CN107423836B - Short-term load prediction method based on body sensing temperature - Google Patents
Short-term load prediction method based on body sensing temperature Download PDFInfo
- Publication number
- CN107423836B CN107423836B CN201710224770.8A CN201710224770A CN107423836B CN 107423836 B CN107423836 B CN 107423836B CN 201710224770 A CN201710224770 A CN 201710224770A CN 107423836 B CN107423836 B CN 107423836B
- Authority
- CN
- China
- Prior art keywords
- load
- day
- predicted
- temperature
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The invention discloses a short-term load forecasting method based on body sensing temperature, which is used for forecasting loads in summer and winter and comprises the following steps: reading historical data; quantification of historical data; calculating the somatosensory temperature; collecting information of a day to be predicted; predicting the daily maximum load; selecting a load trend curve: selecting a historical day closest to the day to be predicted as a trend similar day by a similar day method, and taking the load curve line type of the day as the line type of the day to be predicted; and (3) daily load prediction: and selecting the highest and lowest load values of the day with similar trends, subtracting the lowest load value from the load data of 96 points on the day, dividing the difference between the highest load and the lowest load, multiplying the normalized data by the difference between the predicted highest load and the predicted lowest load of the day to be predicted, and adding the predicted lowest load value to obtain the all-day load predicted value. The invention greatly improves the load forecasting speed and ensures the precision of the load forecasting.
Description
Technical Field
The invention relates to the technical field of electric power, in particular to a short-term load prediction method based on somatosensory temperature.
Background
The power load prediction work is the main basis of the planning, construction, operation and maintenance work of the conductive network. Short-term load prediction is an important component of load prediction. Accurate short-term load prediction can guide a power transmission and distribution system to adjust the operation mode in time, and a power failure and power transmission maintenance plan is reasonably arranged. Therefore, the high-quality load prediction can guide a power grid company to most reasonably utilize various resources such as people, properties, things and the like and obtain the optimal social and economic benefits under the condition of meeting the power supply quality requirement. The importance of short-term load forecasting work is becoming more prominent in the power market environment.
The power load prediction method is continuously developed for many years, and mainly comprises a similar daily method, a time series method, a neural network method, a fuzzy theory method, a trend analysis method and the like.
The above method and the problems are as follows:
the principle of the similarity daily method is simple, but the selection of the similarity evaluation function seriously restricts the prediction precision of the similarity evaluation function.
The time series method is high in calculation speed, but factors influencing loads cannot be embodied in the model, and the model is built by abundant experience.
The neural network method has high prediction accuracy, but has the defects of long learning time and easy limitation to local minimum points.
The fuzzy prediction method can well process the uncertainty of the load change, but the prediction precision of the fuzzy prediction method used alone is not high.
If the trend analysis method can select a proper model and well fit the actual load curve, the prediction result is better, and the difference of the prediction results among different models is larger, so that the selection of the proper model is very difficult.
In summary, the prior art lacks an effective solution to the short-term load prediction problem.
Disclosure of Invention
In order to solve the defects of the prior art, the short-term load prediction method based on the somatosensory temperature is provided, a model which is simple in structure and can embody main factors influencing the load is built according to the characteristics of the load, the calculation speed can be effectively improved, the calculation accuracy is guaranteed, and the problem of local minimum points can be avoided.
The short-term load prediction method based on the somatosensory temperature aims at predicting loads in summer and winter, and comprises the following steps:
reading of historical data: acquiring historical solar weather data and power load data;
quantification of historical data: quantifying different factors influencing the change of the power load according to the influence degree of the factors, wherein the factors comprise the quantification of the day type and the quantification of the meteorological information;
and (3) calculating the somatosensory temperature: calculating the somatosensory temperature of the history every day according to the quantized historical data and a somatosensory temperature calculation formula;
collecting information of days to be predicted: acquiring meteorological information and power load information of a day to be predicted;
predicting the daily maximum load: inputting the acquired day information to be predicted to a daily maximum load prediction model for daily maximum load prediction;
selecting a load trend curve: selecting a historical day closest to the day to be predicted by a similar day method, and taking the load curve line type of the day as the load curve line type of the day to be predicted;
and (3) daily load prediction: and selecting the highest and lowest load values of the day with similar trends, subtracting the lowest load value from the load data, dividing the load data by the difference between the highest load and the lowest load, multiplying the normalized data by the predicted difference between the highest load and the lowest load, and adding the predicted lowest load value to obtain the predicted value of the load all day.
