New energy load prediction system and method based on low-voltage bus

By setting up electrical energy and natural detection devices on the low-voltage busbar, combining social factors to generate frequency band load prediction formulas, the problem of neglecting environmental factors in the existing technology of new energy load prediction is solved, the prediction accuracy is improved, and the rationality of new energy is promoted. Utilization and grid stability.

CN119009938APending Publication Date: 2024-11-22STATE GRID ZHEJIANG ELECTRIC POWER CO LTD SHAOXING POWER SUPPLY CO +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410905411.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-08
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing new energy load prediction methods ignore the influence of environmental factors, resulting in a large gap between the forecast results and the actual situation, resulting in waste of power resources and grid stability problems.

Method used

Set up electricity detection and natural detection devices at each low-voltage busbar to obtain electricity consumption and natural data, combine social factors, generate frequency band data sets and establish frequency band load prediction formulas to generate new energy load prediction models.

Benefits of technology

By considering environmental factors, the accuracy and comprehensiveness of load forecasting are improved, helping to optimize the generation and distribution of new energy, and promoting rational utilization and sustainable development.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119009938A_ABST
    Figure CN119009938A_ABST
Patent Text Reader

Abstract

A new energy load prediction method based on low-voltage buses overcomes the defects in the prior art and comprises the steps that 1, an electric energy detection device and a natural detection device are arranged at each low-voltage bus, the electricity consumption of the low-voltage buses is obtained regularly through the electric energy detection devices, and a natural data set of the low-voltage buses is obtained regularly through the natural detection devices; step 2, obtaining a characteristic time zone of each low-voltage bus according to the electricity consumption; generating a plurality of frequency band regions according to the characteristic time zone of each low-voltage bus; 3, obtaining social factors, regularly obtaining a social data set according to the social factors, and generating a correlation coefficient set according to the frequency band region, the natural data set and the social data set; step 4, generating a frequency band data set, generating a frequency band load prediction formula according to the frequency band data set, and generating a new energy load prediction model according to the frequency band load prediction formula; and step 5, predicting the new energy load by the new energy load prediction model.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of electric load, and in particular to a new energy load prediction system and method based on a low-voltage bus. Background Art

[0002] With the continuous increase in global energy demand and the improvement of environmental protection awareness, the proportion of renewable energy such as wind power and solar energy in the power system has gradually increased. However, these new energy sources are intermittent, volatile and uncontrollable, which brings new challenges to the stable operation of the power system. Accurately predicting the load of new energy sources is an important means to optimize the allocation of power resources and ensure the stable operation of the power grid.

[0003] Environmental factors have a certain direct effect on the prediction of renewable energy load. At the same time, the mutual influence among environmental factors has a certain indirect effect on the prediction of renewable energy load. The existing renewable energy load prediction methods often ignore the impact of environmental factors. The prediction of renewable energy load is quite different from the actual results, which leads to the waste of power resources and affects the stability of power grid operation. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the existing renewable energy load forecasting methods, which often ignore the influence of environmental factors, and the large difference between the forecast and actual results of renewable energy load, resulting in waste of power resources and affecting the stability of power grid operation, and to provide a renewable energy load forecasting system and method based on a low-voltage bus.

[0005] The purpose of the present invention is achieved through the following technical solutions: A new energy load prediction method based on a low-voltage bus comprises the following steps: Step 1: respectively set an electric energy detection device and a natural detection device at each low-voltage bus, regularly obtain the power consumption of the low-voltage bus through the electric energy detection device, and regularly obtain the natural data set of the low-voltage bus through the natural detection device; Step 2, obtaining a characteristic time zone of each low-voltage bus according to the power consumption; generating a plurality of frequency bands according to the characteristic time zone of each low-voltage bus, wherein the frequency bands include a high frequency band, a flat frequency band and a low frequency band; Step 3, obtaining social factors, regularly obtaining social data sets based on social factors, and generating a correlation coefficient set based on the frequency band area, natural data set and social data set; Step 4: Generate a frequency band data set based on the power consumption, natural data set, social data set and correlation coefficient set, generate a frequency band load prediction formula based on the frequency band data set, and generate a new energy load prediction model based on the frequency band load prediction formula; Step 5: The new energy load prediction model predicts the new energy load.

[0006] Preferably, the step 1 is specifically as follows: The electric energy detection device is composed of an electric energy meter, through which the electric energy meter is used to obtain the power consumption. TIME , where TIME is the time when the power consumption is obtained; The natural detection device includes a thermometer, a hygrometer, an anemometer, a rain gauge, a barometer and a visibility meter, and the corresponding temperature, humidity, wind speed, rainfall, air pressure and visibility are obtained through the thermometer, hygrometer, anemometer, rain gauge, barometer and visibility meter, and the temperature, humidity, wind speed, rainfall, air pressure and visibility are marked as natural factors; Generate natural data sets based on natural factors and natural values TIME , where TIME is the time when the natural data set is obtained; The natural value is generated by the following steps: Step 1a, generating corresponding positive state intervals, neutral state intervals and negative state intervals according to natural factors, wherein the positive state intervals, neutral state intervals and negative state intervals do not intersect with each other; Step 1b: Obtain the value corresponding to the natural factor through the natural detection device; Step 1c: if the value ∈ is in the positive state interval, the natural value of the natural factor is obtained to be positive; if the value ∈ is in the flat state interval, the natural value of the natural factor is obtained to be flat; if the value ∈ is in the negative state interval, the natural value of the natural factor is obtained to be negative; There is a one-to-one correspondence between TIME in electricity consumption and TIME in the natural data set.

[0007] Preferably, in step 2, the characteristic time zone of each low-voltage bus is obtained according to the power consumption: The characteristic time zones include peak time zones, flat time zones and valley time zones; Get the power consumption in chronological order TIME , and mark TIME as the power coordinate; Through the two adjacent power coordinates TIME Get electricity usage variables TIMEAREA , where TIMEAREA is the time zone corresponding to the electricity consumption variable; Set continuous threshold, peak power consumption and valley power consumption, where peak power consumption > valley power consumption; If N consecutive power consumption variables TIMEAREA If all of the N consecutive power consumption variables are greater than the peak power consumption value, and N ≥ the continuous threshold, the time zone corresponding to the N consecutive power consumption variables is marked as the peak time zone; If N consecutive power consumption variables TIMEAREA If all of the N consecutive power consumption variables are less than the power consumption valley value, and N ≥ the continuous threshold, the time zone corresponding to the N consecutive power consumption variables is marked as the valley time zone; Mark the time zones outside the peak and valley time zones as flat time zones.

