KR101645445B1 - Load prediction apparatus and method of railway power feeder system using self organied map - Google Patents

Abstract

The present invention relates to a data collection module for collecting load data and a method for grouping input functions by mapping preset input functions among load data collected from a data collection module to a self organizing map and predicting a railroad load on the basis of the grouped input function And a self-organizing mapping module.

Description

BACKGROUND OF THE INVENTION Field of the Invention [0001] The present invention relates to a load forecasting apparatus for a railroad feed system,

The present invention relates to an apparatus and method for predicting a load of an railroad feed system using a self organization map, and more particularly, to a system and method for predicting load of a railroad feed system by grouping load data of a railroad feed system into a self- To an apparatus and method for predicting a load of a railroad feed system using an organization map.

Generally, railway electric power facilities are divided into a first system that supplies power supplied from electric power companies and substations to railway vehicles, and a second system that supplies electric power to power stations in the station (station).

Extra-high voltage power supplied by electric power companies is divided into electricity supplied to the vehicle from the high-voltage switchboard and power consumed within the history. The electric power supplied to the train is supplied to the train through the rectifier transformer, the rectifier, the DC switchboard and the electric cable. The power consumed in the history is supplied to various loads in history through the distribution transformer, high voltage switchboard, do.

In such a railway power facility, it is very important to predict the power consumed by the railway facility. This is because it is directly linked to the operation of the railway and electricity charges. Conventionally, long-term load data used in railway electric power facilities is statistically analyzed and predicted by season, week, and week based on the analysis results.

However, since the conventional load prediction method estimates the railway load only by the average value of the past load data, the load prediction error is large, and environmental factors such as the current temperature and humidity are not reflected. As a result, the conventional load prediction method has a problem that the accuracy of the prediction result is low.

The background art of the present invention is disclosed in Korean Patent Laid-Open Publication No. 10-2010-0067343 (Jun. 22, 2010).

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and it is an object of the present invention to provide a railway feeding system using a self organization map, which can predict a railway load in real time by grouping load data of a railway feeding system into a self- And a method for predicting the load of the apparatus.

Another object of the present invention is to provide a load prediction apparatus for a railroad feed system using a self organization map, which can relatively accurately perform railroad load prediction by setting environment factors such as temperature and humidity as input functions of a self- And a method thereof.

It is still another object of the present invention to provide an apparatus and a method for predicting a load of a railroad feeding system using a magnetic organization map, which can reduce a peak power by predicting a railroad load relatively accurately through a self organization map, .

An apparatus for predicting a load of a railroad feed system using a self-organization map according to an aspect of the present invention includes a data collection module for collecting load data; And a self organization mapping module for mapping the input function among the load data collected from the data collection module to a self organization map to group the input function and predicting a railway load based on the grouped input function .

In the present invention, the load data may include power data of a substation supplied from a substation to a railway vehicle and a historical facility, railway vehicle schedule data on the operation of the railway vehicle, historical facility schedule data on a historical facility load operation, And the internal environment data on the internal environment of the history.

In the present invention, the self-organizing mapping module calculates a Euclidean distance between the input function and a neuron, searches for a neuron having a minimum distance among the neurons calculated, corrects a weight of the input function, And the input functions are grouped according to the degree of similarity until the weights become equal to each other.

In the present invention, the self-organizing mapping module maps the input function to the self-organizing map, groups the input function, and selects a railroad load based on any one of the grouped input functions having the highest similarity.

According to an aspect of the present invention, there is provided a method of predicting a load of a railroad feeding system using a self-organizing map, comprising: collecting load data; Grouping the input function by mapping a predetermined input function among the collected load data to a self organizing map; And predicting a railway load based on the grouped input function.

In the present invention, the load data may include power data of a substation supplied from a substation to a railway vehicle and a historical facility, railway vehicle schedule data on the operation of the railway vehicle, historical facility schedule data on a historical facility load operation, And the internal environment data on the internal environment of the history.

