A new energy vehicle air conditioning control method

Abstract

The present invention provides a new energy vehicle air conditioning control method. The method calculates the solar radiation heat load based on the dynamic change of light intensity, and fully considers the coupling relationship between various heat loads in the vehicle, including the solar radiation heat load, the air conditioning duct heat load, the internal and external circulation wind heat load, the external environment heat load, and the total metabolic heat load of the vehicle occupants. By constructing an improved long short-term memory recurrent neural network model, the power of the new energy vehicle air conditioning compressor is controlled, and more accurate and efficient air conditioning control can be achieved. The present invention can ensure that the air conditioning control system achieves optimal energy efficiency while providing a comfortable riding environment.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy vehicles, and in particular to a method for controlling an air conditioner of a new energy vehicle. Background Art

[0002] New energy vehicles have attracted widespread attention from researchers and developers due to their zero-emission and high-energy efficiency. However, in the development of new energy vehicles, the energy density of power batteries limits their range, which has always been a difficult problem that the industry needs to solve. At present, the industry is using various innovative methods to improve the range of electric vehicles, among which the energy consumption control of the air-conditioning system has become an important research focus. While ensuring the air-conditioning comfort in the passenger compartment, the economy of the entire vehicle must also be considered. Therefore, it is particularly important to develop a method to accurately control the energy consumption of air-conditioning. This need to balance comfort and economy poses higher technical challenges to the control of the air-conditioning system of new energy vehicles.

[0003] The Chinese patent application number CN202311438625.1 discloses a method, device, storage medium and equipment for controlling automobile air conditioners. The technology adjusts the operating state of the air conditioner by comparing the difference between the ambient temperature and the target temperature set in the car. However, the heat load of the automobile air conditioner is determined by multiple factors. The solution does not consider the coupling relationship between the various heat loads in the car, nor the dynamic changes in light intensity, making it difficult to achieve accurate and efficient air conditioner control. Summary of the invention

[0004] The object of the present invention is to provide a new energy vehicle air conditioning control method to achieve accurate and efficient air conditioning control.

[0005] A new energy vehicle air conditioning control method comprises the following steps:

[0006] Step S1, using the built-in light sensor of the new energy vehicle to measure the light intensity outside the vehicle, and calculating the light projection area in combination with the vehicle three-dimensional model, constructing a solar light heat load model according to the measured light intensity and the calculated light projection area, and calculating the solar light heat load through the solar light heat load model;

[0007] Step S2, obtaining the outside environment temperature and the passenger compartment temperature of the new energy vehicle through the temperature sensor built into the vehicle, and determining the air volume of the blower according to the set blower gear, and establishing the air conditioning duct heat load model based on the temperature difference between the inside and outside of the vehicle and the blower air volume, and calculating the air conditioning duct heat load through the air conditioning duct heat load model;

[0008] Step S3, reading the opening of the internal and external circulation air door of the new energy vehicle, and building an internal and external circulation air heat load model based on the temperature difference between the inside and outside of the vehicle in step S2, and calculating the internal and external circulation air heat load through the internal and external circulation air heat load model;

[0009] Step S4, using the vehicle model of the new energy vehicle to calculate the surface area of ​​the vehicle, combining the temperature difference between the inside and outside of the vehicle, establishing an external environment heat load model, and calculating the external environment heat load through the external environment heat load model;

[0010] Step S5, calculating the total metabolic heat load of the occupants in the new energy vehicle;

[0011] Step S6, measuring the inlet and outlet temperature difference of the evaporator by temperature sensors installed before and after the evaporator, and calculating the cooling capacity of the vehicle in combination with the air volume determined by the blower gear position;

[0012] Step S7, constructing an improved long short-term memory recurrent neural network model, normalizing the solar heat load, the air conditioning duct heat load, the internal and external circulation wind heat load, the external environment heat load, the total metabolic heat load, and the vehicle cooling capacity, taking the normalized solar heat load, the air conditioning duct heat load, the internal and external circulation wind heat load, the external environment heat load, and the total metabolic heat load as input values, taking the normalized vehicle cooling capacity as output values, training the long short-term memory recurrent neural network model, and obtaining a trained long short-term memory recurrent neural network model;

[0013] Step S8, using the trained long short-term memory recurrent neural network model to output the heat load result, and controlling the power of the new energy vehicle air-conditioning compressor based on the heat load result.