[0008] Preferably, the process of generating a plurality of frequency bands according to the characteristic time zone of each low-voltage bus comprises: Establishing a one-dimensional time coordinate axis, obtaining characteristic time zones of each low-voltage bus, and mapping the characteristic time zones to the one-dimensional time coordinate axis; Set the overlap threshold; Obtain N consecutive peak time zones with overlapping time zones at the beginning and the end, obtain the total time zone corresponding to the peak time zone, and mark the total time zone as a high frequency band area; wherein N≥overlap threshold; Obtain N consecutive trough time zones with overlapping head and tail time zones, obtain the time period corresponding to the trough time zone, and mark the time period as a low frequency band area; wherein N≥overlap threshold; The time period other than the high frequency band area and the low frequency band area is marked as a flat frequency band area; if there is an overlapping time zone between the high frequency band area and the low frequency band area, the overlapping time zone is marked as a flat frequency band area.

[0009] Preferably, in step 3, generating a set of correlation coefficients according to the frequency band area, the natural data set and the social data set is specifically: The collection process includes: Acquire a natural data set and a social data set in a high-frequency band of TIME∈, and mark the natural data set as high-frequency natural data, and mark the social data set as high-frequency social data; Obtain explicit factors from high-frequency natural data and high-frequency social data, and mark the explicit factors as high explicit factors; obtain implicit factors from high-frequency natural data and high-frequency social data, and mark the implicit factors as high implicit factors; obtain normal factors from high-frequency natural data and high-frequency social data, and mark the normal factors as high normal factors; Acquire a natural data set and a social data set in the TIME∈ flat frequency band area, and mark the natural data set as standard natural data, and mark the social data set as standard social data; Obtain explicit factors from standard natural data and standard social data, and mark the explicit factors as marked explicit factors; obtain normal factors from standard natural data and standard social data, and mark the normal factors as marked normal factors; obtain implicit factors from standard natural data and standard social data, and mark the implicit factors as marked implicit factors; Obtain a natural data set and a social data set in the low-frequency band of TIME∈, and mark the natural data set as low-frequency natural data, and mark the social data set as low-frequency social data; Obtaining explicit factors from low-frequency natural data and low-frequency social data, and marking the explicit factors as low explicit factors; obtaining normal factors from low-frequency natural data and low-frequency social data, and marking the normal factors as low normal factors; obtaining implicit factors from low-frequency natural data and low-frequency social data, and marking the implicit factors as low latent factors; Obtaining a non-correlation factor, wherein the non-correlation factor = high constant factor ∩ standard constant factor ∩ low constant factor; Obtain a strong correlation factor, wherein the strong correlation factor=high-dominance factor+low-dominance factor-high-dominance factor∩low-dominance factor-average-dominance factor∩high-dominance factor-average-dominance factor∩low-dominance factor-no-correlation factor∩high-dominance factor-no-correlation factor∩low-dominance factor; Obtain a weak correlation factor, wherein the weak correlation factor=high latent factor+low latent factor-high latent factor∩low latent factor-average latent factor∩high latent factor-average latent factor∩low latent factor-uncorrelated factor∩high latent factor-uncorrelated factor∩low latent factor; Obtain the dominant factors that originally belonged to the highly dominant factors among the strongly correlated factors, mark the dominant factors as strongly correlated type A factors, and mark the other dominant factors among the strongly correlated factors as strongly correlated type B factors; obtain the recessive factors that originally belonged to the highly recessive factors among the weakly correlated factors, mark the recessive factors as weakly correlated type A factors, and mark the other recessive factors among the weakly correlated factors as weakly correlated type B factors; A strong confrontation index, a weak confrontation index and a non-confrontation index are set, among which the strong confrontation index>weak confrontation index>non-confrontation index; a correlation coefficient set is generated according to the strong correlation type A factor, the strong correlation type B factor, the weak correlation type A factor, the weak correlation type B factor, the non-correlation factor, the strong confrontation index, the weak confrontation index and the non-confrontation index.

[0010] Preferably, in step 4, the process of generating the frequency band data set according to the power consumption, the natural data set, the social data set and the correlation coefficient set includes: The frequency band dataset TIME∈D ={Amount of electricity consumption TIME , several natural datasets TIME , several social datasets TIME , correlation coefficient set}, where D includes a high frequency band area, a flat frequency band area and a low frequency band area.

[0011] Preferably, in step 4, the process of generating a frequency band load prediction formula according to the frequency band data set includes: Step 4a: Get the frequency band dataset TIME∈高频段区 ; Step 4b: Set the buffer time; Get the current time, generate the standard time zone = [current time - buffer time, current time); Step 4c: Get the frequency band data set TIME∈高频段区Electricity consumption in the standard time zone of TIME TIME and label the electricity consumption TIME as general electricity consumption; obtain the general value in the general electricity consumption, where the general value is the average of the values in the general electricity consumption whose integer part is the mode. If there is no mode in the integer part, obtain the median; Step 4d: Obtain the frequency band dataset TIME∈高频段区 The two natural datasets and social datasets closest to the current time in TIME, and label the natural dataset and social dataset as the general natural dataset and general social dataset respectively; Step 4e: Obtain the potential values of each natural factor in the general natural dataset, where the potential values include growth and decrease; Step 4f: Obtain the potential values of each social factor in the general social dataset; Step 4g: Set the standard index and pass rate value; obtain the electricity consumption TIME , natural dataset TIME and social dataset TIME closest to the current time in TIME, and label them as the pre-electricity consumption, pre-natural dataset and pre-social dataset respectively; generate high-frequency load prediction formulas, medium-frequency load prediction formulas and low-frequency load prediction formulas according to the pre-electricity consumption, pre-natural dataset, pre-social dataset, general value of the general electricity consumption, potential values of each natural factor in the general natural dataset, potential values of each social factor in the general social dataset and the correlation coefficient set.