Further, in the short-term load prediction method based on sensible temperature, it is necessary to acquire historical meteorological data and power load data first.
Further, the quantification of the day type: day types are divided in total: and on working days, common holidays (including weekends) and special holidays, the quantized day type numerical values also represent the influence degree of the day type on load prediction, and the specific numerical values of all holidays and days before and after the holidays are determined by analyzing historical power load data by a trend analysis method.
Further, the quantifying of the weather information includes quantifying the weather condition and the air temperature, and the quantifying of the air temperature specifically includes: firstly, adding a first-order filtering link to the air temperature, and fully considering the accumulation of the temperature and the time delay effect; secondly, low temperature is converted into corresponding high temperature.
Further, for historical power load data, power loads need to be classified firstly, and the power loads are mainly classified into basic loads, main transformer loads and fluctuating loads.
Further, through trend analysis of 2013-2015 year load historical data, particularly holidays and days before and after the holidays, the fluctuation load value, namely L is determinedf1。
Further, the base load refers to a part of the load that is substantially unchanged or slightly changed within a short time; the main transformer load refers to the part of load with larger load variation amplitude; the fluctuating load refers to the part of the load with small load change amplitude, and mainly refers to small load fluctuation caused by holidays or emergencies.
Further, wherein the predicted value of the main transformer load is
Lf2＝k*TA
Wherein, TA: maximum sensible temperature, T: maximum influence temperature, highest temperature in summer, lowest temperature in winter, RH: relative humidity, V: wind speed, k is the proportionality coefficient.
Further, the base load Lf3The method has consistency with a k coefficient used in main transformer load prediction and β in a short time, so that Latin hypercube sampling is adopted to analyze historical data according to months to obtain the optimal L suitable for the historical dataf3K and β:
Lfmax＝(Lf1+Lf2+Lf3)*(1-β)+Lf0max*β
wherein L isfmaxThe maximum load prediction value for the day to be predicted, Lf0maxFor the previous day maximum load, Lf1、Lf2And Lf3Fluctuating load, main variable load and basic load respectively, and β is an adaptive coefficient.
Further, selecting a day with similar trend of the daily load curve to be predicted by a similar day method. Specifically, the similarity of weather conditions and day types is determined, the date interval is considered, the historical day with the highest similarity is selected as the trend similar day, and the load curve line type of the day is taken as the load curve line type of the day to be predicted.
Further, since the minimum load variation is small throughout the year and negligible in adjacent days, the day-before-day minimum load to be predicted is taken as the day-to-be-predicted minimum load.
Further, the predicted load value of the ith point is as follows:
Lf(i)＝(Lfs(i)-Lfsmin)/(Lfsmax-Lfsmin)*(Lfmax-Lf0min)+Lf0min
wherein i is 1-96, Lf(i)The predicted value of the load at the ith point of the day to be predicted, Lfs(i)Load value L of the ith point of the day with similar trendfsminMinimum load value, L for day with similar trendfsmaxMaximum load value for day with similar trend, LfmaxThe maximum load prediction value for the day to be predicted, Lf0minThe minimum load value of the previous day.
