Real-time vehicle carbon dioxide emission prediction method and system based on neural network

Abstract

The present invention discloses a neural network-based method and system for predicting real-time vehicle carbon dioxide emissions, relating to the field of vehicle emissions. The method comprises collecting actual road emission test data of a prototype vehicle based on emissions testing conducted on the prototype vehicle; calculating the prototype vehicle's instantaneous acceleration and instantaneous carbon dioxide emission rate based on the collected actual road emission test data; using the prototype vehicle's speed, acceleration, and engine speed as input feature data, and windowing the input feature data to form a training set; building an LSTM deep learning real-time carbon dioxide emission prediction model and training it using the training set; and predicting the instantaneous carbon dioxide emission rate of vehicles of the same model as the prototype vehicle based on the trained real-time carbon dioxide emission prediction model. The present invention can achieve high-precision instantaneous carbon dioxide emission prediction.

Description

Technical Field

[0001] The present invention relates to the field of vehicle emissions, and in particular to a method and system for predicting vehicle carbon dioxide emissions in real time based on a neural network. Background Art

[0002] As the number of vehicles continues to increase, carbon dioxide emissions from vehicle exhaust have also increased dramatically, causing serious impacts on the environment.

[0003] Because current fuel consumption tests are conducted in laboratories, the complex environment of real roads and the operating curves used in laboratory testing do not accurately reflect actual operating conditions, resulting in final fuel consumption test results that are far lower than the vehicle's actual fuel consumption. In other words, laboratory test results do not truly reflect the vehicle's actual CO2 emissions. Furthermore, calls for full lifecycle regulation of vehicle CO2 emissions are growing, but relying solely on current fuel consumption certification testing, production consistency checks, and in-use compliance checks cannot effectively achieve this goal.

[0004] With the development of deep learning technology, numerous researchers have conducted extensive research on deep learning-based vehicle emission prediction models. A paper on vehicle CO2 emissions prediction proposed an LSTM vehicle CO2 emissions prediction model, which achieves high-accuracy CO2 emissions predictions. However, during the data preprocessing phase, the paper used the Min-Max method to normalize the training dataset. Due to the inconsistent scales of the feature data in the training set, this normalization method resulted in inconsistent resolution for each feature. To achieve high prediction accuracy, the model requires strong feature extraction capabilities, so the model constructed in the paper adopts a complex architecture. The proposed model consists of three LSTM layers and one fully connected layer, with 120, 240, and 500 LSTM units in each layer, respectively. The large number of model parameters places high demands on the computer performance required to run the model. Furthermore, the starting data points of adjacent windows in the input features are significantly separated, making the model unsuitable for predicting instantaneous CO2 emissions per second. Summary of the Invention

[0005] In view of the defects existing in the prior art, the purpose of the present invention is to provide a method and system for predicting vehicle carbon dioxide emissions in real time based on a neural network, which can achieve high-precision prediction of instantaneous carbon dioxide emissions.

[0006] To achieve the above objectives, the present invention provides a method for predicting vehicle carbon dioxide emissions in real time based on a neural network, which specifically includes the following steps:

[0007] Based on the emission test conducted on the prototype vehicle, actual road emission test data of the prototype vehicle is collected;

[0008] Based on the collected actual road emission test data, the instantaneous acceleration and instantaneous carbon dioxide emission rate of the prototype vehicle are calculated;

[0009] The speed, acceleration and engine speed of the sample vehicle are used as input feature data, and the input feature data is divided into windows to form a training set;

[0010] Build a real-time carbon dioxide emission prediction model based on LSTM deep learning and use the training set for training;

[0011] Based on the trained real-time carbon dioxide emission prediction model, the instantaneous carbon dioxide emission rate of vehicles of the same model as the sample vehicle can be predicted.

[0012] On the basis of the above technical solution, the actual road emission test data of the sample vehicle is collected based on the emission test conducted on the sample vehicle. The specific steps include:

[0013] The prototype vehicle is subjected to emission testing based on the on-board emission testing equipment carried by the prototype vehicle, and actual road emission test data of the prototype vehicle is collected.

[0014] On the basis of the above technical solution, after taking the speed, acceleration and engine speed of the sample vehicle as input feature data, the following is also included:

[0015] The vehicle speed and engine speed of the sample vehicle in the input feature data are converted and processed, and then the input feature data is divided into windows.

[0016] On the basis of the above technical solutions,

[0017] The speed of the sample vehicle is converted as follows:

[0018] V1=log(V)

[0019] Wherein, V1 represents the vehicle speed after the conversion process, V represents the vehicle speed before the conversion process, and log represents the logarithmic calculation;

[0020] The engine speed of the prototype vehicle is converted as follows:

[0021] RPM1=RPM / 100

[0022] Here, RPM1 represents the engine speed after the conversion process, and RPM represents the engine speed before the conversion process.

