CN105184057A - Weather forecast information based bus bar load prediction method - Google Patents

Abstract

The present invention provides a bus load forecasting method based on weather forecast information, comprising the following steps: step 1: extracting weather forecast information closely corresponding to the geographic location of the bus load according to the geographical coordinates of the area under which the bus load belongs; step 2: analyzing and determining the difference Correlation between bus load and meteorological factors, classify bus load; Step 3: Classify weather forecast information matching bus load according to occurrence time, extract short-term and ultra-short-term bus load historical data and weather forecast information at corresponding time ; Step 4: Select similar days according to historical load data, weather forecast information and date types; Step 5: Classify and predict bus load. The invention constructs a bus load prediction model based on weather forecast and load classification prediction by extracting weather forecast information closely corresponding to the geographical location of the bus, thereby effectively improving the accuracy of the bus load forecast.

Description

A Bus Load Forecasting Method Based on Weather Forecast Information

technical field

The invention relates to an automatic dispatching method of an electric power system, in particular to a bus load forecasting method based on weather forecast information.

Background technique

At present, weather forecasting has been widely used in new energy forecasting. Taking wind power forecasting as an example, the location information of the wind farm site has been determined at the beginning of its establishment, including precise latitude and longitude. When wind power is predicted by the wind field, the weather forecast data of the wind field can be obtained directly from the meteorological center through latitude and longitude. The bus load is the sum of the terminal loads in a relatively small area, and it is generally impossible to determine the latitude and longitude of the terminal loads within its power supply range.

Traditional bus load forecasting usually selects the meteorological information of prefecture-level cities as a reference for bus load forecasting. The scope of prefecture-level cities is relatively large, and the power supply range of a single bus load is usually at the district or county level. The actual weather conditions in different districts and counties may be different. , so it is unreasonable to select the same weather forecast information as a reference for all bus loads in the region.

Contents of the invention

In order to overcome the above-mentioned deficiencies in the prior art, the present invention provides a bus load forecasting method based on weather forecast, by obtaining the geographical location of the administrative area to which the bus load belongs, such as the geographical location of a county, township or city, and extracting the location closely corresponding to the bus geographical location. High-precision weather forecast information, by studying the correlation characteristics of bus load and meteorological factors, a bus load forecasting model based on weather forecast and load classification forecast is constructed, which can effectively improve the accuracy of bus load forecasting.

The solution adopted to achieve the above purpose is:

A bus load forecasting method based on weather forecast, the method comprises the following steps:

Step 1: According to the geographic coordinates of the jurisdiction where the bus load belongs, extract high-precision weather forecast information closely corresponding to the geographical location of the bus load; obtain the matching weather forecast information such as temperature, air pressure, humidity, and rainfall;

Step 2: Analyze and determine the correlation between different bus loads and weather forecast information, and classify the bus loads;

Step 3: Classify the weather forecast information matching the bus load according to the time of occurrence, extract the historical data of short-term and ultra-short-term bus load and the weather forecast information at the corresponding time;

Step 4: Select similar days based on historical load data, weather forecast information and date types;

Step 5: Classify and predict the bus load.

Preferably, according to the largest administrative division that can be covered by the bus load power supply range, such as the geographical longitude and latitude coordinates of a county, township or municipal district, the weather forecast information in the administrative division is extracted.

Preferably, the correlation between each bus load and meteorological factors is analyzed to determine the type of bus load. The calculation formula is as follows:

$\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; cov</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>}{\sqrt{\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}\sqrt{\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

In the formula: X is the historical load of the bus load, Y is the weather forecast information, R is the correlation coefficient, cov(X,Y) is the covariance of X and Y; D(X), D(Y) are X and Y respectively mean square error of . By analyzing the correlation between the bus load and meteorological factors such as temperature and rainfall, the load is divided into different types of loads such as temperature-sensitive load, rainfall-sensitive load, and weather-independent load.

Preferably, in the step 3, the short-term bus load forecast extracts the historical load in units of days, and the ultra-short-term bus load forecast extracts the historical load corresponding to the current day and the forecast period, and scrolls to update them.

Preferably, in the step 4, historical load data, weather forecast information and date types are combined to form daily sample data, similar days are selected by cluster analysis, meteorological elements of forecast days and similar days are fully considered, and analysis is aimed at The characteristics of the bus load and the meteorological information of the forecast day select the input sample data, and the specific steps are as follows:

Define reference samples, comparison samples, and distances between samples

$\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\sqrt{\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

Among them, i and j represent the sample number, n is the number of selected sample days, x _ik and x _jk are the sample data; the center of gravity of the class Among them, m is the number of samples in this class, and Xi is the i-th sample in this class. Take each busbar load as the analysis object.

