KR20190131845A - System and method for predicting error of electric rail car - Google Patents

Abstract

The present application discloses a system and method for predicting a failure of an electric vehicle. According to an embodiment of the present application, provided is a system for predicting a failure of an electric vehicle, which comprises: a data generation unit for generating an integrated failure code table based on collected fault information, analyzing the integrated failure code table, and generation a frequency of failure data by extracting a frequency of each failure code occurred within a certain reference day; a data processing unit for generating a time series graph based on the failure frequency data and converting the time series graph into a stable time series graph; and a data prediction unit for analyzing the time series graph to drive an optimal parameter for application to an ARIMA model, and obtaining failure prediction result data by substituting the ARIMA model completed through the optimal parameter into an ARIMA prediction function.

Description

System and method for predicting failure of electric vehicle {SYSTEM AND METHOD FOR PREDICTING ERROR OF ELECTRIC RAIL CAR}

The present invention relates to a system and method for predicting a failure of an electric vehicle, and more particularly, to a system and a method for predicting a failure of an electric vehicle based on data obtained by extracting a frequency of failure of an electric vehicle generated within a specific reference day.

As an important means of public transportation, the subway used by many people is formed by connecting a plurality of electric cars. Electric cars are made up of a variety of electronic components, from relatively simple sensors such as lighting to critical power lines that power electric cars. As such, most of the components of the electric vehicle are electronic components and thus are not free from failures such as signal transmission error or power interruption.

In the subway, which is used by many people, minor failures of the electronic components that make up the electric train can lead to an accident such as a fire or a human accident. In addition, this may cause a large accident due to the nature of public transportation, it is necessary to monitor the breakdown of the train 24 hours.

Accordingly, the subway is equipped with a control center for each station to monitor the failure of the train in real time. However, various failure-related data obtained through such monitoring are used only for the purpose of determining a faulty part and repairing the corresponding part only once, and are not used as a means for predicting a recurrence or failure of the fault.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and a technical problem to be achieved by the present invention is a system and method for predicting a failure of an electric vehicle by analyzing data using a Arima model, which extracts a frequency of failure of the electric vehicle occurring within a specific reference date. To provide.

The technical problem to be achieved by the present invention is not limited to the technical problem mentioned above, and other technical problems not mentioned above may be clearly understood by those skilled in the art from the following description. There will be.

In order to solve the above technical problem, an embodiment of the present disclosure provides a system for predicting a failure of an electric vehicle, and generates an integrated fault code table for the electric vehicle based on previously collected fault information of the electric vehicle and generates the integrated fault code. A data generation unit for generating failure frequency data based on a failure frequency of the electric vehicle generated within a specific reference date by analyzing a table; and generating a time series graph based on the failure frequency data and converting the time series graph into a stable time series graph. A processing unit and data for analyzing the time series graph to derive an optimal parameter for applying to the Arima model, and substituting the Arima model completed through the optimum parameter into an Arima prediction function to obtain failure prediction result data for the train. Predictive Part Provides a train failure prediction system.

In the present embodiment, the collected fault information includes train fault code information, train driving record information, train maintenance information, and wheel correction information of the electric vehicle, and the integrated fault code table may include the train fault code information and a train. It may include fault data of at least one of the number of faults, fault dates, fault codes, fault spots and fault names of the train according to driving record information, train maintenance information and wheel correction information.

In the present embodiment, the data generation unit may analyze the integrated failure code table, and may generate failure frequency data by extracting a failure frequency of the electric vehicle generated within three days based on the number of failures generated per day.

In the present embodiment, the data processing unit may convert the time series graph into a log function and then convert the log function into a differential function to convert the time series graph into a stable time series graph.

In the present embodiment, the failure prediction result data generated by the data prediction unit is formed as a graph in which the x-axis is a date and the y-axis is a failure frequency, and the data prediction unit extracts data of a date having the failure frequency of 1 or more. can do.