In the present invention, the step of grouping the input function may include: calculating a Euclidean distance between the input function and a neuron to search for a neuron having a minimum distance among the neurons calculated; The input function is grouped according to the degree of similarity until the weight and the next weight have the same value.

In the present invention, the step of predicting the railway load is characterized by mapping the input function to the self organization map, and selecting a railway load based on any one of the grouped input functions having the highest similarity .

The present invention relates to a system and method for predicting a railway load by grouping load data of a railway feed system into a self organization map so as to predict a railway load in real time and setting environmental factors such as temperature and humidity as an input function of a self- So that prediction can be performed.

The present invention makes it possible to predict the railway load relatively accurately through the self-organization map to reduce the peak power, thereby reducing the power operation cost.

1 is a block diagram of a load prediction apparatus of a railroad feed system using a self organization map according to an embodiment of the present invention.
2 is a view showing load data according to an embodiment of the present invention.
3 is an exemplary diagram illustrating an example of load data grouped by a self organization map according to an embodiment of the present invention.
4 is a flowchart of a method of predicting a load of a railroad feed system using a self organization map according to an embodiment of the present invention.
5 is a flowchart illustrating a learning process of a self organization map according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an apparatus and method for predicting a load of a railroad feed system using a self-organizing map according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings. In this process, the thicknesses of the lines and the sizes of the components shown in the drawings may be exaggerated for clarity and convenience of explanation. Further, the terms described below are defined in consideration of the functions of the present invention, which may vary depending on the user, the intention or custom of the operator. Therefore, definitions of these terms should be made based on the contents throughout this specification.

FIG. 1 is a block diagram of a load predicting apparatus for a railroad feed system using a self-organizing map according to an embodiment of the present invention. FIG. 2 is a diagram illustrating load data according to an embodiment of the present invention. FIG. 8 is a diagram illustrating an example of load data grouped by a self organization map according to an embodiment of the present invention. FIG.

1, a load prediction apparatus for a railroad feed system using a self-organizing map according to an embodiment of the present invention includes a data collection module 10, a load data storage unit 20, and a magnetic tissue mapping module 30 .

The data collection module 10 collects load data that is input to the self organization map. The data acquisition module 10 may be installed at various locations such as a substation of a railway power supply system, a railway vehicle, a historical facility, a rail, and the like.

2, the load data collected by the data collection module 10 includes power data of a substation supplied from a substation (not shown) to a railway vehicle (not shown) and a historical facility (not shown) Railway vehicle schedule data for the historical facility operation, historical facility schedule data for the historical facility operation, external environment data for the environment outside the history, and internal environment data for the internal environment for the history.

Power data of a substation supplied from a substation to a railway vehicle includes power data on power supplied to a railway vehicle, power data on power recovered from a railway vehicle, peak power, and power data May be included.

The power data of the substation supplied from the substation to the historical facility can include all of the power data on the power required for the operation of the historical facility.

The schedule data of the railway vehicle for the operation of the railway vehicle may include all the data related to the operation of the railway vehicle, such as the data on the running time of the railway vehicle with respect to the running time of the railway vehicle,

The historical facility schedule data includes the availability of the history load facility and the operation time data. In addition, the railway vehicle schedule data and the historical facility schedule data described above include a railway system operation server (not shown) that integrally manages and manages the entire railway, or a history management server installed in the history and operating and managing the history ) Or the like.

Here, the historical facility schedule data includes schedule data such as air conditioning equipment, lighting equipment, computer equipment, and communication equipment installed in the history.

External environment data on the environment outside the historical facilities and internal environment data on the environment inside the history facilities include temperature and humidity in the history room, temperature and humidity outside the room, and historical room congestion information.

Particularly, the external environmental data and internal environment data are not limited to the environmental information of the railway vehicle and the historical indoor and outdoor, but may include various natural environment information related to the operation and management of the historical facilities. The environmental information can be detected through a railway system operation server (not shown) that operates and manages the entire railway, or a history management server (not shown) installed in the history and operating and managing the history.