Compared with the prior art, the invention has the beneficial effects that:
according to the method, through deep mining of historical data, the approximate linear relation between the load and the body sensing temperature is found, and the body sensing temperature is introduced into load prediction for the first time. And classifying the loads according to the load characteristics, and building an image and intuitive load prediction model by combining prediction methods such as a trend analysis method and the like. The method comprises the steps of deeply mining historical data through Latin hypercube sampling, and determining the basic load of a certain area, the proportion coefficient of load change and the adaptive coefficient. The load forecasting speed is greatly improved, and the load forecasting precision is ensured.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the application and, together with the description, serve to explain the application and are not intended to limit the application.
FIG. 1 is a schematic view of a load prediction job classification;
FIG. 2 is a schematic view of load classification;
FIG. 3 is a flow chart of summer and winter load prediction;
FIG. 42013 shows the maximum load change curve around the national festival of Qing of the year 2015 for 7 days;
fig. 5 shows the effect of air temperature on load.
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. 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 application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
As described in the background art, the short-term load prediction accuracy is insufficient in the prior art, and in order to solve the technical problems, the application provides a short-term load prediction method based on the body-sensing temperature.
In a typical implementation manner of the application, the short-term load prediction work has various characteristics such as seasonality, periodicity, tendency, holiday and the like, and the characteristic of strong systematicness requires that the prediction method has comprehensiveness and adaptivity. The temperature change is small in spring and autumn, and the daily load curves are similar; in summer and winter, because the load is greatly influenced by weather factors such as temperature and the like, the similarity of daily load curves is low, and the prediction difficulty is high.
In view of the above characteristics, as shown in fig. 1, the prediction method first divides the prediction work into two types, i.e., spring and autumn load prediction and summer and winter load prediction, according to the prediction target. Then, each prediction is divided into three types, namely, working day load prediction, common holiday load prediction and special holiday load prediction. The method is characterized in that a novel intelligent prediction method based on the body sensing temperature is adopted for load prediction in summer and winter, and a similar day load prediction method is adopted for load prediction in spring and autumn.
The technical scheme for load prediction in summer and winter is as follows: the intelligent prediction algorithm based on the body sensing temperature is a novel intelligent prediction method developed on the basis of a similar day prediction method, a neural network prediction method, a trend analysis method and a time series prediction method, and in the algorithm, the concept of the body sensing temperature in meteorology is introduced into load prediction for the first time. The method is a comprehensive parameter formed by comprehensively considering the influence factors such as the highest daily temperature, the lowest daily temperature, the wind speed, the humidity and the like. On the other hand, the algorithm incorporates adaptive factors that take into account economic, humanistic, and climatic factors.
In this method, the loads are first classified into a base load, a main change load, and a fluctuating load, as shown in fig. 2. Wherein, the basic load refers to the part of load which is basically unchanged or slightly changed in a short time, and mainly refers to normal production, domestic electric load and the like; the main transformer load refers to the part of load with large load change amplitude, and mainly comprises a refrigeration load in summer, a heating load in winter, a large-scale power failure accident and the like. The fluctuating load refers to the part of the load with small load change amplitude, and mainly refers to small load fluctuation generated by holidays (including weekends) or emergencies.
Due to unpredictability of the emergency, the fluctuating load only takes into account holiday factors in the prediction work, i.e. depending on the day type. The fluctuating load value Lf1 can therefore be determined before prediction, according to the statutory holiday schedule promulgated by the country in advance. The fluctuation load of the common working day is 0, but there are exceptions, such as small change of the load in 1-2 days before and after the common holiday, and an obvious gradual change process of the fluctuation load in 3-4 days before and after the special holiday, wherein the fluctuation of holidays such as spring festival, national day and the like is most obvious, and the cycle is long and large. The maximum load change curve of 7 days before and after the national festival of 2013 + 2015 is shown in fig. 4.
The specific numerical value of the fluctuating load before and after the ordinary holiday and the special holiday is determined by a trend analysis method. And carrying out trend analysis on weekends according to months, and fitting a corresponding curve to carry out load numerical value setting. Similarly, the same is true for holidays, the maximum loads of the same holiday in the history, including the 4 days before and after, are connected to form a curve, and the corresponding curve is fitted to three data formed by the data of three years to serve as a fluctuating load setting value.