In the present embodiment, the optimal parameter corresponds to ARIMA (p, d, q) for analyzing the time series graph and applying it to the Arima model, where d is the difference order, p is the autoregressive (AR) model order, q may be a moving average (MA) model order.

In the present embodiment, the Arima model may be a combination of a regression model based on a weighted sum of historical data and a moving average model based on a weighted sum of errors of past predictions.

In addition, another embodiment of the present application to solve the technical problem, in the electric vehicle failure prediction method using an electric vehicle failure prediction system comprising a data generation unit, a data processing unit and a data prediction unit, the data generation unit previously collected A data generation step of generating an integrated fault code table for the electric vehicle based on fault information of the electric vehicle, and generating fault frequency data based on a fault frequency of the electric vehicle generated within a specific reference date by analyzing the integrated fault code table; A data processing step in which a processing unit generates a time series graph based on the failure frequency data and converts the time series graph into a stable time series graph, and the data predictor analyzes the time series graph to derive an optimal parameter for applying to an Arima model. , The Arima model completed with the optimal parameter provides a failure prediction result train failure prediction comprising the prediction data comprising: obtaining a data for the train by applying the prediction function Arima.

In the present embodiment, the collected fault information includes train fault code information, train driving record information, train maintenance information, and wheel correction information of the electric vehicle, and the integrated fault code table may include the train fault code information and a train. It may include fault data of at least one of the number of faults, fault dates, fault codes, fault spots and fault names of the train according to driving record information, train maintenance information and wheel correction information.

In the present embodiment, the step of generating data includes analyzing the integrated failure code table, and extracting a failure frequency of the electric vehicle generated within three days based on the number of failures generated per day to generate failure frequency data. can do.

In the present embodiment, the data processing may include converting the time series graph into a logarithmic function and then converting the log function into a differential function to convert the time series graph into a stable time series graph.

In the present embodiment, the failure prediction result data generated through the data prediction step is a graph in which the x axis is a date and the y axis is a failure frequency, and the data prediction step is a data of a date having the failure frequency of 1 or more. It may include the process of extraction.

In the present embodiment, the optimal parameter corresponds to ARIMA (p, d, q) for analyzing the time series graph and applying it to the Arima model, where d is the difference order, p is the autoregressive (AR) model order, q may be a moving average (MA) model order.

In the present embodiment, the Arima model may be a combination of a regression model based on a weighted sum of historical data and a moving average model based on a weighted sum of errors of past predictions.

According to the present invention, it is possible to predict the failure of the electric vehicle by analyzing data using the Arima model, which extracts the frequency of failure of the electric vehicle occurring within a specific reference day.

In addition, according to the present invention, various electric vehicle-related fault data used only once can be used as a means for predicting whether a fault recurs or failure, and integrated fault-related observation data and prediction data by using big data technology. Can manage

The effects of the present invention are not limited to the above-described effects, but should be understood to include all the effects deduced from the configuration of the invention described in the detailed description or claims of the present invention.

1 is a block diagram showing a schematic configuration of an electric vehicle failure prediction system according to an embodiment of the present invention.
2 and 3 are diagrams for explaining in more detail the data generation process according to an embodiment of the present invention.
4 and 5 are diagrams for explaining a stable time series graph according to an embodiment of the present invention.
6 and 7 are diagrams for explaining the optimal parameter and Arima model according to an embodiment of the present invention.
8 is a diagram illustrating a prediction function according to an embodiment of the present invention.
9 is a flowchart illustrating a procedure of a method for predicting a failure of an electric vehicle according to another exemplary embodiment of the present invention.

Hereinafter, with reference to the accompanying drawings will be described in detail the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. In addition, the accompanying drawings are intended to facilitate understanding of the embodiments disclosed herein, but are not limited to the technical spirit disclosed herein by the accompanying drawings, all changes included in the spirit and scope of the present invention. It is to be understood to include water, equivalents and substitutes. And the part not related to the description in order to clearly describe the present invention in the drawings are omitted, the size, shape, shape of each component shown in the drawings may be variously modified, the same / similar parts for the entire specification Identical / similar reference numerals are used.