[0023] On the basis of the above technical solution, the window division of the input feature data is specifically as follows:

[0024] The input feature data is divided into windows with the first second data as the starting data point, the window step size is 5 seconds, and the interval between the starting data points of adjacent windows is 1 second.

[0025] Based on the above technical solution, the real-time carbon dioxide emission prediction model includes two neural layers, namely an LSTM layer and a fully connected layer.

[0026] Based on the above technical solution, the LSTM layer includes 5 LSTM units, and the fully connected layer includes 1 unit.

[0027] The present invention provides a vehicle carbon dioxide real-time emission prediction system based on a neural network, comprising:

[0028] A collecting unit, configured to collect actual road emission test data of the sample vehicle based on the emission test performed on the sample vehicle;

[0029] A calculation unit, which is used to calculate the instantaneous acceleration and instantaneous carbon dioxide emission rate of the sample vehicle based on the collected actual road emission test data;

[0030] a partitioning unit, which is used to take the speed, acceleration and engine speed of the sample vehicle as input feature data and perform window partitioning on the input feature data to form a training set;

[0031] A training unit, which is used to build an LSTM deep learning real-time carbon dioxide emission prediction model and use the training set for training;

[0032] The prediction unit is used to predict the instantaneous emission rate of carbon dioxide of vehicles of the same model as the sample vehicle based on the trained real-time carbon dioxide emission prediction model.

[0033] Based on the above technical solution, the real-time carbon dioxide emission prediction model includes two neural layers, namely an LSTM layer and a fully connected layer.

[0034] Based on the above technical solution, the LSTM layer includes 5 LSTM units, and the fully connected layer includes 1 unit.

[0035] Compared with the existing technology, the advantages of the present invention are: by building a simple, high-precision, real-time carbon dioxide emissions prediction model, to reduce model complexity, different normalization methods are adopted for different input features in the training set in the preprocessing stage to ensure that the model has basically consistent resolution for each feature. After the three parameters of normalized vehicle real-time speed, vehicle real-time acceleration, and engine real-time speed are windowed, they constitute input feature data to realize the training of the carbon dioxide emissions prediction model, so that the trained model can achieve high-precision carbon dioxide instantaneous emissions prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0037] Figure 1 The flowchart of a method for predicting real-time vehicle carbon dioxide emissions based on a neural network in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0039] See also Figure 1 As shown, an embodiment of the present invention provides a method for predicting vehicle carbon dioxide emissions in real time based on a neural network, which specifically includes the following steps:

[0040] S1: Based on the emission test conducted on the prototype vehicle, actual road emission test data of the prototype vehicle is collected; the actual road emission test data includes carbon dioxide emission related data and vehicle driving related data.

[0041] In an embodiment of the present invention, actual road emission test data of a sample vehicle is collected based on an emission test performed on the sample vehicle. The specific steps include: performing an emission test on the sample vehicle based on an on-board emission test device carried by the sample vehicle, and collecting actual road emission test data of the sample vehicle.

[0042] S2: Based on the collected actual road emission test data, calculate the instantaneous acceleration and instantaneous carbon dioxide emission rate of the prototype vehicle;

[0043] S3: The speed, acceleration, and engine speed of the sample vehicle are used as input feature data, and the input feature data are divided into windows to form a training set;

[0044] In an embodiment of the present invention, after the speed, acceleration and engine speed of the sample vehicle are used as input feature data, the method further includes: converting the speed and engine speed of the sample vehicle in the input feature data, and then dividing the input feature data into windows.

[0045] The speed of the sample vehicle is converted as follows:

[0046] V1=log(V)

[0047] Wherein, V1 represents the vehicle speed after the conversion process, V represents the vehicle speed before the conversion process, and log represents the logarithmic calculation;

[0048] The engine speed of the prototype vehicle is converted as follows:

[0049] RPM1=RPM / 100

[0050] Here, RPM1 represents the engine speed after the conversion process, and RPM represents the engine speed before the conversion process.

[0051] That is, vehicle speed, acceleration and engine speed are selected as input feature data, and different methods are used to perform normalization conversion processing on different features. The vehicle speed and engine speed are processed using the above formula, and the acceleration is not processed to ensure that the model has basically consistent resolution for each feature. Then, the three parameters of normalized vehicle real-time speed, vehicle real-time acceleration and engine real-time speed are divided into windows with a time step of 5 seconds and a time interval of 1 second between adjacent window starting points to form a training set.

[0052] In the embodiment of the present invention, the input feature data is divided into windows, specifically:

[0053] The input feature data is windowed, starting with the first second, with a window step of 5 seconds and a 1-second interval between the starting data points of adjacent windows. This means that emissions at the tth second are only related to the vehicle's operating status within the previous 4 seconds.

