KR101344755B1 - Apparatus and method for measuring velocity by using kalman filter - Google Patents

Abstract

Disclosed are a speed measuring apparatus and method using a Kalman filter. The first adaptive Kalman filter unit may reduce the influence of the first sensors on the basis of the sensor values input from the plurality of first sensors of the same type and the sensor values of the main sensor selected from the first sensors. Calculate the sensor value. The second adaptive Kalman filter part may be caused by a measurement error of the second sensors based on sensor values input from a plurality of second sensors of the same type as the first sensor and sensor values of the main sensor selected from the second sensors. The second sensor value with reduced effect is calculated. The first synthesis unit calculates an error value by subtracting the first sensor value from the second sensor value. The Kalman filter unit calculates an error estimate value from the first sensor value, the second sensor value, and the error value. The second synthesizer calculates the speed by subtracting the error estimate value from the first sensor value. According to the present invention, it is possible to measure the speed of a moving object with a high degree of reliability without any deterioration in performance even with a failure of a single sensor.

Description

Apparatus and method for measuring velocity by using Kalman filter}

The present invention relates to a speed measuring device and method using a Kalman filter, and more particularly, to an apparatus and method for measuring the speed of a traveling body traveling at high speed by combining data input from a plurality of sensors by the Kalman filter. It is about.

Train systems that run on rails, including high-speed railways, have dynamic characteristics that require a braking distance of at least 1 km when traveling at 100 km / h. Therefore, the train control system, which is introduced to prevent collisions and collisions of trains, aims to control the course of the trains and control the distances to ensure a safe braking distance between vehicles. Detection must be premised.

In general, the train control system consists of a ground control system that controls the operation of the train on the ground and a vehicle control system that supports the safe operation of the engineer. The ground control system is interlocking to control the course to ensure the safe operation of all trains operating on the track, automatic train protection (ATP) device to prevent collision with the preceding train, and information between ground vehicles. It consists of line side communication equipment for transmitting and receiving. In addition, the vehicle control system is composed of a driving speed detection device, a vehicle control device, and a communication device.

It is common to use a fault-tolerant multi-sensor rather than a single sensor in systems requiring high safety, such as railway and aviation facilities, to ensure the safety of driving and reliability of speed detection. In case of domestic railways, the driver is manually operating on the national railway section with a driving speed of 150 km / h or less based on the on-site signal information, and the instrument panel speed information is displayed by using a tachometer sensor.

The tachometer sensor, which is widely used to realize the speed detection function of a vehicle, detects speed and position by counting the number of revolutions of the axle or wheel through sensors such as a magnetic sensor and a light quantity sensor. Such a tachometer sensor has the advantage of detecting the distance and speed of the vehicle continuously, but if the slip or slip of the wheel of the train occurs during acceleration or braking of the train, an error occurs in the speed position information. There is a problem that is difficult to be applied to the automatic unmanned operation system that requires high precision control alone.

On the other hand, the BALI's wireless communication system installed on the track for transmitting and receiving information between on-board vehicles transmits signal control information of the ground equipment for train control and unique location information on which the device is installed to the onboard controller. Therefore, the on-vehicle train control system receives the control information of the ground facilities and the location information of the current train through the Bally antenna on the vehicle body and utilizes it as information for train control. This method has the disadvantage of not being able to detect train position continuously, but it can be used for absolute position correction when mixed with other sensors such as tachometer.

The Doppler sensor is a sensor that measures the speed by using a Doppler effect in which the frequency of the reflected wave increases on the coming object and the frequency of the reflected wave decreases on the moving object. When the Doppler sensor is mounted on the train and a certain frequency is shot on the floor, the reflected wave returns and the speed is measured by measuring the frequency. However, the Doppler sensor has a disadvantage that the error can be exceeded within the allowable range when the sensor is measured in a non-contact manner, so when used alone, the safety of the train control cannot be guaranteed.

Currently, the urban railway train control system or KTX high-speed railway signal system that supports unmanned automatic train operation combines the data input from the aforementioned Doppler sensors, tachometers, and ballistics devices to provide high speed and location information. Employing a device to obtain. 1 shows the configuration of an apparatus for detecting the speed and position of a train. However, the core equipment for controlling the onboard train including the speed position detection device is not developed as a localized product at present, but relies on imports. Therefore, it is urgent to localize the core equipment for on-board train control such as speed detection device, and the speed detection device developed at this time should have high reliability without degrading performance even in the failure of a single sensor when compared with existing facilities.