The suffixes "module" and "unit" for components used in the following description are given or used in consideration of ease of specification, and do not have distinct meanings or roles from each other. In addition, in describing the embodiments disclosed herein, when it is determined that the detailed description of the related known technology may obscure the gist of the embodiments disclosed herein, the detailed description is omitted.

Throughout the specification, when a part is said to be "connected (connected, contacted or coupled) with another part, it is not only when it is" directly connected (connected, contacted or coupled) ", but also in between. This includes cases in which "indirectly connected (connecting, contacting or coupling)" therebetween. Also, when a part is said to "include (or prepare)" a component, it is not to exclude other components, but to "include (or prepare)" other components, unless specifically stated otherwise. That means you can.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. As used herein, the terms "comprise" or "have" are intended to indicate that there is a feature, number, step, action, component, part, or combination thereof described on the specification, and one or more other features. It is to be understood that the present invention does not exclude the possibility of the presence or the addition of numbers, steps, operations, components, components, or a combination thereof.

1 is a block diagram showing a schematic configuration of an electric vehicle failure prediction system (hereinafter, "electric vehicle failure prediction system 10") according to an embodiment of the present invention.

Referring to FIG. 1, the electric vehicle failure prediction system 10 may include a data generator 11, a data processor 12, and a data predictor 13, and may further include a database 14.

The data generating unit 11 generates an integrated fault code table for the electric vehicle based on previously collected fault information of the electric vehicle, analyzes the integrated fault code table and analyzes the fault frequency based on the fault frequency of the electric vehicle generated within a specific reference date. Generate data.

In addition, the data generation unit 11 may analyze the integrated failure code table, and generate failure frequency data by extracting a failure frequency of the electric vehicle generated within three days based on the number of failures generated per day.

The collected fault information used by the data generator 11 may be at least one or more pieces of information, such as train fault code information, train driving record information, train maintenance information, and wheel cutting information.

In addition, the integrated failure code table generated by the data generation unit 11 is the number of failures of each information according to the train failure code information, train driving record information, train maintenance information and wheel correction information, failure date, failure code, failure site And at least one record of the fault name.

The data processor 12 generates a time series graph based on the failure frequency data, and converts the time series graph into a stable time series graph.

In addition, the data processing unit 12 may convert the time series graph into a log function and then convert the log function into a differential function to convert the time series graph into a stable time series graph.

The data predictor 13 analyzes the time series graph to derive an optimal parameter for applying to the Arima model, and obtains a failure prediction result data for the electric vehicle by substituting the Arima model completed through the optimal parameter into an Arima prediction function. do.

The Arima model described herein is an auto-regressive integrated moving average, which corresponds to an integration of an auto-regressive model and a moving average model, and corresponds to a data analysis method. .

The failure prediction result data generated by the data predictor 13 may be formed as a graph in which the x-axis is a date and the y-axis is a failure frequency, and the data predictor 13 is data of a date having the failure frequency of 1 or more. Can be extracted.

The database 14 may store data generated and processed through the data generator 11, the data processor 12, and the data predictor 13, and index and digitize the data to record the data later. You can perform the generation.

2 and 3 are diagrams for explaining in more detail the data generation process according to an embodiment of the present invention.

The data generator 11 described with reference to FIG. 1 generates an integrated fault code table based on data of an event generated in an electric vehicle such as train fault code information, train driving record information, train maintenance information, abnormality detection information, and wheel correction information. do. The integrated fault code table may be generated to include additionally generated fault information as well as previously collected fault information. For example, in the case of the train fault code information, the data generator 11 inquires the fault record of the corresponding electric vehicle in real time using a separate train data collection program, and integrates the fault code information corresponding to the fault record into the fault code table. You can register at

Specifically, referring to FIG. 2 showing an example of the above-described train driving record information, the data generator 11 determines a logic error of the train driving record information through a separate device such as a PC or a kiosk ( Rule), collect train operation record information of a specific train, and extract train operation record information of the corresponding part of the collected data from the collected data and reflect the fault code of the extracted information in the integrated fault code table can do.