[0014] According to the air conditioning control method for new energy vehicles provided by the present invention, the solar heat load is calculated based on the dynamic change of light intensity, and the coupling relationship between various heat loads in the car is fully considered, including the coupling relationship between solar heat load, air conditioning air duct heat load, internal and external circulation air heat load, external environment heat load, and total metabolic heat load of the passengers in the car. By constructing an improved long short-term memory recurrent neural network model, the power of the air conditioning compressor of the new energy vehicle is controlled, and more accurate and efficient air conditioning control can be achieved. The present invention uses the average temperature of the breathing surface of the seat in the car as the evaluation standard of comfort, and the power of the air conditioning compressor as the evaluation standard of economy. The method is comprehensively evaluated by combining these two standards. The results show that the present invention can ensure that the air conditioning control system can achieve optimal energy efficiency while providing a comfortable riding environment; in addition, traditional deep neural networks have limitations in processing time series prediction problems because they cannot effectively capture the time dependence of time series data. The present invention adopts an improved long short-term memory recurrent neural network model, which can effectively solve the gradient disappearance problem through special structural design, thereby making more accurate predictions on the time dependence problem of long series. A more accurate and efficient solution is provided for the air conditioning control of new energy vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic flow chart of the air conditioning control method for new energy vehicles of the present invention;

[0016] Figure 2 A comparison diagram of the predicted value and the actual value in the embodiment of the present invention;

[0017] Figure 3 This is a comparison chart of the average temperature of the breathing surface of the car seat;

[0018] Figure 4 This is a comparison chart of air conditioning compressor power. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] See also Figure 1 The new energy vehicle air conditioning control method provided by the embodiment of the present invention includes steps S1 to S8.

[0021] Step S1, using the built-in light sensor of the new energy vehicle to measure the light intensity outside the vehicle, and calculating the light projection area in combination with the vehicle three-dimensional model, constructing a solar light heat load model according to the measured light intensity and the calculated light projection area, and calculating the solar light heat load through the solar light heat load model.

[0022] In step S1, the solar heat load is calculated using the following formula: :

[0023] ;

[0024] in, K It indicates the light intensity outside the vehicle measured by the light sensor. A 1 is the illumination projection area, dt Indicates time t The differential of .

[0025] In this embodiment, in order to test the stability of data, the CLTC-P working condition is carried out on a four-wheel drive chassis dynamometer system with environmental simulation, and relevant data is collected at the same time. The environmental simulation system is equipped with a sunlight simulation system, and the light intensity outside the vehicle is read through the built-in light sensor of the vehicle. K The sunlight simulation system light intensity is set to 1000w / m 2 , K The value reading is a stable 975w / m 2 , identify the light projection area through the vehicle digital model A 1. The vehicle in this embodiment is a pure electric SUV of a certain brand, and its light projection area A 1 is 8.867m 2 , build a solar heat load model based on the light intensity and the vehicle projection area, and calculate the solar heat load .

[0026] Step S2, obtain the external ambient temperature and the passenger compartment temperature through the temperature sensor built into the new energy vehicle, and at the same time, determine the blower air volume according to the set blower gear, and establish the air conditioning duct heat load model based on the temperature difference between the inside and outside of the vehicle and the blower air volume, and calculate the air conditioning duct heat load through the air conditioning duct heat load model.

[0027] In step S2, the heat load of the air conditioning duct is calculated using the following formula: :

[0028] ;

[0029] in, is the specific heat capacity of air, is the air density, is the blower air volume, is the ambient temperature outside the vehicle, The temperature of the passenger compartment in the vehicle.

[0030] In this embodiment, in order to ensure the consistency of the test, the test vehicle is exposed to an average temperature of 35°C inside the vehicle, the ambient temperature is set to 35°C through the environmental chamber, and the external ambient temperature is read by the vehicle's built-in temperature sensor. , passenger compartment temperature , set the blower gear to the highest gear 12, at this time the blower air volume The value is 138m 3 / h, 1.29kg / m 3 , It is 1000J / (kg·℃).