Among them, the temperature-sensitive load combines 96 points of historical load data, date type and temperature information to form a daily sample (l1,...li...l96,d,t1.t2....t24), where li is the historical load data of the bus, d is the date type, and ti is the temperature.

The rainfall-sensitive load combines 96 points of historical load data and rainfall in each period into samples (l1,...li...l96, r1.r2...r24), where li is the historical load data of the bus, and ri is the rainfall.

The meteorological irrelevant load selects 96 points of historical load data (l1,...li...l96) to form a daily sample, where li is the historical load data of the bus.

The date types are divided into rest days and working days.

Select n days of data as the comparison sample, calculate the clustering distance between the comparison sample and the reference sample according to formula (2), and select the category closest to the reference sample as the similar day. In addition, when selecting similar days, the operation mode of off-grid loads and outlets should be considered, and similar days with the same operation mode should be selected.

Preferably, in step 5, comprehensively considering relevant information such as weather forecast information and bus load historical data, different bus loads select a prediction algorithm suitable for their load characteristics, and classify and predict bus loads.

Preferably, when predicting different types of loads, different similar daily historical loads and weather forecast information are selected as input variables. Among them, the temperature sensitive load selects historical load, temperature and date type as input variables, the rainfall sensitive load selects historical load, rainfall and date type as input variables, and the meteorological irrelevant load only selects historical load and date type as input variables. Each type of load is forecasted separately.

Compared with the prior art, the present invention has the following beneficial effects:

The present invention obtains the administrative area to which the bus load belongs, extracts high-precision weather forecast information closely corresponding to the geographical location of the bus, and constructs a bus load forecasting model based on high-precision weather forecast and load classification forecast, effectively improving the bus load forecasting accuracy.

Description of drawings

Fig. 1 is the flow chart of busbar load prediction based on weather forecast of the present invention;

Detailed ways

The specific embodiments of the present invention will be further described in detail below in conjunction with the accompanying drawings.

The bus load forecasting method based on weather forecast includes the following steps:

Step 1: According to the longitude and latitude information of the bus load power supply administrative division, extract the weather forecast closely corresponding to the geographical location of the bus load. Obtain matching weather forecast information such as temperature, air pressure, humidity, rainfall, etc.;

According to the largest administrative division that can be covered by the power supply range of the bus load, such as the geographical longitude and latitude coordinates of the county, township or municipal district, the weather forecast information in the administrative division is extracted.

Step 2: Analyze and determine the correlation between different bus loads and weather forecast information, and classify the bus loads.

Analyze the correlation between each bus load and meteorological factors, so as to determine the type of bus load, the calculation formula is as follows:

$\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; cov</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>}{\sqrt{\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}\sqrt{\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

In the formula: X is the historical load of the bus load, Y is the weather forecast information, R is the correlation coefficient, cov(X,Y) is the covariance of X and Y; D(X), D(Y) are X and Y respectively mean square error of . By analyzing the correlation between the bus load and meteorological factors such as temperature and rainfall, the load can be divided into different types of loads such as temperature-sensitive load, rainfall-sensitive load, and weather-independent load.

Step 3: Classify the weather forecast information matching the bus load according to the time of occurrence, form the forecast weather forecast information and historical weather forecast information for short-term and ultra-short-term bus load forecasting, and scroll to refresh.

Among them, the weather forecast information matching the bus load is classified according to the occurrence time, and the weather forecast information and historical weather forecast information of short-term and ultra-short-term bus load forecasting are extracted.

Step 4: Fully consider the meteorological elements of the forecast day and the similar day, and select the similar day.

Combine historical load data, temperature information and date types to form daily sample data, select similar days through cluster analysis, fully consider the meteorological elements of forecast days and similar days, and analyze the characteristics of the bus load and the weather forecast on the forecast day The information selects the input sample data, and the specific steps are as follows:

Define the distance between 2 samples and

$\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\sqrt{\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

Among them, i and j represent the sample number, n is the number of selected sample days, x _ik and x _jk are the sample data; the center of gravity of the class Among them, m is the number of samples in this class, and Xi is the i-th sample in this class. Take each busbar load as the analysis object.

Among them, the temperature-sensitive load combines 96 points of historical load data, date type and temperature information to form a daily sample (l1,...li...l96,d,t1.t2....t24), where li is the historical load data of the bus, d is the date type, and ti is the temperature.