For example, referring to FIG. 2, the data generator 11 sets rules and registers failure codes corresponding to various failure names, failure results, and the like in Rule information. In addition, the data generation unit 11 inputs one of the information lists in Val1 of the Rule detail part, inputs an information list and a number in Val2, inputs two Val formulas in the symbol, and connects the next rule with the next rule. You can register a relationship. Here, the information list means a data list of train driving record information.

In addition, referring to FIG. 3 illustrating an example of the above-described train maintenance information, the data generator 11 may generate a word (filter) for extracting a fault code from the train maintenance information through a separate device such as a PC or a kiosk. After describing and defining the connection division, the data including the corresponding rule can be searched among the text records described in the train maintenance data collected through the preset rule, and the data generator 11 corresponds to the corresponding failure. Codes can be registered in the integrated fault code table.

Similarly, the abnormality detection information, the data generation unit 11 collects the abnormality detection data of the equipment through a separate data collection program, and confirms whether or not the abnormality detection such as heat, vibration, etc. based on the reference data and the abnormal data Fault codes can be registered in the integrated fault code table.

In addition, the data generator 11 collects the wheel cutting data of the electric vehicle through a separate data collection program and checks whether the data is abnormal based on the reference data, and registers the fault code of the faulty data in the integrated fault code table. do.

4 and 5 are diagrams for explaining a stable time series graph according to an embodiment of the present invention.

As described above, the data processing unit 12 generates a time series graph based on the failure frequency data and converts the time series graph into a stable time series graph. At this time,

A stable time series graph refers to a time series graph in which the graph is invariant in time regardless of time, the variance is invariant, or the covariance between two time points on the graph is independent of the reference time point. Therefore, as shown in the graph shown in FIG. 4, when the failure frequency data generated by the data generator 11 includes a date having a failure frequency of 0, the data processing unit 12 is formed based on the failure frequency data. Time series graphs do not correspond to stable time series graphs.

Accordingly, the data processing unit 12 may convert the time series graph into a log function and then convert the log function into a differential function to convert the time series graph into a stable time series graph. At this time, an algorithm as shown in FIG. 5 may be used.

Specifically, referring to FIG. 5, the normal time series has the following characteristics.

First, in the AR model, the ACF has a high initial correlation coefficient and then gradually decreases. Since the autocorrelation coefficient, the past value decreases the correlation. In the moving average model, the ACF decreases completely after time k because the moving average does not take into account after time k. In the autoregressive (AR) model, PACF decreases completely after k. In the MA model, PACF has a high initial correlation coefficient and then gradually decreases.

Through this feature, it is possible to determine the p, d, q of the ARIMA coefficient. In this embodiment, the optimal coefficient can be automatically derived through the existing auto.arima function, where d is the difference order. , p is the autoregressive (AR) model order, q is the moving average (MA) model order.

6 and 7 are diagrams for explaining the optimal parameter and Arima model according to an embodiment of the present invention.

The optimal parameter is extracted through a preset formula for applying the ARIMA model. The preset equation may be formed by an algorithm as shown in FIG. 7. Similarly, in ARIMA (p, d, q), d is the difference order, p is the autoregressive (AR) model order, q is the moving average (MA) model order.

The Arima model is created by applying the optimal parameters derived using the algorithm shown in FIG. 7 to the time series graph described above. As described above, the autoregressive integrated moving average (ARIMA) method combines the weighted sum of past data (AR) and the weighted sum of errors of past prediction (MA, moving average). It means how the model is derived.