[0031] The heat load model of the air conditioning duct is established based on the temperature difference between the inside and outside of the vehicle and the air volume of the blower, and the heat load of the air conditioning duct is calculated. .

[0032] Step S3, reading the opening of the internal and external circulation air door of the new energy vehicle, and combining the temperature difference between the inside and outside of the vehicle in step S2, constructing an internal and external circulation air heat load model, and calculating the internal and external circulation air heat load through the internal and external circulation air heat load model.

[0033] In step S3, the heat load of internal and external circulation air is calculated using the following formula: :

[0034] ;

[0035] in, It is the opening degree of internal and external circulation damper.

[0036] Step S4, using the vehicle model of the new energy vehicle to calculate the surface area of ​​the entire vehicle, combining the temperature difference between the inside and outside of the vehicle, establishing an external environment heat load model, and calculating the external environment heat load through the external environment heat load model.

[0037] In step S4, the external environment heat load is calculated using the following formula: :

[0038] ;

[0039] in, is the heat transfer coefficient, Take 1.926w / (m 2 ℃), A 2 is the surface area of ​​the vehicle.

[0040] In this embodiment, the vehicle surface area is calculated by the vehicle digital model. A 2 is 44.295m 2, and then calculate the external environment heat load .

[0041] Step S5, calculating the total metabolic heat load of the occupants in the new energy vehicle.

[0042] In step S5, the total metabolic heat load of the occupants in the new energy vehicle is calculated using the following formula: :

[0043] ;

[0044] in, is the standard human metabolic heat load, m is the number of occupants in the vehicle.

[0045] In this embodiment, the standard human metabolic heat load Take an average and calculate the total metabolic heat load .

[0046] Step S6, measuring the inlet and outlet temperature difference of the evaporator by temperature sensors installed before and after the evaporator, and calculating the cooling capacity of the vehicle in combination with the air volume determined by the blower gear position.

[0047] In step S6, the vehicle cooling capacity is calculated using the following formula: :

[0048] ;

[0049] in, is the evaporator front temperature, is the temperature at the rear end of the evaporator.

[0050] In this embodiment, the blower gear is set to the highest gear, at which time the blower air volume The value is 138m 3 / h, the vehicle cooling capacity is calculated by the evaporator inlet and outlet air temperature difference and the blower air volume .

[0051] Step S7, construct an improved long short-term memory recurrent neural network model, normalize the solar heat load, air conditioning duct heat load, internal and external circulation wind heat load, external environment heat load, total metabolic heat load, and vehicle cooling capacity, use the normalized solar heat load, air conditioning duct heat load, internal and external circulation wind heat load, external environment heat load, and total metabolic heat load as input values, and use the normalized vehicle cooling capacity as output value to train the long short-term memory recurrent neural network model to obtain a trained long short-term memory recurrent neural network model.

[0052] The improved long short-term memory recurrent neural network model includes a recurrent neural network LSTM layer and a fully connected neural network layer. Dropout regularization is used between layers. In the recurrent neural network LSTM layer, the output information of the previous moment is and the long-term memory cell state at the previous moment Combine to form a long-term memory cell state combination at the previous moment , the present invention converts the long-term memory cell state of the previous moment It also adds three gates (forget gate, input gate, and output gate) into the calculation to ensure that more important information from the previous moment is retained.

[0053] Long-term memory cell state combination based on the previous moment A forget gate, an input gate, and an output gate are set. Specifically, the improved long short-term memory recurrent neural network model satisfies the following formula:

[0054] ;

[0055] ;

[0056] ;

[0057] ;

[0058] ;

[0059] ;

[0060] in, f t Represents the output of the forget gate at the current moment; is the sigmoid function; W f and b f is the weight and bias of the forget gate; X t Represents the input features, which are composed of normalized data; i t Represents the output of the input gate at the current moment; W i and b i is the weight and bias of the input gate; o t Represents the output of the output gate at the current moment; W o and b o is the weight and bias of the output gate; Indicates the candidate state at the current moment; tanh is the hyperbolic tangent function; W C and b C are the weights and biases of the candidate states; C t Indicates the current long-term memory cell state. In order to ensure more retention of valid information at the previous moment, the present invention converts the current long-term memory cell state C t Perform mutation processing and add the output of the previous moment to work together; h t Represents the short-term memory function at the current moment.

