moments before and after it. The less K is close to 0, the more likely it is that the power generation at any moment has a mutation feature. m It represents the sum of the power generation differences between the mth other moments and the adjacent moments within the adaptive neighborhood window at any moment; when L m The closer it is to 0, the more likely it is that the power generation at other times will have mutation characteristics. Indicates the change difference characteristic value; when the change difference characteristic value is larger, it means that the difference in the local change characteristics of power generation at two moments is greater. m It represents the inverse of the time interval between this arbitrary moment and the mth other moment. Since the closer the time interval, the more similar the power generation variation characteristics are, so D m The larger the value, the higher the credibility of the corresponding change difference characteristic value. The degree of change difference represents the degree of difference between the local change characteristics of power generation at any time and other times within the adaptive neighborhood window. The larger the degree of change difference, the greater the difference in the local change characteristics of power generation.

[0046] Step S4: Obtain the robustness weight at any moment based on the power generation similarity and the degree of change difference; Decompose the robustness weights of all moments in the power generation time series using a time series decomposition algorithm, perform prediction and fusion based on the decomposition results, and obtain power generation prediction data; Perform power scheduling based on the power generation prediction data.

[0047] After obtaining the power generation similarity and variation difference, a robustness weight for any moment can be obtained based on the power generation similarity and variation difference. Preferably, in an embodiment of the present invention, the step of obtaining the robustness weight for any moment includes: calculating the sum of the variation difference at all moments within the adaptive neighborhood window at that moment to obtain a fourth value; and calculating the ratio of the variation difference at that moment to the fourth value to obtain a variation difference baseline value. The variation difference baseline value represents the proportion of the variation difference at that moment in the fourth value. The smaller the variation difference at that moment, the smaller the variation difference baseline value, indicating that the local variation characteristics of power generation at other moments within the adaptive neighborhood window are more obvious and exhibit certain mutation characteristics, while the local variation characteristics of power generation at that moment are smaller. Therefore, the mutation characteristics at other moments will cause the calculated power generation similarity and robustness weight at that moment to be too small, affecting the decomposition accuracy. Therefore, the smaller the variation difference baseline value, the larger the robustness weight for that moment should be. The difference between the constant 1 and the variation difference baseline value is calculated to obtain the variation normality. The smaller the variation difference baseline value, the greater the variation normality. The product of power generation similarity and variation normality is calculated and normalized to obtain the robustness weight at any moment. The greater the power generation similarity and variation normality, the more consistent the power generation characteristics at any moment are with the normal power generation trend, and the greater the robustness weight at any moment.

[0048] Furthermore, after obtaining the robustness weights of all moments in the power generation time series, the robustness weights of all moments in the power generation time series can be decomposed using a time series decomposition algorithm, and prediction and fusion can be performed based on the decomposition results to obtain power generation forecast data. It should be noted that the STL time series decomposition algorithm belongs to the existing technology, and the specific decomposition steps will not be repeated. The robustness weights obtained based on the data characteristics of the power generation time series are more accurate than the robustness weights obtained based on the residual characteristics. The volatility and periodic variation characteristics of the power generation of clean energy power stations are taken into account in the process of obtaining the robustness weights, ultimately improving the accuracy of the decomposition results. After the decomposition is completed, the seasonal term, trend term, and residual term are obtained. In the embodiment of the present invention, the seasonal term, trend term, and residual term are predicted separately using the existing LSTM long short-term memory neural network, thereby improving the prediction accuracy. It should be noted that the LSTM neural network belongs to the existing technology, and the specific prediction steps will not be repeated. All forecast results are combined to obtain power generation forecast data; ultimately, power dispatch is performed based on this forecast data. This forecast data can be matched with power consumption in different grid regions, selecting the region with the closest power generation and consumption for power transmission, thus avoiding energy waste. Implementers can customize power dispatch rules based on the forecast data, without further restrictions. Thus, by analyzing historical power generation time series, robustness weights are obtained for each moment, thereby improving the accuracy of power generation time series decomposition and forecast results, and making power dispatch more rational.

[0049] In summary, the embodiments of the present invention provide a method for energy scheduling and energy conservation optimization of a smart grid; obtain information richness based on spectrum data of a power generation time series; obtain an adaptive neighborhood window based on spectrum data and information richness; obtain a power generation reference weight based on time interval characteristics; obtain power generation similarity based on power generation difference characteristics and power generation reference weights at any time and at the same time on other dates within the adaptive neighborhood window; obtain a degree of change difference based on power generation change difference characteristics and time interval characteristics at any time and other times within the adaptive neighborhood window. The present invention obtains a robustness weight at any time based on power generation similarity and degree of change difference and performs time series decomposition, performs prediction and fusion based on the decomposition results, and performs power scheduling based on power generation prediction data, thereby improving the accuracy of data decomposition and the rationality of power scheduling.

