CN107506872A - A kind of residential block part throttle characteristics and the Categorical research method of model prediction - Google Patents

Abstract

The invention relates to a classification research method for load characteristics of residential areas and model prediction. By establishing the state space of the residential area, the residential area is divided into different categories, and the load characteristics of other residential areas of the same type are used to fit the load situation of the residential area to be predicted. The invention grasps the characteristic development law, scientifically formulates power consumption planning; classifies and establishes a prediction model, which can conveniently and effectively predict loads, is helpful for scientific deployment of power supply networks, and better enables economical and safe operation of power supply networks.

Description

A Classification Research Method of Residential Area Load Characteristics and Model Prediction

technical field

The invention relates to the field of residential electricity loads, in particular to a classification research method for residential area load characteristics and model prediction.

Background technique

With the rapid development of our country's economy and the further deepening of power system reform, power market analysis is becoming more and more important to the operation and planning of power companies. Load characteristic analysis and investigation are the basis of power system planning, and are necessary means to understand and predict users and markets within the jurisdiction. Residential load is an important part of urban load. It is of great significance to investigate and study the characteristics of urban resident load, to forecast resident load, to formulate power supply schemes in residential areas, to plan urban power grids, to operate power grids economically, and to market electricity. Through in-depth analysis of the regional load characteristics and an in-depth understanding of the region's load characteristics, effective load monitoring can be carried out to improve the power consumption on the demand side and make the entire grid load more stable, thereby improving social benefits. Power system load forecasting can be divided into long-term, medium-term, short-term and ultra-short-term forecasting according to different time periods. The research methods include regression prediction method, trend extrapolation method, time series method, neural network method and so on. If the criterion for selecting a forecasting model is to maximize the forecasting accuracy, it is best to choose a time series model. The basic idea of the time series method to predict the power load is: collect a large amount of accurate historical data, according to the similarity between the future and the past time series, reveal the law of its change over time through the historical load data, establish a scientific model, and conduct a large number of In order to continuously improve the model and achieve the best prediction results.

There are many fruitful research works on the load characteristic analysis of power system. Although these research results have certain reference significance for the regional load characteristic analysis, they also have their own deficiencies and limitations.

Contents of the invention

The object of the present invention is to provide a classification research method for load characteristics and model prediction of residential areas. The method predicts the load of new residential areas or detects the load anomalies of existing residential areas through the load data of existing residential areas, greatly improving Reducing the workload is helpful for the scientific deployment of the power supply network and better economic and safe operation of the power supply network.

In order to achieve the above object, the technical solution of the present invention is: a classification research method of residential area load characteristics and model prediction, comprising the following steps,

S1. Establish the characteristic space of the residential area and classify the categories: according to the characteristic data of the residential area, including the location of the area, the floor area, the floor area ratio, the total number of households, the average price of houses, the opening time, property fees, building categories, educational resources, and living Recreational resources, traffic travel data, these data are processed, so that each residential area is represented by a set of numerical features, and then K-MEANS cluster analysis is performed on each residential area to complete the classification of residential areas;

S2. Establishing an ARIMA prediction model to effectively predict load;

S3. Establish load forecasting of the same type of residential area: for new residential areas, quantify the load indicators of new residential areas according to the load data of existing residential areas with similar characteristic indicators, and predict the load index of new residential areas through the development law of existing residential areas load development.

In an embodiment of the present invention, for two residential areas with similar characteristic indexes, it is necessary to carry out the inspection of abnormal load data, that is, if A and B are two residential areas with similar characteristic indexes, extract characteristics, and establish an ARIMA prediction model for A and B; if the characteristics of A and B cells are similar, the load characteristics will be similar, and vice versa; if the characteristics are similar but the load characteristics are far apart, it is considered abnormal.

In an embodiment of the present invention, the specific implementation process of the step S1 is as follows:

The characteristic data of residential areas are processed as follows:

Regional location: given a unique number according to the community and street;

Land area, floor area ratio, total number of households, average house price, property fee: normalized so that the value range is 0~1;

Opening time: converted into the time difference with the current time, normalize the time difference so that the value range is 0~1;

Building category: divide the building category into one or more combinations of villas, bungalows, small high-rises, and high-rises, and give each form a unique number;

Educational resources: Educational resources are divided into the number of ordinary primary schools, the number of key primary schools, the number of ordinary junior high schools, and the number of key junior high schools. The weights are respectively assigned to 3, 3.5, 4, and 4.5. After calculation, they are normalized so that the value range is 0~1 ;

Life and entertainment resources: Life and entertainment resources are divided into the number of banks, restaurants, movie theaters, shopping malls, and complexes. The weights are assigned to 1, 1, 5, 10, and 20 respectively. 0~1;

Traffic travel: traffic travel is divided into buses and train stations. When the bus line is <5, the value is 0; when the bus line is >5, the value is 0.5; when there is a train station, the value is 1;

After the above-mentioned processing of the characteristic data of residential areas, each residential area is represented by a set of numerical features, and then K-MEANS cluster analysis is performed on each residential area to complete the classification of residential areas.