The load data of 2013-2015 year is researched and found that the main transformer load and the sensible temperature are in a linear relation in winter and summer, so that the predicted value of the main transformer load is as follows:
Lf2＝k*TA(1)
wherein, TA: maximum sensible temperature, T: maximum influence temperature (highest temperature in summer and lowest temperature in winter), RH: relative humidity, V: wind speed, k is the proportionality coefficient. In the method, firstly, the sensible temperature of the day is calculated according to weather information (including the highest temperature of the day, the lowest temperature of the day, weather conditions, wind speed and the like) obtained by weather forecast and the formula.
It should be noted that, the invention does not predict the loads at 96 points per day one by one, but predicts the daily load by predicting the maximum load and selecting the daily curve with similar trend. And the main transformer load and the sensible temperature are in a linear relation, so that the sensible temperature actually calculated every day is the sensible temperature with the largest influence on the load, the highest temperature is taken in summer, and the lowest temperature is taken in winter. Additional weather conditions affect humidity and fluctuating loads and may be the dominant factor in load prediction in spring and autumn, and so need to be obtained.
The main transformer load is in linear relation with the sensible temperature in winter and summer by taking care of equivalently converting low temperature into corresponding high temperature. The influence of the air temperature on the load is divided into two parts: a maximum air temperature and a minimum air temperature. The influence of the highest air temperature on the load is mainly reflected in summer, and the lowest air temperature is reflected in winter. The specific degree of influence is shown in fig. 5. The conversion here means that the lowest air temperature is converted to an equivalent high temperature according to the degree of influence.
Since the predicted load value of each day has a great correlation with the load of the previous day, the adaptive coefficient β is defined.
Base load Lf3The method has consistency with a k coefficient used in main transformer load prediction and β in a short time, so that Latin hypercube sampling is adopted to analyze historical data according to months to obtain the optimal L suitable for the historical dataf3K and β.
Lfmax＝(Lf1+Lf2+Lf3)*(1-β)+Lf0max*β (3)
Wherein L isfmaxIs the maximum load predicted value of the day to be predicted,Lf0maxfor the previous day maximum load, Lf1、Lf2And Lf3Fluctuating load, main variable load and basic load respectively, and β is an adaptive coefficient.
Further, selecting a day with similar trend of the daily load curve to be predicted by a similar day method. Specifically, the similarity of weather conditions and day types is determined, the date interval is considered, the historical day with the highest similarity is selected as the trend similar day, and the load curve line type of the day is taken as the line type of the day to be predicted.
Further, since the minimum load variation is small throughout the year, it is negligible in adjacent days. Therefore, the day-ahead and day-to-day minimum load to be predicted is taken as the day-to-day minimum load to be predicted.
Further, the predicted load value of the ith point is as follows:
Lf(i)＝(Lfs(i)-Lfsmin)/(Lfsmax-Lfsmin)*(Lfmax-Lf0min)+Lf0min(4)
wherein i is 1-96, Lf(i)The predicted value of the load at the ith point of the day to be predicted, Lfs(i)Load value L of the ith point of the day with similar trendfsminMinimum load value, L for day with similar trendfsmaxMaximum load value for day with similar trend, LfmaxThe maximum load prediction value for the day to be predicted, Lf0minThe minimum load value of the previous day.
In another exemplary embodiment of the present application, the summer and winter load prediction is as described above, and the technical solution of the spring and autumn load prediction is as follows:
the main factors influencing the load change are: day type (whether it is weekend, common holiday, special holiday), date distance, highest air temperature, lowest air temperature, rainfall. The main factors affecting load changes are also different in different situations, and there are generally one to two dominant factors. And quantizing the influence factors to obtain a similarity array. When prediction is carried out, the similarity of the historical days and the days to be predicted is sequenced, the day with the highest similarity is the similar day, and the load of each point of the similar day is set to be Lfs(i)(i＝1～96)。
The daily load to be predicted is
Lf(i)＝Lfs(i)*(1-β)+Lf0(i)*β (4)
Wherein L isf0(i)The load of the ith point in the previous day is 96 load points, β is an adaptive coefficient, and i is 1-96.