[0054] S4: Build a real-time carbon dioxide emission prediction model based on LSTM deep learning and use the training set for training;

[0055] In this embodiment of the present invention, the real-time carbon dioxide emissions prediction model includes two neural layers: an LSTM layer and a fully connected layer. The LSTM layer includes five LSTM units, and the fully connected layer includes one unit. The real-time carbon dioxide emissions prediction model does not include an input layer.

[0056] For the real-time CO2 emissions prediction model, the validation set and training set account for 20% and 80% of the total training set, respectively. The training batch size is 64, the Adam optimizer is used, the initial learning rate is 0.002, and the learning rate is decayed by 0.98 times every 500 training epochs. The mean absolute error is the loss function, and the total number of model training epochs is set to 500. To prevent overfitting, the Early Stop strategy (a method to avoid overfitting when training complex machine learning models) is also adopted. After model training is complete, it can be deployed to predict the instantaneous CO2 emissions rate for the same model.

[0057] S5: Based on the trained real-time CO2 emission prediction model, the instantaneous CO2 emission rate of vehicles of the same model as the sample vehicle is predicted.

[0058] In order to verify the accuracy of the vehicle carbon dioxide emission real-time prediction method of the present invention, the following is an illustration with reference to specific examples.

[0059] Four gasoline vehicles of different types that meet the National VI emission standards were selected for actual road emission tests. Each vehicle was tested three times with different driving behaviors (smooth, normal, and intense), and each test lasted more than 6,000 seconds.

[0060] The training data for the real-time CO2 emissions prediction model consists of actual road emissions data under normal driving behavior. After model training, the model is used to predict the instantaneous CO2 emission rate under smooth and aggressive driving conditions. To analyze the instantaneous emission prediction results, we randomly selected a set of prediction data and compared the actual and predicted instantaneous CO2 emission rates for the first 2000 seconds. We found that the overall trends of the predicted and actual CO2 emission rates are consistent, and they are basically consistent for most of the time. To analyze the final CO2 prediction error, the correlation error between the actual and predicted total CO2 emissions was compared. The results show that the maximum relative error between the predicted and measured CO2 emission rates is 8.25%, the minimum relative error is 0.58%, and the average relative error over eight runs is 4.11%.

[0061] The neural network-based vehicle carbon dioxide real-time emission prediction method of the embodiment of the present invention builds a simple, high-precision, real-time carbon dioxide real-time emission prediction model. In order to reduce the complexity of the model, different normalization methods are adopted for different input features in the training set in the preprocessing stage to ensure that the model has basically consistent resolution of each feature. After the three parameters of the normalized vehicle real-time speed, vehicle real-time acceleration and engine real-time speed are divided into windows, they constitute input feature data to realize the training of the carbon dioxide real-time emission prediction model, so that the trained model can achieve high-precision carbon dioxide instantaneous emission prediction.

[0062] In one possible implementation, an embodiment of the present invention further provides a readable storage medium, which is located in a PLC (Programmable Logic Controller) controller. The readable storage medium stores a computer program, which, when executed by a processor, implements the following steps of the method for predicting vehicle carbon dioxide emissions in real time:

[0063] Based on the emission test conducted on the prototype vehicle, actual road emission test data of the prototype vehicle is collected;

[0064] Based on the collected actual road emission test data, the instantaneous acceleration and instantaneous carbon dioxide emission rate of the prototype vehicle are calculated;

[0065] The speed, acceleration and engine speed of the sample vehicle are used as input feature data, and the input feature data is divided into windows to form a training set;

[0066] Build a real-time carbon dioxide emission prediction model based on LSTM deep learning and use the training set for training;

[0067] Based on the trained real-time carbon dioxide emission prediction model, the instantaneous carbon dioxide emission rate of vehicles of the same model as the sample vehicle can be predicted.

[0068] The storage medium can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by an instruction execution system, device or device or used in combination with it.

[0069] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination thereof.

[0070] Computer program code for performing the operations of the present invention can be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0071] An embodiment of the present invention provides a neural network-based vehicle carbon dioxide real-time emission prediction system, which includes a collection unit, a calculation unit, a division unit, a training unit, and a prediction unit.

[0072] The collection unit is used to collect actual road emission test data of the sample vehicle based on the emission test conducted on the sample vehicle; the calculation unit is used to calculate the instantaneous acceleration and instantaneous carbon dioxide emission rate of the sample vehicle based on the collected actual road emission test data; the division unit is used to use the vehicle speed, acceleration and engine speed of the sample vehicle as input feature data, and perform window division on the input feature data to form a training set; the training unit is used to build an LSTM deep learning real-time carbon dioxide emission prediction model, and use the training set for training; the prediction unit is used to predict the instantaneous carbon dioxide emission rate of vehicles of the same model as the sample vehicle based on the trained real-time carbon dioxide emission prediction model.