[0012] Preferably, in step 4, generating the new energy load prediction model according to the frequency band load prediction formula specifically includes: obtaining the current time as the standard time, and obtaining the three electricity consumptions closest to the standard time in TIME at each low-voltage busbar TIME and accumulate and divide the electricity consumption TIME according to time to generate electricity consumption TIME-1 and electricity consumption TIME-2 and electricity consumption TIME-3 to generate electricity consumption where TIME - 3 < TIME - 2 < TIME - 1 < standard time; Set the high-frequency growth rate interval, medium-frequency growth rate interval and low-frequency growth rate interval, where the high-frequency growth rate interval, medium-frequency growth rate interval and low-frequency growth rate interval do not intersect; set the high-value interval, medium-value interval and low-value interval, where the high-value interval, medium-value interval and low-value interval do not intersect; If the electricity consumption growth rate ∈ the high-frequency growth rate and the electricity consumption TIME-1 ∈ the high-value interval ∪ medium-value interval, then predict the new energy load according to the high-frequency load prediction formula; if the electricity consumption growth rate ∈ the high-frequency growth rate and the electricity consumption TIME-1∈ low value interval, the new energy load is predicted according to the flat frequency load prediction formula; If the electricity consumption growth rate ∈ flat frequency growth rate, the electricity consumption TIME-1 ∈ high value interval, the new energy load is predicted according to the high frequency load prediction formula; if the electricity consumption growth rate ∈ flat frequency growth rate, the electricity consumption TIME-1 ∈ flat value interval, the new energy load is predicted by the flat frequency load prediction formula; if the electricity consumption growth rate ∈ flat frequency growth rate, the electricity consumption TIME-1 ∈ low value interval, the new energy load is predicted by the low-frequency load prediction formula; If the electricity consumption growth rate ∈ low-frequency growth rate, the electricity consumption TIME-1 ∈ high value interval ∪ flat value interval, the new energy load is predicted according to the flat frequency load prediction formula; if the electricity consumption growth rate ∈ low frequency growth rate, the electricity consumption TIME-1 ∈ low value range, the new energy load is predicted by the low-frequency load prediction formula.

[0013] A new energy load forecasting system based on a low-voltage bus is used to implement a new energy load forecasting method based on a low-voltage bus, comprising: The data acquisition module is used to respectively set an electric energy detection device and a natural detection device at each low-voltage bus; regularly obtain the power consumption of the low-voltage bus through the electric energy detection device, and regularly obtain the natural data set of the low-voltage bus through the natural detection device; obtain social factors, and regularly obtain social data sets according to the social factors; The data processing module is used to obtain the characteristic time zone of each low-voltage bus according to the power consumption; generate a number of frequency bands according to the characteristic time zone of each low-voltage bus, wherein the frequency bands include a high frequency band, a flat frequency band and a low frequency band; generate a correlation coefficient set according to the frequency band, the natural data set and the social data set; The data analysis module is used to generate a frequency band data set based on electricity consumption, a natural data set, a social data set and a correlation coefficient set; obtain the frequency band data set corresponding to the high-frequency band area to generate a high-frequency load prediction formula; obtain the frequency band data set corresponding to the flat frequency band area to generate a flat-frequency load prediction formula; obtain the frequency band data set corresponding to the low-frequency band area to generate a low-frequency load prediction formula; at the same time, generate corresponding early warning information according to the prediction formula; generate a new energy load prediction model according to the high-frequency load prediction formula, the flat-frequency load prediction formula and the low-frequency load prediction formula.

[0014] Preferably, the new energy load forecasting system based on the low-voltage bus also includes an early warning processing module for processing early warning information.

[0015] The beneficial effects of the present invention are as follows: by setting up an electric energy detection device and a natural detection device at each low-voltage bus, the present invention can obtain the power consumption and natural data sets in real time and accurately, providing accurate basic data for load forecasting; the introduction of social factors and social data sets takes into account the impact of environmental factors on load forecasting, making the forecasting model more comprehensive and accurate. The prediction of new energy loads through this solution helps to better plan and dispatch the power generation and distribution of new energy, and promote the rational use and sustainable development of new energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0017] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0018] Example: A new energy load forecasting method based on low voltage bus, such as Figure 1 As shown, the following steps are included: Step 1: respectively set an electric energy detection device and a natural detection device at each low-voltage bus, regularly obtain the power consumption of the low-voltage bus through the electric energy detection device, and regularly obtain the natural data set of the low-voltage bus through the natural detection device; Step 2, obtaining a characteristic time zone of each low-voltage bus according to the power consumption; generating a plurality of frequency bands according to the characteristic time zone of each low-voltage bus, wherein the frequency bands include a high frequency band, a flat frequency band and a low frequency band; Step 3, obtaining social factors, regularly obtaining social data sets based on social factors, and generating a correlation coefficient set based on the frequency band area, natural data set and social data set; Step 4: Generate a frequency band data set based on power consumption, natural data set, social data set and correlation coefficient set, obtain the frequency band data set corresponding to the high frequency band area, and generate a high frequency load prediction formula; obtain the frequency band data set corresponding to the flat frequency band area, and generate a flat frequency load prediction formula; obtain the frequency band data set corresponding to the low frequency band area, and generate a low frequency load prediction formula; generate a new energy load prediction model based on the high frequency load prediction formula, the flat frequency load prediction formula and the low frequency load prediction formula; Step 5: The new energy load prediction model predicts the new energy load.

[0019] The step 1 is specifically as follows: The electric energy detection device is composed of an electric energy meter, through which the electric energy meter is used to obtain the power consumption. TIME , where TIME is the time when the power consumption is obtained; The natural detection device includes a thermometer, a hygrometer, an anemometer, a rain gauge, a barometer and a visibility meter, and the corresponding temperature, humidity, wind speed, rainfall, air pressure and visibility are obtained through the thermometer, hygrometer, anemometer, rain gauge, barometer and visibility meter, and the temperature, humidity, wind speed, rainfall, air pressure and visibility are marked as natural factors; Generate natural data sets based on natural factors and natural values TIME , where TIME is the time when the natural data set is obtained; The natural value is generated by the following steps: Step 1a, generating corresponding positive state intervals, neutral state intervals and negative state intervals according to natural factors, where: That is, the positive state interval, the flat state interval and the negative state interval do not intersect each other; Step 1b: obtaining the value corresponding to the natural factor through the natural detection device; Step 1c: if the value ∈ is in the positive state interval, the natural value of the natural factor is obtained to be positive; if the value ∈ is in the flat state interval, the natural value of the natural factor is obtained to be flat; if the value ∈ is in the negative state interval, the natural value of the natural factor is obtained to be negative; Generate natural data sets based on natural factors and natural values TIME ={Natural Factor 温度 =Natural value 温度 , natural factors 湿度 =Natural value 湿度 , natural factors 风速 =Natural value 风速 , natural factors 降雨量 =Natural value 降雨量 , natural factors 气压 =Natural value 气压 , natural factors 能见度 =Natural value 能见度}, where TIME is the time when the natural data set is obtained; There is a one-to-one correspondence between TIME in electricity consumption and TIME in the natural data set.