8 is a diagram illustrating a prediction function according to an embodiment of the present invention.

Referring to FIG. 8, the data of three days or more may be predicted using a generated Arima model through a preset prediction function, and a value having a frequency of 1 or more may be determined as a possible failure.

In other words, apply an ARIMA model with optimal parameters to an existing prediction forecast.Arima function to obtain a prediction value, find a value with a frequency of 1 or more, and store the data and images in a database to display the predicted failure code. Can be.

9 is a block diagram illustrating an embodiment of an electric vehicle failure prediction system according to an embodiment of the present invention.

9 is a flowchart illustrating a procedure of a method for predicting a failure of an electric vehicle according to another exemplary embodiment of the present invention. The method for predicting a failure of an electric vehicle according to the present exemplary embodiment of the present invention is described with reference to FIGS. 1 to 8. It is a method of predicting failure of a train using Therefore, hereinafter, description overlapping with the above description will be omitted.

Referring to FIG. 9, a method for predicting a failure of an electric vehicle according to the present embodiment may include a fault collected by the data generator in a method of predicting a failure of a vehicle using an electric vehicle fault prediction system including a data generator, a data processor, and a data predictor. A data generation step (S91) of generating an integrated fault code table based on the information, analyzing the integrated fault code table, extracting a frequency of each fault code generated within a specific reference date and generating fault frequency data, and the data processing unit A data processing step of generating the failure frequency data as a time series graph and converting the time series graph into a stable time series graph, and the data predictor analyzing the time series graph to derive an optimal parameter for application of an Arima model, Arima created with optimal parameters By substituting the prediction type to the function it includes a data prediction step (S93) for obtaining a prediction result of failure data.

Here, the collected fault information may include train fault code information, train driving record information, train maintenance information, and wheel cutting information. The integrated fault code table may include at least one or more of the number of faults, fault dates, fault codes, fault spots, and fault names of each information according to the train fault code information, train driving record information, train maintenance information, and wheel correction information. May include a record.

The data generation step S91 may include analyzing the integrated failure code table, and generating failure frequency data by extracting a frequency of the failure code that occurs within 3 days on the basis of the daily failure code.

The data processing step S92 may include converting the time series graph into a log function and then converting the log function into a differential function to form the time series graph as a stable time series graph.

In addition, the failure prediction result data generated through the data prediction step S93 may be a graph in which the x-axis is set as the date and the y-axis is set as the failure frequency. At this time, the data prediction step (S93) may include a process of extracting the data of the date of the failure frequency of 1 or more.

Also in the method for predicting a failure of an electric vehicle according to the present embodiment, the optimum parameter corresponds to ARIMA (p, d, q) for analyzing the time series graph and applying the Arima model, where d is a difference order and p is a self-regression. (AR) model order, q may be a moving average (MA) model order, wherein the Arima model is a regression model based on the weighted sum of historical data and a moving average model based on the weighted sum of the errors of past predictions. It can be a combined model.

The electric vehicle failure prediction system 10 and the electric vehicle failure prediction method may be implemented through a mobile PC and a mobile application, respectively. For example, the electric vehicle failure prediction system 10 may be implemented in the form of a smart phone, and the functions of the data generator 11, the data processor 12, the data predictor 13, and the database 14 are smart. It can be implemented through storage devices such as wireless communication technology, application processor, liquid crystal display, and RAM installed in the phone, and each procedure can be performed through the application for smartphone that is built in or downloaded. Can be.

The description of the present invention described above is for illustrative purposes, and it will be understood by those skilled in the art that the present invention may be easily modified in other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive. The scope of the present invention is represented by the following claims, and it should be construed that all changes or modifications derived from the meaning and scope of the claims and their equivalents are included in the scope of the present invention.