The load data storage unit 20 stores the load data and inputs the load data to the self organizing mapping module 30.

The self organization mapping module 30 maps the load data collected from the data collection module 10 to the self organization map to group the load data and predicts the railroad load based on the grouped load data.

For reference, Neural Network emulates the characteristics of brain function by computer, and artificial neuron (node) that formed network by synaptic connection changes the synaptic bond strength through learning, Lt; / RTI >

In particular, the Self-Organized Map (SOM) is a system in which an information processing system voluntarily modifies an organization in a system based on past experiences and inputs information from outside It is a kind of neural network modeling structure.

This self organizing map is predominant in detecting the regularity and the correlations of the input pairs. In this embodiment, the neurons of the neural network are able to recognize and learn a set of similar input functions. The self organization map learns to classify input functions according to how input functions are grouped in the input space. Here, the load data is grouped in a space represented by several groups by the weights learned by the self organization map. Thus, the self-organizing map can classify input functions into several groups. FIG. 3 shows an example in which the input functions are grouped, that is, the results after the learning.

The self-organization mapping module 30 maps the load data, which are the input functions, to the self organization map, and groups the load data, and predicts the railroad load based on the grouped load data. Here, the input function can be set among the factors necessary for obtaining the above load data, for example, power data.

For example, when at least one of the schedule data and the environment data is set as the input function, the self organizing mapping module 30 groups at least one of the schedule data and the environment data, and based on the grouped data, For example, power data can be obtained.

When the power data of the substation of the load data is set by the input function, the self organizing mapping module 30 groups the power data of the corresponding substation and stores the power data of the substation in the electric power data of the substation, for example, Power data for power supplied or power data for power supplied only to historical facilities can be obtained.

That is, the input function can be variously set among the load data and is not limited to any one of the load data, and the railway load can also be variously determined according to the input function.

On the other hand, when the load data is input from the data acquisition module 10, the self-organizing mapping module 30 calculates the Euclidean distance between the pre-set input function and the neuron among the load data, And the weight of the load data is corrected to group the input function according to the degree of similarity until the weight of the previous stage becomes equal to the value of the next weight. This process is pre-learned using a plurality of past load data stored in the load data storage unit 20. [

Then, when the new load data is inputted, the self-organizing mapping module 30 maps the load data to the self organization map and groups them, and selects the railroad load based on any one of the grouped input functions having the highest similarity. Here, the railway load may be any one of various railroad electric powers that can be obtained based on the above-described input function.

A method of predicting a load of a railroad feed system using a self organization map according to an embodiment of the present invention will be described in detail with reference to FIGS. 4 and 5. FIG.

FIG. 4 is a flowchart illustrating a method of predicting a load of a railroad feed system using a self organization map according to an exemplary embodiment of the present invention, and FIG. 5 is a flowchart illustrating a self-organizing map learning process according to an embodiment of the present invention.

Referring to FIG. 4, the data collection module 10 collects load data (S10). The load data collected by the data collection module 10 includes power data of a substation supplied from a substation to a railway vehicle and a historical facility, railway vehicle schedule data on the operation of the railway vehicle, historical facility schedule data on the operation of the historical facility load, External environmental data on the external environment, and internal environmental data on the internal environment of the history.

In addition, the data collection module 10 stores the collected load data in the load data storage 20.

Meanwhile, when the load data is collected as described above, the self organizing mapping module 30 maps the input function among the collected load data to the self organizing map to group the input functions (S20). Here, the input function can be variously set in the load data. Therefore, the load data in Figs. 4 and 5 should be understood as an input function.

That is, the self-organizing mapping module 30 calculates the Euclidean distance between the input function and the neuron, searches for a neuron having a minimum distance among the calculated neurons, corrects the weight of the input function, The input function is grouped according to the degree of similarity.

Then, the self-organization mapping module 30 predicts the railway load based on the grouped input functions as described above (S30).