Compared with the prior art, the present invention has the following beneficial effects: the present invention predicts the load of newly built residential areas or detects the abnormal load of existing residential areas through the load data of existing residential areas, which greatly reduces the workload and contributes to Scientific deployment of the power supply network will better enable the economic and safe operation of the power supply network.

Description of drawings

Fig. 1 is a flow chart of prediction based on characteristic indicators consistent with the embodiment of the present invention.

detailed description

The technical solution of the present invention will be specifically described below in conjunction with the accompanying drawings.

The classification research method of a kind of residential area load characteristics and model prediction of the present invention, comprises the following steps,

S1. Establish the characteristic space of the residential area and classify the categories: according to the characteristic data of the residential area, including the location of the area, the floor area, the floor area ratio, the total number of households, the average price of houses, the opening time, property fees, building categories, educational resources, and living Recreational resources, traffic travel data, these data are processed, so that each residential area is represented by a set of numerical features, and then K-MEANS cluster analysis is performed on each residential area to complete the classification of residential areas;

S2. Establishing an ARIMA prediction model to effectively predict load;

S3. Establish load forecasting of the same type of residential area: for new residential areas, quantify the load indicators of new residential areas according to the load data of existing residential areas with similar characteristic indicators, and predict the load index of new residential areas through the development law of existing residential areas load development.

For two residential areas with similar characteristic indexes, it is necessary to test the abnormal load data, that is, if A and B are two residential areas with similar characteristic indexes, extract the characteristics of the two residential areas respectively, and establish ARIMA prediction model; if the characteristics of A and B cells are similar, the load characteristics will be similar, and vice versa; if the characteristics are similar but the load characteristics are far away, it is considered abnormal.

The specific implementation process of the step S1 is as follows:

The characteristic data of residential areas are processed as follows:

Regional location: given a unique number according to the community and street;

Land area, floor area ratio, total number of households, average house price, property fee: normalized so that the value range is 0~1;

Opening time: converted into the time difference with the current time, normalize the time difference so that the value range is 0~1;

Building category: divide the building category into one or more combinations of villas, bungalows, small high-rises, and high-rises, and give each form a unique number;

Educational resources: Educational resources are divided into the number of ordinary primary schools, the number of key primary schools, the number of ordinary junior high schools, and the number of key junior high schools. The weights are respectively assigned to 3, 3.5, 4, and 4.5. After calculation, they are normalized so that the value range is 0~1 ;

Life and entertainment resources: Life and entertainment resources are divided into the number of banks, restaurants, movie theaters, shopping malls, and complexes. The weights are assigned to 1, 1, 5, 10, and 20 respectively. 0~1;

Traffic travel: traffic travel is divided into buses and train stations. When the bus line is <5, the value is 0; when the bus line is >5, the value is 0.5; when there is a train station, the value is 1;

After the above-mentioned processing of the characteristic data of residential areas, each residential area is represented by a set of numerical features, and then K-MEANS cluster analysis is performed on each residential area to complete the classification of residential areas.

The following is the specific implementation process of the present invention.

Residential area load characteristic of the present invention and the classification research method of model prediction, comprise the following steps:

(1) Establish the characteristic space of the residential area and divide it into categories

The characteristic data of residential areas include: regional location, floor area, floor area ratio, total number of households, average house price, opening time, property fees, building categories, educational resources, living and entertainment resources, and transportation. These data can be processed as follows deal with:

Regional location: given a unique number according to the community and street;

Land area, total number of households, average house price, property fee, floor area ratio: normalized so that the value range is 0~1;

Opening time: converted into the time difference with the current time, normalize the time difference so that the value range is 0~1;

Building category: Building categories can be divided into villas, bungalows, small high-rises, and high-rises. There are always 15 combination forms, and each form is given a unique number;

Educational resources: Educational resources are divided into the number of ordinary primary schools, the number of key primary schools, the number of ordinary junior high schools, and the number of key junior high schools. The weights are respectively assigned to 3, 3.5, 4, and 4.5. After calculation, they are normalized so that the value range is 0~1 ;

Life and entertainment resources: Life and entertainment resources are divided into the number of banks, restaurants, movie theaters, shopping malls, and complexes. The weights are assigned to 1, 1, 5, 10, and 20 respectively. 0~1;

Traffic travel: traffic travel is divided into buses and train stations. When the bus line is <5, the value is 0; when the bus line is >5, the value is 0.5; when there is a train station, the value is 1.

After the data are processed above, each residential area is described by a set of numerical features, and K-MEANS cluster analysis is performed on the residential area to complete the classification of the residential area.