[0020] In step 2, the characteristic time zone of each low-voltage bus is obtained according to the power consumption: The characteristic time zones include peak time zones, flat time zones and valley time zones; Get the power consumption in chronological order TIME , and mark TIME as the power coordinate; Through the two adjacent power coordinates TIME Get electricity usage variables TIMEAREA , where TIMEAREA is the time zone corresponding to the electricity consumption variable; Said Among them, TIME i -TIME i+1 Indicates that the time zone corresponding to the electricity consumption variable is [TIME i ,TIME i+1 ); Set continuous threshold, peak power consumption and valley power consumption, where peak power consumption > valley power consumption; If N consecutive power consumption variables TIMEAREA If all of the N consecutive power consumption variables are greater than the peak power consumption value, and N ≥ the continuous threshold, the time zone corresponding to the N consecutive power consumption variables is marked as the peak time zone; If there are N electricity consumption variables TIMEAREA , that is, existence are greater than the peak power consumption, and N ≥ continuous threshold, then The corresponding time zone is marked as the peak time zone, that is, the time zone [TIME1, TIME N+1 ) is the peak time zone; If N consecutive power consumption variables TIMEAREA If all of the N consecutive power consumption variables are less than the power consumption valley value, and N ≥ the continuous threshold, the time zone corresponding to the N consecutive power consumption variables is marked as the valley time zone; Mark the time zones other than the peak time zone and the trough time zone as the flat time zone. N ), where [TIME1, TIME i ) is the peak time zone, [TIME i+5 ,TIME N ) is the low time zone, then [TIME i ,TIME i+5 ) is a flat time zone; the process of generating a plurality of frequency bands according to the characteristic time zone of each low-voltage bus includes: Establishing a one-dimensional time coordinate axis, obtaining characteristic time zones of each low-voltage bus, and mapping the characteristic time zones to the one-dimensional time coordinate axis; Set the overlap threshold; Obtain N consecutive peak time zones with overlapping time zones at the beginning and the end, obtain the total time zone corresponding to the peak time zone, and mark the total time zone as a high frequency band area; wherein N≥overlap threshold; N consecutive peak time zones with overlapping time zones at the beginning and the end, that is, the peak time zones all have overlapping time zones at the beginning and the end, and the total time zones corresponding to the peak time zones are continuous time zones; The overlapping time zone at the beginning and the end is a time zone where both the beginning and the end of the time zone overlap with other time zones; that is, there is a time zone [TIME1, TIME i+5 ),[TIME i ,TIME j+5 ) and [TIMEj , TIME n ), i < j, where the time zone [TIME i , TIME j+5 ), the head of the time zone [TIME1, TIME i+5 ) overlaps with the time zone [TIME i , TIME i+5 ), and the tail of the time zone [TIME i , TIME j+5 ) overlaps with the time zone [TIME j , TIME n ), then the time zone [TIME j , TIME j+5 ) has head-to-tail overlapping time zones; i , TIME j+5 ) has head-to-tail overlapping time zones; The total time zone corresponding to the peak time zone is a continuous time zone, that is, there are peak time zones [TIME1, TIME i+5 ), [TIME i , TIME j+5 ), and [TIME j , TIME n ), and the total time zone corresponding to the peak time zone is [TIME1, TIME n ). Since the total time zone is continuous, this total time zone is a continuous time zone; Obtain N consecutive trough time zones with head-to-tail overlapping time zones, obtain the time periods corresponding to the trough time zones, and mark the time periods as low-frequency band regions; where N ≥ overlap threshold; Mark the time periods other than the high-frequency band region and the low-frequency band region as the flat-frequency band region; if there is an overlapping time zone between the high-frequency band region and the low-frequency band region, mark the overlapping time zone as the flat-frequency band region.

[0021] If there is an overlapping time zone between the high-frequency band region and the low-frequency band region, mark the overlapping time zone as the flat-frequency band region; If there is a high-frequency band region [TIME1, TIME5) and a low-frequency band region [TIME3, TIME7), then the overlapping time zone [TIME3, TIME5) of the high-frequency band region and the low-frequency band region is the flat-frequency band region; In this embodiment, the process of obtaining social factors and regularly obtaining social data sets according to social factors includes: Social factors include holidays, working days, total population, total industrial output value, and employment rate; Obtain the social values corresponding to the social factors, and generate social data sets according to the social factors and social values TIME ={social factor 节假日 =social value 节假日, social factors 工作日 = Social Value 工作日 , social factors 人口总数 = Social Value 人口总数 , social factors 工业总产值 = Social Value 工业总产值 , social factors 就业率 = Social Value 就业率}; TIME is the time when the social data set is obtained; The social value corresponding to holidays includes yes and no. If the TIME is a holiday, the social value is yes, otherwise it is no; The social values ​​corresponding to working days include yes and no. If the TIME is a working day, the social value is yes, otherwise it is no. The social values ​​corresponding to the total population, total industrial output value and employment rate include positive, neutral and negative. Obtain specific values ​​of total population, total industrial output value and employment rate, and obtain corresponding social values ​​of the specific values ​​according to the principle process of obtaining natural values ​​in step a1 and step a3; It should be noted that there is a one-to-one correspondence between the TIME in the natural dataset and the TIME in the social dataset, that is, the natural dataset and the social dataset of the low-voltage bus are obtained at the same time, that is, the power consumption, natural dataset and social dataset of each low-voltage bus are obtained at the same time.