In order to make the technical solutions of the present application more clearly understood by those skilled in the art, the technical solutions of the present application will be described in detail below with reference to specific examples and comparative examples.
As shown in fig. 3, the specific design steps of the present invention are:
step 1, reading historical data: and acquiring historical day weather data and load data.
Quantification of day type: the types of days are divided into four types, namely working days, common holidays (including weekends) and special holidays. The quantized day type value also represents the influence degree of the day type on the load prediction, for example, the quantized day type value of a working day is 0, saturday in weekend is 1, sunday is 2, and the specific values of all holidays and days before and after the holidays are determined by analyzing historical load data by a trend analysis method.
Quantification of weather conditions: since the data given by the weather conditions do not include humidity and rainfall, the procedure is replaced by weather conditions. The weather conditions are mainly classified into fine, cloudy, fog, haze, sand, light rain, rain fall, thunderstorm, light to medium rain, medium to heavy rain, heavy to heavy rain, sleet, snow, light snow, small to medium snow, medium to heavy snow, heavy snow and heavy snow. The influence degrees of rain and snow on the load are respectively larger, and the influence degrees of other loads are not large. Specifically, the method comprises the following steps: in high-temperature weather, the gust of rain can form damp and hot weather, the load can be increased by a small amount, while the ordinary rainfall can cause the temperature reduction to further reduce the load, and the reduction amplitude is different according to the difference of rainfall; in low-temperature weather, the load is increased due to rainfall, but the amplitude is smaller than that at high temperature; when the temperature is moderate, the rainfall has no influence on the load.
Quantification of air temperature: quantification of air temperature takes into account two factors, namely cumulative effects and non-linearities. The influence of the air temperature on the load is divided into two parts: a maximum air temperature and a minimum air temperature. The highest air temperature mainly affects the maximum load in summer days, and the lowest air temperature mainly affects the maximum load in winter days. The specific treatment steps mainly comprise two steps: firstly, adding a first-order filtering link to the air temperature, and fully considering the accumulation of the temperature and the time delay effect; secondly, low temperature is converted into corresponding high temperature.
Step 3, calculating the body sensing temperature:
wherein, TA: maximum sensible temperature, T: maximum influence temperature (highest temperature in summer and lowest temperature in winter), RH: relative humidity, V: wind speed.
And obtaining the historical body sensing temperature of each day according to the formula and the quantized historical data.
Lf2＝k*TA
Lfmax＝(Lf1+Lf2+Lf3)*(1-β)+Lf0max*β
wherein k is a proportionality coefficient, LfmaxThe maximum load prediction value for the day to be predicted, Lf0maxFor the previous day maximum load, Lf1、Lf2And Lf3Fluctuating load, main variable load and basic load respectively, and β is an adaptive coefficient.
When predicted, the load L of the previous dayf0Determining the value, the daily sensible temperature T to be predictedAAccording to the weatherReporting is carried out. L isf1、Lf3And k and β are determined by Latin hypercube mining historical load data during model training, and are determined values during specific prediction, wherein the specific determination method is monthly mining.
And 6, selecting a load trend curve: and selecting the historical day closest to the day to be predicted by a similar day method. Specifically, the similarity of weather conditions and day types is determined, the date interval is considered, the historical day with the highest similarity is selected, and the load curve line type of the day is taken as the line type of the day to be predicted.
Step 7, load prediction on days: the highest load of the day to be predicted is obtained by prediction in the step 5, and the lowest load is selected as the lowest load of the previous day. The profile was selected as the loading profile for days with similar trends.