The rainfall-sensitive load combines 96 points of historical load data and rainfall in each period into samples (l1,...li...l96, r1.r2...r24), where li is the historical load data of the bus, and ri is the rainfall.

The meteorological irrelevant load selects 96 points of historical load data (l1,...li...l96) to form a daily sample, where li is the historical load data of the bus.

Select the data of n days as the comparison sample, calculate the clustering distance between each comparison sample and the reference sample according to formula (2), and select the category with the closest distance to the reference sample as the similar day. In addition, when selecting similar days, the operation mode of off-grid loads and outlets should be considered, and similar days with the same operation mode should be selected.

Step 5: Comprehensively consider relevant information such as weather forecast information and bus load history data, select input quantities and prediction algorithms suitable for their load characteristics for different bus loads, and classify and predict bus loads.

When forecasting, different types of loads select different similar daily historical loads and weather forecast information as input variables. Among them, the temperature sensitive load selects historical load, temperature and date type as input variables, the rainfall sensitive load selects historical load, rainfall and date type as input variables, and the meteorological irrelevant load only selects historical load and date type as input variables. Each type of load is forecasted separately.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application rather than to limit the scope of protection thereof. Although the present application has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: After reading this application, those skilled in the art can still make various changes, modifications or equivalent replacements to the specific implementation methods of the application, but these changes, modifications or equivalent replacements are all within the protection scope of the pending claims of the application.

Claims (7)

1. based on a bus load Forecasting Methodology for weather forecast information, it is characterized in that, described method comprises the steps:

Step 1: the geographic coordinate in area under one's jurisdiction belonging to bus load, extracts the weather forecast information closely corresponding with bus load geographic position;

Step 2: analyze and determine different bus load and weather forecast information correlativity, bus load is classified;

Step 3: sorted out respectively by time of origin by the weather forecast information mated with bus load, extracts short-term, the historical data of ultra-short term bus load and the weather forecast information of corresponding time;

Step 4: according to historical load data, weather forecast information and date type, carries out choosing of similar day;

Step 5: classification prediction is carried out to bus load.

2. Forecasting Methodology as claimed in claim 1, is characterized in that, in described step 1, according to the maximum administrative partition that bus load service area can be contained, extract the weather forecast information in this administrative partition.

3. Forecasting Methodology as claimed in claim 1, it is characterized in that, in described step 2, the computing formula of described analysis is:

$R=\frac{\mathrm{cov}(X,Y)}{\sqrt{D\left(X\right)}\sqrt{D\left(Y\right)}}---\left(1\right)$

In formula: R is different bus load and weather forecast information related coefficient, and X is bus load historical load, and Y is weather forecast information, the covariance that cov (X, Y) is X and Y; D (X), D (Y) are respectively the mean square deviation of X and Y;

Bus load is divided into responsive to temperature type load, rainfall responsive type load, meteorological uncorrelated type load by described being categorized as.

4. Forecasting Methodology as claimed in claim 1, is characterized in that, in described step 3, the historical load in units of day is extracted in the prediction of short-term bus load, and the same day and the historical load corresponding with prediction period are extracted in the prediction of ultra-short term bus load.

5. Forecasting Methodology as claimed in claim 1, is characterized in that, in described step 4, historical load data, weather forecast information and date type are combined to form a day sample data, choose similar day by the method for cluster analysis;

Concrete steps are as follows:

(4-1) distance between reference sample, comparative sample and sample is defined

$d(i,j)=\sqrt{\underset{k=1}{\overset{n}{\&Sigma;}}{(x_{ik}-x_{jk})}^2/n}---\left(2\right)$

Wherein i, j represent catalogue number(Cat.No.), and n is selected sample number of days, x _ik, x _jkfor sample data, k=1,2 ..., n;

(4-2) choose n day data as comparative sample, according to formula (2), calculate the clustering distance of each comparative sample and reference sample, choose the sample data nearest with reference sample as with reference to load;

Described date type is divided into off-day and working day.

6. Forecasting Methodology as claimed in claim 1, it is characterized in that, in described step 5, consider weather forecast information, bus load historical data and date type, different bus load chooses different input variables, carries out classification prediction to bus load.

7. Forecasting Methodology as claimed in claim 6, it is characterized in that, responsive to temperature type load chooses historical load, temperature and date type as input variable;

Rainfall responsive type load chooses historical load, rainfall and date type as input variable;

Meteorological uncorrelated load then chooses historical load and date type as input variable.