[0022] In step 3, generating a set of correlation coefficients according to the frequency band area, the natural data set and the social data set is specifically as follows: The collection process includes: Acquire a natural data set and a social data set in a high-frequency band of TIME∈, and mark the natural data set as high-frequency natural data, and mark the social data set as high-frequency social data; Obtain explicit factors from high-frequency natural data and high-frequency social data, and mark the explicit factors as high explicit factors; obtain implicit factors from high-frequency natural data and high-frequency social data, and mark the implicit factors as high implicit factors; obtain normal factors from high-frequency natural data and high-frequency social data, and mark the normal factors as high normal factors; The high-frequency natural data and high-frequency social data are composed of several factors and their corresponding values; explicit intervals, implicit intervals and normal intervals are set, where the explicit intervals, implicit intervals and normal intervals do not intersect with each other; if there is a factor corresponding to values ​​including L1, L2, ..., Ln, the proportion of L1 is The proportion of L2 is ..., the proportion of Ln is When the value's proportion ∈ explicit interval, the value is marked as explicit data; when the value's proportion ∈ implicit interval, the value is marked as implicit data; when the value's proportion ∈ normal interval, the value is marked as normal data; the factor = explicit data is marked as explicit factor, the factor = implicit data is marked as implicit factor, and the factor = normal data is marked as normal factor; if there are N values ​​that all belong to the explicit interval, the values ​​are invalid; if there are N values ​​that all belong to the implicit interval, the values ​​are invalid; where N ≥ 2; Acquire a natural data set and a social data set in the TIME∈ flat frequency band area, and mark the natural data set as standard natural data, and mark the social data set as standard social data; Obtain explicit factors from standard natural data and standard social data, and mark the explicit factors as marked explicit factors; obtain normal factors from standard natural data and standard social data, and mark the normal factors as marked normal factors; obtain implicit factors from standard natural data and standard social data, and mark the implicit factors as marked implicit factors; According to the principle of obtaining explicit factors, implicit factors and normal factors in high-frequency natural data and high-frequency social data, obtain explicit factors, implicit factors and normal factors in standard natural data and standard social data; Obtain a natural data set and a social data set in the low-frequency band of TIME∈, and mark the natural data set as low-frequency natural data, and mark the social data set as low-frequency social data; Obtaining explicit factors from low-frequency natural data and low-frequency social data, and marking the explicit factors as low explicit factors; obtaining normal factors from low-frequency natural data and low-frequency social data, and marking the normal factors as low normal factors; obtaining implicit factors from low-frequency natural data and low-frequency social data, and marking the implicit factors as low latent factors; Based on the principle of obtaining explicit factors, implicit factors and normal factors in high-frequency natural data and high-frequency social data, obtain explicit factors, implicit factors and normal factors in low-frequency natural data and low-frequency social data; Obtaining a non-correlation factor, wherein the non-correlation factor = high constant factor ∩ standard constant factor ∩ low constant factor; Obtain a strong correlation factor, wherein the strong correlation factor=high-dominance factor+low-dominance factor-high-dominance factor∩low-dominance factor-average-dominance factor∩high-dominance factor-average-dominance factor∩low-dominance factor-no-correlation factor∩high-dominance factor-no-correlation factor∩low-dominance factor; Obtain a weak correlation factor, wherein the weak correlation factor=high latent factor+low latent factor-high latent factor∩low latent factor-average latent factor∩high latent factor-average latent factor∩low latent factor-uncorrelated factor∩high latent factor-uncorrelated factor∩low latent factor; Obtain the dominant factors that originally belonged to the highly dominant factors among the strongly correlated factors, mark the dominant factors as strongly correlated type A factors, and mark the other dominant factors among the strongly correlated factors as strongly correlated type B factors; obtain the recessive factors that originally belonged to the highly recessive factors among the weakly correlated factors, mark the recessive factors as weakly correlated type A factors, and mark the other recessive factors among the weakly correlated factors as weakly correlated type B factors; A strong confrontation index, a weak confrontation index and a non-confrontation index are set, among which the strong confrontation index>weak confrontation index>non-confrontation index; a correlation coefficient set is generated according to the strong correlation type A factor, the strong correlation type B factor, the weak correlation type A factor, the weak correlation type B factor, the non-correlation factor, the strong confrontation index, the weak confrontation index and the non-confrontation index.

[0023] The correlation coefficient set = {<several strongly correlated type A factors, several strongly correlated type B factors, strong confrontation index>, <several weakly correlated type A factors, several weakly correlated type B factors, weak confrontation index>, <several uncorrelated factors, no confrontation index>}.

[0024] In step 4, the process of generating a frequency band data set according to the power consumption, the natural data set, the social data set and the correlation coefficient set includes: The frequency band dataset TIME∈D ={Amount of electricity consumption TIME , several natural datasets TIME , several social datasets TIME , correlation coefficient set}, where D includes a high frequency band area, a flat frequency band area and a low frequency band area.

[0025] When D is in the high-frequency band area, the power consumption in the frequency band data set, the TIME∈ high-frequency band area in the natural data set and the social data set; When D is a flat frequency band, the TIME∈ of the electricity consumption, natural data set and social data set in the frequency band dataset is in a flat frequency band; when D is a low frequency band, the TIME∈ of the electricity consumption, natural data set and social data set in the frequency band dataset is in a low frequency band.

[0026] In step 4, the process of generating a frequency band load prediction formula according to the frequency band data set includes: Step 4a: Get the frequency band dataset TIME∈D , where D is the high frequency band area, that is, the frequency band data set TIME∈高频段区 ; Step 4b: Set the buffer time; Get the current time, generate the standard time zone = [current time - buffer time, current time); Step 4c: Get the frequency band data set TIME∈高频段区 Electricity consumption in TIME® standard time zones TIME , and the power consumption TIMEMarked as general power consumption; obtaining a general value in the general power consumption, the general value being the average value of the values ​​of the general power consumption whose integer part is the mode, and if the integer part does not have a mode, obtaining the median thereof; Step 4d: Get the frequency band dataset TIME∈高频段区 The two natural datasets and social datasets closest to the current time in TIME are respectively labeled as pan-natural dataset and pan-social dataset; Step 4e: Obtain the potential value of each natural factor in the pan-natural data set, wherein the potential value includes growth and decrease. If the natural value corresponding to the natural factor is a positive transition, then the potential value corresponding to the natural factor is growth; if the natural value corresponding to the natural factor is a reverse transition, then the potential value corresponding to the natural factor is a decrease; the positive transition includes the natural value changing from negative to flat, from flat to positive, from negative to positive, and no transition; the reverse transition includes the natural value changing from positive to flat, from flat to negative, and from positive to negative; Step 4f: Obtain the potential value of each social factor in the pan-social data set. If the social value of the potential value object is positive, neutral or negative, then obtain the potential value corresponding to the social factor according to step b5; if the social value of the potential value object is yes or no, and the social value corresponding to the social factor changes to NTY, then the potential value corresponding to the social factor increases; if the social value corresponding to the social factor changes to YTN, then the potential value corresponding to the social factor decreases; the NTY changes to a social value from no to yes and no change is sent, and the YTN changes to a social value from yes to no; Step 4g: Set the standard index and pass rate value; obtain the power consumption closest to the current time TIME , natural datasets TIME and social datasets TIME , marked as pre-electricity data set, pre-natural data set and pre-social data set respectively; according to the pre-electricity data set, pre-natural data set, pre-social data set, the pan-value of pan-electricity consumption, the potential value of each natural factor in the pan-natural data set, the potential value of each social factor in the pan-social data set and the correlation coefficient set, the high-frequency load prediction formula, the flat-frequency load prediction formula and the low-frequency load prediction formula are generated respectively.