Firstly, normalization is carried out: and selecting the highest and lowest load values of days with similar trends, subtracting the lowest load value from the 96-point load data, and dividing by the difference between the highest load and the lowest load. Then, the normalized 96-point data is multiplied by the difference between the predicted highest load and the predicted lowest load, and the predicted lowest load value is added to obtain the predicted value of the all-day load.
The predicted load value of the ith point is as follows:
Lf(i)＝(Lfs(i)-Lfsmin)/(Lfsmax-Lfsmin)*(Lfmax-Lf0min)+Lf0min
wherein i is 1-96, Lf(i)The predicted value of the load at the ith point of the day to be predicted, Lfs(i)Load value L of the ith point of the day with similar trendfsminMinimum load value, L for day with similar trendfsmaxMaximum load value for day with similar trend, LfmaxThe maximum load prediction value for the day to be predicted, Lf0minThe minimum load value of the previous day.
The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (5)
1. The short-term load forecasting method based on the body sensing temperature is characterized by being used for forecasting loads in summer and winter, and comprises the following steps:
reading of historical data: acquiring historical solar weather data and power load data;
quantification of historical data: quantifying different factors influencing the change of the power load according to the influence degree of the factors, wherein the factors comprise the quantification of the day type and the quantification of the meteorological information;
and (3) calculating the somatosensory temperature: calculating the somatosensory temperature of the history every day according to the quantized historical data and a somatosensory temperature calculation formula;
collecting information of days to be predicted: acquiring meteorological information and power load information of a day to be predicted;
predicting the daily maximum load: inputting the acquired day information to be predicted to a daily maximum load prediction model for daily maximum load prediction;
selecting a load trend curve: selecting a historical day closest to the day to be predicted by a similar day method, and taking the load curve line type of the day as the load curve line type of the day to be predicted;
and (3) daily load prediction: selecting the highest and lowest load values of the day with similar trends, subtracting the lowest load value from the load data, dividing the load data by the difference between the highest load and the lowest load, multiplying the normalized data by the predicted difference between the highest load and the lowest load, and adding the predicted lowest load value to obtain the predicted value of the load all day;
aiming at historical power load data, firstly, power loads need to be classified into basic loads, main transformer loads and fluctuating loads;
the base load refers to the part of the load which is basically unchanged or slightly changed in a short time; the main transformer load refers to the part of load with larger load variation amplitude; the fluctuating load refers to the part of load with small load change amplitude, and refers to small load fluctuation caused by holidays or emergencies;
wherein the predicted value of the main transformer load is
Lf2＝k*TA
Wherein, TA: maximum sensible temperature, T: maximum influence temperature, highest temperature in summer, lowest temperature in winter, RH: relative humidity, V: wind speed, k is a proportionality coefficient;
base load Lf3The method has consistency with a k coefficient used in main transformer load prediction and β in a short time, so that Latin hypercube sampling is adopted to analyze historical data according to months to obtain the optimal L suitable for the historical dataf3K and β:
Lfmax＝(Lf1+Lf2+Lf3)*(1-β)+Lf0max*β
wherein L isfmaxThe maximum load prediction value for the day to be predicted, Lf0maxFor the previous day maximum load, Lf1、Lf2And Lf3Respectively, fluctuating load, main transformer load and basic load, wherein β is an adaptive coefficient;
the predicted load value of the ith point is as follows:
Lf(i)＝(Lfs(i)-Lfsmin)/(Lfsmax-Lfsmin)*(Lfmax-Lf0min)+Lf0min
wherein i is 1-96, Lf(i)The predicted value of the load at the ith point of the day to be predicted, Lfs(i)Load value L of the ith point of the day with similar trendfsminMinimum load value, L for day with similar trendfsmaxMaximum load value for day with similar trend, LfmaxThe maximum load prediction value for the day to be predicted, Lf0minThe minimum load value of the previous day.