Korean Patent No. 10-0387074 (Invention name: Vehicle speed measuring device and method and communication method between vehicle speed measuring device and traffic system, Publication date: 2002.02.09) Korean Patent Publication No. 10-2011-0105010 (name of the invention: a method for adjusting a self-mixing laser sensor system for measuring the speed of a vehicle, publication date: September 23, 2011) Korea Patent Registration No. 10-1035533 (Invention name: mobile unit positioning device, published date: 2011.03.29)

SUMMARY OF THE INVENTION The present invention has been made in an effort to provide a speed measuring apparatus and method using a Kalman filter having high reliability without degrading performance even in the failure of a single sensor.

SUMMARY OF THE INVENTION The present invention has been made in an effort to provide a computer-readable recording medium having recorded thereon a program for executing a speed measuring method using a Kalman filter with high reliability without a performance penalty even in the case of a single sensor failure. .

In order to achieve the above technical problem, a speed measuring device using a Kalman filter according to the present invention, the sensor value input from the plurality of first sensors of the same type and the sensor value of the main sensor selected from the first sensors A first adaptive Kalman filter unit for calculating a first sensor value on which the influence of the first sensors is reduced by a measurement error; The influence of the measurement error of the second sensors is reduced based on a sensor value input from a plurality of second sensors of the same type different from the first sensor and a sensor value of a main sensor selected from the second sensors. A second adaptive Kalman filter unit for calculating a second sensor value; A first synthesizer configured to calculate an error value by subtracting the first sensor value from the second sensor value; A Kalman filter unit calculating an error estimate value from the first sensor value, the second sensor value, and the error value; And a second synthesizer configured to calculate a speed by subtracting the error estimate value from the first sensor value.

According to another aspect of the present invention, there is provided a speed measuring method using a Kalman filter according to the present invention, (a) a sensor value input from a plurality of first sensors of the same type and a main sensor selected from the first sensors. Calculating a first sensor value on the basis of a sensor value of the first sensor value, the effect of which is reduced by a measurement error of the first sensors; (b) Influence of measurement errors of the second sensors based on sensor values input from a plurality of second sensors of the same type as the first sensor and sensor values of a main sensor selected from the second sensors. Calculating a second sensor value having a reduced value; (c) calculating an error value by subtracting the first sensor value from the second sensor value; (d) calculating an error estimate value from the first sensor value, the second sensor value, and the error value; And (e) calculating the speed by subtracting the error estimate value from the first sensor value.

According to the speed measuring apparatus and method using the Kalman filter according to the present invention, it is possible to measure the speed of the moving object with a high degree of reliability without deterioration even in the failure of a single sensor. In addition, there is an advantage in that it is robust to sensor defects by multiplexing the same type of sensors while using multiple types of measurement sensors in preparation for sensor defects. Furthermore, when integrating the estimates of the local Kalman filter, it is possible to cope with the defective sensor by assigning an evaluation weight in consideration of the possibility of the defect of the sensor, and using the high contribution among the sensors as the main sensor without fixing the main sensor. By replacing and using it, more reliable speed measurement is possible.

1 is a diagram showing the configuration of an apparatus for detecting the speed and position of a train;
2 is a view showing the configuration of a preferred embodiment of the speed measuring device using a Kalman filter according to the present invention,
3 is a view showing the configuration of a preferred embodiment of an adaptive Kalman filter applied to a speed measuring device according to the present invention;
4 is a diagram illustrating an example of a Kalman filter algorithm used in the present invention;
5 is a view showing a detailed configuration of the sensor value calculation unit 340,
6 is a view showing a process of performing a preferred embodiment of the method for measuring the speed using a Kalman filter according to the present invention, and
7 is a diagram illustrating a process of calculating a sensor value by an adaptive Kalman filter applied to a speed measuring method using a Kalman filter according to the present invention.

Hereinafter, exemplary embodiments of a speed measuring apparatus and method using a Kalman filter according to the present invention will be described in detail with reference to the accompanying drawings.

2 is a view showing the configuration of a preferred embodiment of the speed measuring device using a Kalman filter according to the present invention.

2, the speed measurement apparatus using the Kalman filter according to the present invention, the first adaptive Kalman filter unit 210, the second adaptive Kalman filter unit 220, the first synthesis unit 230, Kalman The filter unit 240, the second synthesis unit 250, the first sensor unit 260, and the second sensor unit 270 are provided.

The first adaptive Kalman filter unit 210 inputs the sensor values measured from the plurality of homogeneous sensors provided in the first sensor unit 260 to minimize the influence of the sensor values input from the failed sensor. Output the first sensor value. In addition, the second adaptive Kalman filter unit 210 inputs sensor values measured from a plurality of homogeneous sensors provided in the second sensor unit 270 to minimize the influence of the sensor values input from the sensor in which the failure occurs. Outputs the second sensor value. In the present embodiment illustrated in FIG. 2, the Doppler sensor and the tachometer are provided in the first sensor unit 260 and the second sensor unit 270, respectively. 3 shows a preferred embodiment of an adaptive Kalman filter provided in the first adaptive Kalman filter unit 210 and the second adaptive Kalman filter unit 220.

Referring to FIG. 3, the adaptive Kalman filter provided in the speed measuring device according to the present invention includes a plurality of third synthesis units 320-1 to 320 -N and a plurality of local Kalman filters 330-1 to 330 -N. ), A sensor value calculator 340, a main sensor selector 350, and a main sensor value provider 360.

Each of the plurality of third synthesizing units 320-1 to 320 -N has a first sensor 310-1 to an N-th sensor at a main sensor value input from a sensor selected as the main sensor from the main sensor value providing unit 360. The sensor value is subtracted from 310-N and output. The result value calculated by each of the third synthesis units 320-1 to 320 -N is the first local Kalman filter 330-1 corresponding to each of the third synthesis units 320-1 to 320 -N. To the N-th region Kalman filter 330-N. The first region Kalman filters 330-1 to 330 -N are each of the first sensors 310-1 to 310-based on a result value input from the corresponding third synthesis unit 320-1 to 320 -N. The error estimation value of the sensor value measured from N) is output. 4 shows the (regional) Kalman filter algorithm used in the present invention.

The sensor value calculator 340 is configured based on an error estimation value for the sensor values of the first sensors 310-1 to 310-N input from the respective Kalman filters 330-1 to 330-N. The final sensor value is minimized with the effect of the sensor. 5 illustrates a detailed configuration of the sensor value calculator 340. Referring to FIG. 5, the sensor value calculator 340 includes a plurality of fourth synthesis units 410-1 to 410 -N, an error calculator 420, and a sensor value determiner 430. The fourth synthesis unit 410-1 to 410 -N receives an error input from each of the first region Kalman filters 330-1 to 330 -N corresponding to the main sensor value input from the main sensor value providing unit 360. The estimated value is subtracted to output estimated sensor values of the respective sensors 310-1 to 310 -N. The error calculator 420 calculates an error corresponding to each sensor 310-1 to 310 -N defined as the sum of deviations of the estimated sensor values of each sensor based on each estimated sensor value. Calculate.

는 제k센서의 추정센서값, 그리고, S_i는 제i센서의 추정센서값이다.Here, SD _k is the error of the k-th sensor defined as the sum of the difference between the estimated sensor value of the k-th sensor and the estimated sensor value of the other sensors,

Is an estimated sensor value of the k-th sensor, and S _i is an estimated sensor value of the i-th sensor.

The sensor value determining unit 430 calculates the weighted average of the sensor estimation values and the errors of the respective sensors 310-1 to 310 -N input from the error calculating unit 420 according to the following equation and adaptive Kalman. Output the final sensor value of the filter.

는 제i센서의 추정센서값이다.Where S _out is the final sensor value of the adaptive Kalman filter, SD _i is the error of the i-sensor, and

Is an estimated sensor value of the i-th sensor.

3, the main sensor selecting unit 350 supplies a sensor having a minimum difference between an error of each sensor 310-1 to 310-N input from the sensor value calculating unit 340 and a final sensor value. Select it as a sensor. The main sensor value providing unit 360 converts the sensor value input from the sensor selected by the main sensor selecting unit 350 among the sensor values input from the respective sensors 310-1 to 310 -N as the main sensor value. The third synthesis units 320-1 to 320 -N and the fourth synthesis units 410-1 to 410 -N are provided.

The first synthesis unit 230 subtracts the first filter output value input from the first adaptive Kalman filter from the second filter output value input from the second adaptive Kalman filter and outputs a difference value. The output of the first synthesis unit 230 is provided to the Kalman filter unit 240. The Kalman filter unit 240 inputs a difference value input from the first synthesis unit 230, a first filter output value input from the first adaptive Kalman filter unit 210, and a second adaptive Kalman filter unit 220. An estimation error is calculated based on the second filter output value. The second synthesis unit 250 subtracts the estimated error calculated by the Kalman filter unit 240 from the first filter output value input from the first adaptive Kalman filter unit 210 to output the speed of the moving object.

As described above, the speed measuring device according to the present invention may be applied to a speed detection unit (SDU) of a speed measuring device of a train requiring high reliability, and at the same time using a plurality of measuring sensors to prepare for defects of a sensor Multiplexing the sensor has the advantage of being robust against sensor defects.

In order to estimate the state of the system, the installation of a sensor to ensure an adequate level of observation must be prioritized, and the survey information provided by the installed sensor provides an optimal state estimate through a filter with the dynamic model and the observation model of the system. Done. In this process, in order to obtain a high confidence level estimate, the information provided by the applied sensor must also have high reliability.

However, the use of highly reliable advanced sensors is directly related to cost and skill level, so it is often impossible to use more than a certain level of sensors. In particular, in systems where precision and reliability are very important for safety reasons, such as train speed measurement devices, the information provided by one sensor is often not sufficient, so that several or several types of sensors can be used simultaneously. In this case, a filter for integrating the information is required in order to integrate the information of the sensors to obtain an optimal state estimate.

The Kalman filter is an algorithm that calculates recursive, linear optimal, and real-time data that estimates the dynamic state of a system in a noisy environment by appropriately using the statistical properties of the system model and observed model measurements to estimate the state of the system. It is widely used for estimating the state of a system through fusion. Kalman filters for integrating sensor information can be configured in many forms. The most common form is centralized Kalman Filtering (CKF), which processes all sensor measurements into a single filter. This method is the simplest to implement, but it is difficult to cope when there is a problem with the measurement, and when there are many measurements, the covariance matrix of the measurement becomes large and there is a problem in the calculation process. The decentralized Kalman Filter (DKF) is a form designed to compensate for the shortcomings of the concentrated Kalman filter.

The distributed structure is a form of integrating information provided from a local filter in a master filter after performing an estimation process in a local filter for mutually independent sensors. This structure is robust to defects because it can handle defects in individual sensors at the local filter level and provide continuous solutions even in the event of a fault. Decentralized filters are more robust to defects than centralized ones, but the master filter integrates information using only the least-squares method, so that nonlinearities between state variables are more accurate than centralized ones. In some cases, they may fall off or shed in the worst case.

In order to overcome the nonlinearity of the system model or the observation model, various methods such as the Extended Kalman Filter and the Unscented Kalman Filter have been proposed, but most of them focus only on the centralized filter, but the structure of the filter suitable for the distributed filter is not suggested. not. However, the adaptive Kalman filter according to the present invention as described with reference to FIGS. 3 and 4 can overcome the disadvantages of the simple distributed Kalman filter in order to estimate information with fault tolerance from multiple sensors of the same type. Filter. In other words, the low contribution of the individual local filter means that the error of the estimated value of the sensor is relatively high, so that the probability of the defect is increased, so that when the sensor value calculator integrates the estimate of the local Kalman filter, By considering the possibility and assigning an evaluation weight, it is possible to cope with a defective sensor. In particular, the adaptive Kalman filter according to the present invention applies a competitive selection strategy of the main sensor so that the high contribution among the sensors can be replaced with the main sensor without fixing the main sensor.

6 is a view showing the implementation of a preferred embodiment for the method of measuring the speed using a Kalman filter according to the present invention.

Referring to FIG. 6, the first adaptive Kalman filter unit 210 measures an error based on sensor values input from a plurality of Doppler sensors and a main sensor value input from a sensor selected as a main sensor among the plurality of Doppler sensors. The first sensor value is minimized and the influence of the large sensor value is output (S600). In addition, the second adaptive Kalman filter unit 220 influences the sensor value having a large measurement error based on the sensor values input from the plurality of tachometers and the main sensor value input from the tachometer selected as the main sensor among the plurality of tachometers. The reduced second sensor value is output (S610). Next, the first synthesis unit 230 outputs a difference value between the second sensor value and the first sensor value (S620). Next, the Kalman filter unit 240 receives the first sensor value, the second sensor value, and a difference value between the second sensor value and the first sensor value and outputs an error estimation value (S630). Finally, the second synthesizer 250 subtracts the error estimate value from the first sensor value and outputs a final speed (S640).

7 is a diagram illustrating a process of calculating a sensor value by an adaptive Kalman filter applied to a speed measuring method using a Kalman filter according to the present invention. Since the first adaptive Kalman filter and the second adaptive Kalman filter differ only in the type of sensor to be connected, the sensor value calculation process is the same. Hereinafter, the adaptive Kalman filter described with reference to FIGS. 3 and 4 will be described as an example. do.

Referring to FIG. 7, first, an initial main sensor (eg, the first sensor 310-1) is selected from the N sensors 310-1 to 310 -N (S700). Next, the third synthesis units 320-1 to 320 -N corresponding to each of the N sensors 310-1 to 310 -N are inputted from the first sensor 310-1 selected as the main sensor. A difference value is output by subtracting the sensor value output from the first to Nth sensors 310-1 to 310 -N from the sensor value, respectively (S710). Next, the local Kalman filters 330-1 to 330-N corresponding to the N sensors 310-1 to 310 -N respectively apply the Kalman filter algorithm to each input difference value to each sensor 310-1. 310-N) to calculate an error estimation value (S720).

Next, the second synthesis units 410-1 to 410-N corresponding to each of the N sensors 310-1 to 310-N subtract the error estimation value from the main sensor value, respectively. The estimated sensor value of N) is output (S730). The error calculator 420 calculates an error corresponding to each sensor 310-1 to 310 -N defined as the sum of deviations of the estimated sensor values of each sensor based on each estimated sensor value. (S740). Next, the sensor value determiner 430 adaptively calculates the weighted average of the sensor estimates and the errors of the respective sensors 310-1 to 310-N input from the error calculator 420 by Equation 2 The final sensor value of the Kalman filter is output (S750). On the other hand, the main sensor selecting unit 350 selects the sensor having the smallest difference between the difference between the final sensor value and the estimated sensor value of each sensor (S760). The main sensor value providing unit 360 receives a sensor value input from a sensor selected by the main sensor selecting unit 350 among the N sensor values input from the N sensors 310-1 to 310 -N. The value is provided to the third synthesizing units 320-1 to 320 -N and the fourth synthesizing units 410-1 to 410 -N when the new sensor value is determined (S770).

The present invention can also be embodied as computer-readable codes on a computer-readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer apparatus is stored. Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like in the form of a carrier wave . In addition, the computer-readable recording medium may be distributed to computer devices connected to a wired / wireless communication network, and a computer-readable code may be stored and executed in a distributed manner.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, It will be understood by those skilled in the art that various changes may be made and equivalents may be substituted without departing from the scope of the appended claims.

Claims (11)

A first sensor value for reducing the influence due to a measurement error of the first sensors is calculated based on a sensor value input from a plurality of first sensors of the same type and a sensor value of a main sensor selected from the first sensors. A first adaptive Kalman filter unit;
The influence of the measurement error of the second sensors is reduced based on a sensor value input from a plurality of second sensors of the same type different from the first sensor and a sensor value of a main sensor selected from the second sensors. A second adaptive Kalman filter unit for calculating a second sensor value;
A first synthesizer configured to calculate an error value by subtracting the first sensor value from the second sensor value;
A Kalman filter unit calculating an error estimate value from the first sensor value, the second sensor value, and the error value; And
And a second synthesizer configured to calculate a speed by subtracting the error estimate value from the first sensor value. The method of claim 1,
The first adaptive Kalman filter unit and the second adaptive Kalman filter unit,
A third synthesis unit configured to subtract sensor values inputted from the respective sensors from the main sensor values inputted from a selected sensor among a plurality of sensors and output a difference value corresponding to each of the sensors;
A local Kalman filter unit for calculating an error estimation value of each of the sensors based on the respective difference values; And
The estimated sensor value of each sensor is calculated by subtracting the respective error estimation value from the main sensor value, and corresponds to each of the sensors defined as the sum of deviations of the estimated sensor values of each sensor based on the estimated sensor values. And a sensor value calculator configured to calculate an error of the sensor and calculate a weighted average of the sensor estimates of each sensor and the error of each sensor, and output a final sensor value. 3. The method of claim 2,
The sensor value calculation unit,
A fourth synthesizer configured to calculate an estimated sensor value of each sensor by subtracting the error estimation value from the main sensor value;
An error calculator configured to calculate an error corresponding to each of the sensors defined as the sum of deviations of the estimated sensor values of the respective sensors based on the estimated sensor values; And
And a sensor value determiner configured to calculate a weighted average of the sensor estimates of each of the sensors and the error of each of the sensors, and output a final sensor value. The method of claim 3, wherein
The error calculating unit calculates an error corresponding to each of the sensors by the following equation (A):
[Mathematical formula A]

Here, SD _k is the error of the k-th sensor defined as the sum of the difference between the estimated sensor value of the k-th sensor and the estimated sensor value of the other sensors,

Is an estimated sensor value of the k-th sensor, and S _i is an estimated sensor value of the i-th sensor. The method of claim 3, wherein
The sensor value determining unit calculates the final sensor value by the following equation (B):
[Mathematical expression B]

Where S _out is the final sensor value of the adaptive Kalman filter, SD _i is the error of the i-sensor, and

Is an estimated sensor value of the i-th sensor. 3. The method of claim 2,
A main sensor selecting unit selecting a sensor having a minimum difference value between the final sensor value and the estimated sensor value as a main sensor; And
And a main sensor value providing unit which changes a sensor value input from a sensor selected as a main sensor among the sensors into a main sensor value and provides the sensor to the third synthesis unit. (a) a first sensor value that reduces the influence of a measurement error of the first sensors based on a sensor value input from a plurality of first sensors of the same type and a sensor value of a main sensor selected from the first sensors; Calculating;
(b) Influence of measurement errors of the second sensors based on sensor values input from a plurality of second sensors of the same type as the first sensor and sensor values of a main sensor selected from the second sensors. Calculating a second sensor value having a reduced value;
(c) calculating an error value by subtracting the first sensor value from the second sensor value;
(d) calculating an error estimate value from the first sensor value, the second sensor value, and the error value; And
(e) calculating the speed by subtracting the error estimate value from the first sensor value. 8. The method of claim 7,
Step (a) and step (b),
(a1) subtracting sensor values input from each sensor from a main sensor value input from a selected sensor among a plurality of sensors and outputting a difference value corresponding to each of the sensors;
(a2) calculating an error estimation value of each sensor based on each difference value; And
(a3) subtracting the respective error estimation value from the main sensor value to calculate an estimated sensor value of each sensor, and the sensor defined as the sum of deviations of the estimated sensor values of each sensor based on the respective estimated sensor value Calculating an error corresponding to each of the sensors, calculating a weighted average of the sensor estimation values of each sensor and the error of each sensor, and outputting a final sensor value. The method of claim 8,
The step (e)
(e1) calculating the estimated sensor value of each sensor by subtracting the respective error estimation value from the main sensor value;
(e2) calculating an error corresponding to each of the sensors defined as the sum of deviations of the estimated sensor values of each of the sensors based on the respective estimated sensor values; And
(e3) calculating a weighted average of the sensor estimates of each of the sensors and the error of each of the sensors, and outputting a final sensor value. The method of claim 8,
(f) selecting a sensor having a minimum difference between the final sensor value and the estimated sensor value as a main sensor; And
(g) changing a sensor value input from a sensor selected as a main sensor among the sensors to a main sensor value. A computer-readable recording medium having recorded thereon a program for executing the speed measuring method according to any one of claims 7 to 10 on a computer.

Images