Claims (14)

In the system which predicts the failure of an electric vehicle,
Generating an integrated fault code table for the electric vehicle based on previously collected fault information of the electric vehicle, generating data for generating fault frequency data based on the fault frequency of the electric vehicle generated within a specific reference date by analyzing the integrated fault code table. part;
A data processing unit generating a time series graph based on the failure frequency data and converting the time series graph into a stable time series graph;
And analyzing the time series graph to derive an optimal parameter for applying to the Arima model, and substituting the Arima model completed through the optimal parameter into an Arima prediction function to obtain failure prediction result data for the train. Electric vehicle failure prediction system characterized in that.
The method of claim 1,
The collected fault information includes train fault code information, train driving record information, train maintenance information, and wheel correction information of the electric vehicle,
The integrated fault code table may include at least one of the number of faults of the electric vehicle according to the train fault code information, train driving record information, train maintenance information, and wheel cutting information, a fault date, a fault code, a fault region, and a fault name. An electric vehicle failure prediction system, comprising fault data of an electric vehicle.
The method of claim 1,
The data generator analyzes the integrated fault code table, and extracts a fault frequency of the train generated within 3 days based on the number of faults generated per day to generate fault frequency data.
The method of claim 1,
And the data processing unit converts the time series graph into a log function and then converts the log function into a differential function to convert the time series graph into a stable time series graph.
The method of claim 1,
The failure prediction result data generated by the data prediction unit is formed as a graph in which the x axis is a date and the y axis is a failure frequency, and the data prediction unit extracts data of a date having the failure frequency of 1 or more. Fault prediction system.
The method of claim 1,
The optimal parameter corresponds to ARIMA (p, d, q) for analyzing the time series graph and applying it to the Arima model,
Wherein d is a difference order, p is an autoregressive (AR) model order, q is a moving average (MA) model order.
The method of claim 1,
The Arima model is a model for combining a regression model based on a weighted sum of historical data and a moving average model based on a weighted sum of errors of past predictions.
In a train failure prediction method using a train failure prediction system comprising a data generation unit, a data processing unit and a data prediction unit,
The data generator generates an integrated fault code table for the electric vehicle based on the fault information of the electric vehicle collected previously, analyzes the integrated fault code table, and analyzes the fault frequency data based on the fault frequency of the electric vehicle generated within a specific reference date. Generating data generation step;
A data processing step of the data processing unit generating a time series graph based on the failure frequency data and converting the time series graph into a stable time series graph; And
The data prediction unit analyzes the time series graph to derive an optimal parameter for applying to the Arima model, and substitutes the Arima model completed through the optimum parameter into an Arima prediction function to obtain failure prediction result data for the train. Electric vehicle failure prediction method comprising the step.
The method of claim 8,
The collected fault information includes train fault code information, train driving record information, train maintenance information, and wheel correction information of the electric vehicle,
The integrated fault code table may include at least one of the number of faults of the electric vehicle according to the train fault code information, train driving record information, train maintenance information, and wheel cutting information, a fault date, a fault code, a fault region, and a fault name. An electric vehicle fault prediction method comprising fault data of an electric car.
The method of claim 8,
The data generation step may include analyzing the integrated fault code table, and generating fault frequency data by extracting a fault frequency of the train occurring within 3 days based on the number of breakdowns generated per day. Failure prediction method.
The method of claim 8,
The data processing step includes converting the time series graph into a log function and then converting the log function into a differential function to convert the time series graph into a stable time series graph.
The method of claim 8,
The failure prediction result data generated through the data prediction step is a graph in which the x-axis is set to the date and the y-axis is set to the failure frequency.
The data predicting step includes the step of extracting the data of the date of the failure frequency of 1 or more.
The method of claim 8,
The optimal parameter corresponds to ARIMA (p, d, q) for analyzing the time series graph and applying it to the Arima model,
Wherein d is a difference order, p is an autoregressive (AR) model order, and q is a moving average (MA) model order.
The method of claim 8,
The Arima model is a model for combining a regression model based on a weighted sum of historical data and a moving average model based on a weighted sum of errors of past predictions.

Images