[0027] Specifically, the Several strongly correlated type A factors, several strongly correlated type B factors), Among them, the PFYC (X, Y) function is to calculate the value of Y when the calculated value of X is greater than the qualified rate value, and use the calculated value of Y as the final result, otherwise, an early warning message is generated; the ZSLJ (A, B, C, D) function is to obtain the coincidence rate of the set A∪B and the set C∪D, that is, the same rate of elements in the set A∪B and the elements in the set C∪D; The function determines whether the values corresponding to the factors in P need to be changed according to the potential values of the factors in Q, and outputs the changed P. If the potential value of a certain factor in Q is increasing, the value corresponding to this factor in P does not need to be changed. If the potential value of a certain factor in Q is decreasing, the value corresponding to this factor in P needs to be downgraded. If the value is positive, it is changed to flat. If the value is flat, it is changed to negative. If the value is negative, it does not need to be changed. If the value is yes, it is changed to no. If the value is no, it is changed to yes; The function is used to obtain the coincidence rate of the set E∪S and the set K, and obtain the total number of elements in the set E∪S, denoted as AE. If there are e factors and their corresponding values in the set E∪S that are the same as the factors and their corresponding values in the set K, the coincidence rate of the set E∪S and the set K is Obtain the frequency band data set corresponding to the flat frequency band area, and generate the flat frequency load prediction formula according to the principle process of generating the high-frequency load prediction formula from step 4a to step 4g; the Obtain the frequency band data set corresponding to the low frequency band area, and generate the low frequency load prediction formula according to the principle process of generating the high-frequency load prediction formula from step 4a to step 4g; the In step 4 mentioned above, generating the new energy load prediction model according to the frequency band load prediction formula is specifically as follows: Obtain the current time and record it as the standard time, and obtain the three power consumptions closest to the standard time at each low-voltage busbar TIME , and sum up the power consumptions TIME Accumulate and divide according to time to generate power consumption TIME-1 , power consumption TIME-2 and power consumption TIME-3 , where, TIME-3<TIME-2<TIME-1<standard time, power consumption TIME-1 is the sum of the power consumptions closest to the standard time at each low-voltage busbar TIME sum, power consumption TIME-2 is the sum of the power consumptions at each low-voltage busbar that are the second closest to the standard time TIME sum, power consumption TIME-3 is the sum of the power consumptions at each low-voltage busbar that are the third closest to the standard time TIME sum, generate Set the high-frequency growth rate interval, the flat-frequency growth rate interval and the low-frequency growth rate interval, where the high-frequency growth rate interval, the flat-frequency growth rate interval and the low-frequency growth rate interval do not intersect; set the high-value interval, the flat-value interval and the low-value interval, where the high-value interval, the flat-value interval and the low-value interval do not intersect; If the electricity consumption growth rate ∈ high-frequency growth rate, electricity consumptionTIME-1 ∈ high value interval ∪ average value interval, the new energy load is predicted according to the high-frequency load prediction formula; if the electricity consumption growth rate ∈ high-frequency growth rate, the electricity consumption TIME-1 ∈ low value interval, the new energy load is predicted according to the flat frequency load prediction formula; If the electricity consumption growth rate ∈ flat frequency growth rate, the electricity consumption TIME-1 ∈ high value interval, the new energy load is predicted according to the high frequency load prediction formula; if the electricity consumption growth rate ∈ flat frequency growth rate, the electricity consumption TIME-1 ∈ flat value interval, the new energy load is predicted by the flat frequency load prediction formula; if the electricity consumption growth rate ∈ flat frequency growth rate, the electricity consumption TIME-1 ∈ low value interval, the new energy load is predicted by the low-frequency load prediction formula; If the electricity consumption growth rate ∈ low-frequency growth rate, the electricity consumption TIME-1 ∈ high value interval ∪ flat value interval, the new energy load is predicted according to the flat frequency load prediction formula; if the electricity consumption growth rate ∈ low frequency growth rate, the electricity consumption TIME-1 ∈ low value range, the new energy load is predicted by the low-frequency load prediction formula.

[0028] A new energy load forecasting system based on a low-voltage bus is used to implement a new energy load forecasting method based on a low-voltage bus, comprising: The data acquisition module is used to respectively set an electric energy detection device and a natural detection device at each low-voltage bus; regularly obtain the power consumption of the low-voltage bus through the electric energy detection device, and regularly obtain the natural data set of the low-voltage bus through the natural detection device; obtain social factors, and regularly obtain social data sets according to the social factors; The data processing module is used to obtain the characteristic time zone of each low-voltage bus according to the power consumption; generate a number of frequency bands according to the characteristic time zone of each low-voltage bus, wherein the frequency bands include a high frequency band, a flat frequency band and a low frequency band; generate a correlation coefficient set according to the frequency band, the natural data set and the social data set; The data analysis module is used to generate a frequency band data set based on electricity consumption, a natural data set, a social data set and a correlation coefficient set; obtain the frequency band data set corresponding to the high-frequency band area to generate a high-frequency load prediction formula; obtain the frequency band data set corresponding to the flat frequency band area to generate a flat-frequency load prediction formula; obtain the frequency band data set corresponding to the low-frequency band area to generate a low-frequency load prediction formula; at the same time, generate corresponding early warning information according to the prediction formula; generate a new energy load prediction model according to the high-frequency load prediction formula, the flat-frequency load prediction formula and the low-frequency load prediction formula.

[0029] The new energy load forecasting system based on the low-voltage bus also includes an early warning processing module for processing early warning information.

[0030] The above-described embodiment is only a preferred solution of the present invention and does not limit the present invention in any form. There are other variations and modifications without exceeding the technical solution described in the claims.