2. The method for predicting the short-term load based on the somatosensory temperature as claimed in claim 1, wherein the quantification of the day types is that the day types are divided into: and on working days, common holidays including weekends and special holidays, the quantized day type numerical values also represent the influence degree of the day type on load prediction, and the specific numerical values of all holidays and days before and after the holidays are determined by analyzing historical power load data by a trend analysis method.
3. The method for predicting short-term load based on somatosensory temperature according to claim 1, wherein the quantification of weather information comprises quantification of weather conditions and quantification of air temperature, and the quantification of air temperature specifically comprises: firstly, adding a first-order filtering link to the air temperature, and fully considering the accumulation of the temperature and the time delay effect; secondly, low temperature is converted into corresponding high temperature.
4. The short-term load prediction method based on somatosensory temperature as claimed in claim 1, wherein the trend similar day of the daily load curve to be predicted is selected by a similar day method: specifically, the similarity of weather conditions and day types is determined, the date interval is considered, the historical day with the highest similarity is selected as the trend similar day, and the load curve line type of the day is taken as the load curve line type of the day to be predicted.
5. The method for predicting short-term load based on sensible temperature according to claim 1, wherein the minimum load of the day before the day to be predicted is taken as the minimum load of the day to be predicted, because the minimum load variation in the whole year is small and can be ignored in adjacent days.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710224770.8A CN107423836B (en) | 2017-04-07 | 2017-04-07 | Short-term load prediction method based on body sensing temperature |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710224770.8A CN107423836B (en) | 2017-04-07 | 2017-04-07 | Short-term load prediction method based on body sensing temperature |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107423836A CN107423836A (en) | 2017-12-01 |
| CN107423836B true CN107423836B (en) | 2020-04-28 |
Family
ID=60423670
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710224770.8A Active CN107423836B (en) | 2017-04-07 | 2017-04-07 | Short-term load prediction method based on body sensing temperature |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107423836B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108647471B (en) * | 2018-05-30 | 2022-05-17 | 新奥泛能网络科技有限公司 | Computing method and device for cold and heat combined supply equipment installation |
| CN109034504B (en) * | 2018-09-14 | 2021-06-25 | 云南电网有限责任公司 | Method and device for establishing short-term load prediction model |
| CN109886567B (en) * | 2019-01-31 | 2022-09-30 | 国电南瑞科技股份有限公司 | Short-term load prediction method considering somatosensory temperature and radiation intensity |
| CN110334366B (en) * | 2019-03-14 | 2023-07-14 | 华电电力科学研究院有限公司 | Building instantaneous cold load prediction method based on Monte Carlo method using Latin hypercube sampling |
| CN109978262A (en) * | 2019-03-27 | 2019-07-05 | 新奥数能科技有限公司 | The selection method and device of similar day for photovoltaic power prediction |
| CN111598303A (en) * | 2020-04-20 | 2020-08-28 | 广东电网有限责任公司电力调度控制中心 | Summer short-term load prediction method based on meteorological component decomposition |
| CN116817416B (en) * | 2023-02-24 | 2024-01-23 | 国网能源研究院有限公司 | Method and system for predicting maximum load of whole society |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102819770A (en) * | 2012-08-02 | 2012-12-12 | 积成电子股份有限公司 | Method for forecasting short-term load of distribution network based on human body comfort index |
| CN103065201A (en) * | 2012-12-19 | 2013-04-24 | 福建省电力有限公司 | Electric load prediction method used for electric power based on factors of temperature and festivals and holidays |
| CN103093288A (en) * | 2013-02-21 | 2013-05-08 | 江苏省电力公司电力科学研究院 | Partition power grid bus load prediction system based on weather information |
