Automobile air conditioner energy consumption prediction method, system, medium and equipment

Abstract

The invention discloses an automobile air conditioner energy consumption prediction method and system, a medium and equipment, and relates to the technical field of new energy automobiles. The method comprises the steps; acquiring automobile operation condition parameters and environment parameters as energy consumption data; the energy consumption prediction model is used for processing the energy consumption data, an automobile air conditioner energy consumption prediction result is obtained, in the energy consumption prediction model, a deep learning algorithm is used for conducting change rule learning on time sequence data in the energy consumption data, meanwhile, spatial features in the energy consumption data are captured, and a new data sample is generated according to the learning result; the new data sample is used for optimizing parameters of the energy consumption prediction model; and controlling and adjusting the running state of the automobile air-conditioning system according to the automobile air-conditioning energy consumption prediction result. According to the method, deep learning of the time sequence can be carried out, and prediction generation of high-dimensional energy consumption data is realized by using the adversarial network, so that accurate prediction of the energy consumption of the automobile air conditioning system is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy vehicles, and in particular to a method, system, medium and equipment for predicting energy consumption of automobile air conditioners. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] The car air conditioning system can provide a comfortable riding environment for passengers, reduce driver fatigue and improve driving safety. But at the same time, the car air conditioning system is also one of the energy consumption of new energy vehicles. Reducing the energy consumption of the car is of great significance to improving the efficiency of energy utilization.

[0004] However, the existing method for predicting air conditioning energy consumption is mainly to establish a physical simulation model of automobile energy consumption, and to revise the simulation model based on the experimental test data of the actual vehicle to improve the accuracy of the simulation prediction. This method has two disadvantages. First, the air conditioning system simulation model converges slowly, has a long calculation cycle, and is inefficient; second, the analysis and judgment of the actual vehicle test data is highly dependent on the engineering experience of engineers. The engineering experience and engineering ability of engineers determine the accuracy of the air conditioning energy consumption prediction value, which has great limitations and is prone to errors.

[0005] Long short-term memory network (LSTM) is a deep learning model suitable for time series prediction, which can handle long-term dependencies. In terms of air-conditioning system energy consumption calculation, LSTM can be used to analyze the real-time data of the air-conditioning system and calculate the energy consumption of the air-conditioning system by learning the change patterns of these data over time.

[0006] The inventors found that the current single long short-term memory network method for calculating the energy consumption of automobile air conditioning systems cannot extract accurate feature information, which in turn affects the identification of energy consumption anomalies. In addition, unlike central air conditioning systems such as buildings, the driving environment of automobiles is constantly changing, the cabin space is small and the height is low, the temperature in the car changes frequently, and the air volume distribution is uneven, making the prediction process of automobile air conditioning energy consumption more difficult. Summary of the invention

[0007] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a method, system, medium and equipment for predicting the energy consumption of automobile air conditioning, which can perform deep learning of time series through a CNN-LSTM hybrid model and use an adversarial network to predict and generate high-dimensional energy consumption data, thereby realizing accurate prediction of the energy consumption of the automobile air-conditioning system.

[0008] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0009] A first aspect of the present invention provides a method for predicting energy consumption of an automobile air conditioner, comprising the following steps:

[0010] Obtaining vehicle operating condition parameters and environmental parameters as energy consumption data;

[0011] The energy consumption data is processed using the energy consumption prediction model to obtain the prediction results of automobile air conditioning energy consumption. In the energy consumption prediction model, a deep learning algorithm is used to learn the changing rules of the time series data in the energy consumption data, and the spatial features in the energy consumption data are captured at the same time. New data samples are generated according to the learning results, and the new data samples are used to optimize the parameters of the energy consumption prediction model;

[0012] The operating state of the automobile air-conditioning system is controlled and adjusted according to the automobile air-conditioning energy consumption prediction results.

[0013] Furthermore, the vehicle operating condition parameters include vehicle speed, air conditioning mode, passenger compartment target temperature, compressor power consumption at the current temperature of the passenger compartment, and front end fan and blower power consumption.