[0073] In the embodiment of the present invention, the real-time carbon dioxide emission prediction model includes two neural layers: an LSTM layer and a fully connected layer. The LSTM layer includes five LSTM units, and the fully connected layer includes one unit.

[0074] The foregoing is merely a list of specific embodiments of the present application, intended to enable those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the broadest scope consistent with the principles and novel features of the present application.

[0075] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

Claims

1. A method for predicting vehicle carbon dioxide emissions in real time based on a neural network, characterized in that: The specific steps include: Based on the emission test conducted on the prototype vehicle, actual road emission test data of the prototype vehicle is collected; Based on the collected actual road emission test data, the instantaneous acceleration and instantaneous carbon dioxide emission rate of the prototype vehicle are calculated; The speed, acceleration and engine speed of the sample vehicle are used as input feature data, and the input feature data is divided into windows to form a training set; Build a real-time carbon dioxide emission prediction model based on LSTM deep learning and use the training set for training; Based on the trained real-time CO2 emission prediction model, the instantaneous CO2 emission rate of vehicles of the same model as the prototype vehicle can be predicted; Wherein, after the speed, acceleration and engine speed of the sample vehicle are used as input feature data, the method further includes: converting the speed and engine speed of the sample vehicle in the input feature data, and then dividing the input feature data into windows; Among them, the speed of the sample vehicle is converted as follows: V1=log(V) Wherein, V1 represents the vehicle speed after conversion, V represents the vehicle speed before conversion, and log represents logarithmic calculation; The engine speed of the prototype vehicle is converted as follows: RPM1=RPM / 100 Here, RPM1 represents the engine speed after the conversion process, and RPM represents the engine speed before the conversion process.

2. The method for predicting vehicle carbon dioxide emissions in real time based on a neural network as claimed in claim 1, characterized in that: The actual road emission test data of the sample vehicle is collected based on the emission test conducted on the sample vehicle, and the specific steps include: The prototype vehicle is subjected to emission testing based on the on-board emission testing equipment carried by the prototype vehicle, and actual road emission test data of the prototype vehicle is collected.

3. The method for predicting vehicle carbon dioxide emissions in real time based on a neural network as claimed in claim 1, characterized in that: The window division of the input feature data is specifically as follows: The input feature data is divided into windows with the first second data as the starting data point, the window step size is 5 seconds, and the interval between the starting data points of adjacent windows is 1 second.

4. The method for predicting vehicle carbon dioxide emissions in real time based on a neural network according to claim 1, wherein: The real-time carbon dioxide emission prediction model includes two neural layers, namely an LSTM layer and a fully connected layer.

5. The method for predicting vehicle carbon dioxide emissions in real time based on a neural network according to claim 4, characterized in that: The LSTM layer includes 5 LSTM units, and the fully connected layer includes 1 unit.

6. A real-time vehicle carbon dioxide emission prediction system based on neural network, characterized in that: include: A collecting unit, configured to collect actual road emission test data of the sample vehicle based on the emission test performed on the sample vehicle; A calculation unit, which is used to calculate the instantaneous acceleration and instantaneous carbon dioxide emission rate of the sample vehicle based on the collected actual road emission test data; a partitioning unit, which is used to take the speed, acceleration and engine speed of the sample vehicle as input feature data and perform window partitioning on the input feature data to form a training set; A training unit, which is used to build an LSTM deep learning real-time carbon dioxide emission prediction model and use the training set for training; A prediction unit, which is used to predict the instantaneous carbon dioxide emission rate of vehicles of the same model as the sample vehicle based on the trained carbon dioxide real-time emission prediction model; Wherein, after the speed, acceleration and engine speed of the sample vehicle are used as input feature data, the method further includes: converting the speed and engine speed of the sample vehicle in the input feature data, and then dividing the input feature data into windows; Among them, the speed of the sample vehicle is converted as follows: V1=log(V) Wherein, V1 represents the vehicle speed after conversion, V represents the vehicle speed before conversion, and log represents logarithmic calculation; The engine speed of the prototype vehicle is converted as follows: RPM1=RPM / 100 Here, RPM1 represents the engine speed after the conversion process, and RPM represents the engine speed before the conversion process.

7. The neural network-based vehicle carbon dioxide emission real-time prediction system according to claim 6, characterized in that: The real-time carbon dioxide emission prediction model includes two neural layers, namely an LSTM layer and a fully connected layer.

8. The neural network-based vehicle carbon dioxide emission real-time prediction system according to claim 7, characterized in that: The LSTM layer includes 5 LSTM units, and the fully connected layer includes 1 unit.

Images