| CN105701570A (en) * | 2016-01-11 | 2016-06-22 | 国网浙江省电力公司经济技术研究院 | Short-term electric power demand analysis method based on overall process technology improvement |
| CN106408223A (en) * | 2016-11-30 | 2017-02-15 | 华北电力大学（保定） | Short-term load prediction based on meteorological similar day and error correction |
-
2017
- 2017-04-07 CN CN201710224770.8A patent/CN107423836B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102819770A (en) * | 2012-08-02 | 2012-12-12 | 积成电子股份有限公司 | Method for forecasting short-term load of distribution network based on human body comfort index |
| CN103065201A (en) * | 2012-12-19 | 2013-04-24 | 福建省电力有限公司 | Electric load prediction method used for electric power based on factors of temperature and festivals and holidays |
| CN103093288A (en) * | 2013-02-21 | 2013-05-08 | 江苏省电力公司电力科学研究院 | Partition power grid bus load prediction system based on weather information |
| CN105701570A (en) * | 2016-01-11 | 2016-06-22 | 国网浙江省电力公司经济技术研究院 | Short-term electric power demand analysis method based on overall process technology improvement |
| CN106408223A (en) * | 2016-11-30 | 2017-02-15 | 华北电力大学（保定） | Short-term load prediction based on meteorological similar day and error correction |
Non-Patent Citations (3)
| Title |
|---|
| 基于体感温度的电力系统负荷分类及负荷预测;谢家安;《电网与清洁能源》;20120831;第28卷(第8期);第24-28页 * |
| 基于相似日搜索的空调短期负荷预测方法;王小刚等;《华中科技大学学报》;20111231;第39卷(第12期);第76-80页 * |
| 电力短期负荷预测相似日选取算法;黎灿兵等;《电力系统自动化》;20080510;第32卷(第9期);第69-72页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107423836A (en) | 2017-12-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107423836B (en) | Short-term load prediction method based on body sensing temperature | |
| CN105069525B (en) | Round-the-clock 96 Day Load Curve Forecastings and optimization update the system | |
| Chikobvu et al. | Regression-SARIMA modelling of daily peak electricity demand in South Africa | |
| CN109886567B (en) | Short-term load prediction method considering somatosensory temperature and radiation intensity | |
| CN110110912B (en) | Photovoltaic power multi-model interval prediction method | |
| CN105631558A (en) | BP neural network photovoltaic power generation system power prediction method based on similar day | |
| CN107730031A (en) | A kind of ultra-short term peak load forecasting method and its system | |
| CN110796307B (en) | Distributed load prediction method and system for comprehensive energy system | |
| WO2015081660A1 (en) | Method for forecasting residential quarter short-term load | |
| CN104156786A (en) | Non-working day maximum daily load forecasting system considering meteorological multi-factor influences | |
| CN105701570A (en) | Short-term electric power demand analysis method based on overall process technology improvement | |
| CN102147839A (en) | Method for forecasting photovoltaic power generation quantity | |
| CN114372360A (en) | Method, terminal and storage medium for power load prediction | |
| CN114021420A (en) | Distributed photovoltaic ultra-short-term power generation prediction method and system | |
| CN115545333A (en) | Method for predicting load curve of multi-load daily-type power distribution network | |
| CN115689146A (en) | Multi-energy intelligent building load scheduling method and system | |
| CN115222106A (en) | User day-ahead load prediction method of self-adaptive model | |
| CN108346009A (en) | A kind of power generation configuration method and device based on user model self study | |
| CN111815039A (en) | Weekly scale wind power probability prediction method and system based on weather classification | |
| CN104915727B (en) | Various dimensions allomer BP neural network optical power ultra-short term prediction method | |
| Ku et al. | Building electric energy prediction modeling for BEMS using easily obtainable weather factors with Kriging model and data mining | |
| CN113887833A (en) | Distributed energy user side time-by-time load prediction method and system | |
| CN104252647B (en) | Electro-load forecast method based on anti-distance weighting interpolation method | |
| CN106022542A (en) | Enterprise gateway load prediction method based on operation load characteristics | |
| CN110489893B (en) | Variable weight-based bus load prediction method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |