KR101925988B1 - Method For Calculating Energy Consumption Of Car By Utilizing Deep Learning For Implementing The Reduction Of Carbon Discharge - Google Patents

Abstract

Disclosed is a method for calculating an energy consumption amount of a vehicle, wherein the method comprises the steps of: collecting data by measuring a speed of an experimental vehicle, an inclination angle of a road on which the experimental vehicle runs, and an actual energy consumption amount of the experimental vehicle at every predetermined time interval by a data collecting device installed in the experimental vehicle which is actually driven; allowing an artificial neural network server to determine values of parameters applied to an artificial neural network when an output value calculated by a learning process using the speed of the experimental vehicle and the inclination angle of the road on which the experimental vehicle runs is compared with an actual energy consumption amount of the experimental vehicle, and when the comparison is within a predetermined error range; allowing an artificial neural network energy consumption amount calculation device installed in a driving vehicle to measure the speed of the driving vehicle and the inclination angle of the road on which the driving vehicle runs, at every predetermined time interval; and calculating the energy consumption amount of the driving vehicle by using the same as an input of the artificial neural network and by using the determined values of the parameters received from the artificial neural network server. The method for calculating an energy consumption amount of a vehicle according to the present invention can more accurately and more flexibly calculate an energy consumption amount of the vehicle by using the artificial neural network, in particular, deep learning.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a method for calculating energy consumption of a vehicle using deep running to reduce carbon emission,

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for calculating energy consumption of a vehicle, and more particularly, to a method for calculating energy consumption of a vehicle using deep learning, which is a kind of machine learning by artificial neural network for performing carbon emission reduction.

In recent years, interest in eco-driving has been increasing in order to improve the fuel efficiency of vehicles and reduce environmental pollution. In order for the driver to be able to drive eco-driving, it is necessary to measure fuel consumption in real time and display it to the driver. However, in order to accurately measure and display the instantaneous fuel consumption rate, the fuel injection quantity measuring device must be installed separately in the vehicle. It is undesirable to mount such a device on a finished vehicle that has been shipped without mounting such a device, since this would unnecessarily increase the price of the automobile and is not for a specific purpose. Even when such a device is mounted in the assembling step of the finished vehicle, it is not preferable because it is required to pay a higher cost than the added function.

Korea Registered Patent No. 10-1205983 (Registered on November 22, 2012) Registered Eco Drive Indicator for Automobile using Smart Phone and Korea Registered Patent No. 10-1308264 (Registered on Mar. 09, 2013) considers energy consumption rate And a Korean Patent Registration No. 10-1249421 (registered on Mar. 26, 2013) discloses an eco-drive induction device by real-time input of only a GPS data.

In the above inventions, the fuel consumption amount of the vehicle is expressed by the product of the vehicle output and the conversion coefficient, the vehicle output is expressed by the product of the vehicle driving force and the vehicle speed, and the vehicle driving force is represented by the rolling resistance, air resistance, Can be expressed as a sum. That is, the fuel consumption amount of the vehicle can be calculated from the vehicle driving force. In order to calculate the vehicle driving force, it is necessary to know not only the speed of the vehicle and the inclination of the road but also the weight of the vehicle, the rolling resistance coefficient, and the air resistance coefficient. However, since the accurate values of the rolling resistance coefficient and the air resistance coefficient of the vehicle are generally unknown to the vehicle manufacturer and the tire manufacturer, there is a limitation in that the fuel consumption amount of the vehicle can not be accurately calculated from the vehicle driving force.

The above-mentioned eco-driving is also consistent with the efforts of the governments to reduce GHG emissions, especially carbon emissions, based on the agreed Kyoto Protocol as a way to implement the GHG reduction goals pursuant to the Convention on Climate Change. Nationally Appropriate Mitigation Actions (NAMA), which means voluntary greenhouse gas reduction measures that developing countries are expected to report directly to the United Nations Framework Convention on Climate Change (UNFCCC) As well as a number of other issues. Greenhouse gas mitigation actions in developed and developing countries should be measurable, reportable and verifiable (MRV). Eco-driving can be a powerful means of implementing NAMA in the transportation sector, but there is a limit to the fact that the MRV method has not yet been developed sufficiently.

On the other hand, Korean Patent Registration No. 10-0274579 (registered on Mar. 14, 2000) discloses a vehicle identification method using a neural network model technique, and Korean Patent Registration No. 10-0837244 (registered on June 06, 2008) Korean Patent Registration No. 10-1703163 (registered on Jan. 31, 2013) discloses an apparatus and a method for predicting a complex failure of a vehicle, 10-2018-0029543 (published on Mar. 21, 2018) discloses a method of diagnosing the state of the vehicle through deep running, and Korean Utility Model Registration No. 20-0173957 (registered on December 23, 1999) And a Korean Utility Model Laid Open Publication No. 1999-0028544 (published on June 15, 1999) discloses a system for controlling the air-fuel ratio of a vehicle.

Until now, there has been no attempt to measure the fuel consumption of a vehicle in real time by artificial neural network, especially using deep running.

An object of the present invention is to provide a vehicle capable of calculating the energy consumption of the vehicle more accurately and more flexibly by calculating the energy consumption of the vehicle such as the fuel consumption amount of the vehicle or the electric consumption amount of the vehicle by utilizing the artificial neural network, And a method for calculating the energy consumption of the apparatus.

It is also an object of the present invention to provide a method for calculating the energy consumption of a vehicle that can effectively implement the MRV method in the transportation sector NAMA.

According to an aspect of the present invention, there is provided a method of calculating energy consumption of a vehicle, the method comprising: installing a data collection device including a GPS receiver and an energy consumption measurement unit on an experimental vehicle; The speed of the test vehicle and the inclination angle of the road on which the test vehicle travels are measured at every predetermined time interval by the GPS receiver, Wherein the control unit measures the energy consumption of the test vehicle at every interval, calculates the speed of the test vehicle measured by the GPS receiver, the inclination angle of the road, and the energy of the test vehicle measured by the energy consumption measuring unit Collecting experimental data by matching consumptions; Delivering the experimental data collected by the data collection device to an artificial neural network server having an artificial neural network including an input layer, at least one hidden layer and an output layer; Wherein the artificial neural network server inputs the velocity of the experimental vehicle and the inclination angle of the road as input values to the input layer, and converts the input layer into a first hidden layer, when the plurality of hidden layers exist, And arbitrarily setting parameters to be applied to conversion from the final hidden layer to the output layer, and comparing an output value calculated in the output layer with an energy consumption amount of the experimental vehicle received from the data collection device; When the neural network server determines that an error between the output value and the energy consumption amount of the experimental vehicle is out of a predetermined allowable range, calculating the output value after changing the parameters applied to the artificial neural network; Determining the parameter applied to the artificial neural network when the artificial neural network server judges that an error between the output value and the energy consumption amount of the experimental vehicle is within a predetermined allowable range; Wherein the artificial neural network server stores parameter data matched with the determined parameters for at least one of a vehicle type of the experimental vehicle, a type of the road applied to the experiment, and an interval of the road applied to the experiment; Transmitting the parameter data calculated by the artificial neural network server to an artificial neural network energy consumption calculating apparatus mounted on a traveling vehicle; Wherein the artificial neural network energy consumption calculation device of the traveling vehicle has an artificial neural network that is the same as the artificial neural network applied to the artificial neural network server, and the type of the traveling vehicle, the type of the road on which the traveling vehicle travels, And applying the parameter data to at least one of the sections of the roads applied to the driving vehicle to calculate the energy consumption of the driving vehicle according to the speed of the driving vehicle and the inclination angle of the driving road.

The artificial neural network preferably has two or more hidden layers.

The artificial neural network energy consumption calculation device of the traveling vehicle measures the speed of the traveling vehicle and the inclination angle of the road on which the traveling vehicle travels at every predetermined time interval by the GPS receiver.

Wherein the artificial neural network energy consumption calculation device of the traveling vehicle has map data, and combines the map data with the position of the traveling vehicle detected by the GPS receiver to determine the type of the road on which the traveling vehicle travels, It is preferable to determine the interval and calculate the energy consumption of the traveling vehicle accordingly.

When the artificial neural network server determines that the error between the output value and the energy consumption of the experimental vehicle is out of a predetermined allowable range, the parameters applied to the artificial neural network may be changed by the following equation (i).

x Gi ----- (i)Wi = Wi -

x Gi ----- (i)

는 학습률을 나나내며, Gi는 Wi의 기울기로서 다음 수학식 (ii)에 의하여 산출됨)(In the above equation (i), Wi represents the i-th parameter,

Where Gi is the slope of Wi and is calculated by the following equation (ii)

Gi = [L (Wi + h) - L (Wi - h)] / (2 x h)

(In the above equation (ii), h is a very small arbitrary value that is mathematically convergent to 0, for example, 0.0001, and L is calculated by the following equation (iii)

L = 0.5 x? (Y - t) ² ----- (iii)

(Where, Y represents the final output value of the artificial neural network and t represents an experimental measurement of the energy consumption of the experimental vehicle)

The method of calculating energy consumption of a vehicle according to the present invention can more accurately and more flexibly calculate the energy consumption of a vehicle by utilizing an artificial neural network, in particular, deep running. The method of the present invention can also effectively implement the MRV approach in the transportation sector NAMA.

1 is a diagram showing a schematic configuration of a system for calculating the energy consumption of a vehicle implementing the method for calculating the energy consumption of a vehicle according to the present invention.
FIG. 2 is a diagram showing a schematic configuration of a data collecting apparatus applied to the vehicle energy consumption calculating system shown in FIG. 1. FIG.
3 is a diagram showing a schematic configuration of an artificial neural network server applied to the system for calculating the energy consumption of the vehicle shown in FIG.
4 is a diagram showing a schematic configuration of an artificial neural network energy consumption calculation apparatus applied to the vehicle energy consumption calculation system shown in FIG.
5 is a view schematically showing an example of an artificial neural network applied to an artificial neural network server applied to the system for calculating the energy consumption of the vehicle shown in FIG.
FIG. 6 is a flowchart illustrating a learning process in which an artificial neural network server applied to the vehicle's energy consumption calculation system shown in FIG. 1 determines the values of parameters by an artificial neural network.
7 is a diagram for explaining a concept of partitioning a road section and applying a parameter value to each road section in the present invention.

Hereinafter, the present invention will be described in detail with reference to the drawings.

A schematic configuration of a vehicle energy consumption calculation system implementing a method for calculating the energy consumption of a vehicle according to the present invention is shown in Fig. The system for calculating the energy consumption of a vehicle according to the present invention includes a data collecting apparatus 100 mounted on an experimental vehicle 12, an artificial neural network server 200 and an artificial neural network energy consumption calculating apparatus (300).

2, the data collecting apparatus 100 includes a GPS receiving unit 110, an energy consumption measuring unit 120, a controller 130, a memory 130, (140) and a communication unit (150). 3, the artificial neural network server 200 includes a communication unit 210, a control unit 220, and a memory unit 230. The artificial neural network server 200 may further include a display unit 240. As shown in FIG. 4, the artificial neural network energy consumption calculating apparatus includes a GPS receiver 310, a controller 320, a memory 330, and a communication unit 340.

In the method for calculating the energy consumption of a vehicle according to the present invention, the data collecting apparatus 100 is mounted on the test vehicle 12 to measure the speed V of the test vehicle 12 and the speed of the test vehicle 12 when the test vehicle 12 actually runs. (T) of the experimental vehicle 12 corresponding to the inclination ([theta]) of the road on which the vehicle 12 traveled. That is, the data collecting apparatus 100 constructs experimental data including the speed V of the test vehicle, the inclination angle? Of the road, and the actual energy consumption t of the test vehicle as a database, . At this time, the data collection device 100 also includes the type of the experimental vehicle 12. [ Climate conditions, such as snow, rain, fog, etc., may also be included in the experimental data as the experimental vehicle 12 runs the experiment, if necessary.

The data collection device 100 may also include latitude, longitude and altitude, which are the original data acquired by the GPS receiver 110. The velocity V of the test vehicle 12 and the tilt angle? Of the road are actually calculated from the latitude, longitude and altitude data obtained by the GPS receiving unit 110.

The data collecting apparatus 100 measures the data at each predetermined time interval and stores the measured data in the memory unit 140. That is, the data collecting apparatus 100 may calculate the speed V of the test vehicle 12 and the test vehicle 12 at each predetermined time interval set by the GPS receiving unit 110 when the test vehicle 12 travels on an actual road ) Of the road on which the vehicle runs. The data collecting apparatus 100 may measure the speed of the test vehicle 12 and the inclination angle of the road on which the test vehicle 12 travels, for example, at intervals of one second.

The data collecting apparatus 100 also measures the actual energy consumption t of the test vehicle 12 at the same time intervals by the energy consumption measuring unit 120. [ The energy consumption measuring unit 120 is a fuel injection quantity measuring device when the test vehicle 12 has an internal combustion engine and is a fuel injection quantity measuring device for measuring the current value from the power line connecting the main battery using the constant voltage to the motor when the test vehicle 12 is an electric vehicle , And calculating the amount of electric power according to the current value and accumulating the amount of electric power with time.

The controller 130 of the data collecting apparatus 100 matches the data such as the speed V of the test vehicle, the inclination angle? Of the road and the actual energy consumption t of the test vehicle, And stores the data in the memory unit 140. The memory unit 140 stores the data in the database. The control unit 130 of the data collecting apparatus 100 also transmits the database to the artificial neural network server 200 through the communication unit 150.

The actual road driving test by the experimental vehicle 12 equipped with the data collecting apparatus 100 is preferably performed in all the sections of all the existing roads. However, in order to reduce the actual amount of experimentation, it is possible to omit experiments on other roads of the same kind, such as highways, national roads, and local roads, and to avoid any experimentation on the same road (for example, a specific highway) Experiments may be omitted in other road sections where the road conditions are considered to be the same. As will be described in more detail below, the parameters for the artificial neural network, which will be applied to the roads and road sections that are omitted in the above-described experiments, are the artificial neural networks that are applied to the similar roads and road sections Can be replaced with the parameters for.

In the present invention, the artificial neural network server 200 receives an input of the speed V of the experimental vehicle and the inclination angle? Of the road on which the experimental vehicle travels, and outputs the actual energy consumption t of the experimental vehicle as an output. And determine the correct parameters for each type of vehicle and type of road or road section by learning.

An artificial neural network generally comprises an input layer, at least one hidden layer, and an output layer, and FIG. 5 shows an example of an artificial neural network with two hidden layers applied to the present invention. 6 is a flow chart illustrating an operation of the artificial neural network server 200, that is, a learning process in which the artificial neural network server 200 determines the values of parameters by the artificial neural network.

As illustrated in FIG. 5, the artificial neural network applied in the present invention includes an input layer having two input values, that is, the velocity V of the experimental vehicle and the inclination angle? Of the road on which the experimental vehicle travels, For example, two hidden layers, and an output layer having one output value. The output value of one node of the hidden layer is a result obtained by multiplying the input value input to the node by the connection strength of the input value to the input of the transfer function. Likewise, the output value of the output layer is the result of multiplying the output value of the hidden layer input to the output layer by the connection strength to the value, and inputting the result of the transfer function. The transfer function may typically employ a sigmoid function, and a function such as f (x) = x may be applied to the output value. The connection strengths in the artificial neural network are represented by the parameters W ₁ , W ₂ and W ₃ in FIG.

In the present invention, the input X to be input to the input layer of the artificial neural network may be a single data set composed of a matrix (V,?) Of 1 row and 2 columns. However, as shown in FIG. 5, It is preferable that the data set is a plurality of data sets composed of two columns of matrices. These plurality of data sets are used as input values to be input to the input layer to determine the parameters (W ₁ , W ₂ and W ₃ ) applied to the artificial neural network. The number of rows of the data set can be determined arbitrarily or appropriately as needed.

Such a date set may be representative of only the specific section of the road on which the experimental vehicle 12 traveled, provided that the experimental vehicle 12 is obtained as a result of running the continuous section of the road. On the other hand, if the experimental set 12 is a combination of the data obtained as a result of running the discrete specific intervals of a certain road, such a date set may be a representative of the entire road or a corresponding It may be representative of any section of the road.

5, an input matrix X of (100 x 2) is transformed into a matrix P of (100 x 20) first hidden layers matrixed with a (2 x 20) parameter matrix W ₁ , of the matrix P is the matrix Q is a (10 x 1) of claim 2 are converted by the matrix Q of the hidden layer, (100 x 10) of (20 x 10) parameter matrix W ₂ and the matrix dot (100 x 10) of And is transformed into an output power Y of (100 x 1) matrix-wise with the parameter power W ₃ . Here, (ixj) means a matrix of i-th row and j-th column. In this transformation, P = sigmoid (X · W ₁ ) is applied, Q = sigmoid (P · W ₂ ) is applied, and Y = Q · W ₃ can be applied.

6, the artificial neural network server 200 includes a vehicle type and speed V of the test vehicle 12, an inclination angle? Of the road on which the test vehicle 12 travels, 12 from the data collection device 100 (step S100). In this case, the reception of the experiment data is preferably performed by wireless communication, but it may be performed by other means, for example, wired communication. For example, the data collection device 100 may store experimental data in an external USB memory And the artificial neural network server 200 copies experiment data stored in such a USB memory. Preferably, the experimental data includes latitude, longitude, and altitude data obtained by the GPS receiver 110.

The artificial neural network server 200 inputs the velocity V of the experimental vehicle 12 and the inclination angle? Of the road on which the experimental vehicle 12 travels among the received experimental data to the input layer of the artificial neural network as described above value and sets the value of the input (X) and s (step S110), the parameter (W _1, W ₂ and W _3), optionally (step S120). Then, the output value Y of the output layer is calculated according to the above-described method according to the input values V and? Of the input layer and the values of the parameters W ₁ , W ₂ and W ₃ (step S 130) . The output value Y thus calculated is compared with the actual energy consumption amount t of the actually measured experimental vehicle 12 to calculate the error L and it is determined whether or not the error L is below the threshold value S140). At this time, L is calculated by the following equation (iii).

If the error L is larger than the threshold value, the values of the parameters W ₁ , W ₂ and W ₃ are changed by, for example, the following equation (i) (step S 150) And the output value Y is calculated again.

x Gi ----- (i)Wi = Wi -

x Gi ----- (i)

는 학습률을 나타낸다. 학습률(

)은 매우 작은 값으로 적절하게 설정될 수 있다. Gi는 Wi의 기울기로서 다음 수학식 (ii)에 의하여 산출된다.In the above equation (i), Wi represents the i-th parameter,

Represents the learning rate. Learning rate

) Can be appropriately set to a very small value. Gi is the slope of Wi and is calculated by the following equation (ii).

Gi = [L (Wi + h) - L (Wi - h)] / (2 x h)

In the above equation (ii), h may be any value much smaller than 1, particularly a very small arbitrary value converging to zero, for example, 0.0001. An example in which the gradient Gi is calculated by the slope descending method is expressed in Equation (ii), but various gradient calculation methods such as an error back propagation method can be applied. L is calculated by the following equation (iii).

L = 0.5 x? (Y - t) ² ----- (iii)

In the equation (iii), Y represents a final output value of the artificial neural network, and t represents an experimental measurement of the actual energy consumption of the experimental vehicle. For an input matrix X of (100 x 2), the output matrix Y has the form (100 x 1). Correspondingly, the actual energy consumption t of the actually measured experimental vehicle 12 also has the form of (100 x 1). Therefore, the equation (iii) can be solved as follows.

_{L = 0.5 x [(Y 1} - t 1) 2 + (Y 2 - t 2) 2 + ······ + (Y 100 - t 100) 2]

On the other hand, if the artificial neural network server 200 determines that the error L between the output value Y and the actual energy consumption t of the experimental vehicle 12 is within the predetermined allowable range, that is, within the threshold value, s to confirm the value of (W _1, W ₂ and W ₃₎ (step S160). Then, parameter data including the values of the determined parameters (W ₁ , W ₂ and W ₃ ) is stored in its own memory unit 230, and such parameter data is also stored in the artificial neural network energy consumption calculating apparatus (300). At this time, the parameter data may be values of parameters determined for at least one of the type of the experimental vehicle, the type of the road applied to the experiment, and the specific section of the road applied to the experiment. The specific section of the road applied to the experiment can be specified by a series of positions (latitude, longitude and altitude) of the experimental vehicle.

The artificial neural network energy consumption calculation device 300 receives the above parameter data from the artificial neural network server 200 and uses it. To this end, the artificial neural network energy consumption calculation apparatus 300 has the same artificial neural network as the artificial neural network applied to the artificial neural network server 200. Thus, the artificial neural network energy consumption calculation device 300 applies the parameter data applied to at least one of the type of the traveling vehicle, the type of the road on which the traveling vehicle travels, and the section of the road applied to the experiment to the artificial neural network, The speed of the vehicle and the energy consumption of the traveling vehicle according to the inclination angle of the traveling road are calculated. It is preferable that the energy consumption calculation of the traveling vehicle in the artificial neural network energy consumption calculation apparatus 300 is performed in real time.

More specifically, the controller 320 of the artificial neural network energy consumption calculation apparatus 300 calculates the speed of the traveling vehicle 14 at every predetermined time interval determined by the GPS receiving unit 310 and the road speed at which the traveling vehicle 14 travels Is measured. The control unit 320 sets the velocity V of the traveling vehicle 14 and the inclination angle? Of the road on which the traveling vehicle 14 travels as the input X of the input layer in the artificial neural network, The output Y of the output layer is obtained by applying the values of the corresponding parameters W ₁ , W ₂ and W ₃ received from the neural network server 200 and the output value is estimated as the energy consumption of the traveling vehicle. Therefore, if the energy consumption for each time interval of the traveling vehicle is all accumulated, the energy consumption for the total traveling of the traveling vehicle can be calculated.

It is theoretically most preferable that the present invention divide the road section and determine the values of the parameters W ₁ , W ₂ and W _{3 for} each section of the road. 7, in a situation where all the roads are divided by intervals and the values of the parameters W ₁ , W ₂ and W ₃ are determined for each section of the road, the parameters W ₁ , W ₂ and W ₃ ) are applied. Fig. 7 does not specifically show the sections of the road on a route other than the route on which the traveling vehicle travels.

It is assumed that a driving vehicle follows the path of ①-②-③-④-⑤-⑥-⑦-⑧ or the path of ①-②-③-④-⑤-⑨-⑩-⑪. At this time, the values of the parameters (W ₁ , W _2, and W ₃ ) are different from each other for each section. When the traveling vehicle is traveling in the section (1), the artificial neural network energy consumption calculation device (300) calculates the energy consumption of the traveling vehicle by applying the values of the parameters (W ₁ , W ₂ and W ₃ ) The same is true for other sections. This means that the cumulative energy consumption of the traveling vehicle depends on the speed of the traveling vehicle and the inclination angle of the road on which the traveling vehicle travels, as well as the path of successive sections. Therefore, according to the present invention, more accurate energy consumption can be calculated.

Although there is no particular criterion for determining the road section, it is desirable to make one road section as short as possible in order to increase the accuracy. However, in this case, the artificial neural network server 200 has a disadvantage that the number of operations to be processed is increased. Accordingly, it may be practical to set the interval from one node of the road to another neighboring node as a road section for the parameters W ₁ , W ₂ and W ₃ . At this time, the node of the road means the branching point of the road.

Applying the same parameters (W ₁ , W ₂ and W ₃ ) to one road section means that the corresponding road section is averaged. That is, each point in the road segment loses its characteristic and is replaced with the entire characteristic of the corresponding road segment.

For example, assuming that the same parameters (W ₁ , W _2, and W ₃ ) are applied to the entire section of a specific highway to calculate the energy consumption of the vehicle, The energy consumption calculated by the artificial neural network energy consumption calculation apparatus 300 may be significantly different from the energy consumption actually consumed by the traveling vehicle. This is because the parameters (W ₁ , W ₂ and W ₃ ) reflecting the characteristics of the road sections of the specific highway that the traveling vehicle does not actually travel are used for calculating the energy consumption of the traveling vehicle. Therefore, in order to apply the parameters (W ₁ , W ₂ and W ₃ ) reflecting only the characteristic of the road section on the path actually traveled by the traveling vehicle, it is desirable to divide the road into road sections of an appropriate size .

In the present invention, if an experiment in which the experimental vehicle 12 runs in all the sections of all the roads existing is performed, the values of the parameters are calculated for all sections of all the roads for the type of the experimental vehicle 12. That is, a map (hereinafter referred to as "parameter map" ) of the values of the parameters is completed for each vehicle type and for all the roads of all the roads. If the experiment can not be performed on all roads, there will be roads where the parameter map is not complete regarding the values of the parameters. For uncompleted roads and road sections, The values of the parameters (W ₁ , W _2, and W ₃ ) for the interval can be applied.

In the present invention, the artificial neural network energy consumption calculation apparatus 300 calculates the energy consumption amount of the traveling vehicle and characteristics of the traveling vehicle in the case where there is no identical experimental vehicle among the experimental vehicles included in the parameter data received from the artificial neural network server 200 A similar experimental vehicle can be regarded as a driving vehicle and the energy consumption can be calculated. For example, if the driving vehicle is a specific brand vehicle of company A and there is no parameter data for the vehicle because the experiment has not yet been performed for the vehicle, the vehicle can be arbitrarily selected according to a predetermined method The corresponding experimental vehicle can be selected.

The artificial neural network energy consumption calculation apparatus 300 may have map data and may combine the map data and the position of the vehicle detected by the GPS receiver 310 to determine the type of road or the section of the road on which the vehicle is traveling You can decide faster and more accurately. That is, the artificial neural network energy consumption calculation apparatus 300 can utilize general map data when it is necessary to more accurately analyze the parameter map represented by the parameter data received from the artificial neural network server 200.

The above-mentioned items can be summarized as follows.

No. Item Contents Row x column Remarks One Input value X = (V, &thetas;) ¹⁾ 1 x 2 GPS data (1Hz) 2 Parameter 1 W ₁ 2 x 20 Initial value = random 3 Intermediate output 1 P = sigmoid (X · W ₁ ) 1 x 20 Matrix internal and sigmoid ²⁾ activation function 4 Parameter 2 W ₂ 20 x 10 Initial value = random 5 Intermediate output value 2 Q = sigmoid (P? W ₂ ) 1 x 10 Matrix intrinsic and sigmoid activation functions 6 Parameter 3 W ₃ 10 x 1 Initial value = random 7 Final output value Y = Q · W ₃ 1 x 1 Energy consumption calculation result value 8 Label ³⁾ t 1 x 1 Energy consumption experiment measurement 9 Loss (error) L = 0.5 x? (Yt) ² Difference between computed and measured values 10 inclination G ₁ = [L (W ₁ + h) -L (W ₁ -h)] / (2 xh)
G ₂ = [L (W ₂ + h) -L (W ₂ -h)] / (2 xh)
G ₃ = [L (W ₃ + h) -L (W ₃ -h)] / (2 xh) 2 x 20
20 x 10
10 x 1 Calculation by slope descent method ⁴⁾
: a value as small as h = 0.0001
: Mathematically calculated for h → 0 11 Update parameters

x G₁
W₂ = W₂ -

x G₂
W₃ = W₃ -

x G₃ W ₁ = W ₁ -

x G ₁
W ₂ = W ₂ -

x G ₂
W ₃ = W ₃ -

x G ₃ 2 x 20
20 x 10
10 x 1 Parameters that minimize loss (

= Learning rate) 12 Iterative operation 13 Determining Parameters W ₁ , W ₂ , W ₃ Remember final value 14 Actual car driving Speed, inclination measurement (X)
→ Energy consumption (Y) calculation Y = sigmoid (sigmoid (X · W 1) · W 2) · W 3

Explanation of each note in the above table

¹⁾ Input value (X): If the experiment time is 100 seconds, it is 100 rows x 2 columns, 200 seconds 200 rows x 2 columns

²⁾ Sigmoid function: Sigmoid (x) = 1 / (1 + exp (-x))

³⁾ Label (t): Value corresponding to correct answer (energy consumption measured by experiment)

⁴⁾ Slope descent method: It is possible to apply various slope calculation methods such as the slope calculation of the numerical differential method (the rate of change of the loss when the parameter is finely changed) → the error back propagation method

The method of the present invention can not completely construct the experimental data from the beginning because it is based on experimental data obtained by running at very different speeds for a large number of vehicles and a large number of roads. Once the experimental data is constructed to some extent, the ANN server 200 constructs parameter data for specific vehicles, specific roads, and specific road sections, and the ANN apparatus 300 calculates the parameters Data can be used to calculate the energy consumption of the traveling vehicle. The data collection device 100 may then perform more sophisticated and more diverse experiments, and the neural network server 200 may then further refine the parameter data by performing a learning process on additional experimental data have. The artificial neural network energy consumption calculation apparatus 300 can more accurately calculate the energy consumption of the traveling vehicle using the updated parameter data.

10: Vehicle Energy Consumption Calculation System 12: Experimental Vehicle
14: Driving vehicle 100: Data collecting device
110, 310: GPS receiving unit 120: Energy consumption measuring unit
130, 220, and 320: Controllers 140, 230, and 330:
150, 210, 340: communication unit 200: artificial neural network server
240:
300: Artificial neural network energy consumption calculation device

Claims (5)

Installing a data collecting apparatus including a GPS receiving unit and an energy consumption measuring unit in an experimental vehicle;
The speed of the test vehicle and the inclination angle of the road on which the test vehicle travels are measured at every predetermined time interval by the GPS receiver, The actual energy consumption of the test vehicle is measured at every interval, the speed of the test vehicle measured by the GPS receiver, the inclination angle of the road and the tilt angle of the road measured by the energy consumption measuring unit Collecting experimental data by matching actual energy consumption;
Delivering the experimental data collected by the data collection device to an artificial neural network server having an artificial neural network including an input layer, at least one hidden layer and an output layer;
Wherein the artificial neural network server inputs the velocity of the experimental vehicle and the inclination angle of the road as input values to the input layer, and converts the input layer into a first hidden layer, when the plurality of hidden layers exist, And arbitrarily setting parameters to be applied to conversion from the final hidden layer to the output layer, and comparing an output value calculated in the output layer with an actual energy consumption amount of the experimental vehicle received from the data collection device;
If the error between the output value and the actual energy consumption of the experimental vehicle is determined to be out of a predetermined allowable range, changing the parameters applied to the artificial neural network and calculating the output value again;
Determining the parameters applied to the artificial neural network when the artificial neural network server determines that an error between the output value and the actual energy consumption of the experimental vehicle is within a predetermined allowable range;
Wherein the artificial neural network server stores parameter data matched with the determined parameters for at least one of a vehicle type of the experimental vehicle, a type of the road applied to the experiment, and an interval of the road applied to the experiment;
Transmitting the parameter data calculated by the artificial neural network server to an artificial neural network energy consumption calculating apparatus mounted on a traveling vehicle;
Wherein the artificial neural network energy consumption calculation device of the traveling vehicle has an artificial neural network that is the same as the artificial neural network applied to the artificial neural network server, and the type of the traveling vehicle, the type of the road on which the traveling vehicle travels, And calculating the energy consumption amount of the vehicle based on the speed of the traveling vehicle and the inclination angle of the traveling road by applying the parameter data applied to at least one of the sections of the road applied to the road Way. The method according to claim 1,
Wherein the artificial neural network has two or more hidden layers. The method according to claim 1,
Wherein the artificial neural network energy consumption calculation device of the traveling vehicle measures the speed of the traveling vehicle and the inclination angle of the road on which the traveling vehicle travels at every predetermined time interval determined by the GPS receiver Way. The method of claim 3,
Wherein the artificial neural network energy consumption calculation device of the traveling vehicle has map data, and combines the map data with the position of the traveling vehicle detected by the GPS receiver to determine the type of the road on which the traveling vehicle travels, And calculating an energy consumption amount of the traveling vehicle based on the determined energy consumption amount. 5. The method according to any one of claims 1 to 4,
Wherein the artificial neural network server changes the parameters applied to the artificial neural network by the following equation (i) when it is determined that the error between the output value and the energy consumption amount of the experimental vehicle exceeds a predetermined allowable range: To calculate the energy consumption of the vehicle.
Wi = Wi -

x Gi ----- (i)
(In the above equation (i), Wi represents the i-th parameter,

Where Gi is the slope of Wi and is calculated by the following equation (ii)
Gi = [L (Wi + h) -L (Wi-h)] / (2 xh)
(In the above equation (ii), h is a very small arbitrary value that is mathematically convergent to 0, for example, 0.0001, and L is calculated by the following equation (iii)
L = 0.5 x? (Y - t) ² ----- (iii)
(Where, Y represents the final output value of the artificial neural network and t represents an experimental measurement of the energy consumption of the experimental vehicle)

Images