[0050] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0051] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for energy dispatching and energy saving optimization of a smart grid, characterized in that: The method comprises the following steps: Obtain the power generation time series of the power station; Obtaining information richness based on amplitude difference characteristics of different frequencies in the spectrum data of the power generation time series; obtaining an adaptive neighborhood window based on the spectrum data and the information richness; Obtain a power generation reference weight based on the time interval characteristics between any moment in the power generation time series and the same moment on other dates; obtain power generation similarity based on the power generation difference characteristics between any moment and the same moment on other dates within the adaptive neighborhood window and the power generation reference weight at the same moment on other dates; obtain the degree of change difference based on the difference characteristics and time interval characteristics of the power generation changes between any moment and other moments in the adaptive neighborhood window; The robustness weight at any moment is obtained according to the power generation similarity and the degree of change difference; the robustness weights of all moments in the power generation time series are decomposed through a time series decomposition algorithm, and prediction and fusion are performed according to the decomposition results to obtain power generation prediction data; and power scheduling is performed according to the power generation prediction data.

2. The method for energy dispatching and energy saving optimization of a smart grid according to claim 1, characterized in that: The step of obtaining information richness according to the amplitude difference characteristics of different frequencies in the spectrum data of the power generation time series includes: In the formula, R represents the information richness, G represents the number of amplitudes in the spectrum data that are not constant 0, e represents the natural constant, and F max Indicates the maximum amplitude in the spectrum data, F g Indicates the gth amplitude in the spectrum data that is not a constant 0 and is not the maximum value. Indicates the information contribution of the g-th amplitude, H max Indicates the frequency corresponding to the maximum amplitude in the spectrum data, H g Indicates the frequency corresponding to the g-th amplitude, |H max -H g | indicates frequency difference.

3. The method for energy dispatching and energy saving optimization of a smart grid according to claim 1, characterized in that: The step of obtaining an adaptive neighborhood window according to the spectrum data and the information richness comprises: The ratio of the preset sampling frequency of the power generation time series to the frequency corresponding to the maximum amplitude value in the spectrum data is calculated to obtain a range reference; the information richness is negatively correlated with the information richness through a preset function, and the product with the range reference is calculated and the nearest odd number is taken to obtain the data length; any analysis moment in the power generation time series is used as the center of the adaptive neighborhood window, and the data length is used as the window length of the adaptive neighborhood window to obtain the adaptive neighborhood window at the arbitrary analysis moment.

4. The method for energy dispatching and energy saving optimization of a smart grid according to claim 1, characterized in that: The step of obtaining a power generation reference weight according to the time interval characteristics between any moment in the power generation time series and the same moment on other dates comprises: Calculate the inverse of the time interval between the arbitrary moment and the same moment on any other date to obtain a first value; calculate the sum of all first values corresponding to the arbitrary moment to obtain a second value; calculate the ratio of the first value to the second value to obtain the power generation reference weight of the same moment on any other date for the arbitrary moment.

5. The method for energy dispatching and energy saving optimization of a smart grid according to claim 1, characterized in that: The step of obtaining power generation similarity based on power generation difference characteristics between the arbitrary time and the same time on other dates within the adaptive neighborhood window and the power generation reference weights at the same time on other dates includes: Where Q represents the power generation similarity at any time, N represents the number of the same time on other dates at any time, e represents a natural constant, Y represents the power generation time series segment within the adaptive neighborhood window at any time, X n represents the time series fragment of power generation within the adaptive neighborhood window at the same time on the nth other date, DTW(Y,X n ) represents Y and X n The dynamic time warping distance between Indicates the change similarity; T n It represents the reference weight of power generation at the same time on the nth other date.

6. The method for energy dispatching and energy saving optimization of a smart grid according to claim 1, characterized in that: The step of obtaining the degree of difference in power generation according to the difference characteristics and time interval characteristics of the power generation change at any moment and other moments in the adaptive neighborhood window includes: Where W represents the degree of change difference, M represents the number of moments in the adaptive neighborhood window, e represents a natural constant, and K represents the sum of the power generation differences between any moment and the adjacent moments before and after it; L m represents the sum of the power generation differences between the mth other moment and the adjacent moments before and after within the adaptive neighborhood window at any moment; Indicates the change difference characteristic value; D m Represents the reciprocal of the time interval between the arbitrary moment and the mth other moment.

7. The method for energy dispatching and energy saving optimization of a smart grid according to claim 1, characterized in that: The step of obtaining the robustness weight at any moment according to the power generation similarity and the degree of change difference includes: Calculate the sum of the degree of change difference at all moments within the adaptive neighborhood window at the arbitrary moment to obtain a fourth value; calculate the ratio of the degree of change difference at the arbitrary moment to the fourth value to obtain a change difference reference value; calculate the difference between a constant 1 and the change difference reference value to obtain a change normality; calculate the product of the power generation similarity and the change normality and normalize them to obtain the robustness weight at the arbitrary moment.

Images