In this case, the input function can be variously set among the load data, and the self organizing mapping module 30 can acquire various railroad loads, for example, power data, of the railway electric power system based on the input function.

Since the load data must be grouped in advance in order to predict the railway load based on the current load data through the self organization map, the self organizing mapping module 30 calculates the load data stored in the load data storage unit 20, In other words, the input functions are grouped in advance by mapping the input functions to the self organization map. This will be described with reference to FIG.

Referring to FIG. 5, the self-organization mapping module 30 receives past load data from the load data storage unit 20 (S210).

Upon receiving the load data from the load data storage unit 20, the self organizing mapping module 30 calculates the Euclidean distance between the neuron and the preset input function in the load data (S220).

Next, the self-organizing mapping module 30 selects a neuron having a minimum distance among the neurons calculated in the step (S230), corrects the weight of the input function, and adjusts the weights until the weight of the front end becomes equal to the next weight (S240), and the input functions are grouped according to the degree of similarity (S250).

Here, the self-organization mapping module 30 maps the load data to the self-organizing map and groups the self-organization map, and the present invention is not limited to the above embodiment, and various methods can be applied as needed.

In addition, when the current load data is collected at the present time point, the load data becomes the past load data at the next time point. Accordingly, the load data storage unit 20 continuously stores the load data each time the load data is input. At this time, the self organizing mapping module 30 performs grouping It is possible to predict the railroad load at the next time point based on the load data.

In this embodiment, the load data of the railway feeding system is grouped into a self-organizing map so that the railroad load can be predicted in real time, and environmental factors such as temperature and humidity are set as input functions of the self- So that accurate railroad load prediction can be performed.

In addition, the present embodiment predicts the railway load relatively accurately through the self-organizing map to reduce the peak power, thereby reducing the power operation cost.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation, I will understand. Accordingly, the true scope of the present invention should be determined by the following claims.

10: Data acquisition module
20: Load data storage unit
30: Self-organizing mapping module

Claims (8)

A data acquisition module for collecting load data; And
A self-organizing mapping module for mapping a predetermined input function among the load data collected from the data collection module to a self-organizing map to group the input functions and predicting a railroad load based on the grouped input function, An apparatus for predicting the load of a railroad feed system using an organizational map.
2. The method of claim 1,
Electric power data of a substation supplied from a substation to a railway vehicle and a historical facility, railway vehicle schedule data on the operation of the railway vehicle, historical facility schedule data on the operation of the historical facility load, external environmental data on the environment outside the history, And internal environment data on the internal environment of the railroad feed system.
2. The method of claim 1, wherein the self-
Calculating a Euclidean distance between the input function and the neuron, searching for a neuron having a minimum distance among the neurons calculated by the Euclidean distance calculating unit, and correcting the weight of the input function, And the grouping of the input functions is performed according to the self-organizing map.
4. The method of claim 3, wherein the self-
Wherein the input function is mapped to the self organization map, and the railroad load is selected based on any one of the grouped input functions having the highest similarity.
Collecting load data;
Grouping the input function by mapping a predetermined input function among the collected load data to a self organizing map; And
And predicting a railway load based on the grouped input function.
6. The method according to claim 5,
Electric power data of a substation supplied from a substation to a railway vehicle and a historical facility, railway vehicle schedule data on the operation of the railway vehicle, historical facility schedule data on the operation of the historical facility load, external environment data on the environment outside the history, And internal environment data on the internal environment of the railroad feed system.
6. The method of claim 5, wherein grouping the input function comprises:
Calculating a Euclidean distance between the input function and the neuron, searching for a neuron having a minimum distance among the neurons calculated by the Euclidean distance calculating unit, and correcting the weight of the input function, Wherein the input function is grouped according to the self-organizing map.
6. The method of claim 5, wherein predicting the railroad load comprises:
Wherein the input function is mapped to the self organization map, and the railway load is selected based on any one of the grouped input functions having the highest similarity.

Images