[0061] It should be pointed out that, in this embodiment, before 1150 seconds, the air-conditioning compressor and breathing surface temperature data vary greatly because the cab temperature does not reach the target value, and are not suitable for training the model. Therefore, when training and testing the model, the test data from 1150 seconds to 1800 seconds are used as the available data set. In specific implementation, the solar heat load, air-conditioning duct heat load, internal and external circulation air heat load, external environment heat load, total metabolic heat load, and vehicle cooling capacity are first pre-processed to remove obviously abnormal data, and then normalized to obtain a value range of (0,1). The normalized solar heat load, air-conditioning duct heat load, internal and external circulation air heat load, external environment heat load, and total metabolic heat load are used as input values. All normalized data are segmented based on the time window. Since the data is short, the length of each time window is the length of 1 time point, and all data obtained after window segmentation are used as training data sets. The training of the model can be done by autonomous programming in MATLAB or Python, or by using a mature deep learning framework such as MindSpore. In this embodiment, autonomous programming in MATLAB is used. The training data set is split into three parts: a training set, a validation set, and a test set. The training set consists of the first 50% of the data in the training data set, the validation set consists of the middle 25% of the data in the training data set, and the test set consists of the last 25% of the data in the training data set. The long short-term memory recurrent neural network model is trained.

[0062] Step S8, using the trained long short-term memory recurrent neural network model to output the heat load result, and controlling the power of the new energy vehicle air-conditioning compressor based on the heat load result.

[0063] The trained model is verified with the test set data. , , , , As input, solve , and Perform denormalization and find The predicted value of The predicted value is compared with the actual value, and the results are as follows Figure 2 As shown, from Figure 2 It can be seen that the model has excellent prediction ability. The trained long short-term memory recurrent neural network model can be used to predict the air-conditioning compressor power and then used for the control of new energy vehicle air-conditioning.

[0064] The method of the present invention is tested below. Under the same working conditions and test environment, the comfort and economy of an air-conditioning system controlled by a traditional method and an air-conditioning system controlled by the method of the present invention are tested and compared.

[0065] Comparison of average temperature of the breathing surface of the car seat Figure 3 As shown, from Figure 3 It can be seen that the present invention has a fast cooling rate and more stable control; the air conditioning compressor power comparison is as follows Figure 4 As shown, from Figure 4 It can be seen that the compression control is smoother, the impact on the compressor is small, and the compressor energy consumption is lower.

[0066] In summary, according to the air conditioning control method for new energy vehicles provided by the present invention, the solar heat load is calculated based on the dynamic change of light intensity, and the coupling relationship between various heat loads in the car is fully considered, including the coupling relationship between solar heat load, air conditioning air duct heat load, internal and external circulation air heat load, external environment heat load, and total metabolic heat load of the passengers in the car. By constructing an improved long short-term memory recurrent neural network model, the power of the air conditioning compressor of the new energy vehicle is controlled, and more accurate and efficient air conditioning control can be achieved. The present invention uses the average temperature of the breathing surface of the seat in the car as the evaluation standard of comfort, and the power of the air conditioning compressor as the evaluation standard of economy. The method is comprehensively evaluated by combining these two standards. The results show that the present invention can ensure that the air conditioning control system can achieve optimal energy efficiency while providing a comfortable riding environment; in addition, traditional deep neural networks have limitations in processing time series prediction problems because they cannot effectively capture the time dependence of time series data. The present invention adopts an improved long short-term memory recurrent neural network model, which can effectively solve the gradient disappearance problem through special structural design, thereby more accurately predicting the time dependence problem of long series. A more accurate and efficient solution is provided for the air conditioning control of new energy vehicles.