(2) Establish ARIMA forecasting model to effectively forecast load

The so-called ARIMA model refers to the model established by converting the non-stationary time series into a stationary time series, and then regressing the dependent variable only on its lag value and the present value and lag value of the random error item. The ARIMA model includes moving average process (MA), autoregressive process (AR), autoregressive moving average process (ARMA) and ARIMA process according to whether the original sequence is stationary or not and the part included in the regression.

(3) Establish load forecasting for the same type of cell

For newly-built communities, quantifying the load indicators of newly-built communities through the load data of existing communities will help reduce the load analysis work of newly-built communities, and accurately grasp the load development of newly-built communities through the development rules of existing communities;

For two cells with similar characteristic indexes, the inspection of abnormal load data can be carried out. As shown in the figure, the features of A and B are extracted respectively, and a prediction model is established for A and B. If the characteristics of A and B cells are similar, the load characteristics will be similar, and vice versa. If the characteristics are similar and the load characteristics are far away, consider an abnormal situation.

Principle of the present invention is:

The present invention divides the categories by establishing the characteristic space of the residential area. If a suitable forecasting model is established through the historical data of a community, then this model is also applicable to the load forecasting of similar communities. With the change of time, the power consumption characteristics of the same community also change, and this change rule can also be classified and analyzed, which greatly reduces the workload.

The beneficial effects of the present invention are:

The present invention predicts the load of newly built residential areas or detects the abnormal load of existing residential areas through the load data of existing residential areas, which greatly reduces the workload, helps to scientifically deploy the power supply network, and better makes the power supply Safe operation of network economy.

Through the above description of the implementation manners, those skilled in the art can clearly understand that the above embodiments can be implemented by software, or by means of software plus a necessary general hardware platform. Based on this understanding, the technical solutions of the above embodiments can be embodied in the form of software products, which can be stored in a non-volatile storage medium (which can be CD-ROM, U disk, mobile hard disk, etc.), including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute the methods described in various embodiments of the present invention.

The above are the preferred embodiments of the present invention, and all changes made according to the technical solution of the present invention, when the functional effect produced does not exceed the scope of the technical solution of the present invention, all belong to the protection scope of the present invention.

Claims (3)

1. a kind of residential block part throttle characteristics and the Categorical research method of model prediction, it is characterised in that：Comprise the following steps,

S1, residential block feature space is established, divide classification：According to the characteristic of residential block, including regional location, occupation of land face Product, plot ratio, total amount, house average price, opening quotation time, property fees, building classification, educational resource, life ＆ amusement resource, traffic Trip data, those data are handled so that each residential block is by one group of character representation to quantize, then to each residence People area carries out K-MEANS cluster analyses, completes the category division of residential block；

S2, establish ARIMA forecast models and effectively predict load；

S3, establish the prediction of same type cell load：For area for settlement, according to the negative of the similar residential block of existing characteristic index The load index of lotus data quantification area for settlement, and predict that the load of area for settlement is sent out by the rule of development of existing residential block Exhibition.

2. a kind of residential block part throttle characteristics according to claim 1 and the Categorical research method of model prediction, its feature exist In：For the similar residential block of two characteristic index, the inspection of load abnormal data need to be carried out, even A and B are two spies Feature is extracted, and establish ARIMA forecast models to A and B in two residential blocks by the property similar residential block of index respectively；If A and B The feature of cell is close, then part throttle characteristics also can be close, and vice versa；If occurring, feature is close, the mutually remote situation of part throttle characteristics, Then think exception occur.

3. a kind of residential block part throttle characteristics according to claim 1 and the Categorical research method of model prediction, its feature exist In：The specific implementation process of the step S1 is as follows：

Following processing is done to the characteristic data of residential block：

Regional location：Unique numbering is given according to community, street；

Floor space, plot ratio, total amount, house average price, property fees：Normalization, it is 0 ~ 1 to make its codomain；

Open the set the time：The time difference with current time is converted into, the time difference is normalized, it is 0 ~ 1 to make its codomain；

Build classification：Building classification is divided into one or more combination forms in villa, foreign-style house, small high-rise, high level, it is every kind of Form gives unique numbering；

Educational resource：Educational resource is divided into primary school's quantity, emphasis primary school quantity, common junior middle school's quantity, emphasis junior middle school quantity, Weights normalize after calculating not assign 3,3.5,4,4.5, make its codomain for 0 ~ 1；

Life ＆ amusement resource：Life ＆ amusement resource is divided into bank quantity, restaurant quantity, cinema's quantity, market quantity, synthesis Quantity weights normalize after calculating not assign 1,1,5,10,20, make its codomain for 0 ~ 1；

Traffic trip：Traffic trip is divided into bus, railway station, works as bus routes<When 5, it is worth for 0；Bus routes>When 5, It is worth for 0.5；It is 1 to have railway station duration；

After the above-mentioned processing of characteristic data progress to residential block so that each residential block is by one group of mark sheet to quantize Show, K-MEANS cluster analyses then are carried out to each residential block, complete the category division of residential block.