Claims

1. A new energy load prediction method based on low-voltage bus, Its characteristics are: The following steps are involved: Step 1: respectively set an electric energy detection device and a natural detection device at each low-voltage bus, regularly obtain the power consumption of the low-voltage bus through the electric energy detection device, and regularly obtain the natural data set of the low-voltage bus through the natural detection device; Step 2, obtaining the characteristic time zone of each low-voltage bus according to the power consumption; generating a plurality of frequency bands according to the characteristic time zone of each low-voltage bus; Step 3, obtaining social factors, regularly obtaining social data sets based on social factors, and generating a correlation coefficient set based on the frequency band area, natural data set and social data set; Step 4: Generate a frequency band data set based on the power consumption, natural data set, social data set and correlation coefficient set, generate a frequency band load prediction formula based on the frequency band data set, and generate a new energy load prediction model based on the frequency band load prediction formula; Step 5: The new energy load prediction model predicts the new energy load.

2. A new energy load prediction method based on low-voltage bus according to claim 1, Its characteristics are: The step 1 is specifically as follows: The electric energy detection device is composed of an electric energy meter, through which the electric energy meter is used to obtain the power consumption. TIME , where TIME is the time when the power consumption is obtained; The natural detection device includes a thermometer, a hygrometer, an anemometer, a rain gauge, a barometer and a visibility meter, and the corresponding temperature, humidity, wind speed, rainfall, air pressure and visibility are obtained through the thermometer, hygrometer, anemometer, rain gauge, barometer and visibility meter, and the temperature, humidity, wind speed, rainfall, air pressure and visibility are marked as natural factors; Generate natural data sets based on natural factors and natural values TIME , where TIME is the time when the natural data set is obtained; The natural value is generated by the following steps: Step 1a, generating corresponding positive state intervals, neutral state intervals and negative state intervals according to natural factors, wherein the positive state intervals, neutral state intervals and negative state intervals do not intersect with each other; Step 1b: Obtain the value corresponding to the natural factor through the natural detection device; Step 1c: if the value ∈ is in the positive state interval, the natural value of the natural factor is obtained to be positive; if the value ∈ is in the flat state interval, the natural value of the natural factor is obtained to be flat; if the value ∈ is in the negative state interval, the natural value of the natural factor is obtained to be negative; There is a one-to-one correspondence between TIME in electricity consumption and TIME in the natural data set.

3. A new energy load prediction method based on low voltage bus according to claim 2, Its characteristics are: In step 2, the characteristic time zone of each low-voltage bus is obtained according to the power consumption: The characteristic time zones include peak time zones, flat time zones and valley time zones; Get the power consumption in chronological order TIME , and mark TIME as the power coordinate; Through the two adjacent power coordinates TIME Get electricity usage variables TIMEAREA , where TIMEAREA is the time zone corresponding to the electricity consumption variable; Set continuous threshold, peak power consumption and valley power consumption, where peak power consumption > valley power consumption; If N consecutive power consumption variables TIMEAREA If all of the N consecutive power consumption variables are greater than the peak power consumption value, and N ≥ the continuous threshold, the time zone corresponding to the N consecutive power consumption variables is marked as the peak time zone; If N consecutive power consumption variables TIMEAREA If all of the N consecutive power consumption variables are less than the power consumption valley value, and N ≥ the continuous threshold, the time zone corresponding to the N consecutive power consumption variables is marked as the valley time zone; Mark the time zones outside the peak and valley time zones as flat time zones.

4. A new energy load prediction method based on low voltage bus according to claim 3, Its characteristics are: The process of generating a plurality of frequency bands according to the characteristic time zone of each low-voltage bus includes: Establishing a one-dimensional time coordinate axis, obtaining characteristic time zones of each low-voltage bus, and mapping the characteristic time zones to the one-dimensional time coordinate axis; Set the overlap threshold; Obtain N consecutive peak time zones with overlapping time zones at the beginning and the end, obtain the total time zone corresponding to the peak time zone, and mark the total time zone as a high frequency band area; wherein N≥overlap threshold; Obtain N consecutive trough time zones with overlapping head and tail time zones, obtain the time period corresponding to the trough time zone, and mark the time period as a low frequency band area; wherein N≥overlap threshold; The time period other than the high frequency band area and the low frequency band area is marked as a flat frequency band area; if there is an overlapping time zone between the high frequency band area and the low frequency band area, the overlapping time zone is marked as a flat frequency band area.

5. A new energy load prediction method based on low voltage bus according to claim 4, Its characteristics are: In step 3, generating a set of correlation coefficients according to the frequency band area, the natural data set and the social data set is specifically as follows: The collection process includes: Acquire a natural data set and a social data set in a high-frequency band of TIME∈, and mark the natural data set as high-frequency natural data, and mark the social data set as high-frequency social data; Obtain explicit factors from high-frequency natural data and high-frequency social data, and mark the explicit factors as high explicit factors; obtain implicit factors from high-frequency natural data and high-frequency social data, and mark the implicit factors as high implicit factors; obtain normal factors from high-frequency natural data and high-frequency social data, and mark the normal factors as high normal factors; Acquire a natural data set and a social data set in the TIME∈ flat frequency band area, and mark the natural data set as standard natural data, and mark the social data set as standard social data; Obtain explicit factors from standard natural data and standard social data, and mark the explicit factors as marked explicit factors; obtain normal factors from standard natural data and standard social data, and mark the normal factors as marked normal factors; obtain implicit factors from standard natural data and standard social data, and mark the implicit factors as marked implicit factors; Obtain a natural data set and a social data set in the low-frequency band of TIME∈, and mark the natural data set as low-frequency natural data, and mark the social data set as low-frequency social data; Obtaining explicit factors from low-frequency natural data and low-frequency social data, and marking the explicit factors as low explicit factors; obtaining normal factors from low-frequency natural data and low-frequency social data, and marking the normal factors as low normal factors; obtaining implicit factors from low-frequency natural data and low-frequency social data, and marking the implicit factors as low latent factors; Obtaining a non-correlation factor, wherein the non-correlation factor=high constant factor∩standard constant factor∩low constant factor; Obtain a strong correlation factor, wherein the strong correlation factor=high-dominance factor+low-dominance factor-high-dominance factor∩low-dominance factor-average-dominance factor∩high-dominance factor-average-dominance factor∩low-dominance factor-no-correlation factor∩high-dominance factor-no-correlation factor∩low-dominance factor; Obtain a weak correlation factor, wherein the weak correlation factor=high latent factor+low latent factor-high latent factor∩low latent factor-average latent factor∩high latent factor-average latent factor∩low latent factor-uncorrelated factor∩high latent factor-uncorrelated factor∩low latent factor; Obtain the dominant factors that originally belonged to the highly dominant factors among the strongly correlated factors, mark the dominant factors as strongly correlated type A factors, and mark the other dominant factors among the strongly correlated factors as strongly correlated type B factors; obtain the recessive factors that originally belonged to the highly recessive factors among the weakly correlated factors, mark the recessive factors as weakly correlated type A factors, and mark the other recessive factors among the weakly correlated factors as weakly correlated type B factors; A strong confrontation index, a weak confrontation index and a non-confrontation index are set, among which the strong confrontation index>weak confrontation index>non-confrontation index; a correlation coefficient set is generated according to the strong correlation type A factor, the strong correlation type B factor, the weak correlation type A factor, the weak correlation type B factor, the non-correlation factor, the strong confrontation index, the weak confrontation index and the non-confrontation index.

6. A new energy load prediction method based on low voltage bus according to claim 5, Its characteristics are: In step 4, the process of generating a frequency band data set according to the power consumption, natural data set, social data set and correlation coefficient set includes: the frequency band data set TIME∈D ={Amount of electricity consumption TIME , several natural datasets TIME , several social datasets TIME , correlation coefficient set}, where D includes a high frequency band area, a flat frequency band area and a low frequency band area.

7. A new energy load prediction method based on low voltage bus according to claim 6, Its characteristics are: In step 4, the process of generating a frequency band load prediction formula according to the frequency band data set includes: Step 4a: Get the frequency band dataset TIME∈高频段区 ; Step 4b: Set the buffer time; obtain the current time, generate the standard time zone = [current time - buffer time, current time); Step 4c: Get the frequency band dataset TIME∈高频段区 Electricity consumption in TIME® standard time zones TIME , and the power consumption TIME Marked as general power consumption; obtaining a general value in the general power consumption, the general value being the average value of the values ​​of the general power consumption whose integer part is the mode, and if the integer part does not have a mode, obtaining the median thereof; Step 4d: Get the frequency band dataset TIME∈高频段区 The two natural datasets and social datasets closest to the current time in TIME are respectively labeled as pan-natural dataset and pan-social dataset; Step 4e: Obtain the potential value of each natural factor in the pan-natural data set, wherein the potential value includes growth and decrease; Step 4f: Obtain the potential value of each social factor in the pan-social data set; Step 4g: Set the standard index and pass rate value; obtain the power consumption closest to the current time TIME , natural datasets TIME and social datasets TIME , marked as pre-electricity data set, pre-natural data set and pre-social data set respectively; according to the pre-electricity data set, pre-natural data set, pre-social data set, the pan-value of pan-electricity consumption, the potential value of each natural factor in the pan-natural data set, the potential value of each social factor in the pan-social data set and the correlation coefficient set, the high-frequency load prediction formula, the flat-frequency load prediction formula and the low-frequency load prediction formula are generated respectively.

8. A new energy load prediction method based on low voltage bus according to claim 7, Its characteristics are: In step 4, the new energy load forecasting model is generated according to the frequency band load forecasting formula as follows: Obtain the current time as the standard time, and obtain the three electricity consumption amounts closest to the standard time at each low-voltage busbar TIME and accumulate and divide the electricity consumption amounts TIME according to time to generate electricity consumption amount TIME-1 , electricity consumption amount TIME-2 and electricity consumption amount TIME-3 , where TIME - 3 < TIME - 2 < TIME - 1 < standard time; Set the high-frequency growth rate interval, the flat-frequency growth rate interval and the low-frequency growth rate interval, where the high-frequency growth rate interval, the flat-frequency growth rate interval and the low-frequency growth rate interval do not intersect with each other; set the high value interval, the flat value interval and the low value interval, where the high value interval, the flat value interval and the low value interval do not intersect with each other; If the electricity consumption growth rate ∈ high frequency growth rate, the electricity consumption TIME-1 ∈ high value interval ∪ average value interval, the new energy load is predicted according to the high-frequency load prediction formula; if the electricity consumption growth rate ∈ high-frequency growth rate, the electricity consumption TIME-1 ∈ low value interval, the new energy load is predicted according to the flat frequency load prediction formula; If the electricity consumption growth rate ∈ flat frequency growth rate, the electricity consumption TIME-1 ∈ high value interval, the new energy load is predicted according to the high frequency load prediction formula; if the electricity consumption growth rate ∈ flat frequency growth rate, the electricity consumption TIME-1 ∈ flat value interval, the new energy load is predicted by the flat frequency load prediction formula; if the electricity consumption growth rate ∈ flat frequency growth rate, the electricity consumption TIME-1 ∈ low value interval, the new energy load is predicted by the low-frequency load prediction formula; If the electricity consumption growth rate ∈ low-frequency growth rate, the electricity consumption TIME-1 ∈ high value interval ∪ flat value interval, the new energy load is predicted according to the flat frequency load prediction formula; if the electricity consumption growth rate ∈ low frequency growth rate, the electricity consumption TIME-1 ∈ low value range, the new energy load is predicted by the low-frequency load prediction formula.

9. A new energy load forecasting system based on a low-voltage bus, used to implement a new energy load forecasting method based on a low-voltage bus as claimed in any one of claims 1 to 8, Its characteristics are: include: A data acquisition module is used to respectively set an electric energy detection device and a natural detection device at each low-voltage bus; The power consumption of the low-voltage bus is regularly obtained through an electric energy detection device, and the natural data set of the low-voltage bus is regularly obtained through a natural detection device; Obtain social factors and regularly obtain social data sets based on social factors; The data processing module is used to obtain the characteristic time zone of each low-voltage bus according to the power consumption; generate a number of frequency bands according to the characteristic time zone of each low-voltage bus, wherein the frequency bands include a high frequency band, a flat frequency band and a low frequency band; generate a correlation coefficient set according to the frequency band, the natural data set and the social data set; The data analysis module is used to generate a frequency band data set based on the power consumption, natural data set, social data set and correlation coefficient set; obtain the frequency band data set corresponding to the high frequency band area, and generate a high frequency load prediction formula; Obtain the frequency band data set corresponding to the flat frequency band area and generate a flat frequency load prediction formula; obtain the frequency band data set corresponding to the low frequency band area and generate a low frequency load prediction formula; at the same time, generate corresponding warning information according to the prediction formula; A new energy load prediction model is generated based on the high-frequency load prediction formula, the flat-frequency load prediction formula and the low-frequency load prediction formula.

10. A new energy load forecasting system based on low voltage bus according to claim 9, Its characteristics are: It also includes an early warning processing module for processing early warning information.