[0067] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0068] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A new energy vehicle air conditioning control method, characterized in that: include: Step S1, using the built-in light sensor of the new energy vehicle to measure the light intensity outside the vehicle, and calculating the light projection area in combination with the vehicle three-dimensional model, constructing a solar light heat load model according to the measured light intensity and the calculated light projection area, and calculating the solar light heat load through the solar light heat load model; Step S2, obtaining the outside environment temperature and the passenger compartment temperature of the new energy vehicle through the temperature sensor built into the vehicle, and determining the air volume of the blower according to the set blower gear, and establishing the air conditioning duct heat load model based on the temperature difference between the inside and outside of the vehicle and the blower air volume, and calculating the air conditioning duct heat load through the air conditioning duct heat load model; Step S3, reading the opening of the internal and external circulation air door of the new energy vehicle, and building an internal and external circulation air heat load model based on the temperature difference between the inside and outside of the vehicle in step S2, and calculating the internal and external circulation air heat load through the internal and external circulation air heat load model; Step S4, using the vehicle model of the new energy vehicle to calculate the surface area of ​​the vehicle, combining the temperature difference between the inside and outside of the vehicle, establishing an external environment heat load model, and calculating the external environment heat load through the external environment heat load model; Step S5, calculating the total metabolic heat load of the occupants in the new energy vehicle; Step S6, measuring the inlet and outlet temperature difference of the evaporator by temperature sensors installed before and after the evaporator, and calculating the cooling capacity of the vehicle in combination with the air volume determined by the blower gear position; Step S7, constructing an improved long short-term memory recurrent neural network model, normalizing the solar heat load, the air conditioning duct heat load, the internal and external circulation wind heat load, the external environment heat load, the total metabolic heat load, and the vehicle cooling capacity, taking the normalized solar heat load, the air conditioning duct heat load, the internal and external circulation wind heat load, the external environment heat load, and the total metabolic heat load as input values, taking the normalized vehicle cooling capacity as output values, training the long short-term memory recurrent neural network model, and obtaining a trained long short-term memory recurrent neural network model; Step S8, using the trained long short-term memory recurrent neural network model to output the heat load result, and controlling the power of the new energy vehicle air-conditioning compressor based on the heat load result; In step S1, the solar heat load is calculated using the following formula: : ； in, K It indicates the light intensity outside the vehicle measured by the light sensor. A 1 is the illumination projection area, dt Indicates time t The differential of In step S2, the heat load of the air conditioning duct is calculated using the following formula: : ； in, is the specific heat capacity of air, is the air density, is the blower air volume, is the ambient temperature outside the vehicle, is the passenger compartment temperature in the vehicle; In step S3, the heat load of internal and external circulation air is calculated using the following formula: : ； in, is the opening of the internal and external circulation damper; In step S4, the external environment heat load is calculated using the following formula: : ； in, is the heat transfer coefficient, A 2 is the surface area of ​​the vehicle; In step S5, the total metabolic heat load of the occupants in the new energy vehicle is calculated using the following formula: : ； in, is the standard human metabolic heat load, m is the number of passengers in the vehicle; In step S6, the vehicle cooling capacity is calculated using the following formula: : ； in, is the evaporator front temperature, is the rear end temperature of the evaporator; The improved long short-term memory recurrent neural network model includes a recurrent neural network LSTM layer and a fully connected neural network layer. Dropout regularization is used between layers. In the recurrent neural network LSTM layer, the output information of the previous moment is and the long-term memory cell state at the previous moment Combine to form a long-term memory cell state combination at the previous moment , based on the long-term memory cell state combination of the previous moment A forget gate, an input gate, and an output gate are set. Specifically, the improved long short-term memory recurrent neural network model satisfies the following formula: ； ； ； ； ； ； in, f t Represents the output of the forget gate at the current moment; is the sigmoid function; W f and b f is the weight and bias of the forget gate; X t Represents the input features, which are composed of normalized data; i t Represents the output of the input gate at the current moment; W i and b i is the weight and bias of the input gate; o t Represents the output of the output gate at the current moment; W o and b o is the weight and bias of the output gate; Indicates the candidate state at the current moment; tanh is the hyperbolic tangent function; W C and b C are the weights and biases of the candidate states; C t Represents the long-term memory cell state at the current moment; h t Represents the short-term memory function at the current moment.

Images