[0014] Furthermore, the environmental parameters include ambient temperature, humidity and solar radiation.

[0015] Furthermore, after obtaining the energy consumption data, the energy consumption data is preprocessed by cleaning, normalizing and serializing.

[0016] Furthermore, the construction process of the energy consumption prediction model is as follows;

[0017] Build an air conditioning system energy consumption database, and select data from the database to form a training set and a validation set;

[0018] Build the network structure of energy consumption prediction model;

[0019] The energy consumption prediction model is trained using the training set.

[0020] Furthermore, the network structure of the energy consumption prediction model consists of a convolutional neural network, a long short-term memory network and an adversarial network. The convolutional neural network is used to capture the spatial features in the energy consumption data, the long short-term memory network is used to learn the changing patterns of the time series data in the energy consumption data, and the adversarial network is used to generate new data samples based on the learning results.

[0021] Furthermore, an alternating training method is used to optimize and adjust the convolutional neural network, long short-term memory network and adversarial network respectively.

[0022] A second aspect of the present invention provides an automobile air conditioning energy consumption prediction system, comprising:

[0023] A data acquisition module is configured to acquire vehicle operating condition parameters and environmental parameters as energy consumption data;

[0024] The energy consumption prediction module is configured to process the energy consumption data using the energy consumption prediction model to obtain the energy consumption prediction result of the automobile air conditioner, wherein the energy consumption prediction model uses a deep learning algorithm to learn the change rules of the time series data in the energy consumption data, and captures the spatial features in the energy consumption data, and generates new data samples according to the learning results, and the new data samples are used to optimize the parameters of the energy consumption prediction model;

[0025] The air conditioning control module is configured to control and adjust the operating state of the automobile air conditioning system according to the automobile air conditioning energy consumption prediction result.

[0026] The third aspect of the present invention provides a medium having a program stored thereon, which, when executed by a processor, implements the steps in the method for predicting energy consumption of an automobile air conditioner as described in the first aspect of the present invention.

[0027] The fourth aspect of the present invention provides a device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for predicting energy consumption of an automobile air conditioner as described in the first aspect of the present invention are implemented.

[0028] One or more of the above technical solutions have the following beneficial effects:

[0029] The present invention discloses a method, system, medium and device for predicting the energy consumption of automobile air conditioners. An air-conditioning system energy consumption prediction model is established based on an LSTM neural network algorithm and applied to the development of the whole vehicle and the user end. The dependency of time series is captured to realize the prediction of air-conditioning energy consumption, which not only reduces the dependence on the engineering experience of engineers and the phenomenon that risks cannot be identified due to lack of engineering experience, but also provides a reference for the high and low temperature endurance development of electric vehicles by quickly and accurately predicting the energy consumption of the air-conditioning system.

[0030] In order to overcome the prediction accuracy problem of a single LSTM and to adapt to the complex and changeable cabin environment of the automobile air-conditioning system, the present invention integrates a convolutional neural network (CNN) and a generative adversarial network (GAN). CNN is used to extract spatial features of multi-dimensional feature data, and GAN is used to generate new data samples for model optimization, further improving the accuracy of energy consumption prediction.

[0031] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0033] Figure 1 This is a flow chart of a method for predicting energy consumption of an automobile air conditioner in Embodiment 1 of the present invention;

[0034] Figure 2 This is the LSTM model construction and training method in Example 1 of the present invention. DETAILED DESCRIPTION

[0035] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0036] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "include" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or their combinations;

[0037] Embodiment 1:

[0038] Embodiment 1 of the present invention provides a method for predicting energy consumption of automobile air conditioner, such as Figure 1 As shown, the following steps are included:

[0039] Step 1: Obtain vehicle operating condition parameters and environmental parameters as energy consumption data.

[0040] Step 1.1: Energy consumption data is obtained through sensors and other equipment installed on the vehicle.

[0041] In this embodiment, the vehicle operating condition parameters include vehicle speed, air conditioning mode, passenger compartment target temperature, compressor power consumption at the current temperature of the passenger compartment, and front fan and blower power consumption. Environmental parameters include ambient temperature, humidity, and solar radiation.

[0042] It can be seen that the data affecting automobile air conditioning energy consumption not only has multiple dimensions but also has complex relationships. In addition, unlike buildings, automobiles have small internal spaces and fast air flow. Air conditioning energy consumption is not only affected by changes in passengers, but also by environmental factors and heat generated by internal equipment in the automobile. Therefore, based on the above complex multi-dimensional data and the special riding environment in the car, this embodiment uses a CNN-LSTM-GAN hybrid network for data processing, which can not only mine the relationship between features, but also solve the problem of small samples of energy consumption data of different types of automobiles and easy overfitting in the training model process based on the real data generated by GAN, greatly improving the accuracy of energy consumption prediction.

[0043] Step 1.2: After obtaining the energy consumption data, perform preprocessing operations such as cleaning, normalization and serialization on the energy consumption data.

[0044] Data normalization and serialization processing is to uniformly scale the cleaned data to the [0,1] scale interval, and split it into a series of training samples, each of which contains continuous time steps, to meet the training requirements of the LSTM model and improve the convergence of the model to avoid underfitting.

[0045] Step 2: Use the energy consumption prediction model to process the energy consumption data and obtain the automobile air conditioning energy consumption prediction results.

[0046] The process of building the energy consumption prediction model is as follows;

[0047] Step 2.1: Build an air conditioning system energy consumption database and select data from the database to form a training set and a validation set.

[0048] The key feature data is collected and preprocessed based on the input format of the CNN-LSTM-GAN hybrid network to build an air conditioning system energy consumption database. Through simulation and experimental means, key feature parameter information such as ambient temperature, humidity, solar radiation, vehicle speed, air conditioning mode, passenger compartment target temperature, compressor power consumption at the current temperature of the passenger compartment, front fan and blower power consumption, etc. is obtained to form an air conditioning system energy consumption database and normalize and sequence it. With the continuous development of electric vehicles, a more complete data system can be obtained to improve the reliability of the database.

[0049] Step 2.2: Build the network structure of the energy consumption prediction model. In the energy consumption prediction model, a CNN-LSTM-GAN hybrid network framework is constructed, and the deep learning algorithm in LSTM is used to learn the changing rules of the time series data in the energy consumption data. At the same time, CNN is used to capture the spatial features in the energy consumption data, and GAN is used to generate new data samples based on the learning results. The new data samples are used to optimize the parameters of the energy consumption prediction model.

[0050] The training data is extracted from the above database and randomly divided into two parts, where the training set (80%) is used to train the model and the validation set (20%) is used to evaluate the performance of the model during training.

[0051] Among them, CNN includes convolutional layers, pooling layers and fully connected layers. CNN is used to process image data, extract features and classify them. After training, the weights are adjusted through back propagation to minimize the loss function.

[0052] Input data includes vehicle speed, air conditioning mode, passenger compartment target temperature, compressor power consumption at current passenger compartment temperature, front fan, blower power consumption, ambient temperature, humidity, and solar radiation.

[0053] High-dimensional energy consumption data involves energy consumption information in multiple dimensions, and each observation value contains many features, which may include time-type data, equipment operation data, energy consumption, etc. When processing high-dimensional energy consumption data, you may face the problem of "dimensionality disaster", such as overfitting, increased computational complexity, and sample sparsity. Therefore, the CNN model is needed to capture the relevant features of energy consumption data from the above input data, learn the complex relationship between the data, enhance the features of energy consumption correlation, and use the convolution layer to extract local features.

[0054] The LSTM structure is as follows Figure 2 As shown in the figure, it includes input layer, hidden layer and output layer. In order to prevent underfitting and overfitting problems during training, it is necessary to adjust the number of hidden layers and the number of neurons in each layer according to the training results of other models such as CNN and GAN models, and finally establish the network architecture.

[0055] This example builds an LSTM neural network architecture based on Python. First, define the network structure, including the input layer, one or more hidden layers, a fully connected layer, and an output layer, and select a suitable activation function; second, determine the loss function and optimizer, and set the training parameters; finally, import the preprocessed data into the model to complete the training. After the training is completed, the descent algorithm is used to optimize the network parameters.

[0056] The data after the CNN network is input into the LSTM network for learning the time relationship. LSTM is used to extract high-level spatiotemporal features in time series data, capture long-term dependencies, and reflect energy consumption trends. The output data of LSTM is then directly input into the GAN data simulation.

[0057] In this embodiment, the GAN model includes a discriminator and a generator. The generator of the GAN is attached to the output layer of the LSTM network, and the prediction and generation of high-dimensional energy consumption data are completed through the generation ability of the GAN.

[0058] The steps to train a GAN model include:

[0059] Design a GAN generator whose input dimension should match the output dimension of the LSTM network. The goal of the generator is to learn how to convert the predictive features of these LSTM outputs into realistic high-dimensional energy consumption data samples.

[0060] The output of the LSTM network is directly used as the input of the GAN generator, so that the features extracted by LSTM can be used by the generator to generate energy consumption data.

[0061] Alternately train the discriminator and generator:

[0062] Training the discriminator: fix the generator parameters, input the actual energy consumption data and the energy consumption data generated by the generator, and maximize the loss of the discriminator so that it can distinguish between real energy consumption data and fake energy consumption data.

[0063] Train the generator: fix the discriminator parameters, input random noise to generate fake energy consumption data, minimize the generator's loss, and make the images generated by the generator more realistic to confuse the discriminator.

[0064] After training, the mean square error (MSE) or other appropriate loss functions are used for optimization.

[0065] The network structure of the energy consumption prediction model in this embodiment is composed of a convolutional neural network, a long short-term memory network and an adversarial network. The convolutional neural network is used to capture the spatial features in the energy consumption data, the long short-term memory network is used to learn the changing rules of the time series data in the energy consumption data, and the adversarial network is used to generate new data samples according to the learning results, so as to simulate the energy consumption data in different environments and expand the data set.

[0066] It should be noted that in this embodiment, the new energy consumption data generated by the GAN network is input into the CNN model and the LSTM model for two-part iteration. The new energy consumption data passes through the CNN model to the LSTM model to obtain a new output, and passes through the LSTM model directly to obtain a new output. The two outputs are compared, and the parameters of the CNN and LSTM models are adjusted preferentially. This can avoid the problem of data prediction direction deviation caused by a single feedback.

[0067] Step 2.3: Use the training set to train the energy consumption prediction model.

[0068] This embodiment uses an alternating training method to optimize and adjust the convolutional neural network, the long short-term memory network, and the adversarial network respectively, which can gradually optimize the performance of the entire model and realize complex data processing and generation tasks.

[0069] Step 3: Control and adjust the operating state of the automobile air-conditioning system according to the automobile air-conditioning energy consumption prediction result.

[0070] Embodiment 2:

[0071] Embodiment 2 of the present invention provides a vehicle air conditioning energy consumption prediction system, including:

[0072] A data acquisition module is configured to acquire vehicle operating condition parameters and environmental parameters as energy consumption data;

[0073] The energy consumption prediction module is configured to process the energy consumption data using the energy consumption prediction model to obtain the energy consumption prediction result of the automobile air conditioner, wherein the energy consumption prediction model uses a deep learning algorithm to learn the change rules of the time series data in the energy consumption data, and captures the spatial features in the energy consumption data, and generates new data samples according to the learning results, and the new data samples are used to optimize the parameters of the energy consumption prediction model;

[0074] The air conditioning control module is configured to control and adjust the operating state of the automobile air conditioning system according to the automobile air conditioning energy consumption prediction result.

[0075] Embodiment three:

[0076] Embodiment 3 of the present invention provides a medium on which a program is stored. When the program is executed by a processor, the steps in the automobile air-conditioning energy consumption prediction method as described in Embodiment 1 of the present invention are implemented.

[0077] Embodiment 4:

[0078] Embodiment 4 of the present invention provides a device, including a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps in the automobile air conditioning energy consumption prediction method as described in Embodiment 1 of the present invention are implemented.

[0079] The steps involved in the above embodiments 2, 3 and 4 correspond to the method embodiment 1. For the specific implementation methods, please refer to the relevant description part of embodiment 1.

[0080] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0081] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting energy consumption of automobile air conditioner, characterized in that: The following steps are involved: Obtaining vehicle operating condition parameters and environmental parameters as energy consumption data; The energy consumption data is processed using the energy consumption prediction model to obtain the prediction results of automobile air conditioning energy consumption. In the energy consumption prediction model, a deep learning algorithm is used to learn the changing rules of the time series data in the energy consumption data, and the spatial features in the energy consumption data are captured at the same time. New data samples are generated according to the learning results, and the new data samples are used to optimize the parameters of the energy consumption prediction model; The operating state of the automobile air-conditioning system is controlled and adjusted according to the automobile air-conditioning energy consumption prediction results.

2. The automobile air conditioning energy consumption prediction method according to claim 1, characterized in that: The vehicle operating condition parameters include vehicle speed, air conditioning mode, passenger compartment target temperature, compressor power consumption at the current passenger compartment temperature, and front end fan and blower power consumption.

3. The automobile air conditioning energy consumption prediction method according to claim 1, characterized in that: Environmental parameters include ambient temperature, humidity and solar radiation.

4. The automobile air conditioning energy consumption prediction method according to claim 1, characterized in that: After obtaining the energy consumption data, the energy consumption data is preprocessed by cleaning, normalizing and serializing.

5. The automobile air conditioning energy consumption prediction method according to claim 1, characterized in that: The process of building the energy consumption prediction model is as follows; Build an air conditioning system energy consumption database, and select data from the database to form a training set and a validation set; Build the network structure of energy consumption prediction model; The energy consumption prediction model is trained using the training set.

6. The automobile air conditioning energy consumption prediction method according to claim 5, characterized in that: The network structure of the energy consumption prediction model consists of a convolutional neural network, a long short-term memory network and an adversarial network. The convolutional neural network is used to capture the spatial features in the energy consumption data, the long short-term memory network is used to learn the changing rules of the time series data in the energy consumption data, and the adversarial network is used to generate new data samples based on the learning results.

7. The automobile air conditioning energy consumption prediction method according to claim 6, characterized in that: The convolutional neural network, long short-term memory network and adversarial network are optimized and adjusted respectively by alternating training.

8. A vehicle air conditioning energy consumption prediction system, characterized in that: include: A data acquisition module is configured to acquire vehicle operating condition parameters and environmental parameters as energy consumption data; The energy consumption prediction module is configured to process the energy consumption data using the energy consumption prediction model to obtain the energy consumption prediction result of the automobile air conditioner, wherein the energy consumption prediction model uses a deep learning algorithm to learn the change rules of the time series data in the energy consumption data, and captures the spatial features in the energy consumption data, and generates new data samples according to the learning results, and the new data samples are used to optimize the parameters of the energy consumption prediction model; The air conditioning control module is configured to control and adjust the operating state of the automobile air conditioning system according to the automobile air conditioning energy consumption prediction result.

9. A computer-readable storage medium, characterized in that: A plurality of instructions are stored therein, and the instructions are suitable for being loaded by a processor of a terminal device and executing the automobile air conditioning energy consumption prediction method described in any one of claims 1-7.

10. A terminal device, characterized in that: It includes a processor and a computer-readable storage medium, the processor is used to implement various instructions; the computer-readable storage medium is used to store multiple instructions, and the instructions are suitable for being loaded by the processor and executing the automobile air conditioning energy consumption prediction method according to any one of claims 1-7.

Images